Muhammad Khalifa
commited on
Commit
•
39c097c
1
Parent(s):
877878e
first commit
Browse files- 1a_52k/best_model/adapter_config.json +21 -0
- 1a_52k/best_model/adapter_model.bin +3 -0
- 1a_52k/best_model/optimizer.pt +3 -0
- 1a_52k/best_model/rng_state.pth +3 -0
- 1a_52k/best_model/scheduler.pt +3 -0
- 1a_52k/best_model/trainer_state.json +3987 -0
- 1a_52k/best_model/training_args.bin +3 -0
- 1b_52k/best_model/adapter_config.json +21 -0
- 1b_52k/best_model/adapter_model.bin +3 -0
- 1b_52k/best_model/optimizer.pt +3 -0
- 1b_52k/best_model/rng_state.pth +3 -0
- 1b_52k/best_model/scheduler.pt +3 -0
- 1b_52k/best_model/trainer_state.json +3987 -0
- 1b_52k/best_model/training_args.bin +3 -0
- alpaca_text_52K/best_model/adapter_config.json +21 -0
- alpaca_text_52K/best_model/adapter_model.bin +3 -0
- alpaca_text_52K/best_model/optimizer.pt +3 -0
- alpaca_text_52K/best_model/rng_state.pth +3 -0
- alpaca_text_52K/best_model/scheduler.pt +3 -0
- alpaca_text_52K/best_model/trainer_state.json +3859 -0
- alpaca_text_52K/best_model/training_args.bin +3 -0
1a_52k/best_model/adapter_config.json
ADDED
|
@@ -0,0 +1,21 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"base_model_name_or_path": "huggyllama/llama-7b",
|
| 3 |
+
"bias": "none",
|
| 4 |
+
"enable_lora": null,
|
| 5 |
+
"fan_in_fan_out": false,
|
| 6 |
+
"inference_mode": true,
|
| 7 |
+
"init_lora_weights": true,
|
| 8 |
+
"lora_alpha": 16,
|
| 9 |
+
"lora_dropout": 0.05,
|
| 10 |
+
"merge_weights": false,
|
| 11 |
+
"modules_to_save": null,
|
| 12 |
+
"peft_type": "LORA",
|
| 13 |
+
"r": 16,
|
| 14 |
+
"target_modules": [
|
| 15 |
+
"q_proj",
|
| 16 |
+
"k_proj",
|
| 17 |
+
"v_proj",
|
| 18 |
+
"o_proj"
|
| 19 |
+
],
|
| 20 |
+
"task_type": "CAUSAL_LM"
|
| 21 |
+
}
|
1a_52k/best_model/adapter_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6e5871c6d2fa0fb90709370431883c7b20149cacf4175bd9f4766aaf362d279f
|
| 3 |
+
size 67201357
|
1a_52k/best_model/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a64ca6663da69a276883fb6a52102ace936f93df4122e0382ef35b92c0ad6755
|
| 3 |
+
size 134433093
|
1a_52k/best_model/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:75c52bef46030966bb63ec38778f49d8a8b7d813cc4b4b7b3ba6d83599ed5777
|
| 3 |
+
size 14575
|
1a_52k/best_model/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:25fbc276672b2f61745ccf7554abaf6f5ce340e9f74caad0df42e27f32a75c00
|
| 3 |
+
size 627
|
1a_52k/best_model/trainer_state.json
ADDED
|
@@ -0,0 +1,3987 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.6827040314674377,
|
| 3 |
+
"best_model_checkpoint": "checkpoints/1a_52k/checkpoint-6200",
|
| 4 |
+
"epoch": 1.9838412926965843,
|
| 5 |
+
"eval_steps": 200,
|
| 6 |
+
"global_step": 6200,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.0,
|
| 13 |
+
"learning_rate": 2.6999999999999996e-05,
|
| 14 |
+
"loss": 2.0256,
|
| 15 |
+
"step": 10
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"epoch": 0.01,
|
| 19 |
+
"learning_rate": 5.399999999999999e-05,
|
| 20 |
+
"loss": 1.9236,
|
| 21 |
+
"step": 20
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"epoch": 0.01,
|
| 25 |
+
"learning_rate": 7.8e-05,
|
| 26 |
+
"loss": 1.712,
|
| 27 |
+
"step": 30
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.01,
|
| 31 |
+
"learning_rate": 0.00010799999999999998,
|
| 32 |
+
"loss": 1.2747,
|
| 33 |
+
"step": 40
|
| 34 |
+
},
|
| 35 |
+
{
|
| 36 |
+
"epoch": 0.02,
|
| 37 |
+
"learning_rate": 0.000138,
|
| 38 |
+
"loss": 1.0657,
|
| 39 |
+
"step": 50
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.02,
|
| 43 |
+
"learning_rate": 0.000168,
|
| 44 |
+
"loss": 0.973,
|
| 45 |
+
"step": 60
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.02,
|
| 49 |
+
"learning_rate": 0.000198,
|
| 50 |
+
"loss": 0.8178,
|
| 51 |
+
"step": 70
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.03,
|
| 55 |
+
"learning_rate": 0.00022799999999999999,
|
| 56 |
+
"loss": 0.7901,
|
| 57 |
+
"step": 80
|
| 58 |
+
},
|
| 59 |
+
{
|
| 60 |
+
"epoch": 0.03,
|
| 61 |
+
"learning_rate": 0.000258,
|
| 62 |
+
"loss": 0.7289,
|
| 63 |
+
"step": 90
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.03,
|
| 67 |
+
"learning_rate": 0.00028799999999999995,
|
| 68 |
+
"loss": 0.7429,
|
| 69 |
+
"step": 100
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.04,
|
| 73 |
+
"learning_rate": 0.0002998840579710145,
|
| 74 |
+
"loss": 0.7631,
|
| 75 |
+
"step": 110
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"epoch": 0.04,
|
| 79 |
+
"learning_rate": 0.0002996908212560386,
|
| 80 |
+
"loss": 0.7339,
|
| 81 |
+
"step": 120
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.04,
|
| 85 |
+
"learning_rate": 0.00029949758454106277,
|
| 86 |
+
"loss": 0.7435,
|
| 87 |
+
"step": 130
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.04,
|
| 91 |
+
"learning_rate": 0.00029930434782608696,
|
| 92 |
+
"loss": 0.7333,
|
| 93 |
+
"step": 140
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.05,
|
| 97 |
+
"learning_rate": 0.0002991111111111111,
|
| 98 |
+
"loss": 0.769,
|
| 99 |
+
"step": 150
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.05,
|
| 103 |
+
"learning_rate": 0.00029891787439613525,
|
| 104 |
+
"loss": 0.7644,
|
| 105 |
+
"step": 160
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 0.05,
|
| 109 |
+
"learning_rate": 0.0002987246376811594,
|
| 110 |
+
"loss": 0.7517,
|
| 111 |
+
"step": 170
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"epoch": 0.06,
|
| 115 |
+
"learning_rate": 0.0002985314009661836,
|
| 116 |
+
"loss": 0.7212,
|
| 117 |
+
"step": 180
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 0.06,
|
| 121 |
+
"learning_rate": 0.00029833816425120773,
|
| 122 |
+
"loss": 0.745,
|
| 123 |
+
"step": 190
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"epoch": 0.06,
|
| 127 |
+
"learning_rate": 0.00029814492753623187,
|
| 128 |
+
"loss": 0.7023,
|
| 129 |
+
"step": 200
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.06,
|
| 133 |
+
"eval_loss": 0.7244767546653748,
|
| 134 |
+
"eval_runtime": 150.1014,
|
| 135 |
+
"eval_samples_per_second": 13.324,
|
| 136 |
+
"eval_steps_per_second": 1.666,
|
| 137 |
+
"step": 200
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 0.07,
|
| 141 |
+
"learning_rate": 0.000297951690821256,
|
| 142 |
+
"loss": 0.6783,
|
| 143 |
+
"step": 210
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.07,
|
| 147 |
+
"learning_rate": 0.00029775845410628016,
|
| 148 |
+
"loss": 0.7327,
|
| 149 |
+
"step": 220
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.07,
|
| 153 |
+
"learning_rate": 0.00029756521739130435,
|
| 154 |
+
"loss": 0.69,
|
| 155 |
+
"step": 230
|
| 156 |
+
},
|
| 157 |
+
{
|
| 158 |
+
"epoch": 0.08,
|
| 159 |
+
"learning_rate": 0.0002973719806763285,
|
| 160 |
+
"loss": 0.7069,
|
| 161 |
+
"step": 240
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"epoch": 0.08,
|
| 165 |
+
"learning_rate": 0.00029717874396135264,
|
| 166 |
+
"loss": 0.7276,
|
| 167 |
+
"step": 250
|
| 168 |
+
},
|
| 169 |
+
{
|
| 170 |
+
"epoch": 0.08,
|
| 171 |
+
"learning_rate": 0.0002969855072463768,
|
| 172 |
+
"loss": 0.7356,
|
| 173 |
+
"step": 260
|
| 174 |
+
},
|
| 175 |
+
{
|
| 176 |
+
"epoch": 0.09,
|
| 177 |
+
"learning_rate": 0.000296792270531401,
|
| 178 |
+
"loss": 0.7103,
|
| 179 |
+
"step": 270
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 0.09,
|
| 183 |
+
"learning_rate": 0.0002965990338164251,
|
| 184 |
+
"loss": 0.7224,
|
| 185 |
+
"step": 280
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.09,
|
| 189 |
+
"learning_rate": 0.00029640579710144926,
|
| 190 |
+
"loss": 0.6898,
|
| 191 |
+
"step": 290
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.1,
|
| 195 |
+
"learning_rate": 0.0002962125603864734,
|
| 196 |
+
"loss": 0.7222,
|
| 197 |
+
"step": 300
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 0.1,
|
| 201 |
+
"learning_rate": 0.00029601932367149755,
|
| 202 |
+
"loss": 0.685,
|
| 203 |
+
"step": 310
|
| 204 |
+
},
|
| 205 |
+
{
|
| 206 |
+
"epoch": 0.1,
|
| 207 |
+
"learning_rate": 0.00029582608695652175,
|
| 208 |
+
"loss": 0.7389,
|
| 209 |
+
"step": 320
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 0.11,
|
| 213 |
+
"learning_rate": 0.0002956328502415459,
|
| 214 |
+
"loss": 0.6956,
|
| 215 |
+
"step": 330
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 0.11,
|
| 219 |
+
"learning_rate": 0.00029543961352657003,
|
| 220 |
+
"loss": 0.7191,
|
| 221 |
+
"step": 340
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 0.11,
|
| 225 |
+
"learning_rate": 0.0002952463768115942,
|
| 226 |
+
"loss": 0.6938,
|
| 227 |
+
"step": 350
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 0.12,
|
| 231 |
+
"learning_rate": 0.00029505314009661837,
|
| 232 |
+
"loss": 0.695,
|
| 233 |
+
"step": 360
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.12,
|
| 237 |
+
"learning_rate": 0.0002948599033816425,
|
| 238 |
+
"loss": 0.7169,
|
| 239 |
+
"step": 370
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 0.12,
|
| 243 |
+
"learning_rate": 0.00029466666666666666,
|
| 244 |
+
"loss": 0.7313,
|
| 245 |
+
"step": 380
|
| 246 |
+
},
|
| 247 |
+
{
|
| 248 |
+
"epoch": 0.12,
|
| 249 |
+
"learning_rate": 0.0002944734299516908,
|
| 250 |
+
"loss": 0.7016,
|
| 251 |
+
"step": 390
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 0.13,
|
| 255 |
+
"learning_rate": 0.00029428019323671494,
|
| 256 |
+
"loss": 0.7149,
|
| 257 |
+
"step": 400
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 0.13,
|
| 261 |
+
"eval_loss": 0.7104699611663818,
|
| 262 |
+
"eval_runtime": 150.0015,
|
| 263 |
+
"eval_samples_per_second": 13.333,
|
| 264 |
+
"eval_steps_per_second": 1.667,
|
| 265 |
+
"step": 400
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 0.13,
|
| 269 |
+
"learning_rate": 0.00029408695652173914,
|
| 270 |
+
"loss": 0.7133,
|
| 271 |
+
"step": 410
|
| 272 |
+
},
|
| 273 |
+
{
|
| 274 |
+
"epoch": 0.13,
|
| 275 |
+
"learning_rate": 0.0002938937198067633,
|
| 276 |
+
"loss": 0.7568,
|
| 277 |
+
"step": 420
|
| 278 |
+
},
|
| 279 |
+
{
|
| 280 |
+
"epoch": 0.14,
|
| 281 |
+
"learning_rate": 0.0002937004830917874,
|
| 282 |
+
"loss": 0.7159,
|
| 283 |
+
"step": 430
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 0.14,
|
| 287 |
+
"learning_rate": 0.00029350724637681156,
|
| 288 |
+
"loss": 0.7356,
|
| 289 |
+
"step": 440
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.14,
|
| 293 |
+
"learning_rate": 0.00029331400966183576,
|
| 294 |
+
"loss": 0.665,
|
| 295 |
+
"step": 450
|
| 296 |
+
},
|
| 297 |
+
{
|
| 298 |
+
"epoch": 0.15,
|
| 299 |
+
"learning_rate": 0.0002931207729468599,
|
| 300 |
+
"loss": 0.7017,
|
| 301 |
+
"step": 460
|
| 302 |
+
},
|
| 303 |
+
{
|
| 304 |
+
"epoch": 0.15,
|
| 305 |
+
"learning_rate": 0.00029292753623188405,
|
| 306 |
+
"loss": 0.6979,
|
| 307 |
+
"step": 470
|
| 308 |
+
},
|
| 309 |
+
{
|
| 310 |
+
"epoch": 0.15,
|
| 311 |
+
"learning_rate": 0.0002927342995169082,
|
| 312 |
+
"loss": 0.7104,
|
| 313 |
+
"step": 480
|
| 314 |
+
},
|
| 315 |
+
{
|
| 316 |
+
"epoch": 0.16,
|
| 317 |
+
"learning_rate": 0.00029254106280193233,
|
| 318 |
+
"loss": 0.6907,
|
| 319 |
+
"step": 490
|
| 320 |
+
},
|
| 321 |
+
{
|
| 322 |
+
"epoch": 0.16,
|
| 323 |
+
"learning_rate": 0.00029234782608695653,
|
| 324 |
+
"loss": 0.7407,
|
| 325 |
+
"step": 500
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.16,
|
| 329 |
+
"learning_rate": 0.00029215458937198067,
|
| 330 |
+
"loss": 0.7028,
|
| 331 |
+
"step": 510
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.17,
|
| 335 |
+
"learning_rate": 0.0002919613526570048,
|
| 336 |
+
"loss": 0.7102,
|
| 337 |
+
"step": 520
|
| 338 |
+
},
|
| 339 |
+
{
|
| 340 |
+
"epoch": 0.17,
|
| 341 |
+
"learning_rate": 0.00029176811594202896,
|
| 342 |
+
"loss": 0.6956,
|
| 343 |
+
"step": 530
|
| 344 |
+
},
|
| 345 |
+
{
|
| 346 |
+
"epoch": 0.17,
|
| 347 |
+
"learning_rate": 0.00029157487922705315,
|
| 348 |
+
"loss": 0.6926,
|
| 349 |
+
"step": 540
|
| 350 |
+
},
|
| 351 |
+
{
|
| 352 |
+
"epoch": 0.18,
|
| 353 |
+
"learning_rate": 0.0002913816425120773,
|
| 354 |
+
"loss": 0.7114,
|
| 355 |
+
"step": 550
|
| 356 |
+
},
|
| 357 |
+
{
|
| 358 |
+
"epoch": 0.18,
|
| 359 |
+
"learning_rate": 0.00029118840579710144,
|
| 360 |
+
"loss": 0.7066,
|
| 361 |
+
"step": 560
|
| 362 |
+
},
|
| 363 |
+
{
|
| 364 |
+
"epoch": 0.18,
|
| 365 |
+
"learning_rate": 0.0002909951690821256,
|
| 366 |
+
"loss": 0.6853,
|
| 367 |
+
"step": 570
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 0.19,
|
| 371 |
+
"learning_rate": 0.0002908019323671497,
|
| 372 |
+
"loss": 0.7063,
|
| 373 |
+
"step": 580
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 0.19,
|
| 377 |
+
"learning_rate": 0.0002906086956521739,
|
| 378 |
+
"loss": 0.7064,
|
| 379 |
+
"step": 590
|
| 380 |
+
},
|
| 381 |
+
{
|
| 382 |
+
"epoch": 0.19,
|
| 383 |
+
"learning_rate": 0.00029041545893719806,
|
| 384 |
+
"loss": 0.729,
|
| 385 |
+
"step": 600
|
| 386 |
+
},
|
| 387 |
+
{
|
| 388 |
+
"epoch": 0.19,
|
| 389 |
+
"eval_loss": 0.7055845856666565,
|
| 390 |
+
"eval_runtime": 149.8446,
|
| 391 |
+
"eval_samples_per_second": 13.347,
|
| 392 |
+
"eval_steps_per_second": 1.668,
|
| 393 |
+
"step": 600
|
| 394 |
+
},
|
| 395 |
+
{
|
| 396 |
+
"epoch": 0.2,
|
| 397 |
+
"learning_rate": 0.0002902222222222222,
|
| 398 |
+
"loss": 0.7098,
|
| 399 |
+
"step": 610
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 0.2,
|
| 403 |
+
"learning_rate": 0.00029002898550724635,
|
| 404 |
+
"loss": 0.6363,
|
| 405 |
+
"step": 620
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 0.2,
|
| 409 |
+
"learning_rate": 0.0002898357487922705,
|
| 410 |
+
"loss": 0.7128,
|
| 411 |
+
"step": 630
|
| 412 |
+
},
|
| 413 |
+
{
|
| 414 |
+
"epoch": 0.2,
|
| 415 |
+
"learning_rate": 0.0002896425120772947,
|
| 416 |
+
"loss": 0.692,
|
| 417 |
+
"step": 640
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 0.21,
|
| 421 |
+
"learning_rate": 0.00028944927536231883,
|
| 422 |
+
"loss": 0.7068,
|
| 423 |
+
"step": 650
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 0.21,
|
| 427 |
+
"learning_rate": 0.00028925603864734297,
|
| 428 |
+
"loss": 0.7191,
|
| 429 |
+
"step": 660
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.21,
|
| 433 |
+
"learning_rate": 0.0002890628019323671,
|
| 434 |
+
"loss": 0.6865,
|
| 435 |
+
"step": 670
|
| 436 |
+
},
|
| 437 |
+
{
|
| 438 |
+
"epoch": 0.22,
|
| 439 |
+
"learning_rate": 0.0002888695652173913,
|
| 440 |
+
"loss": 0.7331,
|
| 441 |
+
"step": 680
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 0.22,
|
| 445 |
+
"learning_rate": 0.00028867632850241545,
|
| 446 |
+
"loss": 0.652,
|
| 447 |
+
"step": 690
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 0.22,
|
| 451 |
+
"learning_rate": 0.0002884830917874396,
|
| 452 |
+
"loss": 0.7188,
|
| 453 |
+
"step": 700
|
| 454 |
+
},
|
| 455 |
+
{
|
| 456 |
+
"epoch": 0.23,
|
| 457 |
+
"learning_rate": 0.00028828985507246374,
|
| 458 |
+
"loss": 0.6834,
|
| 459 |
+
"step": 710
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 0.23,
|
| 463 |
+
"learning_rate": 0.0002880966183574879,
|
| 464 |
+
"loss": 0.693,
|
| 465 |
+
"step": 720
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 0.23,
|
| 469 |
+
"learning_rate": 0.0002879033816425121,
|
| 470 |
+
"loss": 0.7445,
|
| 471 |
+
"step": 730
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 0.24,
|
| 475 |
+
"learning_rate": 0.0002877101449275362,
|
| 476 |
+
"loss": 0.6904,
|
| 477 |
+
"step": 740
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 0.24,
|
| 481 |
+
"learning_rate": 0.00028751690821256036,
|
| 482 |
+
"loss": 0.7547,
|
| 483 |
+
"step": 750
|
| 484 |
+
},
|
| 485 |
+
{
|
| 486 |
+
"epoch": 0.24,
|
| 487 |
+
"learning_rate": 0.0002873236714975845,
|
| 488 |
+
"loss": 0.7068,
|
| 489 |
+
"step": 760
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 0.25,
|
| 493 |
+
"learning_rate": 0.0002871304347826087,
|
| 494 |
+
"loss": 0.6677,
|
| 495 |
+
"step": 770
|
| 496 |
+
},
|
| 497 |
+
{
|
| 498 |
+
"epoch": 0.25,
|
| 499 |
+
"learning_rate": 0.00028693719806763285,
|
| 500 |
+
"loss": 0.6808,
|
| 501 |
+
"step": 780
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 0.25,
|
| 505 |
+
"learning_rate": 0.000286743961352657,
|
| 506 |
+
"loss": 0.7142,
|
| 507 |
+
"step": 790
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 0.26,
|
| 511 |
+
"learning_rate": 0.00028655072463768113,
|
| 512 |
+
"loss": 0.7126,
|
| 513 |
+
"step": 800
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 0.26,
|
| 517 |
+
"eval_loss": 0.7015364766120911,
|
| 518 |
+
"eval_runtime": 149.6299,
|
| 519 |
+
"eval_samples_per_second": 13.366,
|
| 520 |
+
"eval_steps_per_second": 1.671,
|
| 521 |
+
"step": 800
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 0.26,
|
| 525 |
+
"learning_rate": 0.0002863574879227053,
|
| 526 |
+
"loss": 0.69,
|
| 527 |
+
"step": 810
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 0.26,
|
| 531 |
+
"learning_rate": 0.00028616425120772947,
|
| 532 |
+
"loss": 0.6772,
|
| 533 |
+
"step": 820
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 0.27,
|
| 537 |
+
"learning_rate": 0.0002859710144927536,
|
| 538 |
+
"loss": 0.6881,
|
| 539 |
+
"step": 830
|
| 540 |
+
},
|
| 541 |
+
{
|
| 542 |
+
"epoch": 0.27,
|
| 543 |
+
"learning_rate": 0.00028577777777777776,
|
| 544 |
+
"loss": 0.6509,
|
| 545 |
+
"step": 840
|
| 546 |
+
},
|
| 547 |
+
{
|
| 548 |
+
"epoch": 0.27,
|
| 549 |
+
"learning_rate": 0.0002855845410628019,
|
| 550 |
+
"loss": 0.6924,
|
| 551 |
+
"step": 850
|
| 552 |
+
},
|
| 553 |
+
{
|
| 554 |
+
"epoch": 0.28,
|
| 555 |
+
"learning_rate": 0.00028539130434782604,
|
| 556 |
+
"loss": 0.7116,
|
| 557 |
+
"step": 860
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 0.28,
|
| 561 |
+
"learning_rate": 0.00028519806763285024,
|
| 562 |
+
"loss": 0.711,
|
| 563 |
+
"step": 870
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 0.28,
|
| 567 |
+
"learning_rate": 0.0002850048309178744,
|
| 568 |
+
"loss": 0.6897,
|
| 569 |
+
"step": 880
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 0.28,
|
| 573 |
+
"learning_rate": 0.0002848115942028985,
|
| 574 |
+
"loss": 0.6689,
|
| 575 |
+
"step": 890
|
| 576 |
+
},
|
| 577 |
+
{
|
| 578 |
+
"epoch": 0.29,
|
| 579 |
+
"learning_rate": 0.00028461835748792266,
|
| 580 |
+
"loss": 0.6891,
|
| 581 |
+
"step": 900
|
| 582 |
+
},
|
| 583 |
+
{
|
| 584 |
+
"epoch": 0.29,
|
| 585 |
+
"learning_rate": 0.00028442512077294686,
|
| 586 |
+
"loss": 0.6985,
|
| 587 |
+
"step": 910
|
| 588 |
+
},
|
| 589 |
+
{
|
| 590 |
+
"epoch": 0.29,
|
| 591 |
+
"learning_rate": 0.000284231884057971,
|
| 592 |
+
"loss": 0.6643,
|
| 593 |
+
"step": 920
|
| 594 |
+
},
|
| 595 |
+
{
|
| 596 |
+
"epoch": 0.3,
|
| 597 |
+
"learning_rate": 0.00028403864734299515,
|
| 598 |
+
"loss": 0.7277,
|
| 599 |
+
"step": 930
|
| 600 |
+
},
|
| 601 |
+
{
|
| 602 |
+
"epoch": 0.3,
|
| 603 |
+
"learning_rate": 0.0002838454106280193,
|
| 604 |
+
"loss": 0.703,
|
| 605 |
+
"step": 940
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 0.3,
|
| 609 |
+
"learning_rate": 0.00028365217391304343,
|
| 610 |
+
"loss": 0.6678,
|
| 611 |
+
"step": 950
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 0.31,
|
| 615 |
+
"learning_rate": 0.00028345893719806763,
|
| 616 |
+
"loss": 0.6836,
|
| 617 |
+
"step": 960
|
| 618 |
+
},
|
| 619 |
+
{
|
| 620 |
+
"epoch": 0.31,
|
| 621 |
+
"learning_rate": 0.00028326570048309177,
|
| 622 |
+
"loss": 0.7164,
|
| 623 |
+
"step": 970
|
| 624 |
+
},
|
| 625 |
+
{
|
| 626 |
+
"epoch": 0.31,
|
| 627 |
+
"learning_rate": 0.0002830724637681159,
|
| 628 |
+
"loss": 0.6382,
|
| 629 |
+
"step": 980
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 0.32,
|
| 633 |
+
"learning_rate": 0.00028287922705314006,
|
| 634 |
+
"loss": 0.7,
|
| 635 |
+
"step": 990
|
| 636 |
+
},
|
| 637 |
+
{
|
| 638 |
+
"epoch": 0.32,
|
| 639 |
+
"learning_rate": 0.0002826859903381642,
|
| 640 |
+
"loss": 0.6974,
|
| 641 |
+
"step": 1000
|
| 642 |
+
},
|
| 643 |
+
{
|
| 644 |
+
"epoch": 0.32,
|
| 645 |
+
"eval_loss": 0.6978325843811035,
|
| 646 |
+
"eval_runtime": 149.6863,
|
| 647 |
+
"eval_samples_per_second": 13.361,
|
| 648 |
+
"eval_steps_per_second": 1.67,
|
| 649 |
+
"step": 1000
|
| 650 |
+
},
|
| 651 |
+
{
|
| 652 |
+
"epoch": 0.32,
|
| 653 |
+
"learning_rate": 0.0002824927536231884,
|
| 654 |
+
"loss": 0.6644,
|
| 655 |
+
"step": 1010
|
| 656 |
+
},
|
| 657 |
+
{
|
| 658 |
+
"epoch": 0.33,
|
| 659 |
+
"learning_rate": 0.00028229951690821254,
|
| 660 |
+
"loss": 0.672,
|
| 661 |
+
"step": 1020
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 0.33,
|
| 665 |
+
"learning_rate": 0.0002821062801932367,
|
| 666 |
+
"loss": 0.7036,
|
| 667 |
+
"step": 1030
|
| 668 |
+
},
|
| 669 |
+
{
|
| 670 |
+
"epoch": 0.33,
|
| 671 |
+
"learning_rate": 0.0002819130434782608,
|
| 672 |
+
"loss": 0.7049,
|
| 673 |
+
"step": 1040
|
| 674 |
+
},
|
| 675 |
+
{
|
| 676 |
+
"epoch": 0.34,
|
| 677 |
+
"learning_rate": 0.000281719806763285,
|
| 678 |
+
"loss": 0.7033,
|
| 679 |
+
"step": 1050
|
| 680 |
+
},
|
| 681 |
+
{
|
| 682 |
+
"epoch": 0.34,
|
| 683 |
+
"learning_rate": 0.00028152657004830916,
|
| 684 |
+
"loss": 0.7059,
|
| 685 |
+
"step": 1060
|
| 686 |
+
},
|
| 687 |
+
{
|
| 688 |
+
"epoch": 0.34,
|
| 689 |
+
"learning_rate": 0.0002813333333333333,
|
| 690 |
+
"loss": 0.6915,
|
| 691 |
+
"step": 1070
|
| 692 |
+
},
|
| 693 |
+
{
|
| 694 |
+
"epoch": 0.35,
|
| 695 |
+
"learning_rate": 0.00028114009661835745,
|
| 696 |
+
"loss": 0.7609,
|
| 697 |
+
"step": 1080
|
| 698 |
+
},
|
| 699 |
+
{
|
| 700 |
+
"epoch": 0.35,
|
| 701 |
+
"learning_rate": 0.0002809468599033816,
|
| 702 |
+
"loss": 0.6756,
|
| 703 |
+
"step": 1090
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 0.35,
|
| 707 |
+
"learning_rate": 0.0002807536231884058,
|
| 708 |
+
"loss": 0.7127,
|
| 709 |
+
"step": 1100
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"epoch": 0.36,
|
| 713 |
+
"learning_rate": 0.00028056038647342993,
|
| 714 |
+
"loss": 0.7212,
|
| 715 |
+
"step": 1110
|
| 716 |
+
},
|
| 717 |
+
{
|
| 718 |
+
"epoch": 0.36,
|
| 719 |
+
"learning_rate": 0.00028036714975845407,
|
| 720 |
+
"loss": 0.6974,
|
| 721 |
+
"step": 1120
|
| 722 |
+
},
|
| 723 |
+
{
|
| 724 |
+
"epoch": 0.36,
|
| 725 |
+
"learning_rate": 0.0002801739130434782,
|
| 726 |
+
"loss": 0.6948,
|
| 727 |
+
"step": 1130
|
| 728 |
+
},
|
| 729 |
+
{
|
| 730 |
+
"epoch": 0.36,
|
| 731 |
+
"learning_rate": 0.0002799806763285024,
|
| 732 |
+
"loss": 0.73,
|
| 733 |
+
"step": 1140
|
| 734 |
+
},
|
| 735 |
+
{
|
| 736 |
+
"epoch": 0.37,
|
| 737 |
+
"learning_rate": 0.00027978743961352655,
|
| 738 |
+
"loss": 0.7154,
|
| 739 |
+
"step": 1150
|
| 740 |
+
},
|
| 741 |
+
{
|
| 742 |
+
"epoch": 0.37,
|
| 743 |
+
"learning_rate": 0.0002795942028985507,
|
| 744 |
+
"loss": 0.7007,
|
| 745 |
+
"step": 1160
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 0.37,
|
| 749 |
+
"learning_rate": 0.00027940096618357484,
|
| 750 |
+
"loss": 0.6854,
|
| 751 |
+
"step": 1170
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 0.38,
|
| 755 |
+
"learning_rate": 0.000279207729468599,
|
| 756 |
+
"loss": 0.7075,
|
| 757 |
+
"step": 1180
|
| 758 |
+
},
|
| 759 |
+
{
|
| 760 |
+
"epoch": 0.38,
|
| 761 |
+
"learning_rate": 0.0002790144927536232,
|
| 762 |
+
"loss": 0.6942,
|
| 763 |
+
"step": 1190
|
| 764 |
+
},
|
| 765 |
+
{
|
| 766 |
+
"epoch": 0.38,
|
| 767 |
+
"learning_rate": 0.0002788212560386473,
|
| 768 |
+
"loss": 0.7389,
|
| 769 |
+
"step": 1200
|
| 770 |
+
},
|
| 771 |
+
{
|
| 772 |
+
"epoch": 0.38,
|
| 773 |
+
"eval_loss": 0.6969788670539856,
|
| 774 |
+
"eval_runtime": 149.7859,
|
| 775 |
+
"eval_samples_per_second": 13.352,
|
| 776 |
+
"eval_steps_per_second": 1.669,
|
| 777 |
+
"step": 1200
|
| 778 |
+
},
|
| 779 |
+
{
|
| 780 |
+
"epoch": 0.39,
|
| 781 |
+
"learning_rate": 0.00027862801932367146,
|
| 782 |
+
"loss": 0.6633,
|
| 783 |
+
"step": 1210
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 0.39,
|
| 787 |
+
"learning_rate": 0.0002784347826086956,
|
| 788 |
+
"loss": 0.6741,
|
| 789 |
+
"step": 1220
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 0.39,
|
| 793 |
+
"learning_rate": 0.00027824154589371975,
|
| 794 |
+
"loss": 0.702,
|
| 795 |
+
"step": 1230
|
| 796 |
+
},
|
| 797 |
+
{
|
| 798 |
+
"epoch": 0.4,
|
| 799 |
+
"learning_rate": 0.00027804830917874395,
|
| 800 |
+
"loss": 0.691,
|
| 801 |
+
"step": 1240
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 0.4,
|
| 805 |
+
"learning_rate": 0.0002778550724637681,
|
| 806 |
+
"loss": 0.7049,
|
| 807 |
+
"step": 1250
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 0.4,
|
| 811 |
+
"learning_rate": 0.00027766183574879223,
|
| 812 |
+
"loss": 0.7254,
|
| 813 |
+
"step": 1260
|
| 814 |
+
},
|
| 815 |
+
{
|
| 816 |
+
"epoch": 0.41,
|
| 817 |
+
"learning_rate": 0.0002774685990338164,
|
| 818 |
+
"loss": 0.6732,
|
| 819 |
+
"step": 1270
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 0.41,
|
| 823 |
+
"learning_rate": 0.00027727536231884057,
|
| 824 |
+
"loss": 0.6995,
|
| 825 |
+
"step": 1280
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 0.41,
|
| 829 |
+
"learning_rate": 0.0002770821256038647,
|
| 830 |
+
"loss": 0.687,
|
| 831 |
+
"step": 1290
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 0.42,
|
| 835 |
+
"learning_rate": 0.00027688888888888885,
|
| 836 |
+
"loss": 0.6831,
|
| 837 |
+
"step": 1300
|
| 838 |
+
},
|
| 839 |
+
{
|
| 840 |
+
"epoch": 0.42,
|
| 841 |
+
"learning_rate": 0.000276695652173913,
|
| 842 |
+
"loss": 0.7189,
|
| 843 |
+
"step": 1310
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 0.42,
|
| 847 |
+
"learning_rate": 0.00027650241545893714,
|
| 848 |
+
"loss": 0.6996,
|
| 849 |
+
"step": 1320
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 0.43,
|
| 853 |
+
"learning_rate": 0.00027630917874396134,
|
| 854 |
+
"loss": 0.696,
|
| 855 |
+
"step": 1330
|
| 856 |
+
},
|
| 857 |
+
{
|
| 858 |
+
"epoch": 0.43,
|
| 859 |
+
"learning_rate": 0.0002761159420289855,
|
| 860 |
+
"loss": 0.7237,
|
| 861 |
+
"step": 1340
|
| 862 |
+
},
|
| 863 |
+
{
|
| 864 |
+
"epoch": 0.43,
|
| 865 |
+
"learning_rate": 0.0002759227053140096,
|
| 866 |
+
"loss": 0.7266,
|
| 867 |
+
"step": 1350
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 0.44,
|
| 871 |
+
"learning_rate": 0.00027572946859903376,
|
| 872 |
+
"loss": 0.6745,
|
| 873 |
+
"step": 1360
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 0.44,
|
| 877 |
+
"learning_rate": 0.0002755362318840579,
|
| 878 |
+
"loss": 0.7102,
|
| 879 |
+
"step": 1370
|
| 880 |
+
},
|
| 881 |
+
{
|
| 882 |
+
"epoch": 0.44,
|
| 883 |
+
"learning_rate": 0.0002753429951690821,
|
| 884 |
+
"loss": 0.6922,
|
| 885 |
+
"step": 1380
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 0.44,
|
| 889 |
+
"learning_rate": 0.00027514975845410625,
|
| 890 |
+
"loss": 0.7059,
|
| 891 |
+
"step": 1390
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 0.45,
|
| 895 |
+
"learning_rate": 0.0002749565217391304,
|
| 896 |
+
"loss": 0.7198,
|
| 897 |
+
"step": 1400
|
| 898 |
+
},
|
| 899 |
+
{
|
| 900 |
+
"epoch": 0.45,
|
| 901 |
+
"eval_loss": 0.6949622631072998,
|
| 902 |
+
"eval_runtime": 150.1806,
|
| 903 |
+
"eval_samples_per_second": 13.317,
|
| 904 |
+
"eval_steps_per_second": 1.665,
|
| 905 |
+
"step": 1400
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 0.45,
|
| 909 |
+
"learning_rate": 0.00027476328502415453,
|
| 910 |
+
"loss": 0.6608,
|
| 911 |
+
"step": 1410
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 0.45,
|
| 915 |
+
"learning_rate": 0.00027457004830917873,
|
| 916 |
+
"loss": 0.6699,
|
| 917 |
+
"step": 1420
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"epoch": 0.46,
|
| 921 |
+
"learning_rate": 0.00027437681159420287,
|
| 922 |
+
"loss": 0.6817,
|
| 923 |
+
"step": 1430
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 0.46,
|
| 927 |
+
"learning_rate": 0.000274183574879227,
|
| 928 |
+
"loss": 0.7019,
|
| 929 |
+
"step": 1440
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 0.46,
|
| 933 |
+
"learning_rate": 0.00027399033816425116,
|
| 934 |
+
"loss": 0.6839,
|
| 935 |
+
"step": 1450
|
| 936 |
+
},
|
| 937 |
+
{
|
| 938 |
+
"epoch": 0.47,
|
| 939 |
+
"learning_rate": 0.0002737971014492753,
|
| 940 |
+
"loss": 0.6725,
|
| 941 |
+
"step": 1460
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 0.47,
|
| 945 |
+
"learning_rate": 0.0002736038647342995,
|
| 946 |
+
"loss": 0.7065,
|
| 947 |
+
"step": 1470
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 0.47,
|
| 951 |
+
"learning_rate": 0.00027341062801932364,
|
| 952 |
+
"loss": 0.6728,
|
| 953 |
+
"step": 1480
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 0.48,
|
| 957 |
+
"learning_rate": 0.0002732173913043478,
|
| 958 |
+
"loss": 0.6449,
|
| 959 |
+
"step": 1490
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 0.48,
|
| 963 |
+
"learning_rate": 0.0002730241545893719,
|
| 964 |
+
"loss": 0.7094,
|
| 965 |
+
"step": 1500
|
| 966 |
+
},
|
| 967 |
+
{
|
| 968 |
+
"epoch": 0.48,
|
| 969 |
+
"learning_rate": 0.0002728309178743961,
|
| 970 |
+
"loss": 0.6881,
|
| 971 |
+
"step": 1510
|
| 972 |
+
},
|
| 973 |
+
{
|
| 974 |
+
"epoch": 0.49,
|
| 975 |
+
"learning_rate": 0.00027263768115942026,
|
| 976 |
+
"loss": 0.6804,
|
| 977 |
+
"step": 1520
|
| 978 |
+
},
|
| 979 |
+
{
|
| 980 |
+
"epoch": 0.49,
|
| 981 |
+
"learning_rate": 0.0002724444444444444,
|
| 982 |
+
"loss": 0.6822,
|
| 983 |
+
"step": 1530
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 0.49,
|
| 987 |
+
"learning_rate": 0.00027225120772946855,
|
| 988 |
+
"loss": 0.6816,
|
| 989 |
+
"step": 1540
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 0.5,
|
| 993 |
+
"learning_rate": 0.0002720579710144927,
|
| 994 |
+
"loss": 0.6615,
|
| 995 |
+
"step": 1550
|
| 996 |
+
},
|
| 997 |
+
{
|
| 998 |
+
"epoch": 0.5,
|
| 999 |
+
"learning_rate": 0.0002718647342995169,
|
| 1000 |
+
"loss": 0.6945,
|
| 1001 |
+
"step": 1560
|
| 1002 |
+
},
|
| 1003 |
+
{
|
| 1004 |
+
"epoch": 0.5,
|
| 1005 |
+
"learning_rate": 0.00027167149758454103,
|
| 1006 |
+
"loss": 0.7249,
|
| 1007 |
+
"step": 1570
|
| 1008 |
+
},
|
| 1009 |
+
{
|
| 1010 |
+
"epoch": 0.51,
|
| 1011 |
+
"learning_rate": 0.00027147826086956517,
|
| 1012 |
+
"loss": 0.7061,
|
| 1013 |
+
"step": 1580
|
| 1014 |
+
},
|
| 1015 |
+
{
|
| 1016 |
+
"epoch": 0.51,
|
| 1017 |
+
"learning_rate": 0.0002712850241545893,
|
| 1018 |
+
"loss": 0.6909,
|
| 1019 |
+
"step": 1590
|
| 1020 |
+
},
|
| 1021 |
+
{
|
| 1022 |
+
"epoch": 0.51,
|
| 1023 |
+
"learning_rate": 0.0002710917874396135,
|
| 1024 |
+
"loss": 0.7214,
|
| 1025 |
+
"step": 1600
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 0.51,
|
| 1029 |
+
"eval_loss": 0.6923746466636658,
|
| 1030 |
+
"eval_runtime": 150.1003,
|
| 1031 |
+
"eval_samples_per_second": 13.324,
|
| 1032 |
+
"eval_steps_per_second": 1.666,
|
| 1033 |
+
"step": 1600
|
| 1034 |
+
},
|
| 1035 |
+
{
|
| 1036 |
+
"epoch": 0.52,
|
| 1037 |
+
"learning_rate": 0.00027089855072463765,
|
| 1038 |
+
"loss": 0.7448,
|
| 1039 |
+
"step": 1610
|
| 1040 |
+
},
|
| 1041 |
+
{
|
| 1042 |
+
"epoch": 0.52,
|
| 1043 |
+
"learning_rate": 0.0002707053140096618,
|
| 1044 |
+
"loss": 0.6746,
|
| 1045 |
+
"step": 1620
|
| 1046 |
+
},
|
| 1047 |
+
{
|
| 1048 |
+
"epoch": 0.52,
|
| 1049 |
+
"learning_rate": 0.00027051207729468594,
|
| 1050 |
+
"loss": 0.6952,
|
| 1051 |
+
"step": 1630
|
| 1052 |
+
},
|
| 1053 |
+
{
|
| 1054 |
+
"epoch": 0.52,
|
| 1055 |
+
"learning_rate": 0.0002703188405797101,
|
| 1056 |
+
"loss": 0.6985,
|
| 1057 |
+
"step": 1640
|
| 1058 |
+
},
|
| 1059 |
+
{
|
| 1060 |
+
"epoch": 0.53,
|
| 1061 |
+
"learning_rate": 0.0002701256038647343,
|
| 1062 |
+
"loss": 0.706,
|
| 1063 |
+
"step": 1650
|
| 1064 |
+
},
|
| 1065 |
+
{
|
| 1066 |
+
"epoch": 0.53,
|
| 1067 |
+
"learning_rate": 0.0002699323671497584,
|
| 1068 |
+
"loss": 0.6838,
|
| 1069 |
+
"step": 1660
|
| 1070 |
+
},
|
| 1071 |
+
{
|
| 1072 |
+
"epoch": 0.53,
|
| 1073 |
+
"learning_rate": 0.00026973913043478256,
|
| 1074 |
+
"loss": 0.6809,
|
| 1075 |
+
"step": 1670
|
| 1076 |
+
},
|
| 1077 |
+
{
|
| 1078 |
+
"epoch": 0.54,
|
| 1079 |
+
"learning_rate": 0.0002695458937198067,
|
| 1080 |
+
"loss": 0.7066,
|
| 1081 |
+
"step": 1680
|
| 1082 |
+
},
|
| 1083 |
+
{
|
| 1084 |
+
"epoch": 0.54,
|
| 1085 |
+
"learning_rate": 0.0002693526570048309,
|
| 1086 |
+
"loss": 0.6828,
|
| 1087 |
+
"step": 1690
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 0.54,
|
| 1091 |
+
"learning_rate": 0.00026915942028985505,
|
| 1092 |
+
"loss": 0.6653,
|
| 1093 |
+
"step": 1700
|
| 1094 |
+
},
|
| 1095 |
+
{
|
| 1096 |
+
"epoch": 0.55,
|
| 1097 |
+
"learning_rate": 0.0002689661835748792,
|
| 1098 |
+
"loss": 0.6772,
|
| 1099 |
+
"step": 1710
|
| 1100 |
+
},
|
| 1101 |
+
{
|
| 1102 |
+
"epoch": 0.55,
|
| 1103 |
+
"learning_rate": 0.00026877294685990333,
|
| 1104 |
+
"loss": 0.6798,
|
| 1105 |
+
"step": 1720
|
| 1106 |
+
},
|
| 1107 |
+
{
|
| 1108 |
+
"epoch": 0.55,
|
| 1109 |
+
"learning_rate": 0.00026857971014492753,
|
| 1110 |
+
"loss": 0.6838,
|
| 1111 |
+
"step": 1730
|
| 1112 |
+
},
|
| 1113 |
+
{
|
| 1114 |
+
"epoch": 0.56,
|
| 1115 |
+
"learning_rate": 0.00026838647342995167,
|
| 1116 |
+
"loss": 0.7115,
|
| 1117 |
+
"step": 1740
|
| 1118 |
+
},
|
| 1119 |
+
{
|
| 1120 |
+
"epoch": 0.56,
|
| 1121 |
+
"learning_rate": 0.0002681932367149758,
|
| 1122 |
+
"loss": 0.6907,
|
| 1123 |
+
"step": 1750
|
| 1124 |
+
},
|
| 1125 |
+
{
|
| 1126 |
+
"epoch": 0.56,
|
| 1127 |
+
"learning_rate": 0.00026799999999999995,
|
| 1128 |
+
"loss": 0.6587,
|
| 1129 |
+
"step": 1760
|
| 1130 |
+
},
|
| 1131 |
+
{
|
| 1132 |
+
"epoch": 0.57,
|
| 1133 |
+
"learning_rate": 0.0002678067632850241,
|
| 1134 |
+
"loss": 0.7089,
|
| 1135 |
+
"step": 1770
|
| 1136 |
+
},
|
| 1137 |
+
{
|
| 1138 |
+
"epoch": 0.57,
|
| 1139 |
+
"learning_rate": 0.0002676135265700483,
|
| 1140 |
+
"loss": 0.6947,
|
| 1141 |
+
"step": 1780
|
| 1142 |
+
},
|
| 1143 |
+
{
|
| 1144 |
+
"epoch": 0.57,
|
| 1145 |
+
"learning_rate": 0.00026742028985507244,
|
| 1146 |
+
"loss": 0.698,
|
| 1147 |
+
"step": 1790
|
| 1148 |
+
},
|
| 1149 |
+
{
|
| 1150 |
+
"epoch": 0.58,
|
| 1151 |
+
"learning_rate": 0.0002672270531400966,
|
| 1152 |
+
"loss": 0.7183,
|
| 1153 |
+
"step": 1800
|
| 1154 |
+
},
|
| 1155 |
+
{
|
| 1156 |
+
"epoch": 0.58,
|
| 1157 |
+
"eval_loss": 0.6911550164222717,
|
| 1158 |
+
"eval_runtime": 368.3262,
|
| 1159 |
+
"eval_samples_per_second": 5.43,
|
| 1160 |
+
"eval_steps_per_second": 0.679,
|
| 1161 |
+
"step": 1800
|
| 1162 |
+
},
|
| 1163 |
+
{
|
| 1164 |
+
"epoch": 0.58,
|
| 1165 |
+
"learning_rate": 0.0002670338164251207,
|
| 1166 |
+
"loss": 0.7501,
|
| 1167 |
+
"step": 1810
|
| 1168 |
+
},
|
| 1169 |
+
{
|
| 1170 |
+
"epoch": 0.58,
|
| 1171 |
+
"learning_rate": 0.0002668405797101449,
|
| 1172 |
+
"loss": 0.6472,
|
| 1173 |
+
"step": 1820
|
| 1174 |
+
},
|
| 1175 |
+
{
|
| 1176 |
+
"epoch": 0.59,
|
| 1177 |
+
"learning_rate": 0.00026664734299516906,
|
| 1178 |
+
"loss": 0.6773,
|
| 1179 |
+
"step": 1830
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 0.59,
|
| 1183 |
+
"learning_rate": 0.0002664541062801932,
|
| 1184 |
+
"loss": 0.7195,
|
| 1185 |
+
"step": 1840
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 0.59,
|
| 1189 |
+
"learning_rate": 0.00026626086956521735,
|
| 1190 |
+
"loss": 0.7111,
|
| 1191 |
+
"step": 1850
|
| 1192 |
+
},
|
| 1193 |
+
{
|
| 1194 |
+
"epoch": 0.6,
|
| 1195 |
+
"learning_rate": 0.0002660676328502415,
|
| 1196 |
+
"loss": 0.6921,
|
| 1197 |
+
"step": 1860
|
| 1198 |
+
},
|
| 1199 |
+
{
|
| 1200 |
+
"epoch": 0.6,
|
| 1201 |
+
"learning_rate": 0.0002658743961352657,
|
| 1202 |
+
"loss": 0.6776,
|
| 1203 |
+
"step": 1870
|
| 1204 |
+
},
|
| 1205 |
+
{
|
| 1206 |
+
"epoch": 0.6,
|
| 1207 |
+
"learning_rate": 0.00026568115942028983,
|
| 1208 |
+
"loss": 0.6911,
|
| 1209 |
+
"step": 1880
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"epoch": 0.6,
|
| 1213 |
+
"learning_rate": 0.00026548792270531397,
|
| 1214 |
+
"loss": 0.7253,
|
| 1215 |
+
"step": 1890
|
| 1216 |
+
},
|
| 1217 |
+
{
|
| 1218 |
+
"epoch": 0.61,
|
| 1219 |
+
"learning_rate": 0.0002652946859903381,
|
| 1220 |
+
"loss": 0.672,
|
| 1221 |
+
"step": 1900
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 0.61,
|
| 1225 |
+
"learning_rate": 0.0002651014492753623,
|
| 1226 |
+
"loss": 0.7156,
|
| 1227 |
+
"step": 1910
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 0.61,
|
| 1231 |
+
"learning_rate": 0.00026490821256038645,
|
| 1232 |
+
"loss": 0.6929,
|
| 1233 |
+
"step": 1920
|
| 1234 |
+
},
|
| 1235 |
+
{
|
| 1236 |
+
"epoch": 0.62,
|
| 1237 |
+
"learning_rate": 0.0002647149758454106,
|
| 1238 |
+
"loss": 0.714,
|
| 1239 |
+
"step": 1930
|
| 1240 |
+
},
|
| 1241 |
+
{
|
| 1242 |
+
"epoch": 0.62,
|
| 1243 |
+
"learning_rate": 0.00026452173913043474,
|
| 1244 |
+
"loss": 0.6834,
|
| 1245 |
+
"step": 1940
|
| 1246 |
+
},
|
| 1247 |
+
{
|
| 1248 |
+
"epoch": 0.62,
|
| 1249 |
+
"learning_rate": 0.0002643285024154589,
|
| 1250 |
+
"loss": 0.7413,
|
| 1251 |
+
"step": 1950
|
| 1252 |
+
},
|
| 1253 |
+
{
|
| 1254 |
+
"epoch": 0.63,
|
| 1255 |
+
"learning_rate": 0.0002641352657004831,
|
| 1256 |
+
"loss": 0.6369,
|
| 1257 |
+
"step": 1960
|
| 1258 |
+
},
|
| 1259 |
+
{
|
| 1260 |
+
"epoch": 0.63,
|
| 1261 |
+
"learning_rate": 0.0002639420289855072,
|
| 1262 |
+
"loss": 0.6581,
|
| 1263 |
+
"step": 1970
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 0.63,
|
| 1267 |
+
"learning_rate": 0.00026374879227053136,
|
| 1268 |
+
"loss": 0.644,
|
| 1269 |
+
"step": 1980
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 0.64,
|
| 1273 |
+
"learning_rate": 0.0002635555555555555,
|
| 1274 |
+
"loss": 0.675,
|
| 1275 |
+
"step": 1990
|
| 1276 |
+
},
|
| 1277 |
+
{
|
| 1278 |
+
"epoch": 0.64,
|
| 1279 |
+
"learning_rate": 0.0002633623188405797,
|
| 1280 |
+
"loss": 0.6935,
|
| 1281 |
+
"step": 2000
|
| 1282 |
+
},
|
| 1283 |
+
{
|
| 1284 |
+
"epoch": 0.64,
|
| 1285 |
+
"eval_loss": 0.6894997954368591,
|
| 1286 |
+
"eval_runtime": 384.4995,
|
| 1287 |
+
"eval_samples_per_second": 5.202,
|
| 1288 |
+
"eval_steps_per_second": 0.65,
|
| 1289 |
+
"step": 2000
|
| 1290 |
+
},
|
| 1291 |
+
{
|
| 1292 |
+
"epoch": 0.64,
|
| 1293 |
+
"learning_rate": 0.00026316908212560384,
|
| 1294 |
+
"loss": 0.6739,
|
| 1295 |
+
"step": 2010
|
| 1296 |
+
},
|
| 1297 |
+
{
|
| 1298 |
+
"epoch": 0.65,
|
| 1299 |
+
"learning_rate": 0.000262975845410628,
|
| 1300 |
+
"loss": 0.719,
|
| 1301 |
+
"step": 2020
|
| 1302 |
+
},
|
| 1303 |
+
{
|
| 1304 |
+
"epoch": 0.65,
|
| 1305 |
+
"learning_rate": 0.00026278260869565213,
|
| 1306 |
+
"loss": 0.6724,
|
| 1307 |
+
"step": 2030
|
| 1308 |
+
},
|
| 1309 |
+
{
|
| 1310 |
+
"epoch": 0.65,
|
| 1311 |
+
"learning_rate": 0.0002625893719806763,
|
| 1312 |
+
"loss": 0.6736,
|
| 1313 |
+
"step": 2040
|
| 1314 |
+
},
|
| 1315 |
+
{
|
| 1316 |
+
"epoch": 0.66,
|
| 1317 |
+
"learning_rate": 0.00026239613526570047,
|
| 1318 |
+
"loss": 0.675,
|
| 1319 |
+
"step": 2050
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 0.66,
|
| 1323 |
+
"learning_rate": 0.0002622028985507246,
|
| 1324 |
+
"loss": 0.7289,
|
| 1325 |
+
"step": 2060
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 0.66,
|
| 1329 |
+
"learning_rate": 0.00026200966183574875,
|
| 1330 |
+
"loss": 0.7081,
|
| 1331 |
+
"step": 2070
|
| 1332 |
+
},
|
| 1333 |
+
{
|
| 1334 |
+
"epoch": 0.67,
|
| 1335 |
+
"learning_rate": 0.0002618164251207729,
|
| 1336 |
+
"loss": 0.6529,
|
| 1337 |
+
"step": 2080
|
| 1338 |
+
},
|
| 1339 |
+
{
|
| 1340 |
+
"epoch": 0.67,
|
| 1341 |
+
"learning_rate": 0.0002616231884057971,
|
| 1342 |
+
"loss": 0.6659,
|
| 1343 |
+
"step": 2090
|
| 1344 |
+
},
|
| 1345 |
+
{
|
| 1346 |
+
"epoch": 0.67,
|
| 1347 |
+
"learning_rate": 0.00026142995169082124,
|
| 1348 |
+
"loss": 0.663,
|
| 1349 |
+
"step": 2100
|
| 1350 |
+
},
|
| 1351 |
+
{
|
| 1352 |
+
"epoch": 0.68,
|
| 1353 |
+
"learning_rate": 0.0002612367149758454,
|
| 1354 |
+
"loss": 0.7021,
|
| 1355 |
+
"step": 2110
|
| 1356 |
+
},
|
| 1357 |
+
{
|
| 1358 |
+
"epoch": 0.68,
|
| 1359 |
+
"learning_rate": 0.0002610434782608695,
|
| 1360 |
+
"loss": 0.6856,
|
| 1361 |
+
"step": 2120
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 0.68,
|
| 1365 |
+
"learning_rate": 0.0002608502415458937,
|
| 1366 |
+
"loss": 0.6762,
|
| 1367 |
+
"step": 2130
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 0.68,
|
| 1371 |
+
"learning_rate": 0.00026065700483091786,
|
| 1372 |
+
"loss": 0.6754,
|
| 1373 |
+
"step": 2140
|
| 1374 |
+
},
|
| 1375 |
+
{
|
| 1376 |
+
"epoch": 0.69,
|
| 1377 |
+
"learning_rate": 0.000260463768115942,
|
| 1378 |
+
"loss": 0.7019,
|
| 1379 |
+
"step": 2150
|
| 1380 |
+
},
|
| 1381 |
+
{
|
| 1382 |
+
"epoch": 0.69,
|
| 1383 |
+
"learning_rate": 0.00026027053140096615,
|
| 1384 |
+
"loss": 0.6847,
|
| 1385 |
+
"step": 2160
|
| 1386 |
+
},
|
| 1387 |
+
{
|
| 1388 |
+
"epoch": 0.69,
|
| 1389 |
+
"learning_rate": 0.0002600772946859903,
|
| 1390 |
+
"loss": 0.6971,
|
| 1391 |
+
"step": 2170
|
| 1392 |
+
},
|
| 1393 |
+
{
|
| 1394 |
+
"epoch": 0.7,
|
| 1395 |
+
"learning_rate": 0.0002598840579710145,
|
| 1396 |
+
"loss": 0.6819,
|
| 1397 |
+
"step": 2180
|
| 1398 |
+
},
|
| 1399 |
+
{
|
| 1400 |
+
"epoch": 0.7,
|
| 1401 |
+
"learning_rate": 0.0002596908212560386,
|
| 1402 |
+
"loss": 0.6898,
|
| 1403 |
+
"step": 2190
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 0.7,
|
| 1407 |
+
"learning_rate": 0.00025949758454106277,
|
| 1408 |
+
"loss": 0.698,
|
| 1409 |
+
"step": 2200
|
| 1410 |
+
},
|
| 1411 |
+
{
|
| 1412 |
+
"epoch": 0.7,
|
| 1413 |
+
"eval_loss": 0.6893799304962158,
|
| 1414 |
+
"eval_runtime": 386.4657,
|
| 1415 |
+
"eval_samples_per_second": 5.175,
|
| 1416 |
+
"eval_steps_per_second": 0.647,
|
| 1417 |
+
"step": 2200
|
| 1418 |
+
},
|
| 1419 |
+
{
|
| 1420 |
+
"epoch": 0.71,
|
| 1421 |
+
"learning_rate": 0.0002593043478260869,
|
| 1422 |
+
"loss": 0.6876,
|
| 1423 |
+
"step": 2210
|
| 1424 |
+
},
|
| 1425 |
+
{
|
| 1426 |
+
"epoch": 0.71,
|
| 1427 |
+
"learning_rate": 0.0002591111111111111,
|
| 1428 |
+
"loss": 0.6798,
|
| 1429 |
+
"step": 2220
|
| 1430 |
+
},
|
| 1431 |
+
{
|
| 1432 |
+
"epoch": 0.71,
|
| 1433 |
+
"learning_rate": 0.00025891787439613525,
|
| 1434 |
+
"loss": 0.706,
|
| 1435 |
+
"step": 2230
|
| 1436 |
+
},
|
| 1437 |
+
{
|
| 1438 |
+
"epoch": 0.72,
|
| 1439 |
+
"learning_rate": 0.0002587246376811594,
|
| 1440 |
+
"loss": 0.6495,
|
| 1441 |
+
"step": 2240
|
| 1442 |
+
},
|
| 1443 |
+
{
|
| 1444 |
+
"epoch": 0.72,
|
| 1445 |
+
"learning_rate": 0.00025853140096618354,
|
| 1446 |
+
"loss": 0.6646,
|
| 1447 |
+
"step": 2250
|
| 1448 |
+
},
|
| 1449 |
+
{
|
| 1450 |
+
"epoch": 0.72,
|
| 1451 |
+
"learning_rate": 0.0002583381642512077,
|
| 1452 |
+
"loss": 0.6417,
|
| 1453 |
+
"step": 2260
|
| 1454 |
+
},
|
| 1455 |
+
{
|
| 1456 |
+
"epoch": 0.73,
|
| 1457 |
+
"learning_rate": 0.0002581449275362319,
|
| 1458 |
+
"loss": 0.6681,
|
| 1459 |
+
"step": 2270
|
| 1460 |
+
},
|
| 1461 |
+
{
|
| 1462 |
+
"epoch": 0.73,
|
| 1463 |
+
"learning_rate": 0.000257951690821256,
|
| 1464 |
+
"loss": 0.6689,
|
| 1465 |
+
"step": 2280
|
| 1466 |
+
},
|
| 1467 |
+
{
|
| 1468 |
+
"epoch": 0.73,
|
| 1469 |
+
"learning_rate": 0.00025775845410628016,
|
| 1470 |
+
"loss": 0.6837,
|
| 1471 |
+
"step": 2290
|
| 1472 |
+
},
|
| 1473 |
+
{
|
| 1474 |
+
"epoch": 0.74,
|
| 1475 |
+
"learning_rate": 0.0002575652173913043,
|
| 1476 |
+
"loss": 0.7031,
|
| 1477 |
+
"step": 2300
|
| 1478 |
+
},
|
| 1479 |
+
{
|
| 1480 |
+
"epoch": 0.74,
|
| 1481 |
+
"learning_rate": 0.0002573719806763285,
|
| 1482 |
+
"loss": 0.6746,
|
| 1483 |
+
"step": 2310
|
| 1484 |
+
},
|
| 1485 |
+
{
|
| 1486 |
+
"epoch": 0.74,
|
| 1487 |
+
"learning_rate": 0.00025717874396135264,
|
| 1488 |
+
"loss": 0.6951,
|
| 1489 |
+
"step": 2320
|
| 1490 |
+
},
|
| 1491 |
+
{
|
| 1492 |
+
"epoch": 0.75,
|
| 1493 |
+
"learning_rate": 0.0002569855072463768,
|
| 1494 |
+
"loss": 0.6988,
|
| 1495 |
+
"step": 2330
|
| 1496 |
+
},
|
| 1497 |
+
{
|
| 1498 |
+
"epoch": 0.75,
|
| 1499 |
+
"learning_rate": 0.00025679227053140093,
|
| 1500 |
+
"loss": 0.6541,
|
| 1501 |
+
"step": 2340
|
| 1502 |
+
},
|
| 1503 |
+
{
|
| 1504 |
+
"epoch": 0.75,
|
| 1505 |
+
"learning_rate": 0.0002565990338164251,
|
| 1506 |
+
"loss": 0.6366,
|
| 1507 |
+
"step": 2350
|
| 1508 |
+
},
|
| 1509 |
+
{
|
| 1510 |
+
"epoch": 0.76,
|
| 1511 |
+
"learning_rate": 0.00025640579710144927,
|
| 1512 |
+
"loss": 0.7011,
|
| 1513 |
+
"step": 2360
|
| 1514 |
+
},
|
| 1515 |
+
{
|
| 1516 |
+
"epoch": 0.76,
|
| 1517 |
+
"learning_rate": 0.0002562125603864734,
|
| 1518 |
+
"loss": 0.6935,
|
| 1519 |
+
"step": 2370
|
| 1520 |
+
},
|
| 1521 |
+
{
|
| 1522 |
+
"epoch": 0.76,
|
| 1523 |
+
"learning_rate": 0.00025601932367149755,
|
| 1524 |
+
"loss": 0.6931,
|
| 1525 |
+
"step": 2380
|
| 1526 |
+
},
|
| 1527 |
+
{
|
| 1528 |
+
"epoch": 0.76,
|
| 1529 |
+
"learning_rate": 0.0002558260869565217,
|
| 1530 |
+
"loss": 0.7004,
|
| 1531 |
+
"step": 2390
|
| 1532 |
+
},
|
| 1533 |
+
{
|
| 1534 |
+
"epoch": 0.77,
|
| 1535 |
+
"learning_rate": 0.0002556328502415459,
|
| 1536 |
+
"loss": 0.6556,
|
| 1537 |
+
"step": 2400
|
| 1538 |
+
},
|
| 1539 |
+
{
|
| 1540 |
+
"epoch": 0.77,
|
| 1541 |
+
"eval_loss": 0.6892030239105225,
|
| 1542 |
+
"eval_runtime": 391.4843,
|
| 1543 |
+
"eval_samples_per_second": 5.109,
|
| 1544 |
+
"eval_steps_per_second": 0.639,
|
| 1545 |
+
"step": 2400
|
| 1546 |
+
},
|
| 1547 |
+
{
|
| 1548 |
+
"epoch": 0.77,
|
| 1549 |
+
"learning_rate": 0.00025543961352657003,
|
| 1550 |
+
"loss": 0.664,
|
| 1551 |
+
"step": 2410
|
| 1552 |
+
},
|
| 1553 |
+
{
|
| 1554 |
+
"epoch": 0.77,
|
| 1555 |
+
"learning_rate": 0.0002552463768115942,
|
| 1556 |
+
"loss": 0.7162,
|
| 1557 |
+
"step": 2420
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 0.78,
|
| 1561 |
+
"learning_rate": 0.0002550531400966183,
|
| 1562 |
+
"loss": 0.646,
|
| 1563 |
+
"step": 2430
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 0.78,
|
| 1567 |
+
"learning_rate": 0.0002548599033816425,
|
| 1568 |
+
"loss": 0.6515,
|
| 1569 |
+
"step": 2440
|
| 1570 |
+
},
|
| 1571 |
+
{
|
| 1572 |
+
"epoch": 0.78,
|
| 1573 |
+
"learning_rate": 0.00025466666666666666,
|
| 1574 |
+
"loss": 0.6953,
|
| 1575 |
+
"step": 2450
|
| 1576 |
+
},
|
| 1577 |
+
{
|
| 1578 |
+
"epoch": 0.79,
|
| 1579 |
+
"learning_rate": 0.0002544734299516908,
|
| 1580 |
+
"loss": 0.6887,
|
| 1581 |
+
"step": 2460
|
| 1582 |
+
},
|
| 1583 |
+
{
|
| 1584 |
+
"epoch": 0.79,
|
| 1585 |
+
"learning_rate": 0.00025428019323671494,
|
| 1586 |
+
"loss": 0.6739,
|
| 1587 |
+
"step": 2470
|
| 1588 |
+
},
|
| 1589 |
+
{
|
| 1590 |
+
"epoch": 0.79,
|
| 1591 |
+
"learning_rate": 0.0002540869565217391,
|
| 1592 |
+
"loss": 0.6923,
|
| 1593 |
+
"step": 2480
|
| 1594 |
+
},
|
| 1595 |
+
{
|
| 1596 |
+
"epoch": 0.8,
|
| 1597 |
+
"learning_rate": 0.0002538937198067633,
|
| 1598 |
+
"loss": 0.6877,
|
| 1599 |
+
"step": 2490
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 0.8,
|
| 1603 |
+
"learning_rate": 0.0002537004830917874,
|
| 1604 |
+
"loss": 0.6865,
|
| 1605 |
+
"step": 2500
|
| 1606 |
+
},
|
| 1607 |
+
{
|
| 1608 |
+
"epoch": 0.8,
|
| 1609 |
+
"learning_rate": 0.00025350724637681157,
|
| 1610 |
+
"loss": 0.6337,
|
| 1611 |
+
"step": 2510
|
| 1612 |
+
},
|
| 1613 |
+
{
|
| 1614 |
+
"epoch": 0.81,
|
| 1615 |
+
"learning_rate": 0.0002533140096618357,
|
| 1616 |
+
"loss": 0.7073,
|
| 1617 |
+
"step": 2520
|
| 1618 |
+
},
|
| 1619 |
+
{
|
| 1620 |
+
"epoch": 0.81,
|
| 1621 |
+
"learning_rate": 0.0002531207729468599,
|
| 1622 |
+
"loss": 0.6973,
|
| 1623 |
+
"step": 2530
|
| 1624 |
+
},
|
| 1625 |
+
{
|
| 1626 |
+
"epoch": 0.81,
|
| 1627 |
+
"learning_rate": 0.00025292753623188405,
|
| 1628 |
+
"loss": 0.6719,
|
| 1629 |
+
"step": 2540
|
| 1630 |
+
},
|
| 1631 |
+
{
|
| 1632 |
+
"epoch": 0.82,
|
| 1633 |
+
"learning_rate": 0.0002527342995169082,
|
| 1634 |
+
"loss": 0.6674,
|
| 1635 |
+
"step": 2550
|
| 1636 |
+
},
|
| 1637 |
+
{
|
| 1638 |
+
"epoch": 0.82,
|
| 1639 |
+
"learning_rate": 0.00025254106280193234,
|
| 1640 |
+
"loss": 0.6745,
|
| 1641 |
+
"step": 2560
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 0.82,
|
| 1645 |
+
"learning_rate": 0.0002523478260869565,
|
| 1646 |
+
"loss": 0.6914,
|
| 1647 |
+
"step": 2570
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"epoch": 0.83,
|
| 1651 |
+
"learning_rate": 0.0002521545893719807,
|
| 1652 |
+
"loss": 0.6382,
|
| 1653 |
+
"step": 2580
|
| 1654 |
+
},
|
| 1655 |
+
{
|
| 1656 |
+
"epoch": 0.83,
|
| 1657 |
+
"learning_rate": 0.0002519613526570048,
|
| 1658 |
+
"loss": 0.6644,
|
| 1659 |
+
"step": 2590
|
| 1660 |
+
},
|
| 1661 |
+
{
|
| 1662 |
+
"epoch": 0.83,
|
| 1663 |
+
"learning_rate": 0.00025176811594202896,
|
| 1664 |
+
"loss": 0.6892,
|
| 1665 |
+
"step": 2600
|
| 1666 |
+
},
|
| 1667 |
+
{
|
| 1668 |
+
"epoch": 0.83,
|
| 1669 |
+
"eval_loss": 0.6873727440834045,
|
| 1670 |
+
"eval_runtime": 390.6367,
|
| 1671 |
+
"eval_samples_per_second": 5.12,
|
| 1672 |
+
"eval_steps_per_second": 0.64,
|
| 1673 |
+
"step": 2600
|
| 1674 |
+
},
|
| 1675 |
+
{
|
| 1676 |
+
"epoch": 0.84,
|
| 1677 |
+
"learning_rate": 0.0002515748792270531,
|
| 1678 |
+
"loss": 0.6592,
|
| 1679 |
+
"step": 2610
|
| 1680 |
+
},
|
| 1681 |
+
{
|
| 1682 |
+
"epoch": 0.84,
|
| 1683 |
+
"learning_rate": 0.0002513816425120773,
|
| 1684 |
+
"loss": 0.6827,
|
| 1685 |
+
"step": 2620
|
| 1686 |
+
},
|
| 1687 |
+
{
|
| 1688 |
+
"epoch": 0.84,
|
| 1689 |
+
"learning_rate": 0.00025118840579710144,
|
| 1690 |
+
"loss": 0.6436,
|
| 1691 |
+
"step": 2630
|
| 1692 |
+
},
|
| 1693 |
+
{
|
| 1694 |
+
"epoch": 0.84,
|
| 1695 |
+
"learning_rate": 0.0002509951690821256,
|
| 1696 |
+
"loss": 0.6969,
|
| 1697 |
+
"step": 2640
|
| 1698 |
+
},
|
| 1699 |
+
{
|
| 1700 |
+
"epoch": 0.85,
|
| 1701 |
+
"learning_rate": 0.0002508019323671497,
|
| 1702 |
+
"loss": 0.6747,
|
| 1703 |
+
"step": 2650
|
| 1704 |
+
},
|
| 1705 |
+
{
|
| 1706 |
+
"epoch": 0.85,
|
| 1707 |
+
"learning_rate": 0.0002506086956521739,
|
| 1708 |
+
"loss": 0.697,
|
| 1709 |
+
"step": 2660
|
| 1710 |
+
},
|
| 1711 |
+
{
|
| 1712 |
+
"epoch": 0.85,
|
| 1713 |
+
"learning_rate": 0.00025041545893719807,
|
| 1714 |
+
"loss": 0.7146,
|
| 1715 |
+
"step": 2670
|
| 1716 |
+
},
|
| 1717 |
+
{
|
| 1718 |
+
"epoch": 0.86,
|
| 1719 |
+
"learning_rate": 0.0002502222222222222,
|
| 1720 |
+
"loss": 0.689,
|
| 1721 |
+
"step": 2680
|
| 1722 |
+
},
|
| 1723 |
+
{
|
| 1724 |
+
"epoch": 0.86,
|
| 1725 |
+
"learning_rate": 0.00025002898550724635,
|
| 1726 |
+
"loss": 0.6957,
|
| 1727 |
+
"step": 2690
|
| 1728 |
+
},
|
| 1729 |
+
{
|
| 1730 |
+
"epoch": 0.86,
|
| 1731 |
+
"learning_rate": 0.0002498357487922705,
|
| 1732 |
+
"loss": 0.708,
|
| 1733 |
+
"step": 2700
|
| 1734 |
+
},
|
| 1735 |
+
{
|
| 1736 |
+
"epoch": 0.87,
|
| 1737 |
+
"learning_rate": 0.0002496425120772947,
|
| 1738 |
+
"loss": 0.7306,
|
| 1739 |
+
"step": 2710
|
| 1740 |
+
},
|
| 1741 |
+
{
|
| 1742 |
+
"epoch": 0.87,
|
| 1743 |
+
"learning_rate": 0.00024944927536231883,
|
| 1744 |
+
"loss": 0.6418,
|
| 1745 |
+
"step": 2720
|
| 1746 |
+
},
|
| 1747 |
+
{
|
| 1748 |
+
"epoch": 0.87,
|
| 1749 |
+
"learning_rate": 0.000249256038647343,
|
| 1750 |
+
"loss": 0.6888,
|
| 1751 |
+
"step": 2730
|
| 1752 |
+
},
|
| 1753 |
+
{
|
| 1754 |
+
"epoch": 0.88,
|
| 1755 |
+
"learning_rate": 0.0002490628019323671,
|
| 1756 |
+
"loss": 0.6573,
|
| 1757 |
+
"step": 2740
|
| 1758 |
+
},
|
| 1759 |
+
{
|
| 1760 |
+
"epoch": 0.88,
|
| 1761 |
+
"learning_rate": 0.0002488695652173913,
|
| 1762 |
+
"loss": 0.7008,
|
| 1763 |
+
"step": 2750
|
| 1764 |
+
},
|
| 1765 |
+
{
|
| 1766 |
+
"epoch": 0.88,
|
| 1767 |
+
"learning_rate": 0.00024867632850241546,
|
| 1768 |
+
"loss": 0.689,
|
| 1769 |
+
"step": 2760
|
| 1770 |
+
},
|
| 1771 |
+
{
|
| 1772 |
+
"epoch": 0.89,
|
| 1773 |
+
"learning_rate": 0.0002484830917874396,
|
| 1774 |
+
"loss": 0.6739,
|
| 1775 |
+
"step": 2770
|
| 1776 |
+
},
|
| 1777 |
+
{
|
| 1778 |
+
"epoch": 0.89,
|
| 1779 |
+
"learning_rate": 0.00024828985507246374,
|
| 1780 |
+
"loss": 0.6867,
|
| 1781 |
+
"step": 2780
|
| 1782 |
+
},
|
| 1783 |
+
{
|
| 1784 |
+
"epoch": 0.89,
|
| 1785 |
+
"learning_rate": 0.0002480966183574879,
|
| 1786 |
+
"loss": 0.6874,
|
| 1787 |
+
"step": 2790
|
| 1788 |
+
},
|
| 1789 |
+
{
|
| 1790 |
+
"epoch": 0.9,
|
| 1791 |
+
"learning_rate": 0.0002479033816425121,
|
| 1792 |
+
"loss": 0.6858,
|
| 1793 |
+
"step": 2800
|
| 1794 |
+
},
|
| 1795 |
+
{
|
| 1796 |
+
"epoch": 0.9,
|
| 1797 |
+
"eval_loss": 0.686144232749939,
|
| 1798 |
+
"eval_runtime": 396.2716,
|
| 1799 |
+
"eval_samples_per_second": 5.047,
|
| 1800 |
+
"eval_steps_per_second": 0.631,
|
| 1801 |
+
"step": 2800
|
| 1802 |
+
},
|
| 1803 |
+
{
|
| 1804 |
+
"epoch": 0.9,
|
| 1805 |
+
"learning_rate": 0.0002477101449275362,
|
| 1806 |
+
"loss": 0.659,
|
| 1807 |
+
"step": 2810
|
| 1808 |
+
},
|
| 1809 |
+
{
|
| 1810 |
+
"epoch": 0.9,
|
| 1811 |
+
"learning_rate": 0.00024751690821256037,
|
| 1812 |
+
"loss": 0.6537,
|
| 1813 |
+
"step": 2820
|
| 1814 |
+
},
|
| 1815 |
+
{
|
| 1816 |
+
"epoch": 0.91,
|
| 1817 |
+
"learning_rate": 0.0002473236714975845,
|
| 1818 |
+
"loss": 0.7331,
|
| 1819 |
+
"step": 2830
|
| 1820 |
+
},
|
| 1821 |
+
{
|
| 1822 |
+
"epoch": 0.91,
|
| 1823 |
+
"learning_rate": 0.0002471304347826087,
|
| 1824 |
+
"loss": 0.6855,
|
| 1825 |
+
"step": 2840
|
| 1826 |
+
},
|
| 1827 |
+
{
|
| 1828 |
+
"epoch": 0.91,
|
| 1829 |
+
"learning_rate": 0.00024693719806763285,
|
| 1830 |
+
"loss": 0.7252,
|
| 1831 |
+
"step": 2850
|
| 1832 |
+
},
|
| 1833 |
+
{
|
| 1834 |
+
"epoch": 0.92,
|
| 1835 |
+
"learning_rate": 0.000246743961352657,
|
| 1836 |
+
"loss": 0.7,
|
| 1837 |
+
"step": 2860
|
| 1838 |
+
},
|
| 1839 |
+
{
|
| 1840 |
+
"epoch": 0.92,
|
| 1841 |
+
"learning_rate": 0.00024655072463768113,
|
| 1842 |
+
"loss": 0.6917,
|
| 1843 |
+
"step": 2870
|
| 1844 |
+
},
|
| 1845 |
+
{
|
| 1846 |
+
"epoch": 0.92,
|
| 1847 |
+
"learning_rate": 0.0002463574879227053,
|
| 1848 |
+
"loss": 0.6828,
|
| 1849 |
+
"step": 2880
|
| 1850 |
+
},
|
| 1851 |
+
{
|
| 1852 |
+
"epoch": 0.92,
|
| 1853 |
+
"learning_rate": 0.0002461642512077295,
|
| 1854 |
+
"loss": 0.6747,
|
| 1855 |
+
"step": 2890
|
| 1856 |
+
},
|
| 1857 |
+
{
|
| 1858 |
+
"epoch": 0.93,
|
| 1859 |
+
"learning_rate": 0.0002459710144927536,
|
| 1860 |
+
"loss": 0.696,
|
| 1861 |
+
"step": 2900
|
| 1862 |
+
},
|
| 1863 |
+
{
|
| 1864 |
+
"epoch": 0.93,
|
| 1865 |
+
"learning_rate": 0.00024577777777777776,
|
| 1866 |
+
"loss": 0.6573,
|
| 1867 |
+
"step": 2910
|
| 1868 |
+
},
|
| 1869 |
+
{
|
| 1870 |
+
"epoch": 0.93,
|
| 1871 |
+
"learning_rate": 0.0002455845410628019,
|
| 1872 |
+
"loss": 0.6811,
|
| 1873 |
+
"step": 2920
|
| 1874 |
+
},
|
| 1875 |
+
{
|
| 1876 |
+
"epoch": 0.94,
|
| 1877 |
+
"learning_rate": 0.0002453913043478261,
|
| 1878 |
+
"loss": 0.6992,
|
| 1879 |
+
"step": 2930
|
| 1880 |
+
},
|
| 1881 |
+
{
|
| 1882 |
+
"epoch": 0.94,
|
| 1883 |
+
"learning_rate": 0.00024519806763285024,
|
| 1884 |
+
"loss": 0.6972,
|
| 1885 |
+
"step": 2940
|
| 1886 |
+
},
|
| 1887 |
+
{
|
| 1888 |
+
"epoch": 0.94,
|
| 1889 |
+
"learning_rate": 0.0002450048309178744,
|
| 1890 |
+
"loss": 0.6685,
|
| 1891 |
+
"step": 2950
|
| 1892 |
+
},
|
| 1893 |
+
{
|
| 1894 |
+
"epoch": 0.95,
|
| 1895 |
+
"learning_rate": 0.0002448115942028985,
|
| 1896 |
+
"loss": 0.6531,
|
| 1897 |
+
"step": 2960
|
| 1898 |
+
},
|
| 1899 |
+
{
|
| 1900 |
+
"epoch": 0.95,
|
| 1901 |
+
"learning_rate": 0.0002446183574879227,
|
| 1902 |
+
"loss": 0.6749,
|
| 1903 |
+
"step": 2970
|
| 1904 |
+
},
|
| 1905 |
+
{
|
| 1906 |
+
"epoch": 0.95,
|
| 1907 |
+
"learning_rate": 0.00024442512077294686,
|
| 1908 |
+
"loss": 0.6811,
|
| 1909 |
+
"step": 2980
|
| 1910 |
+
},
|
| 1911 |
+
{
|
| 1912 |
+
"epoch": 0.96,
|
| 1913 |
+
"learning_rate": 0.000244231884057971,
|
| 1914 |
+
"loss": 0.6904,
|
| 1915 |
+
"step": 2990
|
| 1916 |
+
},
|
| 1917 |
+
{
|
| 1918 |
+
"epoch": 0.96,
|
| 1919 |
+
"learning_rate": 0.00024403864734299515,
|
| 1920 |
+
"loss": 0.6819,
|
| 1921 |
+
"step": 3000
|
| 1922 |
+
},
|
| 1923 |
+
{
|
| 1924 |
+
"epoch": 0.96,
|
| 1925 |
+
"eval_loss": 0.6858677864074707,
|
| 1926 |
+
"eval_runtime": 418.7569,
|
| 1927 |
+
"eval_samples_per_second": 4.776,
|
| 1928 |
+
"eval_steps_per_second": 0.597,
|
| 1929 |
+
"step": 3000
|
| 1930 |
+
},
|
| 1931 |
+
{
|
| 1932 |
+
"epoch": 0.96,
|
| 1933 |
+
"learning_rate": 0.0002438454106280193,
|
| 1934 |
+
"loss": 0.6773,
|
| 1935 |
+
"step": 3010
|
| 1936 |
+
},
|
| 1937 |
+
{
|
| 1938 |
+
"epoch": 0.97,
|
| 1939 |
+
"learning_rate": 0.00024365217391304346,
|
| 1940 |
+
"loss": 0.7034,
|
| 1941 |
+
"step": 3020
|
| 1942 |
+
},
|
| 1943 |
+
{
|
| 1944 |
+
"epoch": 0.97,
|
| 1945 |
+
"learning_rate": 0.0002434589371980676,
|
| 1946 |
+
"loss": 0.6804,
|
| 1947 |
+
"step": 3030
|
| 1948 |
+
},
|
| 1949 |
+
{
|
| 1950 |
+
"epoch": 0.97,
|
| 1951 |
+
"learning_rate": 0.00024326570048309177,
|
| 1952 |
+
"loss": 0.6912,
|
| 1953 |
+
"step": 3040
|
| 1954 |
+
},
|
| 1955 |
+
{
|
| 1956 |
+
"epoch": 0.98,
|
| 1957 |
+
"learning_rate": 0.00024307246376811592,
|
| 1958 |
+
"loss": 0.683,
|
| 1959 |
+
"step": 3050
|
| 1960 |
+
},
|
| 1961 |
+
{
|
| 1962 |
+
"epoch": 0.98,
|
| 1963 |
+
"learning_rate": 0.0002428792270531401,
|
| 1964 |
+
"loss": 0.6767,
|
| 1965 |
+
"step": 3060
|
| 1966 |
+
},
|
| 1967 |
+
{
|
| 1968 |
+
"epoch": 0.98,
|
| 1969 |
+
"learning_rate": 0.00024268599033816423,
|
| 1970 |
+
"loss": 0.6683,
|
| 1971 |
+
"step": 3070
|
| 1972 |
+
},
|
| 1973 |
+
{
|
| 1974 |
+
"epoch": 0.99,
|
| 1975 |
+
"learning_rate": 0.00024249275362318837,
|
| 1976 |
+
"loss": 0.6777,
|
| 1977 |
+
"step": 3080
|
| 1978 |
+
},
|
| 1979 |
+
{
|
| 1980 |
+
"epoch": 0.99,
|
| 1981 |
+
"learning_rate": 0.00024229951690821254,
|
| 1982 |
+
"loss": 0.6813,
|
| 1983 |
+
"step": 3090
|
| 1984 |
+
},
|
| 1985 |
+
{
|
| 1986 |
+
"epoch": 0.99,
|
| 1987 |
+
"learning_rate": 0.00024210628019323668,
|
| 1988 |
+
"loss": 0.6878,
|
| 1989 |
+
"step": 3100
|
| 1990 |
+
},
|
| 1991 |
+
{
|
| 1992 |
+
"epoch": 1.0,
|
| 1993 |
+
"learning_rate": 0.00024191304347826085,
|
| 1994 |
+
"loss": 0.6885,
|
| 1995 |
+
"step": 3110
|
| 1996 |
+
},
|
| 1997 |
+
{
|
| 1998 |
+
"epoch": 1.0,
|
| 1999 |
+
"learning_rate": 0.000241719806763285,
|
| 2000 |
+
"loss": 0.6373,
|
| 2001 |
+
"step": 3120
|
| 2002 |
+
},
|
| 2003 |
+
{
|
| 2004 |
+
"epoch": 1.0,
|
| 2005 |
+
"learning_rate": 0.00024152657004830917,
|
| 2006 |
+
"loss": 0.6982,
|
| 2007 |
+
"step": 3130
|
| 2008 |
+
},
|
| 2009 |
+
{
|
| 2010 |
+
"epoch": 1.0,
|
| 2011 |
+
"learning_rate": 0.0002413333333333333,
|
| 2012 |
+
"loss": 0.6485,
|
| 2013 |
+
"step": 3140
|
| 2014 |
+
},
|
| 2015 |
+
{
|
| 2016 |
+
"epoch": 1.01,
|
| 2017 |
+
"learning_rate": 0.00024114009661835748,
|
| 2018 |
+
"loss": 0.6522,
|
| 2019 |
+
"step": 3150
|
| 2020 |
+
},
|
| 2021 |
+
{
|
| 2022 |
+
"epoch": 1.01,
|
| 2023 |
+
"learning_rate": 0.00024094685990338162,
|
| 2024 |
+
"loss": 0.6092,
|
| 2025 |
+
"step": 3160
|
| 2026 |
+
},
|
| 2027 |
+
{
|
| 2028 |
+
"epoch": 1.01,
|
| 2029 |
+
"learning_rate": 0.00024075362318840576,
|
| 2030 |
+
"loss": 0.6362,
|
| 2031 |
+
"step": 3170
|
| 2032 |
+
},
|
| 2033 |
+
{
|
| 2034 |
+
"epoch": 1.02,
|
| 2035 |
+
"learning_rate": 0.00024056038647342993,
|
| 2036 |
+
"loss": 0.6725,
|
| 2037 |
+
"step": 3180
|
| 2038 |
+
},
|
| 2039 |
+
{
|
| 2040 |
+
"epoch": 1.02,
|
| 2041 |
+
"learning_rate": 0.00024036714975845408,
|
| 2042 |
+
"loss": 0.6718,
|
| 2043 |
+
"step": 3190
|
| 2044 |
+
},
|
| 2045 |
+
{
|
| 2046 |
+
"epoch": 1.02,
|
| 2047 |
+
"learning_rate": 0.00024017391304347825,
|
| 2048 |
+
"loss": 0.6738,
|
| 2049 |
+
"step": 3200
|
| 2050 |
+
},
|
| 2051 |
+
{
|
| 2052 |
+
"epoch": 1.02,
|
| 2053 |
+
"eval_loss": 0.6860418915748596,
|
| 2054 |
+
"eval_runtime": 407.1637,
|
| 2055 |
+
"eval_samples_per_second": 4.912,
|
| 2056 |
+
"eval_steps_per_second": 0.614,
|
| 2057 |
+
"step": 3200
|
| 2058 |
+
},
|
| 2059 |
+
{
|
| 2060 |
+
"epoch": 1.03,
|
| 2061 |
+
"learning_rate": 0.0002399806763285024,
|
| 2062 |
+
"loss": 0.658,
|
| 2063 |
+
"step": 3210
|
| 2064 |
+
},
|
| 2065 |
+
{
|
| 2066 |
+
"epoch": 1.03,
|
| 2067 |
+
"learning_rate": 0.00023978743961352656,
|
| 2068 |
+
"loss": 0.6325,
|
| 2069 |
+
"step": 3220
|
| 2070 |
+
},
|
| 2071 |
+
{
|
| 2072 |
+
"epoch": 1.03,
|
| 2073 |
+
"learning_rate": 0.0002395942028985507,
|
| 2074 |
+
"loss": 0.6372,
|
| 2075 |
+
"step": 3230
|
| 2076 |
+
},
|
| 2077 |
+
{
|
| 2078 |
+
"epoch": 1.04,
|
| 2079 |
+
"learning_rate": 0.00023940096618357487,
|
| 2080 |
+
"loss": 0.6075,
|
| 2081 |
+
"step": 3240
|
| 2082 |
+
},
|
| 2083 |
+
{
|
| 2084 |
+
"epoch": 1.04,
|
| 2085 |
+
"learning_rate": 0.000239207729468599,
|
| 2086 |
+
"loss": 0.6212,
|
| 2087 |
+
"step": 3250
|
| 2088 |
+
},
|
| 2089 |
+
{
|
| 2090 |
+
"epoch": 1.04,
|
| 2091 |
+
"learning_rate": 0.00023901449275362315,
|
| 2092 |
+
"loss": 0.6373,
|
| 2093 |
+
"step": 3260
|
| 2094 |
+
},
|
| 2095 |
+
{
|
| 2096 |
+
"epoch": 1.05,
|
| 2097 |
+
"learning_rate": 0.00023882125603864732,
|
| 2098 |
+
"loss": 0.6799,
|
| 2099 |
+
"step": 3270
|
| 2100 |
+
},
|
| 2101 |
+
{
|
| 2102 |
+
"epoch": 1.05,
|
| 2103 |
+
"learning_rate": 0.0002386280193236715,
|
| 2104 |
+
"loss": 0.6769,
|
| 2105 |
+
"step": 3280
|
| 2106 |
+
},
|
| 2107 |
+
{
|
| 2108 |
+
"epoch": 1.05,
|
| 2109 |
+
"learning_rate": 0.00023843478260869564,
|
| 2110 |
+
"loss": 0.6265,
|
| 2111 |
+
"step": 3290
|
| 2112 |
+
},
|
| 2113 |
+
{
|
| 2114 |
+
"epoch": 1.06,
|
| 2115 |
+
"learning_rate": 0.00023824154589371978,
|
| 2116 |
+
"loss": 0.6627,
|
| 2117 |
+
"step": 3300
|
| 2118 |
+
},
|
| 2119 |
+
{
|
| 2120 |
+
"epoch": 1.06,
|
| 2121 |
+
"learning_rate": 0.00023804830917874392,
|
| 2122 |
+
"loss": 0.6602,
|
| 2123 |
+
"step": 3310
|
| 2124 |
+
},
|
| 2125 |
+
{
|
| 2126 |
+
"epoch": 1.06,
|
| 2127 |
+
"learning_rate": 0.0002378550724637681,
|
| 2128 |
+
"loss": 0.671,
|
| 2129 |
+
"step": 3320
|
| 2130 |
+
},
|
| 2131 |
+
{
|
| 2132 |
+
"epoch": 1.07,
|
| 2133 |
+
"learning_rate": 0.00023766183574879226,
|
| 2134 |
+
"loss": 0.6297,
|
| 2135 |
+
"step": 3330
|
| 2136 |
+
},
|
| 2137 |
+
{
|
| 2138 |
+
"epoch": 1.07,
|
| 2139 |
+
"learning_rate": 0.0002374685990338164,
|
| 2140 |
+
"loss": 0.662,
|
| 2141 |
+
"step": 3340
|
| 2142 |
+
},
|
| 2143 |
+
{
|
| 2144 |
+
"epoch": 1.07,
|
| 2145 |
+
"learning_rate": 0.00023727536231884055,
|
| 2146 |
+
"loss": 0.6224,
|
| 2147 |
+
"step": 3350
|
| 2148 |
+
},
|
| 2149 |
+
{
|
| 2150 |
+
"epoch": 1.08,
|
| 2151 |
+
"learning_rate": 0.00023708212560386472,
|
| 2152 |
+
"loss": 0.6527,
|
| 2153 |
+
"step": 3360
|
| 2154 |
+
},
|
| 2155 |
+
{
|
| 2156 |
+
"epoch": 1.08,
|
| 2157 |
+
"learning_rate": 0.00023688888888888889,
|
| 2158 |
+
"loss": 0.6388,
|
| 2159 |
+
"step": 3370
|
| 2160 |
+
},
|
| 2161 |
+
{
|
| 2162 |
+
"epoch": 1.08,
|
| 2163 |
+
"learning_rate": 0.00023669565217391303,
|
| 2164 |
+
"loss": 0.6626,
|
| 2165 |
+
"step": 3380
|
| 2166 |
+
},
|
| 2167 |
+
{
|
| 2168 |
+
"epoch": 1.08,
|
| 2169 |
+
"learning_rate": 0.00023650241545893717,
|
| 2170 |
+
"loss": 0.6522,
|
| 2171 |
+
"step": 3390
|
| 2172 |
+
},
|
| 2173 |
+
{
|
| 2174 |
+
"epoch": 1.09,
|
| 2175 |
+
"learning_rate": 0.0002363091787439613,
|
| 2176 |
+
"loss": 0.6514,
|
| 2177 |
+
"step": 3400
|
| 2178 |
+
},
|
| 2179 |
+
{
|
| 2180 |
+
"epoch": 1.09,
|
| 2181 |
+
"eval_loss": 0.6873291730880737,
|
| 2182 |
+
"eval_runtime": 392.2057,
|
| 2183 |
+
"eval_samples_per_second": 5.099,
|
| 2184 |
+
"eval_steps_per_second": 0.637,
|
| 2185 |
+
"step": 3400
|
| 2186 |
+
},
|
| 2187 |
+
{
|
| 2188 |
+
"epoch": 1.09,
|
| 2189 |
+
"learning_rate": 0.00023611594202898548,
|
| 2190 |
+
"loss": 0.6147,
|
| 2191 |
+
"step": 3410
|
| 2192 |
+
},
|
| 2193 |
+
{
|
| 2194 |
+
"epoch": 1.09,
|
| 2195 |
+
"learning_rate": 0.00023592270531400965,
|
| 2196 |
+
"loss": 0.6282,
|
| 2197 |
+
"step": 3420
|
| 2198 |
+
},
|
| 2199 |
+
{
|
| 2200 |
+
"epoch": 1.1,
|
| 2201 |
+
"learning_rate": 0.0002357294685990338,
|
| 2202 |
+
"loss": 0.6168,
|
| 2203 |
+
"step": 3430
|
| 2204 |
+
},
|
| 2205 |
+
{
|
| 2206 |
+
"epoch": 1.1,
|
| 2207 |
+
"learning_rate": 0.00023553623188405794,
|
| 2208 |
+
"loss": 0.6858,
|
| 2209 |
+
"step": 3440
|
| 2210 |
+
},
|
| 2211 |
+
{
|
| 2212 |
+
"epoch": 1.1,
|
| 2213 |
+
"learning_rate": 0.00023534299516908208,
|
| 2214 |
+
"loss": 0.6219,
|
| 2215 |
+
"step": 3450
|
| 2216 |
+
},
|
| 2217 |
+
{
|
| 2218 |
+
"epoch": 1.11,
|
| 2219 |
+
"learning_rate": 0.00023514975845410628,
|
| 2220 |
+
"loss": 0.6566,
|
| 2221 |
+
"step": 3460
|
| 2222 |
+
},
|
| 2223 |
+
{
|
| 2224 |
+
"epoch": 1.11,
|
| 2225 |
+
"learning_rate": 0.00023495652173913042,
|
| 2226 |
+
"loss": 0.6633,
|
| 2227 |
+
"step": 3470
|
| 2228 |
+
},
|
| 2229 |
+
{
|
| 2230 |
+
"epoch": 1.11,
|
| 2231 |
+
"learning_rate": 0.00023476328502415456,
|
| 2232 |
+
"loss": 0.6526,
|
| 2233 |
+
"step": 3480
|
| 2234 |
+
},
|
| 2235 |
+
{
|
| 2236 |
+
"epoch": 1.12,
|
| 2237 |
+
"learning_rate": 0.0002345700483091787,
|
| 2238 |
+
"loss": 0.6544,
|
| 2239 |
+
"step": 3490
|
| 2240 |
+
},
|
| 2241 |
+
{
|
| 2242 |
+
"epoch": 1.12,
|
| 2243 |
+
"learning_rate": 0.00023437681159420287,
|
| 2244 |
+
"loss": 0.6588,
|
| 2245 |
+
"step": 3500
|
| 2246 |
+
},
|
| 2247 |
+
{
|
| 2248 |
+
"epoch": 1.12,
|
| 2249 |
+
"learning_rate": 0.00023418357487922704,
|
| 2250 |
+
"loss": 0.6477,
|
| 2251 |
+
"step": 3510
|
| 2252 |
+
},
|
| 2253 |
+
{
|
| 2254 |
+
"epoch": 1.13,
|
| 2255 |
+
"learning_rate": 0.0002339903381642512,
|
| 2256 |
+
"loss": 0.6494,
|
| 2257 |
+
"step": 3520
|
| 2258 |
+
},
|
| 2259 |
+
{
|
| 2260 |
+
"epoch": 1.13,
|
| 2261 |
+
"learning_rate": 0.00023379710144927533,
|
| 2262 |
+
"loss": 0.6219,
|
| 2263 |
+
"step": 3530
|
| 2264 |
+
},
|
| 2265 |
+
{
|
| 2266 |
+
"epoch": 1.13,
|
| 2267 |
+
"learning_rate": 0.00023360386473429947,
|
| 2268 |
+
"loss": 0.6369,
|
| 2269 |
+
"step": 3540
|
| 2270 |
+
},
|
| 2271 |
+
{
|
| 2272 |
+
"epoch": 1.14,
|
| 2273 |
+
"learning_rate": 0.00023341062801932367,
|
| 2274 |
+
"loss": 0.6647,
|
| 2275 |
+
"step": 3550
|
| 2276 |
+
},
|
| 2277 |
+
{
|
| 2278 |
+
"epoch": 1.14,
|
| 2279 |
+
"learning_rate": 0.0002332173913043478,
|
| 2280 |
+
"loss": 0.6477,
|
| 2281 |
+
"step": 3560
|
| 2282 |
+
},
|
| 2283 |
+
{
|
| 2284 |
+
"epoch": 1.14,
|
| 2285 |
+
"learning_rate": 0.00023302415458937195,
|
| 2286 |
+
"loss": 0.6721,
|
| 2287 |
+
"step": 3570
|
| 2288 |
+
},
|
| 2289 |
+
{
|
| 2290 |
+
"epoch": 1.15,
|
| 2291 |
+
"learning_rate": 0.0002328309178743961,
|
| 2292 |
+
"loss": 0.6327,
|
| 2293 |
+
"step": 3580
|
| 2294 |
+
},
|
| 2295 |
+
{
|
| 2296 |
+
"epoch": 1.15,
|
| 2297 |
+
"learning_rate": 0.0002326376811594203,
|
| 2298 |
+
"loss": 0.6668,
|
| 2299 |
+
"step": 3590
|
| 2300 |
+
},
|
| 2301 |
+
{
|
| 2302 |
+
"epoch": 1.15,
|
| 2303 |
+
"learning_rate": 0.00023244444444444444,
|
| 2304 |
+
"loss": 0.6422,
|
| 2305 |
+
"step": 3600
|
| 2306 |
+
},
|
| 2307 |
+
{
|
| 2308 |
+
"epoch": 1.15,
|
| 2309 |
+
"eval_loss": 0.6869779229164124,
|
| 2310 |
+
"eval_runtime": 396.5319,
|
| 2311 |
+
"eval_samples_per_second": 5.044,
|
| 2312 |
+
"eval_steps_per_second": 0.63,
|
| 2313 |
+
"step": 3600
|
| 2314 |
+
},
|
| 2315 |
+
{
|
| 2316 |
+
"epoch": 1.16,
|
| 2317 |
+
"learning_rate": 0.00023225120772946858,
|
| 2318 |
+
"loss": 0.6774,
|
| 2319 |
+
"step": 3610
|
| 2320 |
+
},
|
| 2321 |
+
{
|
| 2322 |
+
"epoch": 1.16,
|
| 2323 |
+
"learning_rate": 0.00023205797101449272,
|
| 2324 |
+
"loss": 0.6663,
|
| 2325 |
+
"step": 3620
|
| 2326 |
+
},
|
| 2327 |
+
{
|
| 2328 |
+
"epoch": 1.16,
|
| 2329 |
+
"learning_rate": 0.00023186473429951686,
|
| 2330 |
+
"loss": 0.6903,
|
| 2331 |
+
"step": 3630
|
| 2332 |
+
},
|
| 2333 |
+
{
|
| 2334 |
+
"epoch": 1.16,
|
| 2335 |
+
"learning_rate": 0.00023167149758454106,
|
| 2336 |
+
"loss": 0.6362,
|
| 2337 |
+
"step": 3640
|
| 2338 |
+
},
|
| 2339 |
+
{
|
| 2340 |
+
"epoch": 1.17,
|
| 2341 |
+
"learning_rate": 0.0002314782608695652,
|
| 2342 |
+
"loss": 0.6187,
|
| 2343 |
+
"step": 3650
|
| 2344 |
+
},
|
| 2345 |
+
{
|
| 2346 |
+
"epoch": 1.17,
|
| 2347 |
+
"learning_rate": 0.00023128502415458935,
|
| 2348 |
+
"loss": 0.6256,
|
| 2349 |
+
"step": 3660
|
| 2350 |
+
},
|
| 2351 |
+
{
|
| 2352 |
+
"epoch": 1.17,
|
| 2353 |
+
"learning_rate": 0.0002310917874396135,
|
| 2354 |
+
"loss": 0.6493,
|
| 2355 |
+
"step": 3670
|
| 2356 |
+
},
|
| 2357 |
+
{
|
| 2358 |
+
"epoch": 1.18,
|
| 2359 |
+
"learning_rate": 0.00023089855072463768,
|
| 2360 |
+
"loss": 0.635,
|
| 2361 |
+
"step": 3680
|
| 2362 |
+
},
|
| 2363 |
+
{
|
| 2364 |
+
"epoch": 1.18,
|
| 2365 |
+
"learning_rate": 0.00023070531400966183,
|
| 2366 |
+
"loss": 0.6973,
|
| 2367 |
+
"step": 3690
|
| 2368 |
+
},
|
| 2369 |
+
{
|
| 2370 |
+
"epoch": 1.18,
|
| 2371 |
+
"learning_rate": 0.00023051207729468597,
|
| 2372 |
+
"loss": 0.6562,
|
| 2373 |
+
"step": 3700
|
| 2374 |
+
},
|
| 2375 |
+
{
|
| 2376 |
+
"epoch": 1.19,
|
| 2377 |
+
"learning_rate": 0.0002303188405797101,
|
| 2378 |
+
"loss": 0.6327,
|
| 2379 |
+
"step": 3710
|
| 2380 |
+
},
|
| 2381 |
+
{
|
| 2382 |
+
"epoch": 1.19,
|
| 2383 |
+
"learning_rate": 0.00023012560386473425,
|
| 2384 |
+
"loss": 0.6736,
|
| 2385 |
+
"step": 3720
|
| 2386 |
+
},
|
| 2387 |
+
{
|
| 2388 |
+
"epoch": 1.19,
|
| 2389 |
+
"learning_rate": 0.00022993236714975845,
|
| 2390 |
+
"loss": 0.6178,
|
| 2391 |
+
"step": 3730
|
| 2392 |
+
},
|
| 2393 |
+
{
|
| 2394 |
+
"epoch": 1.2,
|
| 2395 |
+
"learning_rate": 0.0002297391304347826,
|
| 2396 |
+
"loss": 0.6574,
|
| 2397 |
+
"step": 3740
|
| 2398 |
+
},
|
| 2399 |
+
{
|
| 2400 |
+
"epoch": 1.2,
|
| 2401 |
+
"learning_rate": 0.00022954589371980674,
|
| 2402 |
+
"loss": 0.6848,
|
| 2403 |
+
"step": 3750
|
| 2404 |
+
},
|
| 2405 |
+
{
|
| 2406 |
+
"epoch": 1.2,
|
| 2407 |
+
"learning_rate": 0.00022935265700483088,
|
| 2408 |
+
"loss": 0.6478,
|
| 2409 |
+
"step": 3760
|
| 2410 |
+
},
|
| 2411 |
+
{
|
| 2412 |
+
"epoch": 1.21,
|
| 2413 |
+
"learning_rate": 0.00022915942028985508,
|
| 2414 |
+
"loss": 0.6144,
|
| 2415 |
+
"step": 3770
|
| 2416 |
+
},
|
| 2417 |
+
{
|
| 2418 |
+
"epoch": 1.21,
|
| 2419 |
+
"learning_rate": 0.00022896618357487922,
|
| 2420 |
+
"loss": 0.6609,
|
| 2421 |
+
"step": 3780
|
| 2422 |
+
},
|
| 2423 |
+
{
|
| 2424 |
+
"epoch": 1.21,
|
| 2425 |
+
"learning_rate": 0.00022877294685990336,
|
| 2426 |
+
"loss": 0.6744,
|
| 2427 |
+
"step": 3790
|
| 2428 |
+
},
|
| 2429 |
+
{
|
| 2430 |
+
"epoch": 1.22,
|
| 2431 |
+
"learning_rate": 0.0002285797101449275,
|
| 2432 |
+
"loss": 0.636,
|
| 2433 |
+
"step": 3800
|
| 2434 |
+
},
|
| 2435 |
+
{
|
| 2436 |
+
"epoch": 1.22,
|
| 2437 |
+
"eval_loss": 0.6865532398223877,
|
| 2438 |
+
"eval_runtime": 393.5432,
|
| 2439 |
+
"eval_samples_per_second": 5.082,
|
| 2440 |
+
"eval_steps_per_second": 0.635,
|
| 2441 |
+
"step": 3800
|
| 2442 |
+
},
|
| 2443 |
+
{
|
| 2444 |
+
"epoch": 1.22,
|
| 2445 |
+
"learning_rate": 0.00022838647342995165,
|
| 2446 |
+
"loss": 0.654,
|
| 2447 |
+
"step": 3810
|
| 2448 |
+
},
|
| 2449 |
+
{
|
| 2450 |
+
"epoch": 1.22,
|
| 2451 |
+
"learning_rate": 0.00022819323671497584,
|
| 2452 |
+
"loss": 0.6442,
|
| 2453 |
+
"step": 3820
|
| 2454 |
+
},
|
| 2455 |
+
{
|
| 2456 |
+
"epoch": 1.23,
|
| 2457 |
+
"learning_rate": 0.00022799999999999999,
|
| 2458 |
+
"loss": 0.6835,
|
| 2459 |
+
"step": 3830
|
| 2460 |
+
},
|
| 2461 |
+
{
|
| 2462 |
+
"epoch": 1.23,
|
| 2463 |
+
"learning_rate": 0.00022780676328502413,
|
| 2464 |
+
"loss": 0.6339,
|
| 2465 |
+
"step": 3840
|
| 2466 |
+
},
|
| 2467 |
+
{
|
| 2468 |
+
"epoch": 1.23,
|
| 2469 |
+
"learning_rate": 0.00022761352657004827,
|
| 2470 |
+
"loss": 0.6512,
|
| 2471 |
+
"step": 3850
|
| 2472 |
+
},
|
| 2473 |
+
{
|
| 2474 |
+
"epoch": 1.24,
|
| 2475 |
+
"learning_rate": 0.00022742028985507247,
|
| 2476 |
+
"loss": 0.622,
|
| 2477 |
+
"step": 3860
|
| 2478 |
+
},
|
| 2479 |
+
{
|
| 2480 |
+
"epoch": 1.24,
|
| 2481 |
+
"learning_rate": 0.0002272270531400966,
|
| 2482 |
+
"loss": 0.6802,
|
| 2483 |
+
"step": 3870
|
| 2484 |
+
},
|
| 2485 |
+
{
|
| 2486 |
+
"epoch": 1.24,
|
| 2487 |
+
"learning_rate": 0.00022703381642512075,
|
| 2488 |
+
"loss": 0.6911,
|
| 2489 |
+
"step": 3880
|
| 2490 |
+
},
|
| 2491 |
+
{
|
| 2492 |
+
"epoch": 1.24,
|
| 2493 |
+
"learning_rate": 0.0002268405797101449,
|
| 2494 |
+
"loss": 0.6815,
|
| 2495 |
+
"step": 3890
|
| 2496 |
+
},
|
| 2497 |
+
{
|
| 2498 |
+
"epoch": 1.25,
|
| 2499 |
+
"learning_rate": 0.0002266473429951691,
|
| 2500 |
+
"loss": 0.6569,
|
| 2501 |
+
"step": 3900
|
| 2502 |
+
},
|
| 2503 |
+
{
|
| 2504 |
+
"epoch": 1.25,
|
| 2505 |
+
"learning_rate": 0.00022645410628019323,
|
| 2506 |
+
"loss": 0.6363,
|
| 2507 |
+
"step": 3910
|
| 2508 |
+
},
|
| 2509 |
+
{
|
| 2510 |
+
"epoch": 1.25,
|
| 2511 |
+
"learning_rate": 0.00022626086956521738,
|
| 2512 |
+
"loss": 0.6371,
|
| 2513 |
+
"step": 3920
|
| 2514 |
+
},
|
| 2515 |
+
{
|
| 2516 |
+
"epoch": 1.26,
|
| 2517 |
+
"learning_rate": 0.00022606763285024152,
|
| 2518 |
+
"loss": 0.6458,
|
| 2519 |
+
"step": 3930
|
| 2520 |
+
},
|
| 2521 |
+
{
|
| 2522 |
+
"epoch": 1.26,
|
| 2523 |
+
"learning_rate": 0.00022587439613526566,
|
| 2524 |
+
"loss": 0.668,
|
| 2525 |
+
"step": 3940
|
| 2526 |
+
},
|
| 2527 |
+
{
|
| 2528 |
+
"epoch": 1.26,
|
| 2529 |
+
"learning_rate": 0.00022568115942028986,
|
| 2530 |
+
"loss": 0.6676,
|
| 2531 |
+
"step": 3950
|
| 2532 |
+
},
|
| 2533 |
+
{
|
| 2534 |
+
"epoch": 1.27,
|
| 2535 |
+
"learning_rate": 0.000225487922705314,
|
| 2536 |
+
"loss": 0.6692,
|
| 2537 |
+
"step": 3960
|
| 2538 |
+
},
|
| 2539 |
+
{
|
| 2540 |
+
"epoch": 1.27,
|
| 2541 |
+
"learning_rate": 0.00022529468599033814,
|
| 2542 |
+
"loss": 0.6502,
|
| 2543 |
+
"step": 3970
|
| 2544 |
+
},
|
| 2545 |
+
{
|
| 2546 |
+
"epoch": 1.27,
|
| 2547 |
+
"learning_rate": 0.0002251014492753623,
|
| 2548 |
+
"loss": 0.6387,
|
| 2549 |
+
"step": 3980
|
| 2550 |
+
},
|
| 2551 |
+
{
|
| 2552 |
+
"epoch": 1.28,
|
| 2553 |
+
"learning_rate": 0.00022490821256038646,
|
| 2554 |
+
"loss": 0.6475,
|
| 2555 |
+
"step": 3990
|
| 2556 |
+
},
|
| 2557 |
+
{
|
| 2558 |
+
"epoch": 1.28,
|
| 2559 |
+
"learning_rate": 0.00022471497584541063,
|
| 2560 |
+
"loss": 0.6627,
|
| 2561 |
+
"step": 4000
|
| 2562 |
+
},
|
| 2563 |
+
{
|
| 2564 |
+
"epoch": 1.28,
|
| 2565 |
+
"eval_loss": 0.6856961846351624,
|
| 2566 |
+
"eval_runtime": 387.1458,
|
| 2567 |
+
"eval_samples_per_second": 5.166,
|
| 2568 |
+
"eval_steps_per_second": 0.646,
|
| 2569 |
+
"step": 4000
|
| 2570 |
+
},
|
| 2571 |
+
{
|
| 2572 |
+
"epoch": 1.28,
|
| 2573 |
+
"learning_rate": 0.00022452173913043477,
|
| 2574 |
+
"loss": 0.6346,
|
| 2575 |
+
"step": 4010
|
| 2576 |
+
},
|
| 2577 |
+
{
|
| 2578 |
+
"epoch": 1.29,
|
| 2579 |
+
"learning_rate": 0.0002243285024154589,
|
| 2580 |
+
"loss": 0.6593,
|
| 2581 |
+
"step": 4020
|
| 2582 |
+
},
|
| 2583 |
+
{
|
| 2584 |
+
"epoch": 1.29,
|
| 2585 |
+
"learning_rate": 0.00022413526570048305,
|
| 2586 |
+
"loss": 0.6397,
|
| 2587 |
+
"step": 4030
|
| 2588 |
+
},
|
| 2589 |
+
{
|
| 2590 |
+
"epoch": 1.29,
|
| 2591 |
+
"learning_rate": 0.00022394202898550725,
|
| 2592 |
+
"loss": 0.641,
|
| 2593 |
+
"step": 4040
|
| 2594 |
+
},
|
| 2595 |
+
{
|
| 2596 |
+
"epoch": 1.3,
|
| 2597 |
+
"learning_rate": 0.0002237487922705314,
|
| 2598 |
+
"loss": 0.6766,
|
| 2599 |
+
"step": 4050
|
| 2600 |
+
},
|
| 2601 |
+
{
|
| 2602 |
+
"epoch": 1.3,
|
| 2603 |
+
"learning_rate": 0.00022355555555555554,
|
| 2604 |
+
"loss": 0.6931,
|
| 2605 |
+
"step": 4060
|
| 2606 |
+
},
|
| 2607 |
+
{
|
| 2608 |
+
"epoch": 1.3,
|
| 2609 |
+
"learning_rate": 0.00022336231884057968,
|
| 2610 |
+
"loss": 0.6618,
|
| 2611 |
+
"step": 4070
|
| 2612 |
+
},
|
| 2613 |
+
{
|
| 2614 |
+
"epoch": 1.31,
|
| 2615 |
+
"learning_rate": 0.00022316908212560385,
|
| 2616 |
+
"loss": 0.6655,
|
| 2617 |
+
"step": 4080
|
| 2618 |
+
},
|
| 2619 |
+
{
|
| 2620 |
+
"epoch": 1.31,
|
| 2621 |
+
"learning_rate": 0.00022297584541062802,
|
| 2622 |
+
"loss": 0.647,
|
| 2623 |
+
"step": 4090
|
| 2624 |
+
},
|
| 2625 |
+
{
|
| 2626 |
+
"epoch": 1.31,
|
| 2627 |
+
"learning_rate": 0.00022278260869565216,
|
| 2628 |
+
"loss": 0.6337,
|
| 2629 |
+
"step": 4100
|
| 2630 |
+
},
|
| 2631 |
+
{
|
| 2632 |
+
"epoch": 1.32,
|
| 2633 |
+
"learning_rate": 0.0002225893719806763,
|
| 2634 |
+
"loss": 0.6957,
|
| 2635 |
+
"step": 4110
|
| 2636 |
+
},
|
| 2637 |
+
{
|
| 2638 |
+
"epoch": 1.32,
|
| 2639 |
+
"learning_rate": 0.00022239613526570044,
|
| 2640 |
+
"loss": 0.6435,
|
| 2641 |
+
"step": 4120
|
| 2642 |
+
},
|
| 2643 |
+
{
|
| 2644 |
+
"epoch": 1.32,
|
| 2645 |
+
"learning_rate": 0.00022220289855072464,
|
| 2646 |
+
"loss": 0.6272,
|
| 2647 |
+
"step": 4130
|
| 2648 |
+
},
|
| 2649 |
+
{
|
| 2650 |
+
"epoch": 1.32,
|
| 2651 |
+
"learning_rate": 0.00022200966183574878,
|
| 2652 |
+
"loss": 0.6502,
|
| 2653 |
+
"step": 4140
|
| 2654 |
+
},
|
| 2655 |
+
{
|
| 2656 |
+
"epoch": 1.33,
|
| 2657 |
+
"learning_rate": 0.00022181642512077293,
|
| 2658 |
+
"loss": 0.6514,
|
| 2659 |
+
"step": 4150
|
| 2660 |
+
},
|
| 2661 |
+
{
|
| 2662 |
+
"epoch": 1.33,
|
| 2663 |
+
"learning_rate": 0.00022162318840579707,
|
| 2664 |
+
"loss": 0.6625,
|
| 2665 |
+
"step": 4160
|
| 2666 |
+
},
|
| 2667 |
+
{
|
| 2668 |
+
"epoch": 1.33,
|
| 2669 |
+
"learning_rate": 0.00022142995169082124,
|
| 2670 |
+
"loss": 0.6284,
|
| 2671 |
+
"step": 4170
|
| 2672 |
+
},
|
| 2673 |
+
{
|
| 2674 |
+
"epoch": 1.34,
|
| 2675 |
+
"learning_rate": 0.0002212367149758454,
|
| 2676 |
+
"loss": 0.6503,
|
| 2677 |
+
"step": 4180
|
| 2678 |
+
},
|
| 2679 |
+
{
|
| 2680 |
+
"epoch": 1.34,
|
| 2681 |
+
"learning_rate": 0.00022104347826086955,
|
| 2682 |
+
"loss": 0.6425,
|
| 2683 |
+
"step": 4190
|
| 2684 |
+
},
|
| 2685 |
+
{
|
| 2686 |
+
"epoch": 1.34,
|
| 2687 |
+
"learning_rate": 0.0002208502415458937,
|
| 2688 |
+
"loss": 0.6818,
|
| 2689 |
+
"step": 4200
|
| 2690 |
+
},
|
| 2691 |
+
{
|
| 2692 |
+
"epoch": 1.34,
|
| 2693 |
+
"eval_loss": 0.6859603524208069,
|
| 2694 |
+
"eval_runtime": 382.589,
|
| 2695 |
+
"eval_samples_per_second": 5.228,
|
| 2696 |
+
"eval_steps_per_second": 0.653,
|
| 2697 |
+
"step": 4200
|
| 2698 |
+
},
|
| 2699 |
+
{
|
| 2700 |
+
"epoch": 1.35,
|
| 2701 |
+
"learning_rate": 0.00022065700483091784,
|
| 2702 |
+
"loss": 0.6299,
|
| 2703 |
+
"step": 4210
|
| 2704 |
+
},
|
| 2705 |
+
{
|
| 2706 |
+
"epoch": 1.35,
|
| 2707 |
+
"learning_rate": 0.000220463768115942,
|
| 2708 |
+
"loss": 0.6508,
|
| 2709 |
+
"step": 4220
|
| 2710 |
+
},
|
| 2711 |
+
{
|
| 2712 |
+
"epoch": 1.35,
|
| 2713 |
+
"learning_rate": 0.00022027053140096618,
|
| 2714 |
+
"loss": 0.6572,
|
| 2715 |
+
"step": 4230
|
| 2716 |
+
},
|
| 2717 |
+
{
|
| 2718 |
+
"epoch": 1.36,
|
| 2719 |
+
"learning_rate": 0.00022007729468599032,
|
| 2720 |
+
"loss": 0.6781,
|
| 2721 |
+
"step": 4240
|
| 2722 |
+
},
|
| 2723 |
+
{
|
| 2724 |
+
"epoch": 1.36,
|
| 2725 |
+
"learning_rate": 0.00021988405797101446,
|
| 2726 |
+
"loss": 0.6806,
|
| 2727 |
+
"step": 4250
|
| 2728 |
+
},
|
| 2729 |
+
{
|
| 2730 |
+
"epoch": 1.36,
|
| 2731 |
+
"learning_rate": 0.00021969082125603863,
|
| 2732 |
+
"loss": 0.632,
|
| 2733 |
+
"step": 4260
|
| 2734 |
+
},
|
| 2735 |
+
{
|
| 2736 |
+
"epoch": 1.37,
|
| 2737 |
+
"learning_rate": 0.0002194975845410628,
|
| 2738 |
+
"loss": 0.6631,
|
| 2739 |
+
"step": 4270
|
| 2740 |
+
},
|
| 2741 |
+
{
|
| 2742 |
+
"epoch": 1.37,
|
| 2743 |
+
"learning_rate": 0.00021930434782608694,
|
| 2744 |
+
"loss": 0.6729,
|
| 2745 |
+
"step": 4280
|
| 2746 |
+
},
|
| 2747 |
+
{
|
| 2748 |
+
"epoch": 1.37,
|
| 2749 |
+
"learning_rate": 0.00021911111111111109,
|
| 2750 |
+
"loss": 0.6618,
|
| 2751 |
+
"step": 4290
|
| 2752 |
+
},
|
| 2753 |
+
{
|
| 2754 |
+
"epoch": 1.38,
|
| 2755 |
+
"learning_rate": 0.00021891787439613525,
|
| 2756 |
+
"loss": 0.6132,
|
| 2757 |
+
"step": 4300
|
| 2758 |
+
},
|
| 2759 |
+
{
|
| 2760 |
+
"epoch": 1.38,
|
| 2761 |
+
"learning_rate": 0.0002187246376811594,
|
| 2762 |
+
"loss": 0.6457,
|
| 2763 |
+
"step": 4310
|
| 2764 |
+
},
|
| 2765 |
+
{
|
| 2766 |
+
"epoch": 1.38,
|
| 2767 |
+
"learning_rate": 0.00021853140096618357,
|
| 2768 |
+
"loss": 0.6443,
|
| 2769 |
+
"step": 4320
|
| 2770 |
+
},
|
| 2771 |
+
{
|
| 2772 |
+
"epoch": 1.39,
|
| 2773 |
+
"learning_rate": 0.0002183381642512077,
|
| 2774 |
+
"loss": 0.653,
|
| 2775 |
+
"step": 4330
|
| 2776 |
+
},
|
| 2777 |
+
{
|
| 2778 |
+
"epoch": 1.39,
|
| 2779 |
+
"learning_rate": 0.00021814492753623185,
|
| 2780 |
+
"loss": 0.6706,
|
| 2781 |
+
"step": 4340
|
| 2782 |
+
},
|
| 2783 |
+
{
|
| 2784 |
+
"epoch": 1.39,
|
| 2785 |
+
"learning_rate": 0.00021795169082125602,
|
| 2786 |
+
"loss": 0.6684,
|
| 2787 |
+
"step": 4350
|
| 2788 |
+
},
|
| 2789 |
+
{
|
| 2790 |
+
"epoch": 1.4,
|
| 2791 |
+
"learning_rate": 0.00021775845410628016,
|
| 2792 |
+
"loss": 0.6597,
|
| 2793 |
+
"step": 4360
|
| 2794 |
+
},
|
| 2795 |
+
{
|
| 2796 |
+
"epoch": 1.4,
|
| 2797 |
+
"learning_rate": 0.00021756521739130433,
|
| 2798 |
+
"loss": 0.6478,
|
| 2799 |
+
"step": 4370
|
| 2800 |
+
},
|
| 2801 |
+
{
|
| 2802 |
+
"epoch": 1.4,
|
| 2803 |
+
"learning_rate": 0.00021737198067632848,
|
| 2804 |
+
"loss": 0.6411,
|
| 2805 |
+
"step": 4380
|
| 2806 |
+
},
|
| 2807 |
+
{
|
| 2808 |
+
"epoch": 1.4,
|
| 2809 |
+
"learning_rate": 0.00021717874396135265,
|
| 2810 |
+
"loss": 0.657,
|
| 2811 |
+
"step": 4390
|
| 2812 |
+
},
|
| 2813 |
+
{
|
| 2814 |
+
"epoch": 1.41,
|
| 2815 |
+
"learning_rate": 0.0002169855072463768,
|
| 2816 |
+
"loss": 0.663,
|
| 2817 |
+
"step": 4400
|
| 2818 |
+
},
|
| 2819 |
+
{
|
| 2820 |
+
"epoch": 1.41,
|
| 2821 |
+
"eval_loss": 0.6874400973320007,
|
| 2822 |
+
"eval_runtime": 387.4088,
|
| 2823 |
+
"eval_samples_per_second": 5.163,
|
| 2824 |
+
"eval_steps_per_second": 0.645,
|
| 2825 |
+
"step": 4400
|
| 2826 |
+
},
|
| 2827 |
+
{
|
| 2828 |
+
"epoch": 1.41,
|
| 2829 |
+
"learning_rate": 0.00021679227053140096,
|
| 2830 |
+
"loss": 0.6833,
|
| 2831 |
+
"step": 4410
|
| 2832 |
+
},
|
| 2833 |
+
{
|
| 2834 |
+
"epoch": 1.41,
|
| 2835 |
+
"learning_rate": 0.0002165990338164251,
|
| 2836 |
+
"loss": 0.6827,
|
| 2837 |
+
"step": 4420
|
| 2838 |
+
},
|
| 2839 |
+
{
|
| 2840 |
+
"epoch": 1.42,
|
| 2841 |
+
"learning_rate": 0.00021640579710144924,
|
| 2842 |
+
"loss": 0.6789,
|
| 2843 |
+
"step": 4430
|
| 2844 |
+
},
|
| 2845 |
+
{
|
| 2846 |
+
"epoch": 1.42,
|
| 2847 |
+
"learning_rate": 0.0002162125603864734,
|
| 2848 |
+
"loss": 0.6582,
|
| 2849 |
+
"step": 4440
|
| 2850 |
+
},
|
| 2851 |
+
{
|
| 2852 |
+
"epoch": 1.42,
|
| 2853 |
+
"learning_rate": 0.00021601932367149756,
|
| 2854 |
+
"loss": 0.6222,
|
| 2855 |
+
"step": 4450
|
| 2856 |
+
},
|
| 2857 |
+
{
|
| 2858 |
+
"epoch": 1.43,
|
| 2859 |
+
"learning_rate": 0.00021582608695652173,
|
| 2860 |
+
"loss": 0.6314,
|
| 2861 |
+
"step": 4460
|
| 2862 |
+
},
|
| 2863 |
+
{
|
| 2864 |
+
"epoch": 1.43,
|
| 2865 |
+
"learning_rate": 0.00021563285024154587,
|
| 2866 |
+
"loss": 0.6466,
|
| 2867 |
+
"step": 4470
|
| 2868 |
+
},
|
| 2869 |
+
{
|
| 2870 |
+
"epoch": 1.43,
|
| 2871 |
+
"learning_rate": 0.00021543961352657004,
|
| 2872 |
+
"loss": 0.6734,
|
| 2873 |
+
"step": 4480
|
| 2874 |
+
},
|
| 2875 |
+
{
|
| 2876 |
+
"epoch": 1.44,
|
| 2877 |
+
"learning_rate": 0.00021524637681159418,
|
| 2878 |
+
"loss": 0.6552,
|
| 2879 |
+
"step": 4490
|
| 2880 |
+
},
|
| 2881 |
+
{
|
| 2882 |
+
"epoch": 1.44,
|
| 2883 |
+
"learning_rate": 0.00021505314009661835,
|
| 2884 |
+
"loss": 0.7156,
|
| 2885 |
+
"step": 4500
|
| 2886 |
+
},
|
| 2887 |
+
{
|
| 2888 |
+
"epoch": 1.44,
|
| 2889 |
+
"learning_rate": 0.0002148599033816425,
|
| 2890 |
+
"loss": 0.6548,
|
| 2891 |
+
"step": 4510
|
| 2892 |
+
},
|
| 2893 |
+
{
|
| 2894 |
+
"epoch": 1.45,
|
| 2895 |
+
"learning_rate": 0.00021466666666666664,
|
| 2896 |
+
"loss": 0.7265,
|
| 2897 |
+
"step": 4520
|
| 2898 |
+
},
|
| 2899 |
+
{
|
| 2900 |
+
"epoch": 1.45,
|
| 2901 |
+
"learning_rate": 0.0002144734299516908,
|
| 2902 |
+
"loss": 0.6757,
|
| 2903 |
+
"step": 4530
|
| 2904 |
+
},
|
| 2905 |
+
{
|
| 2906 |
+
"epoch": 1.45,
|
| 2907 |
+
"learning_rate": 0.00021428019323671495,
|
| 2908 |
+
"loss": 0.6914,
|
| 2909 |
+
"step": 4540
|
| 2910 |
+
},
|
| 2911 |
+
{
|
| 2912 |
+
"epoch": 1.46,
|
| 2913 |
+
"learning_rate": 0.00021408695652173912,
|
| 2914 |
+
"loss": 0.6746,
|
| 2915 |
+
"step": 4550
|
| 2916 |
+
},
|
| 2917 |
+
{
|
| 2918 |
+
"epoch": 1.46,
|
| 2919 |
+
"learning_rate": 0.00021389371980676326,
|
| 2920 |
+
"loss": 0.7085,
|
| 2921 |
+
"step": 4560
|
| 2922 |
+
},
|
| 2923 |
+
{
|
| 2924 |
+
"epoch": 1.46,
|
| 2925 |
+
"learning_rate": 0.00021370048309178743,
|
| 2926 |
+
"loss": 0.7058,
|
| 2927 |
+
"step": 4570
|
| 2928 |
+
},
|
| 2929 |
+
{
|
| 2930 |
+
"epoch": 1.47,
|
| 2931 |
+
"learning_rate": 0.00021350724637681157,
|
| 2932 |
+
"loss": 0.6599,
|
| 2933 |
+
"step": 4580
|
| 2934 |
+
},
|
| 2935 |
+
{
|
| 2936 |
+
"epoch": 1.47,
|
| 2937 |
+
"learning_rate": 0.00021331400966183571,
|
| 2938 |
+
"loss": 0.6653,
|
| 2939 |
+
"step": 4590
|
| 2940 |
+
},
|
| 2941 |
+
{
|
| 2942 |
+
"epoch": 1.47,
|
| 2943 |
+
"learning_rate": 0.00021312077294685988,
|
| 2944 |
+
"loss": 0.6757,
|
| 2945 |
+
"step": 4600
|
| 2946 |
+
},
|
| 2947 |
+
{
|
| 2948 |
+
"epoch": 1.47,
|
| 2949 |
+
"eval_loss": 0.6850671172142029,
|
| 2950 |
+
"eval_runtime": 410.1023,
|
| 2951 |
+
"eval_samples_per_second": 4.877,
|
| 2952 |
+
"eval_steps_per_second": 0.61,
|
| 2953 |
+
"step": 4600
|
| 2954 |
+
},
|
| 2955 |
+
{
|
| 2956 |
+
"epoch": 1.48,
|
| 2957 |
+
"learning_rate": 0.00021292753623188405,
|
| 2958 |
+
"loss": 0.6644,
|
| 2959 |
+
"step": 4610
|
| 2960 |
+
},
|
| 2961 |
+
{
|
| 2962 |
+
"epoch": 1.48,
|
| 2963 |
+
"learning_rate": 0.0002127342995169082,
|
| 2964 |
+
"loss": 0.6444,
|
| 2965 |
+
"step": 4620
|
| 2966 |
+
},
|
| 2967 |
+
{
|
| 2968 |
+
"epoch": 1.48,
|
| 2969 |
+
"learning_rate": 0.00021254106280193234,
|
| 2970 |
+
"loss": 0.6548,
|
| 2971 |
+
"step": 4630
|
| 2972 |
+
},
|
| 2973 |
+
{
|
| 2974 |
+
"epoch": 1.48,
|
| 2975 |
+
"learning_rate": 0.0002123478260869565,
|
| 2976 |
+
"loss": 0.6361,
|
| 2977 |
+
"step": 4640
|
| 2978 |
+
},
|
| 2979 |
+
{
|
| 2980 |
+
"epoch": 1.49,
|
| 2981 |
+
"learning_rate": 0.00021215458937198065,
|
| 2982 |
+
"loss": 0.6726,
|
| 2983 |
+
"step": 4650
|
| 2984 |
+
},
|
| 2985 |
+
{
|
| 2986 |
+
"epoch": 1.49,
|
| 2987 |
+
"learning_rate": 0.00021196135265700482,
|
| 2988 |
+
"loss": 0.6451,
|
| 2989 |
+
"step": 4660
|
| 2990 |
+
},
|
| 2991 |
+
{
|
| 2992 |
+
"epoch": 1.49,
|
| 2993 |
+
"learning_rate": 0.00021176811594202896,
|
| 2994 |
+
"loss": 0.6939,
|
| 2995 |
+
"step": 4670
|
| 2996 |
+
},
|
| 2997 |
+
{
|
| 2998 |
+
"epoch": 1.5,
|
| 2999 |
+
"learning_rate": 0.0002115748792270531,
|
| 3000 |
+
"loss": 0.6569,
|
| 3001 |
+
"step": 4680
|
| 3002 |
+
},
|
| 3003 |
+
{
|
| 3004 |
+
"epoch": 1.5,
|
| 3005 |
+
"learning_rate": 0.00021138164251207728,
|
| 3006 |
+
"loss": 0.6405,
|
| 3007 |
+
"step": 4690
|
| 3008 |
+
},
|
| 3009 |
+
{
|
| 3010 |
+
"epoch": 1.5,
|
| 3011 |
+
"learning_rate": 0.00021118840579710145,
|
| 3012 |
+
"loss": 0.6261,
|
| 3013 |
+
"step": 4700
|
| 3014 |
+
},
|
| 3015 |
+
{
|
| 3016 |
+
"epoch": 1.51,
|
| 3017 |
+
"learning_rate": 0.0002109951690821256,
|
| 3018 |
+
"loss": 0.6544,
|
| 3019 |
+
"step": 4710
|
| 3020 |
+
},
|
| 3021 |
+
{
|
| 3022 |
+
"epoch": 1.51,
|
| 3023 |
+
"learning_rate": 0.00021080193236714973,
|
| 3024 |
+
"loss": 0.6367,
|
| 3025 |
+
"step": 4720
|
| 3026 |
+
},
|
| 3027 |
+
{
|
| 3028 |
+
"epoch": 1.51,
|
| 3029 |
+
"learning_rate": 0.0002106086956521739,
|
| 3030 |
+
"loss": 0.6691,
|
| 3031 |
+
"step": 4730
|
| 3032 |
+
},
|
| 3033 |
+
{
|
| 3034 |
+
"epoch": 1.52,
|
| 3035 |
+
"learning_rate": 0.00021041545893719804,
|
| 3036 |
+
"loss": 0.6228,
|
| 3037 |
+
"step": 4740
|
| 3038 |
+
},
|
| 3039 |
+
{
|
| 3040 |
+
"epoch": 1.52,
|
| 3041 |
+
"learning_rate": 0.0002102222222222222,
|
| 3042 |
+
"loss": 0.6628,
|
| 3043 |
+
"step": 4750
|
| 3044 |
+
},
|
| 3045 |
+
{
|
| 3046 |
+
"epoch": 1.52,
|
| 3047 |
+
"learning_rate": 0.00021002898550724635,
|
| 3048 |
+
"loss": 0.6678,
|
| 3049 |
+
"step": 4760
|
| 3050 |
+
},
|
| 3051 |
+
{
|
| 3052 |
+
"epoch": 1.53,
|
| 3053 |
+
"learning_rate": 0.0002098357487922705,
|
| 3054 |
+
"loss": 0.6699,
|
| 3055 |
+
"step": 4770
|
| 3056 |
+
},
|
| 3057 |
+
{
|
| 3058 |
+
"epoch": 1.53,
|
| 3059 |
+
"learning_rate": 0.00020964251207729467,
|
| 3060 |
+
"loss": 0.683,
|
| 3061 |
+
"step": 4780
|
| 3062 |
+
},
|
| 3063 |
+
{
|
| 3064 |
+
"epoch": 1.53,
|
| 3065 |
+
"learning_rate": 0.00020944927536231884,
|
| 3066 |
+
"loss": 0.688,
|
| 3067 |
+
"step": 4790
|
| 3068 |
+
},
|
| 3069 |
+
{
|
| 3070 |
+
"epoch": 1.54,
|
| 3071 |
+
"learning_rate": 0.00020925603864734298,
|
| 3072 |
+
"loss": 0.6661,
|
| 3073 |
+
"step": 4800
|
| 3074 |
+
},
|
| 3075 |
+
{
|
| 3076 |
+
"epoch": 1.54,
|
| 3077 |
+
"eval_loss": 0.6855675578117371,
|
| 3078 |
+
"eval_runtime": 385.7747,
|
| 3079 |
+
"eval_samples_per_second": 5.184,
|
| 3080 |
+
"eval_steps_per_second": 0.648,
|
| 3081 |
+
"step": 4800
|
| 3082 |
+
},
|
| 3083 |
+
{
|
| 3084 |
+
"epoch": 1.54,
|
| 3085 |
+
"learning_rate": 0.00020906280193236712,
|
| 3086 |
+
"loss": 0.6605,
|
| 3087 |
+
"step": 4810
|
| 3088 |
+
},
|
| 3089 |
+
{
|
| 3090 |
+
"epoch": 1.54,
|
| 3091 |
+
"learning_rate": 0.00020886956521739126,
|
| 3092 |
+
"loss": 0.7119,
|
| 3093 |
+
"step": 4820
|
| 3094 |
+
},
|
| 3095 |
+
{
|
| 3096 |
+
"epoch": 1.55,
|
| 3097 |
+
"learning_rate": 0.00020867632850241543,
|
| 3098 |
+
"loss": 0.6181,
|
| 3099 |
+
"step": 4830
|
| 3100 |
+
},
|
| 3101 |
+
{
|
| 3102 |
+
"epoch": 1.55,
|
| 3103 |
+
"learning_rate": 0.0002084830917874396,
|
| 3104 |
+
"loss": 0.6484,
|
| 3105 |
+
"step": 4840
|
| 3106 |
+
},
|
| 3107 |
+
{
|
| 3108 |
+
"epoch": 1.55,
|
| 3109 |
+
"learning_rate": 0.00020828985507246375,
|
| 3110 |
+
"loss": 0.6747,
|
| 3111 |
+
"step": 4850
|
| 3112 |
+
},
|
| 3113 |
+
{
|
| 3114 |
+
"epoch": 1.56,
|
| 3115 |
+
"learning_rate": 0.0002080966183574879,
|
| 3116 |
+
"loss": 0.6455,
|
| 3117 |
+
"step": 4860
|
| 3118 |
+
},
|
| 3119 |
+
{
|
| 3120 |
+
"epoch": 1.56,
|
| 3121 |
+
"learning_rate": 0.00020790338164251206,
|
| 3122 |
+
"loss": 0.6591,
|
| 3123 |
+
"step": 4870
|
| 3124 |
+
},
|
| 3125 |
+
{
|
| 3126 |
+
"epoch": 1.56,
|
| 3127 |
+
"learning_rate": 0.00020771014492753623,
|
| 3128 |
+
"loss": 0.6898,
|
| 3129 |
+
"step": 4880
|
| 3130 |
+
},
|
| 3131 |
+
{
|
| 3132 |
+
"epoch": 1.56,
|
| 3133 |
+
"learning_rate": 0.00020751690821256037,
|
| 3134 |
+
"loss": 0.6491,
|
| 3135 |
+
"step": 4890
|
| 3136 |
+
},
|
| 3137 |
+
{
|
| 3138 |
+
"epoch": 1.57,
|
| 3139 |
+
"learning_rate": 0.0002073236714975845,
|
| 3140 |
+
"loss": 0.66,
|
| 3141 |
+
"step": 4900
|
| 3142 |
+
},
|
| 3143 |
+
{
|
| 3144 |
+
"epoch": 1.57,
|
| 3145 |
+
"learning_rate": 0.00020713043478260866,
|
| 3146 |
+
"loss": 0.6423,
|
| 3147 |
+
"step": 4910
|
| 3148 |
+
},
|
| 3149 |
+
{
|
| 3150 |
+
"epoch": 1.57,
|
| 3151 |
+
"learning_rate": 0.00020693719806763285,
|
| 3152 |
+
"loss": 0.6102,
|
| 3153 |
+
"step": 4920
|
| 3154 |
+
},
|
| 3155 |
+
{
|
| 3156 |
+
"epoch": 1.58,
|
| 3157 |
+
"learning_rate": 0.000206743961352657,
|
| 3158 |
+
"loss": 0.6709,
|
| 3159 |
+
"step": 4930
|
| 3160 |
+
},
|
| 3161 |
+
{
|
| 3162 |
+
"epoch": 1.58,
|
| 3163 |
+
"learning_rate": 0.00020655072463768114,
|
| 3164 |
+
"loss": 0.6432,
|
| 3165 |
+
"step": 4940
|
| 3166 |
+
},
|
| 3167 |
+
{
|
| 3168 |
+
"epoch": 1.58,
|
| 3169 |
+
"learning_rate": 0.00020635748792270528,
|
| 3170 |
+
"loss": 0.6388,
|
| 3171 |
+
"step": 4950
|
| 3172 |
+
},
|
| 3173 |
+
{
|
| 3174 |
+
"epoch": 1.59,
|
| 3175 |
+
"learning_rate": 0.00020616425120772942,
|
| 3176 |
+
"loss": 0.6574,
|
| 3177 |
+
"step": 4960
|
| 3178 |
+
},
|
| 3179 |
+
{
|
| 3180 |
+
"epoch": 1.59,
|
| 3181 |
+
"learning_rate": 0.00020597101449275362,
|
| 3182 |
+
"loss": 0.6687,
|
| 3183 |
+
"step": 4970
|
| 3184 |
+
},
|
| 3185 |
+
{
|
| 3186 |
+
"epoch": 1.59,
|
| 3187 |
+
"learning_rate": 0.00020577777777777776,
|
| 3188 |
+
"loss": 0.674,
|
| 3189 |
+
"step": 4980
|
| 3190 |
+
},
|
| 3191 |
+
{
|
| 3192 |
+
"epoch": 1.6,
|
| 3193 |
+
"learning_rate": 0.0002055845410628019,
|
| 3194 |
+
"loss": 0.618,
|
| 3195 |
+
"step": 4990
|
| 3196 |
+
},
|
| 3197 |
+
{
|
| 3198 |
+
"epoch": 1.6,
|
| 3199 |
+
"learning_rate": 0.00020539130434782605,
|
| 3200 |
+
"loss": 0.689,
|
| 3201 |
+
"step": 5000
|
| 3202 |
+
},
|
| 3203 |
+
{
|
| 3204 |
+
"epoch": 1.6,
|
| 3205 |
+
"eval_loss": 0.6850703358650208,
|
| 3206 |
+
"eval_runtime": 372.4881,
|
| 3207 |
+
"eval_samples_per_second": 5.369,
|
| 3208 |
+
"eval_steps_per_second": 0.671,
|
| 3209 |
+
"step": 5000
|
| 3210 |
+
},
|
| 3211 |
+
{
|
| 3212 |
+
"epoch": 1.6,
|
| 3213 |
+
"learning_rate": 0.00020519806763285024,
|
| 3214 |
+
"loss": 0.6865,
|
| 3215 |
+
"step": 5010
|
| 3216 |
+
},
|
| 3217 |
+
{
|
| 3218 |
+
"epoch": 1.61,
|
| 3219 |
+
"learning_rate": 0.0002050048309178744,
|
| 3220 |
+
"loss": 0.6529,
|
| 3221 |
+
"step": 5020
|
| 3222 |
+
},
|
| 3223 |
+
{
|
| 3224 |
+
"epoch": 1.61,
|
| 3225 |
+
"learning_rate": 0.00020481159420289853,
|
| 3226 |
+
"loss": 0.6495,
|
| 3227 |
+
"step": 5030
|
| 3228 |
+
},
|
| 3229 |
+
{
|
| 3230 |
+
"epoch": 1.61,
|
| 3231 |
+
"learning_rate": 0.00020461835748792267,
|
| 3232 |
+
"loss": 0.6517,
|
| 3233 |
+
"step": 5040
|
| 3234 |
+
},
|
| 3235 |
+
{
|
| 3236 |
+
"epoch": 1.62,
|
| 3237 |
+
"learning_rate": 0.00020442512077294681,
|
| 3238 |
+
"loss": 0.6684,
|
| 3239 |
+
"step": 5050
|
| 3240 |
+
},
|
| 3241 |
+
{
|
| 3242 |
+
"epoch": 1.62,
|
| 3243 |
+
"learning_rate": 0.000204231884057971,
|
| 3244 |
+
"loss": 0.6412,
|
| 3245 |
+
"step": 5060
|
| 3246 |
+
},
|
| 3247 |
+
{
|
| 3248 |
+
"epoch": 1.62,
|
| 3249 |
+
"learning_rate": 0.00020403864734299515,
|
| 3250 |
+
"loss": 0.6463,
|
| 3251 |
+
"step": 5070
|
| 3252 |
+
},
|
| 3253 |
+
{
|
| 3254 |
+
"epoch": 1.63,
|
| 3255 |
+
"learning_rate": 0.0002038454106280193,
|
| 3256 |
+
"loss": 0.633,
|
| 3257 |
+
"step": 5080
|
| 3258 |
+
},
|
| 3259 |
+
{
|
| 3260 |
+
"epoch": 1.63,
|
| 3261 |
+
"learning_rate": 0.00020365217391304344,
|
| 3262 |
+
"loss": 0.6473,
|
| 3263 |
+
"step": 5090
|
| 3264 |
+
},
|
| 3265 |
+
{
|
| 3266 |
+
"epoch": 1.63,
|
| 3267 |
+
"learning_rate": 0.00020345893719806764,
|
| 3268 |
+
"loss": 0.6102,
|
| 3269 |
+
"step": 5100
|
| 3270 |
+
},
|
| 3271 |
+
{
|
| 3272 |
+
"epoch": 1.64,
|
| 3273 |
+
"learning_rate": 0.00020326570048309178,
|
| 3274 |
+
"loss": 0.6816,
|
| 3275 |
+
"step": 5110
|
| 3276 |
+
},
|
| 3277 |
+
{
|
| 3278 |
+
"epoch": 1.64,
|
| 3279 |
+
"learning_rate": 0.00020307246376811592,
|
| 3280 |
+
"loss": 0.6743,
|
| 3281 |
+
"step": 5120
|
| 3282 |
+
},
|
| 3283 |
+
{
|
| 3284 |
+
"epoch": 1.64,
|
| 3285 |
+
"learning_rate": 0.00020287922705314006,
|
| 3286 |
+
"loss": 0.6745,
|
| 3287 |
+
"step": 5130
|
| 3288 |
+
},
|
| 3289 |
+
{
|
| 3290 |
+
"epoch": 1.64,
|
| 3291 |
+
"learning_rate": 0.0002026859903381642,
|
| 3292 |
+
"loss": 0.6435,
|
| 3293 |
+
"step": 5140
|
| 3294 |
+
},
|
| 3295 |
+
{
|
| 3296 |
+
"epoch": 1.65,
|
| 3297 |
+
"learning_rate": 0.0002024927536231884,
|
| 3298 |
+
"loss": 0.6675,
|
| 3299 |
+
"step": 5150
|
| 3300 |
+
},
|
| 3301 |
+
{
|
| 3302 |
+
"epoch": 1.65,
|
| 3303 |
+
"learning_rate": 0.00020229951690821255,
|
| 3304 |
+
"loss": 0.6547,
|
| 3305 |
+
"step": 5160
|
| 3306 |
+
},
|
| 3307 |
+
{
|
| 3308 |
+
"epoch": 1.65,
|
| 3309 |
+
"learning_rate": 0.0002021062801932367,
|
| 3310 |
+
"loss": 0.6464,
|
| 3311 |
+
"step": 5170
|
| 3312 |
+
},
|
| 3313 |
+
{
|
| 3314 |
+
"epoch": 1.66,
|
| 3315 |
+
"learning_rate": 0.00020191304347826083,
|
| 3316 |
+
"loss": 0.6894,
|
| 3317 |
+
"step": 5180
|
| 3318 |
+
},
|
| 3319 |
+
{
|
| 3320 |
+
"epoch": 1.66,
|
| 3321 |
+
"learning_rate": 0.00020171980676328503,
|
| 3322 |
+
"loss": 0.6613,
|
| 3323 |
+
"step": 5190
|
| 3324 |
+
},
|
| 3325 |
+
{
|
| 3326 |
+
"epoch": 1.66,
|
| 3327 |
+
"learning_rate": 0.00020152657004830917,
|
| 3328 |
+
"loss": 0.6362,
|
| 3329 |
+
"step": 5200
|
| 3330 |
+
},
|
| 3331 |
+
{
|
| 3332 |
+
"epoch": 1.66,
|
| 3333 |
+
"eval_loss": 0.6850407719612122,
|
| 3334 |
+
"eval_runtime": 377.0156,
|
| 3335 |
+
"eval_samples_per_second": 5.305,
|
| 3336 |
+
"eval_steps_per_second": 0.663,
|
| 3337 |
+
"step": 5200
|
| 3338 |
+
},
|
| 3339 |
+
{
|
| 3340 |
+
"epoch": 1.67,
|
| 3341 |
+
"learning_rate": 0.0002013333333333333,
|
| 3342 |
+
"loss": 0.636,
|
| 3343 |
+
"step": 5210
|
| 3344 |
+
},
|
| 3345 |
+
{
|
| 3346 |
+
"epoch": 1.67,
|
| 3347 |
+
"learning_rate": 0.00020114009661835745,
|
| 3348 |
+
"loss": 0.6731,
|
| 3349 |
+
"step": 5220
|
| 3350 |
+
},
|
| 3351 |
+
{
|
| 3352 |
+
"epoch": 1.67,
|
| 3353 |
+
"learning_rate": 0.00020094685990338165,
|
| 3354 |
+
"loss": 0.6429,
|
| 3355 |
+
"step": 5230
|
| 3356 |
+
},
|
| 3357 |
+
{
|
| 3358 |
+
"epoch": 1.68,
|
| 3359 |
+
"learning_rate": 0.0002007536231884058,
|
| 3360 |
+
"loss": 0.6808,
|
| 3361 |
+
"step": 5240
|
| 3362 |
+
},
|
| 3363 |
+
{
|
| 3364 |
+
"epoch": 1.68,
|
| 3365 |
+
"learning_rate": 0.00020056038647342994,
|
| 3366 |
+
"loss": 0.6728,
|
| 3367 |
+
"step": 5250
|
| 3368 |
+
},
|
| 3369 |
+
{
|
| 3370 |
+
"epoch": 1.68,
|
| 3371 |
+
"learning_rate": 0.00020036714975845408,
|
| 3372 |
+
"loss": 0.6377,
|
| 3373 |
+
"step": 5260
|
| 3374 |
+
},
|
| 3375 |
+
{
|
| 3376 |
+
"epoch": 1.69,
|
| 3377 |
+
"learning_rate": 0.00020017391304347822,
|
| 3378 |
+
"loss": 0.6616,
|
| 3379 |
+
"step": 5270
|
| 3380 |
+
},
|
| 3381 |
+
{
|
| 3382 |
+
"epoch": 1.69,
|
| 3383 |
+
"learning_rate": 0.00019998067632850242,
|
| 3384 |
+
"loss": 0.6571,
|
| 3385 |
+
"step": 5280
|
| 3386 |
+
},
|
| 3387 |
+
{
|
| 3388 |
+
"epoch": 1.69,
|
| 3389 |
+
"learning_rate": 0.00019978743961352656,
|
| 3390 |
+
"loss": 0.6765,
|
| 3391 |
+
"step": 5290
|
| 3392 |
+
},
|
| 3393 |
+
{
|
| 3394 |
+
"epoch": 1.7,
|
| 3395 |
+
"learning_rate": 0.0001995942028985507,
|
| 3396 |
+
"loss": 0.6636,
|
| 3397 |
+
"step": 5300
|
| 3398 |
+
},
|
| 3399 |
+
{
|
| 3400 |
+
"epoch": 1.7,
|
| 3401 |
+
"learning_rate": 0.00019940096618357485,
|
| 3402 |
+
"loss": 0.6621,
|
| 3403 |
+
"step": 5310
|
| 3404 |
+
},
|
| 3405 |
+
{
|
| 3406 |
+
"epoch": 1.7,
|
| 3407 |
+
"learning_rate": 0.00019920772946859904,
|
| 3408 |
+
"loss": 0.6788,
|
| 3409 |
+
"step": 5320
|
| 3410 |
+
},
|
| 3411 |
+
{
|
| 3412 |
+
"epoch": 1.71,
|
| 3413 |
+
"learning_rate": 0.00019901449275362319,
|
| 3414 |
+
"loss": 0.6491,
|
| 3415 |
+
"step": 5330
|
| 3416 |
+
},
|
| 3417 |
+
{
|
| 3418 |
+
"epoch": 1.71,
|
| 3419 |
+
"learning_rate": 0.00019882125603864733,
|
| 3420 |
+
"loss": 0.6413,
|
| 3421 |
+
"step": 5340
|
| 3422 |
+
},
|
| 3423 |
+
{
|
| 3424 |
+
"epoch": 1.71,
|
| 3425 |
+
"learning_rate": 0.00019862801932367147,
|
| 3426 |
+
"loss": 0.6287,
|
| 3427 |
+
"step": 5350
|
| 3428 |
+
},
|
| 3429 |
+
{
|
| 3430 |
+
"epoch": 1.72,
|
| 3431 |
+
"learning_rate": 0.0001984347826086956,
|
| 3432 |
+
"loss": 0.6624,
|
| 3433 |
+
"step": 5360
|
| 3434 |
+
},
|
| 3435 |
+
{
|
| 3436 |
+
"epoch": 1.72,
|
| 3437 |
+
"learning_rate": 0.0001982415458937198,
|
| 3438 |
+
"loss": 0.6657,
|
| 3439 |
+
"step": 5370
|
| 3440 |
+
},
|
| 3441 |
+
{
|
| 3442 |
+
"epoch": 1.72,
|
| 3443 |
+
"learning_rate": 0.00019804830917874395,
|
| 3444 |
+
"loss": 0.6617,
|
| 3445 |
+
"step": 5380
|
| 3446 |
+
},
|
| 3447 |
+
{
|
| 3448 |
+
"epoch": 1.72,
|
| 3449 |
+
"learning_rate": 0.0001978550724637681,
|
| 3450 |
+
"loss": 0.6791,
|
| 3451 |
+
"step": 5390
|
| 3452 |
+
},
|
| 3453 |
+
{
|
| 3454 |
+
"epoch": 1.73,
|
| 3455 |
+
"learning_rate": 0.00019766183574879224,
|
| 3456 |
+
"loss": 0.6537,
|
| 3457 |
+
"step": 5400
|
| 3458 |
+
},
|
| 3459 |
+
{
|
| 3460 |
+
"epoch": 1.73,
|
| 3461 |
+
"eval_loss": 0.683772623538971,
|
| 3462 |
+
"eval_runtime": 380.6447,
|
| 3463 |
+
"eval_samples_per_second": 5.254,
|
| 3464 |
+
"eval_steps_per_second": 0.657,
|
| 3465 |
+
"step": 5400
|
| 3466 |
+
},
|
| 3467 |
+
{
|
| 3468 |
+
"epoch": 1.73,
|
| 3469 |
+
"learning_rate": 0.00019746859903381643,
|
| 3470 |
+
"loss": 0.5705,
|
| 3471 |
+
"step": 5410
|
| 3472 |
+
},
|
| 3473 |
+
{
|
| 3474 |
+
"epoch": 1.73,
|
| 3475 |
+
"learning_rate": 0.00019727536231884058,
|
| 3476 |
+
"loss": 0.634,
|
| 3477 |
+
"step": 5420
|
| 3478 |
+
},
|
| 3479 |
+
{
|
| 3480 |
+
"epoch": 1.74,
|
| 3481 |
+
"learning_rate": 0.00019708212560386472,
|
| 3482 |
+
"loss": 0.6558,
|
| 3483 |
+
"step": 5430
|
| 3484 |
+
},
|
| 3485 |
+
{
|
| 3486 |
+
"epoch": 1.74,
|
| 3487 |
+
"learning_rate": 0.00019688888888888886,
|
| 3488 |
+
"loss": 0.657,
|
| 3489 |
+
"step": 5440
|
| 3490 |
+
},
|
| 3491 |
+
{
|
| 3492 |
+
"epoch": 1.74,
|
| 3493 |
+
"learning_rate": 0.000196695652173913,
|
| 3494 |
+
"loss": 0.6615,
|
| 3495 |
+
"step": 5450
|
| 3496 |
+
},
|
| 3497 |
+
{
|
| 3498 |
+
"epoch": 1.75,
|
| 3499 |
+
"learning_rate": 0.0001965024154589372,
|
| 3500 |
+
"loss": 0.6656,
|
| 3501 |
+
"step": 5460
|
| 3502 |
+
},
|
| 3503 |
+
{
|
| 3504 |
+
"epoch": 1.75,
|
| 3505 |
+
"learning_rate": 0.00019630917874396134,
|
| 3506 |
+
"loss": 0.6286,
|
| 3507 |
+
"step": 5470
|
| 3508 |
+
},
|
| 3509 |
+
{
|
| 3510 |
+
"epoch": 1.75,
|
| 3511 |
+
"learning_rate": 0.0001961159420289855,
|
| 3512 |
+
"loss": 0.657,
|
| 3513 |
+
"step": 5480
|
| 3514 |
+
},
|
| 3515 |
+
{
|
| 3516 |
+
"epoch": 1.76,
|
| 3517 |
+
"learning_rate": 0.00019592270531400963,
|
| 3518 |
+
"loss": 0.6891,
|
| 3519 |
+
"step": 5490
|
| 3520 |
+
},
|
| 3521 |
+
{
|
| 3522 |
+
"epoch": 1.76,
|
| 3523 |
+
"learning_rate": 0.0001957294685990338,
|
| 3524 |
+
"loss": 0.6539,
|
| 3525 |
+
"step": 5500
|
| 3526 |
+
},
|
| 3527 |
+
{
|
| 3528 |
+
"epoch": 1.76,
|
| 3529 |
+
"learning_rate": 0.00019553623188405797,
|
| 3530 |
+
"loss": 0.6335,
|
| 3531 |
+
"step": 5510
|
| 3532 |
+
},
|
| 3533 |
+
{
|
| 3534 |
+
"epoch": 1.77,
|
| 3535 |
+
"learning_rate": 0.0001953429951690821,
|
| 3536 |
+
"loss": 0.6625,
|
| 3537 |
+
"step": 5520
|
| 3538 |
+
},
|
| 3539 |
+
{
|
| 3540 |
+
"epoch": 1.77,
|
| 3541 |
+
"learning_rate": 0.00019514975845410625,
|
| 3542 |
+
"loss": 0.6597,
|
| 3543 |
+
"step": 5530
|
| 3544 |
+
},
|
| 3545 |
+
{
|
| 3546 |
+
"epoch": 1.77,
|
| 3547 |
+
"learning_rate": 0.00019495652173913042,
|
| 3548 |
+
"loss": 0.6753,
|
| 3549 |
+
"step": 5540
|
| 3550 |
+
},
|
| 3551 |
+
{
|
| 3552 |
+
"epoch": 1.78,
|
| 3553 |
+
"learning_rate": 0.0001947632850241546,
|
| 3554 |
+
"loss": 0.6644,
|
| 3555 |
+
"step": 5550
|
| 3556 |
+
},
|
| 3557 |
+
{
|
| 3558 |
+
"epoch": 1.78,
|
| 3559 |
+
"learning_rate": 0.00019457004830917874,
|
| 3560 |
+
"loss": 0.6434,
|
| 3561 |
+
"step": 5560
|
| 3562 |
+
},
|
| 3563 |
+
{
|
| 3564 |
+
"epoch": 1.78,
|
| 3565 |
+
"learning_rate": 0.00019437681159420288,
|
| 3566 |
+
"loss": 0.6604,
|
| 3567 |
+
"step": 5570
|
| 3568 |
+
},
|
| 3569 |
+
{
|
| 3570 |
+
"epoch": 1.79,
|
| 3571 |
+
"learning_rate": 0.00019418357487922702,
|
| 3572 |
+
"loss": 0.6538,
|
| 3573 |
+
"step": 5580
|
| 3574 |
+
},
|
| 3575 |
+
{
|
| 3576 |
+
"epoch": 1.79,
|
| 3577 |
+
"learning_rate": 0.0001939903381642512,
|
| 3578 |
+
"loss": 0.639,
|
| 3579 |
+
"step": 5590
|
| 3580 |
+
},
|
| 3581 |
+
{
|
| 3582 |
+
"epoch": 1.79,
|
| 3583 |
+
"learning_rate": 0.00019379710144927536,
|
| 3584 |
+
"loss": 0.6668,
|
| 3585 |
+
"step": 5600
|
| 3586 |
+
},
|
| 3587 |
+
{
|
| 3588 |
+
"epoch": 1.79,
|
| 3589 |
+
"eval_loss": 0.6843588352203369,
|
| 3590 |
+
"eval_runtime": 378.4765,
|
| 3591 |
+
"eval_samples_per_second": 5.284,
|
| 3592 |
+
"eval_steps_per_second": 0.661,
|
| 3593 |
+
"step": 5600
|
| 3594 |
+
},
|
| 3595 |
+
{
|
| 3596 |
+
"epoch": 1.8,
|
| 3597 |
+
"learning_rate": 0.0001936038647342995,
|
| 3598 |
+
"loss": 0.6375,
|
| 3599 |
+
"step": 5610
|
| 3600 |
+
},
|
| 3601 |
+
{
|
| 3602 |
+
"epoch": 1.8,
|
| 3603 |
+
"learning_rate": 0.00019341062801932364,
|
| 3604 |
+
"loss": 0.6455,
|
| 3605 |
+
"step": 5620
|
| 3606 |
+
},
|
| 3607 |
+
{
|
| 3608 |
+
"epoch": 1.8,
|
| 3609 |
+
"learning_rate": 0.00019321739130434781,
|
| 3610 |
+
"loss": 0.6856,
|
| 3611 |
+
"step": 5630
|
| 3612 |
+
},
|
| 3613 |
+
{
|
| 3614 |
+
"epoch": 1.8,
|
| 3615 |
+
"learning_rate": 0.00019302415458937198,
|
| 3616 |
+
"loss": 0.648,
|
| 3617 |
+
"step": 5640
|
| 3618 |
+
},
|
| 3619 |
+
{
|
| 3620 |
+
"epoch": 1.81,
|
| 3621 |
+
"learning_rate": 0.00019283091787439613,
|
| 3622 |
+
"loss": 0.6005,
|
| 3623 |
+
"step": 5650
|
| 3624 |
+
},
|
| 3625 |
+
{
|
| 3626 |
+
"epoch": 1.81,
|
| 3627 |
+
"learning_rate": 0.00019263768115942027,
|
| 3628 |
+
"loss": 0.6719,
|
| 3629 |
+
"step": 5660
|
| 3630 |
+
},
|
| 3631 |
+
{
|
| 3632 |
+
"epoch": 1.81,
|
| 3633 |
+
"learning_rate": 0.0001924444444444444,
|
| 3634 |
+
"loss": 0.6588,
|
| 3635 |
+
"step": 5670
|
| 3636 |
+
},
|
| 3637 |
+
{
|
| 3638 |
+
"epoch": 1.82,
|
| 3639 |
+
"learning_rate": 0.00019225120772946858,
|
| 3640 |
+
"loss": 0.6517,
|
| 3641 |
+
"step": 5680
|
| 3642 |
+
},
|
| 3643 |
+
{
|
| 3644 |
+
"epoch": 1.82,
|
| 3645 |
+
"learning_rate": 0.00019205797101449275,
|
| 3646 |
+
"loss": 0.6489,
|
| 3647 |
+
"step": 5690
|
| 3648 |
+
},
|
| 3649 |
+
{
|
| 3650 |
+
"epoch": 1.82,
|
| 3651 |
+
"learning_rate": 0.0001918647342995169,
|
| 3652 |
+
"loss": 0.6283,
|
| 3653 |
+
"step": 5700
|
| 3654 |
+
},
|
| 3655 |
+
{
|
| 3656 |
+
"epoch": 1.83,
|
| 3657 |
+
"learning_rate": 0.00019167149758454104,
|
| 3658 |
+
"loss": 0.6523,
|
| 3659 |
+
"step": 5710
|
| 3660 |
+
},
|
| 3661 |
+
{
|
| 3662 |
+
"epoch": 1.83,
|
| 3663 |
+
"learning_rate": 0.0001914782608695652,
|
| 3664 |
+
"loss": 0.5914,
|
| 3665 |
+
"step": 5720
|
| 3666 |
+
},
|
| 3667 |
+
{
|
| 3668 |
+
"epoch": 1.83,
|
| 3669 |
+
"learning_rate": 0.00019128502415458935,
|
| 3670 |
+
"loss": 0.6403,
|
| 3671 |
+
"step": 5730
|
| 3672 |
+
},
|
| 3673 |
+
{
|
| 3674 |
+
"epoch": 1.84,
|
| 3675 |
+
"learning_rate": 0.00019109178743961352,
|
| 3676 |
+
"loss": 0.6426,
|
| 3677 |
+
"step": 5740
|
| 3678 |
+
},
|
| 3679 |
+
{
|
| 3680 |
+
"epoch": 1.84,
|
| 3681 |
+
"learning_rate": 0.00019089855072463766,
|
| 3682 |
+
"loss": 0.6609,
|
| 3683 |
+
"step": 5750
|
| 3684 |
+
},
|
| 3685 |
+
{
|
| 3686 |
+
"epoch": 1.84,
|
| 3687 |
+
"learning_rate": 0.0001907053140096618,
|
| 3688 |
+
"loss": 0.6394,
|
| 3689 |
+
"step": 5760
|
| 3690 |
+
},
|
| 3691 |
+
{
|
| 3692 |
+
"epoch": 1.85,
|
| 3693 |
+
"learning_rate": 0.00019051207729468597,
|
| 3694 |
+
"loss": 0.6546,
|
| 3695 |
+
"step": 5770
|
| 3696 |
+
},
|
| 3697 |
+
{
|
| 3698 |
+
"epoch": 1.85,
|
| 3699 |
+
"learning_rate": 0.00019031884057971014,
|
| 3700 |
+
"loss": 0.6879,
|
| 3701 |
+
"step": 5780
|
| 3702 |
+
},
|
| 3703 |
+
{
|
| 3704 |
+
"epoch": 1.85,
|
| 3705 |
+
"learning_rate": 0.00019012560386473429,
|
| 3706 |
+
"loss": 0.6116,
|
| 3707 |
+
"step": 5790
|
| 3708 |
+
},
|
| 3709 |
+
{
|
| 3710 |
+
"epoch": 1.86,
|
| 3711 |
+
"learning_rate": 0.00018993236714975843,
|
| 3712 |
+
"loss": 0.6477,
|
| 3713 |
+
"step": 5800
|
| 3714 |
+
},
|
| 3715 |
+
{
|
| 3716 |
+
"epoch": 1.86,
|
| 3717 |
+
"eval_loss": 0.6840969920158386,
|
| 3718 |
+
"eval_runtime": 385.7288,
|
| 3719 |
+
"eval_samples_per_second": 5.185,
|
| 3720 |
+
"eval_steps_per_second": 0.648,
|
| 3721 |
+
"step": 5800
|
| 3722 |
+
},
|
| 3723 |
+
{
|
| 3724 |
+
"epoch": 1.86,
|
| 3725 |
+
"learning_rate": 0.0001897391304347826,
|
| 3726 |
+
"loss": 0.6657,
|
| 3727 |
+
"step": 5810
|
| 3728 |
+
},
|
| 3729 |
+
{
|
| 3730 |
+
"epoch": 1.86,
|
| 3731 |
+
"learning_rate": 0.00018954589371980674,
|
| 3732 |
+
"loss": 0.6399,
|
| 3733 |
+
"step": 5820
|
| 3734 |
+
},
|
| 3735 |
+
{
|
| 3736 |
+
"epoch": 1.87,
|
| 3737 |
+
"learning_rate": 0.0001893526570048309,
|
| 3738 |
+
"loss": 0.6587,
|
| 3739 |
+
"step": 5830
|
| 3740 |
+
},
|
| 3741 |
+
{
|
| 3742 |
+
"epoch": 1.87,
|
| 3743 |
+
"learning_rate": 0.00018915942028985505,
|
| 3744 |
+
"loss": 0.6333,
|
| 3745 |
+
"step": 5840
|
| 3746 |
+
},
|
| 3747 |
+
{
|
| 3748 |
+
"epoch": 1.87,
|
| 3749 |
+
"learning_rate": 0.00018896618357487922,
|
| 3750 |
+
"loss": 0.6327,
|
| 3751 |
+
"step": 5850
|
| 3752 |
+
},
|
| 3753 |
+
{
|
| 3754 |
+
"epoch": 1.88,
|
| 3755 |
+
"learning_rate": 0.00018877294685990336,
|
| 3756 |
+
"loss": 0.6788,
|
| 3757 |
+
"step": 5860
|
| 3758 |
+
},
|
| 3759 |
+
{
|
| 3760 |
+
"epoch": 1.88,
|
| 3761 |
+
"learning_rate": 0.0001885797101449275,
|
| 3762 |
+
"loss": 0.6765,
|
| 3763 |
+
"step": 5870
|
| 3764 |
+
},
|
| 3765 |
+
{
|
| 3766 |
+
"epoch": 1.88,
|
| 3767 |
+
"learning_rate": 0.00018838647342995168,
|
| 3768 |
+
"loss": 0.6742,
|
| 3769 |
+
"step": 5880
|
| 3770 |
+
},
|
| 3771 |
+
{
|
| 3772 |
+
"epoch": 1.88,
|
| 3773 |
+
"learning_rate": 0.00018819323671497582,
|
| 3774 |
+
"loss": 0.6462,
|
| 3775 |
+
"step": 5890
|
| 3776 |
+
},
|
| 3777 |
+
{
|
| 3778 |
+
"epoch": 1.89,
|
| 3779 |
+
"learning_rate": 0.000188,
|
| 3780 |
+
"loss": 0.6907,
|
| 3781 |
+
"step": 5900
|
| 3782 |
+
},
|
| 3783 |
+
{
|
| 3784 |
+
"epoch": 1.89,
|
| 3785 |
+
"learning_rate": 0.00018780676328502413,
|
| 3786 |
+
"loss": 0.6645,
|
| 3787 |
+
"step": 5910
|
| 3788 |
+
},
|
| 3789 |
+
{
|
| 3790 |
+
"epoch": 1.89,
|
| 3791 |
+
"learning_rate": 0.0001876135265700483,
|
| 3792 |
+
"loss": 0.6719,
|
| 3793 |
+
"step": 5920
|
| 3794 |
+
},
|
| 3795 |
+
{
|
| 3796 |
+
"epoch": 1.9,
|
| 3797 |
+
"learning_rate": 0.00018742028985507244,
|
| 3798 |
+
"loss": 0.6417,
|
| 3799 |
+
"step": 5930
|
| 3800 |
+
},
|
| 3801 |
+
{
|
| 3802 |
+
"epoch": 1.9,
|
| 3803 |
+
"learning_rate": 0.0001872270531400966,
|
| 3804 |
+
"loss": 0.6383,
|
| 3805 |
+
"step": 5940
|
| 3806 |
+
},
|
| 3807 |
+
{
|
| 3808 |
+
"epoch": 1.9,
|
| 3809 |
+
"learning_rate": 0.00018703381642512076,
|
| 3810 |
+
"loss": 0.6342,
|
| 3811 |
+
"step": 5950
|
| 3812 |
+
},
|
| 3813 |
+
{
|
| 3814 |
+
"epoch": 1.91,
|
| 3815 |
+
"learning_rate": 0.0001868405797101449,
|
| 3816 |
+
"loss": 0.657,
|
| 3817 |
+
"step": 5960
|
| 3818 |
+
},
|
| 3819 |
+
{
|
| 3820 |
+
"epoch": 1.91,
|
| 3821 |
+
"learning_rate": 0.00018664734299516907,
|
| 3822 |
+
"loss": 0.6832,
|
| 3823 |
+
"step": 5970
|
| 3824 |
+
},
|
| 3825 |
+
{
|
| 3826 |
+
"epoch": 1.91,
|
| 3827 |
+
"learning_rate": 0.0001864541062801932,
|
| 3828 |
+
"loss": 0.6658,
|
| 3829 |
+
"step": 5980
|
| 3830 |
+
},
|
| 3831 |
+
{
|
| 3832 |
+
"epoch": 1.92,
|
| 3833 |
+
"learning_rate": 0.00018626086956521738,
|
| 3834 |
+
"loss": 0.6614,
|
| 3835 |
+
"step": 5990
|
| 3836 |
+
},
|
| 3837 |
+
{
|
| 3838 |
+
"epoch": 1.92,
|
| 3839 |
+
"learning_rate": 0.00018606763285024152,
|
| 3840 |
+
"loss": 0.6878,
|
| 3841 |
+
"step": 6000
|
| 3842 |
+
},
|
| 3843 |
+
{
|
| 3844 |
+
"epoch": 1.92,
|
| 3845 |
+
"eval_loss": 0.6828343272209167,
|
| 3846 |
+
"eval_runtime": 392.2472,
|
| 3847 |
+
"eval_samples_per_second": 5.099,
|
| 3848 |
+
"eval_steps_per_second": 0.637,
|
| 3849 |
+
"step": 6000
|
| 3850 |
+
},
|
| 3851 |
+
{
|
| 3852 |
+
"epoch": 1.92,
|
| 3853 |
+
"learning_rate": 0.0001858743961352657,
|
| 3854 |
+
"loss": 0.6408,
|
| 3855 |
+
"step": 6010
|
| 3856 |
+
},
|
| 3857 |
+
{
|
| 3858 |
+
"epoch": 1.93,
|
| 3859 |
+
"learning_rate": 0.00018568115942028984,
|
| 3860 |
+
"loss": 0.6498,
|
| 3861 |
+
"step": 6020
|
| 3862 |
+
},
|
| 3863 |
+
{
|
| 3864 |
+
"epoch": 1.93,
|
| 3865 |
+
"learning_rate": 0.000185487922705314,
|
| 3866 |
+
"loss": 0.6598,
|
| 3867 |
+
"step": 6030
|
| 3868 |
+
},
|
| 3869 |
+
{
|
| 3870 |
+
"epoch": 1.93,
|
| 3871 |
+
"learning_rate": 0.00018529468599033815,
|
| 3872 |
+
"loss": 0.6231,
|
| 3873 |
+
"step": 6040
|
| 3874 |
+
},
|
| 3875 |
+
{
|
| 3876 |
+
"epoch": 1.94,
|
| 3877 |
+
"learning_rate": 0.0001851014492753623,
|
| 3878 |
+
"loss": 0.6906,
|
| 3879 |
+
"step": 6050
|
| 3880 |
+
},
|
| 3881 |
+
{
|
| 3882 |
+
"epoch": 1.94,
|
| 3883 |
+
"learning_rate": 0.00018490821256038646,
|
| 3884 |
+
"loss": 0.6467,
|
| 3885 |
+
"step": 6060
|
| 3886 |
+
},
|
| 3887 |
+
{
|
| 3888 |
+
"epoch": 1.94,
|
| 3889 |
+
"learning_rate": 0.0001847149758454106,
|
| 3890 |
+
"loss": 0.6355,
|
| 3891 |
+
"step": 6070
|
| 3892 |
+
},
|
| 3893 |
+
{
|
| 3894 |
+
"epoch": 1.95,
|
| 3895 |
+
"learning_rate": 0.00018452173913043477,
|
| 3896 |
+
"loss": 0.6794,
|
| 3897 |
+
"step": 6080
|
| 3898 |
+
},
|
| 3899 |
+
{
|
| 3900 |
+
"epoch": 1.95,
|
| 3901 |
+
"learning_rate": 0.00018432850241545891,
|
| 3902 |
+
"loss": 0.6475,
|
| 3903 |
+
"step": 6090
|
| 3904 |
+
},
|
| 3905 |
+
{
|
| 3906 |
+
"epoch": 1.95,
|
| 3907 |
+
"learning_rate": 0.00018413526570048306,
|
| 3908 |
+
"loss": 0.6165,
|
| 3909 |
+
"step": 6100
|
| 3910 |
+
},
|
| 3911 |
+
{
|
| 3912 |
+
"epoch": 1.96,
|
| 3913 |
+
"learning_rate": 0.00018394202898550723,
|
| 3914 |
+
"loss": 0.6281,
|
| 3915 |
+
"step": 6110
|
| 3916 |
+
},
|
| 3917 |
+
{
|
| 3918 |
+
"epoch": 1.96,
|
| 3919 |
+
"learning_rate": 0.0001837487922705314,
|
| 3920 |
+
"loss": 0.6473,
|
| 3921 |
+
"step": 6120
|
| 3922 |
+
},
|
| 3923 |
+
{
|
| 3924 |
+
"epoch": 1.96,
|
| 3925 |
+
"learning_rate": 0.00018355555555555554,
|
| 3926 |
+
"loss": 0.6415,
|
| 3927 |
+
"step": 6130
|
| 3928 |
+
},
|
| 3929 |
+
{
|
| 3930 |
+
"epoch": 1.96,
|
| 3931 |
+
"learning_rate": 0.00018336231884057968,
|
| 3932 |
+
"loss": 0.659,
|
| 3933 |
+
"step": 6140
|
| 3934 |
+
},
|
| 3935 |
+
{
|
| 3936 |
+
"epoch": 1.97,
|
| 3937 |
+
"learning_rate": 0.00018316908212560385,
|
| 3938 |
+
"loss": 0.6821,
|
| 3939 |
+
"step": 6150
|
| 3940 |
+
},
|
| 3941 |
+
{
|
| 3942 |
+
"epoch": 1.97,
|
| 3943 |
+
"learning_rate": 0.00018297584541062802,
|
| 3944 |
+
"loss": 0.6631,
|
| 3945 |
+
"step": 6160
|
| 3946 |
+
},
|
| 3947 |
+
{
|
| 3948 |
+
"epoch": 1.97,
|
| 3949 |
+
"learning_rate": 0.00018278260869565216,
|
| 3950 |
+
"loss": 0.6332,
|
| 3951 |
+
"step": 6170
|
| 3952 |
+
},
|
| 3953 |
+
{
|
| 3954 |
+
"epoch": 1.98,
|
| 3955 |
+
"learning_rate": 0.0001825893719806763,
|
| 3956 |
+
"loss": 0.6561,
|
| 3957 |
+
"step": 6180
|
| 3958 |
+
},
|
| 3959 |
+
{
|
| 3960 |
+
"epoch": 1.98,
|
| 3961 |
+
"learning_rate": 0.00018239613526570045,
|
| 3962 |
+
"loss": 0.6654,
|
| 3963 |
+
"step": 6190
|
| 3964 |
+
},
|
| 3965 |
+
{
|
| 3966 |
+
"epoch": 1.98,
|
| 3967 |
+
"learning_rate": 0.00018220289855072462,
|
| 3968 |
+
"loss": 0.6656,
|
| 3969 |
+
"step": 6200
|
| 3970 |
+
},
|
| 3971 |
+
{
|
| 3972 |
+
"epoch": 1.98,
|
| 3973 |
+
"eval_loss": 0.6827040314674377,
|
| 3974 |
+
"eval_runtime": 411.107,
|
| 3975 |
+
"eval_samples_per_second": 4.865,
|
| 3976 |
+
"eval_steps_per_second": 0.608,
|
| 3977 |
+
"step": 6200
|
| 3978 |
+
}
|
| 3979 |
+
],
|
| 3980 |
+
"logging_steps": 10,
|
| 3981 |
+
"max_steps": 15625,
|
| 3982 |
+
"num_train_epochs": 5,
|
| 3983 |
+
"save_steps": 200,
|
| 3984 |
+
"total_flos": 9.076709518992998e+17,
|
| 3985 |
+
"trial_name": null,
|
| 3986 |
+
"trial_params": null
|
| 3987 |
+
}
|
1a_52k/best_model/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9d7061c7af35fd492ca3f6488de8df547618d812380d9bafbcd899a141e4439a
|
| 3 |
+
size 4027
|
1b_52k/best_model/adapter_config.json
ADDED
|
@@ -0,0 +1,21 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"base_model_name_or_path": "huggyllama/llama-7b",
|
| 3 |
+
"bias": "none",
|
| 4 |
+
"enable_lora": null,
|
| 5 |
+
"fan_in_fan_out": false,
|
| 6 |
+
"inference_mode": true,
|
| 7 |
+
"init_lora_weights": true,
|
| 8 |
+
"lora_alpha": 16,
|
| 9 |
+
"lora_dropout": 0.05,
|
| 10 |
+
"merge_weights": false,
|
| 11 |
+
"modules_to_save": null,
|
| 12 |
+
"peft_type": "LORA",
|
| 13 |
+
"r": 16,
|
| 14 |
+
"target_modules": [
|
| 15 |
+
"q_proj",
|
| 16 |
+
"k_proj",
|
| 17 |
+
"v_proj",
|
| 18 |
+
"o_proj"
|
| 19 |
+
],
|
| 20 |
+
"task_type": "CAUSAL_LM"
|
| 21 |
+
}
|
1b_52k/best_model/adapter_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e0c9ec4bd365e976e6dd3756be389a4ba67a7a717d24e40ab3e92bce8a4aeb9b
|
| 3 |
+
size 67201357
|
1b_52k/best_model/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e09c4a4fffc8d2697b85bc45adb2186c5375b110db55628526cab3c41ee7ec51
|
| 3 |
+
size 134433093
|
1b_52k/best_model/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a7d6f96c7f113a40e55242b894882813a652bd0bf62a9dddb84e671e6c82e98a
|
| 3 |
+
size 14575
|
1b_52k/best_model/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:34396d061373c501d7b53ce6f3a85b6a9bb7a6872aad71fe1d037c07b8516fe0
|
| 3 |
+
size 627
|
1b_52k/best_model/trainer_state.json
ADDED
|
@@ -0,0 +1,3987 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.4666987955570221,
|
| 3 |
+
"best_model_checkpoint": "checkpoints/1b_52k/checkpoint-6200",
|
| 4 |
+
"epoch": 1.9935691318327975,
|
| 5 |
+
"eval_steps": 200,
|
| 6 |
+
"global_step": 6200,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.0,
|
| 13 |
+
"learning_rate": 2.6999999999999996e-05,
|
| 14 |
+
"loss": 1.2981,
|
| 15 |
+
"step": 10
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"epoch": 0.01,
|
| 19 |
+
"learning_rate": 5.1e-05,
|
| 20 |
+
"loss": 1.2597,
|
| 21 |
+
"step": 20
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"epoch": 0.01,
|
| 25 |
+
"learning_rate": 8.1e-05,
|
| 26 |
+
"loss": 1.1419,
|
| 27 |
+
"step": 30
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.01,
|
| 31 |
+
"learning_rate": 0.00011099999999999999,
|
| 32 |
+
"loss": 0.8779,
|
| 33 |
+
"step": 40
|
| 34 |
+
},
|
| 35 |
+
{
|
| 36 |
+
"epoch": 0.02,
|
| 37 |
+
"learning_rate": 0.00014099999999999998,
|
| 38 |
+
"loss": 0.7661,
|
| 39 |
+
"step": 50
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.02,
|
| 43 |
+
"learning_rate": 0.00017099999999999998,
|
| 44 |
+
"loss": 0.7354,
|
| 45 |
+
"step": 60
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.02,
|
| 49 |
+
"learning_rate": 0.000201,
|
| 50 |
+
"loss": 0.5858,
|
| 51 |
+
"step": 70
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.03,
|
| 55 |
+
"learning_rate": 0.00023099999999999998,
|
| 56 |
+
"loss": 0.5491,
|
| 57 |
+
"step": 80
|
| 58 |
+
},
|
| 59 |
+
{
|
| 60 |
+
"epoch": 0.03,
|
| 61 |
+
"learning_rate": 0.000261,
|
| 62 |
+
"loss": 0.5445,
|
| 63 |
+
"step": 90
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.03,
|
| 67 |
+
"learning_rate": 0.00029099999999999997,
|
| 68 |
+
"loss": 0.5566,
|
| 69 |
+
"step": 100
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.04,
|
| 73 |
+
"learning_rate": 0.0002998640776699029,
|
| 74 |
+
"loss": 0.5398,
|
| 75 |
+
"step": 110
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"epoch": 0.04,
|
| 79 |
+
"learning_rate": 0.0002996699029126213,
|
| 80 |
+
"loss": 0.528,
|
| 81 |
+
"step": 120
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.04,
|
| 85 |
+
"learning_rate": 0.0002994757281553398,
|
| 86 |
+
"loss": 0.5818,
|
| 87 |
+
"step": 130
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.05,
|
| 91 |
+
"learning_rate": 0.00029928155339805826,
|
| 92 |
+
"loss": 0.5048,
|
| 93 |
+
"step": 140
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.05,
|
| 97 |
+
"learning_rate": 0.0002990873786407767,
|
| 98 |
+
"loss": 0.5234,
|
| 99 |
+
"step": 150
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.05,
|
| 103 |
+
"learning_rate": 0.00029889320388349516,
|
| 104 |
+
"loss": 0.5353,
|
| 105 |
+
"step": 160
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 0.05,
|
| 109 |
+
"learning_rate": 0.0002986990291262136,
|
| 110 |
+
"loss": 0.5341,
|
| 111 |
+
"step": 170
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"epoch": 0.06,
|
| 115 |
+
"learning_rate": 0.000298504854368932,
|
| 116 |
+
"loss": 0.5185,
|
| 117 |
+
"step": 180
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 0.06,
|
| 121 |
+
"learning_rate": 0.00029831067961165047,
|
| 122 |
+
"loss": 0.5046,
|
| 123 |
+
"step": 190
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"epoch": 0.06,
|
| 127 |
+
"learning_rate": 0.0002981165048543689,
|
| 128 |
+
"loss": 0.4929,
|
| 129 |
+
"step": 200
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.06,
|
| 133 |
+
"eval_loss": 0.5254431366920471,
|
| 134 |
+
"eval_runtime": 503.7885,
|
| 135 |
+
"eval_samples_per_second": 3.97,
|
| 136 |
+
"eval_steps_per_second": 0.496,
|
| 137 |
+
"step": 200
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 0.07,
|
| 141 |
+
"learning_rate": 0.00029792233009708736,
|
| 142 |
+
"loss": 0.5269,
|
| 143 |
+
"step": 210
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.07,
|
| 147 |
+
"learning_rate": 0.0002977281553398058,
|
| 148 |
+
"loss": 0.5073,
|
| 149 |
+
"step": 220
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.07,
|
| 153 |
+
"learning_rate": 0.00029753398058252425,
|
| 154 |
+
"loss": 0.5534,
|
| 155 |
+
"step": 230
|
| 156 |
+
},
|
| 157 |
+
{
|
| 158 |
+
"epoch": 0.08,
|
| 159 |
+
"learning_rate": 0.00029733980582524267,
|
| 160 |
+
"loss": 0.5466,
|
| 161 |
+
"step": 240
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"epoch": 0.08,
|
| 165 |
+
"learning_rate": 0.00029714563106796114,
|
| 166 |
+
"loss": 0.5065,
|
| 167 |
+
"step": 250
|
| 168 |
+
},
|
| 169 |
+
{
|
| 170 |
+
"epoch": 0.08,
|
| 171 |
+
"learning_rate": 0.0002969514563106796,
|
| 172 |
+
"loss": 0.5038,
|
| 173 |
+
"step": 260
|
| 174 |
+
},
|
| 175 |
+
{
|
| 176 |
+
"epoch": 0.09,
|
| 177 |
+
"learning_rate": 0.00029675728155339803,
|
| 178 |
+
"loss": 0.5398,
|
| 179 |
+
"step": 270
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 0.09,
|
| 183 |
+
"learning_rate": 0.0002965631067961165,
|
| 184 |
+
"loss": 0.5068,
|
| 185 |
+
"step": 280
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.09,
|
| 189 |
+
"learning_rate": 0.0002963689320388349,
|
| 190 |
+
"loss": 0.496,
|
| 191 |
+
"step": 290
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.1,
|
| 195 |
+
"learning_rate": 0.0002961747572815534,
|
| 196 |
+
"loss": 0.5276,
|
| 197 |
+
"step": 300
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 0.1,
|
| 201 |
+
"learning_rate": 0.0002959805825242718,
|
| 202 |
+
"loss": 0.5269,
|
| 203 |
+
"step": 310
|
| 204 |
+
},
|
| 205 |
+
{
|
| 206 |
+
"epoch": 0.1,
|
| 207 |
+
"learning_rate": 0.00029578640776699023,
|
| 208 |
+
"loss": 0.504,
|
| 209 |
+
"step": 320
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 0.11,
|
| 213 |
+
"learning_rate": 0.0002955922330097087,
|
| 214 |
+
"loss": 0.5183,
|
| 215 |
+
"step": 330
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 0.11,
|
| 219 |
+
"learning_rate": 0.0002953980582524271,
|
| 220 |
+
"loss": 0.4911,
|
| 221 |
+
"step": 340
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 0.11,
|
| 225 |
+
"learning_rate": 0.0002952038834951456,
|
| 226 |
+
"loss": 0.4937,
|
| 227 |
+
"step": 350
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 0.12,
|
| 231 |
+
"learning_rate": 0.00029500970873786407,
|
| 232 |
+
"loss": 0.4942,
|
| 233 |
+
"step": 360
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.12,
|
| 237 |
+
"learning_rate": 0.0002948155339805825,
|
| 238 |
+
"loss": 0.4763,
|
| 239 |
+
"step": 370
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 0.12,
|
| 243 |
+
"learning_rate": 0.00029462135922330096,
|
| 244 |
+
"loss": 0.5261,
|
| 245 |
+
"step": 380
|
| 246 |
+
},
|
| 247 |
+
{
|
| 248 |
+
"epoch": 0.13,
|
| 249 |
+
"learning_rate": 0.0002944271844660194,
|
| 250 |
+
"loss": 0.5031,
|
| 251 |
+
"step": 390
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 0.13,
|
| 255 |
+
"learning_rate": 0.00029423300970873786,
|
| 256 |
+
"loss": 0.5103,
|
| 257 |
+
"step": 400
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 0.13,
|
| 261 |
+
"eval_loss": 0.5078505873680115,
|
| 262 |
+
"eval_runtime": 523.4102,
|
| 263 |
+
"eval_samples_per_second": 3.821,
|
| 264 |
+
"eval_steps_per_second": 0.478,
|
| 265 |
+
"step": 400
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 0.13,
|
| 269 |
+
"learning_rate": 0.0002940388349514563,
|
| 270 |
+
"loss": 0.5134,
|
| 271 |
+
"step": 410
|
| 272 |
+
},
|
| 273 |
+
{
|
| 274 |
+
"epoch": 0.14,
|
| 275 |
+
"learning_rate": 0.00029384466019417475,
|
| 276 |
+
"loss": 0.5013,
|
| 277 |
+
"step": 420
|
| 278 |
+
},
|
| 279 |
+
{
|
| 280 |
+
"epoch": 0.14,
|
| 281 |
+
"learning_rate": 0.00029365048543689317,
|
| 282 |
+
"loss": 0.5222,
|
| 283 |
+
"step": 430
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 0.14,
|
| 287 |
+
"learning_rate": 0.00029345631067961164,
|
| 288 |
+
"loss": 0.5072,
|
| 289 |
+
"step": 440
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.14,
|
| 293 |
+
"learning_rate": 0.00029326213592233006,
|
| 294 |
+
"loss": 0.4943,
|
| 295 |
+
"step": 450
|
| 296 |
+
},
|
| 297 |
+
{
|
| 298 |
+
"epoch": 0.15,
|
| 299 |
+
"learning_rate": 0.00029306796116504853,
|
| 300 |
+
"loss": 0.4941,
|
| 301 |
+
"step": 460
|
| 302 |
+
},
|
| 303 |
+
{
|
| 304 |
+
"epoch": 0.15,
|
| 305 |
+
"learning_rate": 0.00029287378640776695,
|
| 306 |
+
"loss": 0.4767,
|
| 307 |
+
"step": 470
|
| 308 |
+
},
|
| 309 |
+
{
|
| 310 |
+
"epoch": 0.15,
|
| 311 |
+
"learning_rate": 0.0002926796116504854,
|
| 312 |
+
"loss": 0.5166,
|
| 313 |
+
"step": 480
|
| 314 |
+
},
|
| 315 |
+
{
|
| 316 |
+
"epoch": 0.16,
|
| 317 |
+
"learning_rate": 0.00029248543689320384,
|
| 318 |
+
"loss": 0.487,
|
| 319 |
+
"step": 490
|
| 320 |
+
},
|
| 321 |
+
{
|
| 322 |
+
"epoch": 0.16,
|
| 323 |
+
"learning_rate": 0.0002922912621359223,
|
| 324 |
+
"loss": 0.5165,
|
| 325 |
+
"step": 500
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.16,
|
| 329 |
+
"learning_rate": 0.00029209708737864073,
|
| 330 |
+
"loss": 0.4742,
|
| 331 |
+
"step": 510
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.17,
|
| 335 |
+
"learning_rate": 0.0002919029126213592,
|
| 336 |
+
"loss": 0.496,
|
| 337 |
+
"step": 520
|
| 338 |
+
},
|
| 339 |
+
{
|
| 340 |
+
"epoch": 0.17,
|
| 341 |
+
"learning_rate": 0.0002917087378640776,
|
| 342 |
+
"loss": 0.4934,
|
| 343 |
+
"step": 530
|
| 344 |
+
},
|
| 345 |
+
{
|
| 346 |
+
"epoch": 0.17,
|
| 347 |
+
"learning_rate": 0.0002915145631067961,
|
| 348 |
+
"loss": 0.4915,
|
| 349 |
+
"step": 540
|
| 350 |
+
},
|
| 351 |
+
{
|
| 352 |
+
"epoch": 0.18,
|
| 353 |
+
"learning_rate": 0.00029132038834951457,
|
| 354 |
+
"loss": 0.4783,
|
| 355 |
+
"step": 550
|
| 356 |
+
},
|
| 357 |
+
{
|
| 358 |
+
"epoch": 0.18,
|
| 359 |
+
"learning_rate": 0.000291126213592233,
|
| 360 |
+
"loss": 0.4869,
|
| 361 |
+
"step": 560
|
| 362 |
+
},
|
| 363 |
+
{
|
| 364 |
+
"epoch": 0.18,
|
| 365 |
+
"learning_rate": 0.00029093203883495146,
|
| 366 |
+
"loss": 0.4802,
|
| 367 |
+
"step": 570
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 0.19,
|
| 371 |
+
"learning_rate": 0.0002907378640776699,
|
| 372 |
+
"loss": 0.4953,
|
| 373 |
+
"step": 580
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 0.19,
|
| 377 |
+
"learning_rate": 0.00029054368932038835,
|
| 378 |
+
"loss": 0.4859,
|
| 379 |
+
"step": 590
|
| 380 |
+
},
|
| 381 |
+
{
|
| 382 |
+
"epoch": 0.19,
|
| 383 |
+
"learning_rate": 0.00029034951456310677,
|
| 384 |
+
"loss": 0.5093,
|
| 385 |
+
"step": 600
|
| 386 |
+
},
|
| 387 |
+
{
|
| 388 |
+
"epoch": 0.19,
|
| 389 |
+
"eval_loss": 0.4987016022205353,
|
| 390 |
+
"eval_runtime": 527.0047,
|
| 391 |
+
"eval_samples_per_second": 3.795,
|
| 392 |
+
"eval_steps_per_second": 0.474,
|
| 393 |
+
"step": 600
|
| 394 |
+
},
|
| 395 |
+
{
|
| 396 |
+
"epoch": 0.2,
|
| 397 |
+
"learning_rate": 0.0002901553398058252,
|
| 398 |
+
"loss": 0.4876,
|
| 399 |
+
"step": 610
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 0.2,
|
| 403 |
+
"learning_rate": 0.00028996116504854366,
|
| 404 |
+
"loss": 0.5086,
|
| 405 |
+
"step": 620
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 0.2,
|
| 409 |
+
"learning_rate": 0.0002897669902912621,
|
| 410 |
+
"loss": 0.5089,
|
| 411 |
+
"step": 630
|
| 412 |
+
},
|
| 413 |
+
{
|
| 414 |
+
"epoch": 0.21,
|
| 415 |
+
"learning_rate": 0.00028957281553398056,
|
| 416 |
+
"loss": 0.5143,
|
| 417 |
+
"step": 640
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 0.21,
|
| 421 |
+
"learning_rate": 0.000289378640776699,
|
| 422 |
+
"loss": 0.5274,
|
| 423 |
+
"step": 650
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 0.21,
|
| 427 |
+
"learning_rate": 0.00028918446601941745,
|
| 428 |
+
"loss": 0.4704,
|
| 429 |
+
"step": 660
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.22,
|
| 433 |
+
"learning_rate": 0.0002889902912621359,
|
| 434 |
+
"loss": 0.4951,
|
| 435 |
+
"step": 670
|
| 436 |
+
},
|
| 437 |
+
{
|
| 438 |
+
"epoch": 0.22,
|
| 439 |
+
"learning_rate": 0.00028879611650485434,
|
| 440 |
+
"loss": 0.4949,
|
| 441 |
+
"step": 680
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 0.22,
|
| 445 |
+
"learning_rate": 0.0002886019417475728,
|
| 446 |
+
"loss": 0.465,
|
| 447 |
+
"step": 690
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 0.23,
|
| 451 |
+
"learning_rate": 0.00028840776699029123,
|
| 452 |
+
"loss": 0.4942,
|
| 453 |
+
"step": 700
|
| 454 |
+
},
|
| 455 |
+
{
|
| 456 |
+
"epoch": 0.23,
|
| 457 |
+
"learning_rate": 0.0002882135922330097,
|
| 458 |
+
"loss": 0.4861,
|
| 459 |
+
"step": 710
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 0.23,
|
| 463 |
+
"learning_rate": 0.0002880194174757281,
|
| 464 |
+
"loss": 0.4591,
|
| 465 |
+
"step": 720
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 0.23,
|
| 469 |
+
"learning_rate": 0.0002878252427184466,
|
| 470 |
+
"loss": 0.4983,
|
| 471 |
+
"step": 730
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 0.24,
|
| 475 |
+
"learning_rate": 0.000287631067961165,
|
| 476 |
+
"loss": 0.4883,
|
| 477 |
+
"step": 740
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 0.24,
|
| 481 |
+
"learning_rate": 0.00028743689320388343,
|
| 482 |
+
"loss": 0.485,
|
| 483 |
+
"step": 750
|
| 484 |
+
},
|
| 485 |
+
{
|
| 486 |
+
"epoch": 0.24,
|
| 487 |
+
"learning_rate": 0.0002872427184466019,
|
| 488 |
+
"loss": 0.4901,
|
| 489 |
+
"step": 760
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 0.25,
|
| 493 |
+
"learning_rate": 0.0002870485436893204,
|
| 494 |
+
"loss": 0.5018,
|
| 495 |
+
"step": 770
|
| 496 |
+
},
|
| 497 |
+
{
|
| 498 |
+
"epoch": 0.25,
|
| 499 |
+
"learning_rate": 0.0002868543689320388,
|
| 500 |
+
"loss": 0.4737,
|
| 501 |
+
"step": 780
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 0.25,
|
| 505 |
+
"learning_rate": 0.00028666019417475727,
|
| 506 |
+
"loss": 0.4885,
|
| 507 |
+
"step": 790
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 0.26,
|
| 511 |
+
"learning_rate": 0.0002864660194174757,
|
| 512 |
+
"loss": 0.4958,
|
| 513 |
+
"step": 800
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 0.26,
|
| 517 |
+
"eval_loss": 0.4947653114795685,
|
| 518 |
+
"eval_runtime": 537.1211,
|
| 519 |
+
"eval_samples_per_second": 3.724,
|
| 520 |
+
"eval_steps_per_second": 0.465,
|
| 521 |
+
"step": 800
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 0.26,
|
| 525 |
+
"learning_rate": 0.00028627184466019416,
|
| 526 |
+
"loss": 0.4811,
|
| 527 |
+
"step": 810
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 0.26,
|
| 531 |
+
"learning_rate": 0.0002860776699029126,
|
| 532 |
+
"loss": 0.5244,
|
| 533 |
+
"step": 820
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 0.27,
|
| 537 |
+
"learning_rate": 0.00028588349514563105,
|
| 538 |
+
"loss": 0.4735,
|
| 539 |
+
"step": 830
|
| 540 |
+
},
|
| 541 |
+
{
|
| 542 |
+
"epoch": 0.27,
|
| 543 |
+
"learning_rate": 0.00028568932038834947,
|
| 544 |
+
"loss": 0.5116,
|
| 545 |
+
"step": 840
|
| 546 |
+
},
|
| 547 |
+
{
|
| 548 |
+
"epoch": 0.27,
|
| 549 |
+
"learning_rate": 0.00028549514563106794,
|
| 550 |
+
"loss": 0.5015,
|
| 551 |
+
"step": 850
|
| 552 |
+
},
|
| 553 |
+
{
|
| 554 |
+
"epoch": 0.28,
|
| 555 |
+
"learning_rate": 0.0002853009708737864,
|
| 556 |
+
"loss": 0.4914,
|
| 557 |
+
"step": 860
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 0.28,
|
| 561 |
+
"learning_rate": 0.00028510679611650484,
|
| 562 |
+
"loss": 0.5336,
|
| 563 |
+
"step": 870
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 0.28,
|
| 567 |
+
"learning_rate": 0.0002849126213592233,
|
| 568 |
+
"loss": 0.4975,
|
| 569 |
+
"step": 880
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 0.29,
|
| 573 |
+
"learning_rate": 0.00028471844660194173,
|
| 574 |
+
"loss": 0.4693,
|
| 575 |
+
"step": 890
|
| 576 |
+
},
|
| 577 |
+
{
|
| 578 |
+
"epoch": 0.29,
|
| 579 |
+
"learning_rate": 0.00028452427184466015,
|
| 580 |
+
"loss": 0.4915,
|
| 581 |
+
"step": 900
|
| 582 |
+
},
|
| 583 |
+
{
|
| 584 |
+
"epoch": 0.29,
|
| 585 |
+
"learning_rate": 0.0002843300970873786,
|
| 586 |
+
"loss": 0.4595,
|
| 587 |
+
"step": 910
|
| 588 |
+
},
|
| 589 |
+
{
|
| 590 |
+
"epoch": 0.3,
|
| 591 |
+
"learning_rate": 0.00028413592233009704,
|
| 592 |
+
"loss": 0.5101,
|
| 593 |
+
"step": 920
|
| 594 |
+
},
|
| 595 |
+
{
|
| 596 |
+
"epoch": 0.3,
|
| 597 |
+
"learning_rate": 0.0002839417475728155,
|
| 598 |
+
"loss": 0.4835,
|
| 599 |
+
"step": 930
|
| 600 |
+
},
|
| 601 |
+
{
|
| 602 |
+
"epoch": 0.3,
|
| 603 |
+
"learning_rate": 0.00028374757281553393,
|
| 604 |
+
"loss": 0.4749,
|
| 605 |
+
"step": 940
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 0.31,
|
| 609 |
+
"learning_rate": 0.0002835533980582524,
|
| 610 |
+
"loss": 0.4943,
|
| 611 |
+
"step": 950
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 0.31,
|
| 615 |
+
"learning_rate": 0.0002833592233009709,
|
| 616 |
+
"loss": 0.5031,
|
| 617 |
+
"step": 960
|
| 618 |
+
},
|
| 619 |
+
{
|
| 620 |
+
"epoch": 0.31,
|
| 621 |
+
"learning_rate": 0.0002831650485436893,
|
| 622 |
+
"loss": 0.472,
|
| 623 |
+
"step": 970
|
| 624 |
+
},
|
| 625 |
+
{
|
| 626 |
+
"epoch": 0.32,
|
| 627 |
+
"learning_rate": 0.00028297087378640777,
|
| 628 |
+
"loss": 0.5117,
|
| 629 |
+
"step": 980
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 0.32,
|
| 633 |
+
"learning_rate": 0.0002827766990291262,
|
| 634 |
+
"loss": 0.4917,
|
| 635 |
+
"step": 990
|
| 636 |
+
},
|
| 637 |
+
{
|
| 638 |
+
"epoch": 0.32,
|
| 639 |
+
"learning_rate": 0.00028258252427184466,
|
| 640 |
+
"loss": 0.4764,
|
| 641 |
+
"step": 1000
|
| 642 |
+
},
|
| 643 |
+
{
|
| 644 |
+
"epoch": 0.32,
|
| 645 |
+
"eval_loss": 0.49110785126686096,
|
| 646 |
+
"eval_runtime": 534.8721,
|
| 647 |
+
"eval_samples_per_second": 3.739,
|
| 648 |
+
"eval_steps_per_second": 0.467,
|
| 649 |
+
"step": 1000
|
| 650 |
+
},
|
| 651 |
+
{
|
| 652 |
+
"epoch": 0.32,
|
| 653 |
+
"learning_rate": 0.0002823883495145631,
|
| 654 |
+
"loss": 0.4696,
|
| 655 |
+
"step": 1010
|
| 656 |
+
},
|
| 657 |
+
{
|
| 658 |
+
"epoch": 0.33,
|
| 659 |
+
"learning_rate": 0.00028219417475728155,
|
| 660 |
+
"loss": 0.4927,
|
| 661 |
+
"step": 1020
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 0.33,
|
| 665 |
+
"learning_rate": 0.00028199999999999997,
|
| 666 |
+
"loss": 0.4699,
|
| 667 |
+
"step": 1030
|
| 668 |
+
},
|
| 669 |
+
{
|
| 670 |
+
"epoch": 0.33,
|
| 671 |
+
"learning_rate": 0.0002818058252427184,
|
| 672 |
+
"loss": 0.4746,
|
| 673 |
+
"step": 1040
|
| 674 |
+
},
|
| 675 |
+
{
|
| 676 |
+
"epoch": 0.34,
|
| 677 |
+
"learning_rate": 0.00028161165048543686,
|
| 678 |
+
"loss": 0.4837,
|
| 679 |
+
"step": 1050
|
| 680 |
+
},
|
| 681 |
+
{
|
| 682 |
+
"epoch": 0.34,
|
| 683 |
+
"learning_rate": 0.0002814174757281553,
|
| 684 |
+
"loss": 0.4655,
|
| 685 |
+
"step": 1060
|
| 686 |
+
},
|
| 687 |
+
{
|
| 688 |
+
"epoch": 0.34,
|
| 689 |
+
"learning_rate": 0.00028122330097087375,
|
| 690 |
+
"loss": 0.4977,
|
| 691 |
+
"step": 1070
|
| 692 |
+
},
|
| 693 |
+
{
|
| 694 |
+
"epoch": 0.35,
|
| 695 |
+
"learning_rate": 0.0002810291262135922,
|
| 696 |
+
"loss": 0.4806,
|
| 697 |
+
"step": 1080
|
| 698 |
+
},
|
| 699 |
+
{
|
| 700 |
+
"epoch": 0.35,
|
| 701 |
+
"learning_rate": 0.00028083495145631064,
|
| 702 |
+
"loss": 0.4728,
|
| 703 |
+
"step": 1090
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 0.35,
|
| 707 |
+
"learning_rate": 0.0002806407766990291,
|
| 708 |
+
"loss": 0.4704,
|
| 709 |
+
"step": 1100
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"epoch": 0.36,
|
| 713 |
+
"learning_rate": 0.00028044660194174754,
|
| 714 |
+
"loss": 0.4583,
|
| 715 |
+
"step": 1110
|
| 716 |
+
},
|
| 717 |
+
{
|
| 718 |
+
"epoch": 0.36,
|
| 719 |
+
"learning_rate": 0.000280252427184466,
|
| 720 |
+
"loss": 0.4733,
|
| 721 |
+
"step": 1120
|
| 722 |
+
},
|
| 723 |
+
{
|
| 724 |
+
"epoch": 0.36,
|
| 725 |
+
"learning_rate": 0.00028005825242718443,
|
| 726 |
+
"loss": 0.4797,
|
| 727 |
+
"step": 1130
|
| 728 |
+
},
|
| 729 |
+
{
|
| 730 |
+
"epoch": 0.37,
|
| 731 |
+
"learning_rate": 0.0002798640776699029,
|
| 732 |
+
"loss": 0.4971,
|
| 733 |
+
"step": 1140
|
| 734 |
+
},
|
| 735 |
+
{
|
| 736 |
+
"epoch": 0.37,
|
| 737 |
+
"learning_rate": 0.0002796699029126213,
|
| 738 |
+
"loss": 0.4556,
|
| 739 |
+
"step": 1150
|
| 740 |
+
},
|
| 741 |
+
{
|
| 742 |
+
"epoch": 0.37,
|
| 743 |
+
"learning_rate": 0.0002794757281553398,
|
| 744 |
+
"loss": 0.4836,
|
| 745 |
+
"step": 1160
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 0.38,
|
| 749 |
+
"learning_rate": 0.00027928155339805827,
|
| 750 |
+
"loss": 0.4931,
|
| 751 |
+
"step": 1170
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 0.38,
|
| 755 |
+
"learning_rate": 0.0002790873786407767,
|
| 756 |
+
"loss": 0.4741,
|
| 757 |
+
"step": 1180
|
| 758 |
+
},
|
| 759 |
+
{
|
| 760 |
+
"epoch": 0.38,
|
| 761 |
+
"learning_rate": 0.0002788932038834951,
|
| 762 |
+
"loss": 0.4926,
|
| 763 |
+
"step": 1190
|
| 764 |
+
},
|
| 765 |
+
{
|
| 766 |
+
"epoch": 0.39,
|
| 767 |
+
"learning_rate": 0.0002786990291262136,
|
| 768 |
+
"loss": 0.4791,
|
| 769 |
+
"step": 1200
|
| 770 |
+
},
|
| 771 |
+
{
|
| 772 |
+
"epoch": 0.39,
|
| 773 |
+
"eval_loss": 0.48850855231285095,
|
| 774 |
+
"eval_runtime": 532.9556,
|
| 775 |
+
"eval_samples_per_second": 3.753,
|
| 776 |
+
"eval_steps_per_second": 0.469,
|
| 777 |
+
"step": 1200
|
| 778 |
+
},
|
| 779 |
+
{
|
| 780 |
+
"epoch": 0.39,
|
| 781 |
+
"learning_rate": 0.000278504854368932,
|
| 782 |
+
"loss": 0.4925,
|
| 783 |
+
"step": 1210
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 0.39,
|
| 787 |
+
"learning_rate": 0.00027831067961165047,
|
| 788 |
+
"loss": 0.4947,
|
| 789 |
+
"step": 1220
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 0.4,
|
| 793 |
+
"learning_rate": 0.0002781165048543689,
|
| 794 |
+
"loss": 0.4804,
|
| 795 |
+
"step": 1230
|
| 796 |
+
},
|
| 797 |
+
{
|
| 798 |
+
"epoch": 0.4,
|
| 799 |
+
"learning_rate": 0.00027792233009708736,
|
| 800 |
+
"loss": 0.5022,
|
| 801 |
+
"step": 1240
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 0.4,
|
| 805 |
+
"learning_rate": 0.0002777281553398058,
|
| 806 |
+
"loss": 0.5123,
|
| 807 |
+
"step": 1250
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 0.41,
|
| 811 |
+
"learning_rate": 0.00027753398058252425,
|
| 812 |
+
"loss": 0.5035,
|
| 813 |
+
"step": 1260
|
| 814 |
+
},
|
| 815 |
+
{
|
| 816 |
+
"epoch": 0.41,
|
| 817 |
+
"learning_rate": 0.0002773398058252427,
|
| 818 |
+
"loss": 0.4622,
|
| 819 |
+
"step": 1270
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 0.41,
|
| 823 |
+
"learning_rate": 0.00027714563106796114,
|
| 824 |
+
"loss": 0.4921,
|
| 825 |
+
"step": 1280
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 0.41,
|
| 829 |
+
"learning_rate": 0.0002769514563106796,
|
| 830 |
+
"loss": 0.4908,
|
| 831 |
+
"step": 1290
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 0.42,
|
| 835 |
+
"learning_rate": 0.00027675728155339803,
|
| 836 |
+
"loss": 0.4937,
|
| 837 |
+
"step": 1300
|
| 838 |
+
},
|
| 839 |
+
{
|
| 840 |
+
"epoch": 0.42,
|
| 841 |
+
"learning_rate": 0.0002765631067961165,
|
| 842 |
+
"loss": 0.4787,
|
| 843 |
+
"step": 1310
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 0.42,
|
| 847 |
+
"learning_rate": 0.0002763689320388349,
|
| 848 |
+
"loss": 0.4844,
|
| 849 |
+
"step": 1320
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 0.43,
|
| 853 |
+
"learning_rate": 0.00027617475728155334,
|
| 854 |
+
"loss": 0.4835,
|
| 855 |
+
"step": 1330
|
| 856 |
+
},
|
| 857 |
+
{
|
| 858 |
+
"epoch": 0.43,
|
| 859 |
+
"learning_rate": 0.0002759805825242718,
|
| 860 |
+
"loss": 0.4707,
|
| 861 |
+
"step": 1340
|
| 862 |
+
},
|
| 863 |
+
{
|
| 864 |
+
"epoch": 0.43,
|
| 865 |
+
"learning_rate": 0.00027578640776699024,
|
| 866 |
+
"loss": 0.4695,
|
| 867 |
+
"step": 1350
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 0.44,
|
| 871 |
+
"learning_rate": 0.0002755922330097087,
|
| 872 |
+
"loss": 0.4692,
|
| 873 |
+
"step": 1360
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 0.44,
|
| 877 |
+
"learning_rate": 0.0002753980582524272,
|
| 878 |
+
"loss": 0.4865,
|
| 879 |
+
"step": 1370
|
| 880 |
+
},
|
| 881 |
+
{
|
| 882 |
+
"epoch": 0.44,
|
| 883 |
+
"learning_rate": 0.0002752038834951456,
|
| 884 |
+
"loss": 0.4816,
|
| 885 |
+
"step": 1380
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 0.45,
|
| 889 |
+
"learning_rate": 0.0002750097087378641,
|
| 890 |
+
"loss": 0.4665,
|
| 891 |
+
"step": 1390
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 0.45,
|
| 895 |
+
"learning_rate": 0.0002748155339805825,
|
| 896 |
+
"loss": 0.5001,
|
| 897 |
+
"step": 1400
|
| 898 |
+
},
|
| 899 |
+
{
|
| 900 |
+
"epoch": 0.45,
|
| 901 |
+
"eval_loss": 0.4845764636993408,
|
| 902 |
+
"eval_runtime": 528.8954,
|
| 903 |
+
"eval_samples_per_second": 3.781,
|
| 904 |
+
"eval_steps_per_second": 0.473,
|
| 905 |
+
"step": 1400
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 0.45,
|
| 909 |
+
"learning_rate": 0.00027462135922330097,
|
| 910 |
+
"loss": 0.4748,
|
| 911 |
+
"step": 1410
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 0.46,
|
| 915 |
+
"learning_rate": 0.0002744271844660194,
|
| 916 |
+
"loss": 0.4855,
|
| 917 |
+
"step": 1420
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"epoch": 0.46,
|
| 921 |
+
"learning_rate": 0.00027423300970873786,
|
| 922 |
+
"loss": 0.4921,
|
| 923 |
+
"step": 1430
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 0.46,
|
| 927 |
+
"learning_rate": 0.0002740388349514563,
|
| 928 |
+
"loss": 0.4978,
|
| 929 |
+
"step": 1440
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 0.47,
|
| 933 |
+
"learning_rate": 0.00027384466019417475,
|
| 934 |
+
"loss": 0.4807,
|
| 935 |
+
"step": 1450
|
| 936 |
+
},
|
| 937 |
+
{
|
| 938 |
+
"epoch": 0.47,
|
| 939 |
+
"learning_rate": 0.0002736504854368932,
|
| 940 |
+
"loss": 0.49,
|
| 941 |
+
"step": 1460
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 0.47,
|
| 945 |
+
"learning_rate": 0.0002734563106796116,
|
| 946 |
+
"loss": 0.4812,
|
| 947 |
+
"step": 1470
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 0.48,
|
| 951 |
+
"learning_rate": 0.00027326213592233006,
|
| 952 |
+
"loss": 0.4823,
|
| 953 |
+
"step": 1480
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 0.48,
|
| 957 |
+
"learning_rate": 0.00027306796116504853,
|
| 958 |
+
"loss": 0.4581,
|
| 959 |
+
"step": 1490
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 0.48,
|
| 963 |
+
"learning_rate": 0.00027287378640776695,
|
| 964 |
+
"loss": 0.4975,
|
| 965 |
+
"step": 1500
|
| 966 |
+
},
|
| 967 |
+
{
|
| 968 |
+
"epoch": 0.49,
|
| 969 |
+
"learning_rate": 0.0002726796116504854,
|
| 970 |
+
"loss": 0.4859,
|
| 971 |
+
"step": 1510
|
| 972 |
+
},
|
| 973 |
+
{
|
| 974 |
+
"epoch": 0.49,
|
| 975 |
+
"learning_rate": 0.00027248543689320384,
|
| 976 |
+
"loss": 0.4661,
|
| 977 |
+
"step": 1520
|
| 978 |
+
},
|
| 979 |
+
{
|
| 980 |
+
"epoch": 0.49,
|
| 981 |
+
"learning_rate": 0.0002722912621359223,
|
| 982 |
+
"loss": 0.4699,
|
| 983 |
+
"step": 1530
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 0.5,
|
| 987 |
+
"learning_rate": 0.00027209708737864073,
|
| 988 |
+
"loss": 0.4768,
|
| 989 |
+
"step": 1540
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 0.5,
|
| 993 |
+
"learning_rate": 0.0002719029126213592,
|
| 994 |
+
"loss": 0.5059,
|
| 995 |
+
"step": 1550
|
| 996 |
+
},
|
| 997 |
+
{
|
| 998 |
+
"epoch": 0.5,
|
| 999 |
+
"learning_rate": 0.0002717087378640776,
|
| 1000 |
+
"loss": 0.49,
|
| 1001 |
+
"step": 1560
|
| 1002 |
+
},
|
| 1003 |
+
{
|
| 1004 |
+
"epoch": 0.5,
|
| 1005 |
+
"learning_rate": 0.0002715145631067961,
|
| 1006 |
+
"loss": 0.495,
|
| 1007 |
+
"step": 1570
|
| 1008 |
+
},
|
| 1009 |
+
{
|
| 1010 |
+
"epoch": 0.51,
|
| 1011 |
+
"learning_rate": 0.00027132038834951457,
|
| 1012 |
+
"loss": 0.4662,
|
| 1013 |
+
"step": 1580
|
| 1014 |
+
},
|
| 1015 |
+
{
|
| 1016 |
+
"epoch": 0.51,
|
| 1017 |
+
"learning_rate": 0.000271126213592233,
|
| 1018 |
+
"loss": 0.4845,
|
| 1019 |
+
"step": 1590
|
| 1020 |
+
},
|
| 1021 |
+
{
|
| 1022 |
+
"epoch": 0.51,
|
| 1023 |
+
"learning_rate": 0.00027093203883495146,
|
| 1024 |
+
"loss": 0.4706,
|
| 1025 |
+
"step": 1600
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 0.51,
|
| 1029 |
+
"eval_loss": 0.48333096504211426,
|
| 1030 |
+
"eval_runtime": 527.3194,
|
| 1031 |
+
"eval_samples_per_second": 3.793,
|
| 1032 |
+
"eval_steps_per_second": 0.474,
|
| 1033 |
+
"step": 1600
|
| 1034 |
+
},
|
| 1035 |
+
{
|
| 1036 |
+
"epoch": 0.52,
|
| 1037 |
+
"learning_rate": 0.0002707378640776699,
|
| 1038 |
+
"loss": 0.4668,
|
| 1039 |
+
"step": 1610
|
| 1040 |
+
},
|
| 1041 |
+
{
|
| 1042 |
+
"epoch": 0.52,
|
| 1043 |
+
"learning_rate": 0.0002705436893203883,
|
| 1044 |
+
"loss": 0.4956,
|
| 1045 |
+
"step": 1620
|
| 1046 |
+
},
|
| 1047 |
+
{
|
| 1048 |
+
"epoch": 0.52,
|
| 1049 |
+
"learning_rate": 0.0002703495145631068,
|
| 1050 |
+
"loss": 0.4624,
|
| 1051 |
+
"step": 1630
|
| 1052 |
+
},
|
| 1053 |
+
{
|
| 1054 |
+
"epoch": 0.53,
|
| 1055 |
+
"learning_rate": 0.0002701553398058252,
|
| 1056 |
+
"loss": 0.5262,
|
| 1057 |
+
"step": 1640
|
| 1058 |
+
},
|
| 1059 |
+
{
|
| 1060 |
+
"epoch": 0.53,
|
| 1061 |
+
"learning_rate": 0.00026996116504854367,
|
| 1062 |
+
"loss": 0.4741,
|
| 1063 |
+
"step": 1650
|
| 1064 |
+
},
|
| 1065 |
+
{
|
| 1066 |
+
"epoch": 0.53,
|
| 1067 |
+
"learning_rate": 0.0002697669902912621,
|
| 1068 |
+
"loss": 0.4764,
|
| 1069 |
+
"step": 1660
|
| 1070 |
+
},
|
| 1071 |
+
{
|
| 1072 |
+
"epoch": 0.54,
|
| 1073 |
+
"learning_rate": 0.00026957281553398056,
|
| 1074 |
+
"loss": 0.4791,
|
| 1075 |
+
"step": 1670
|
| 1076 |
+
},
|
| 1077 |
+
{
|
| 1078 |
+
"epoch": 0.54,
|
| 1079 |
+
"learning_rate": 0.00026937864077669903,
|
| 1080 |
+
"loss": 0.4867,
|
| 1081 |
+
"step": 1680
|
| 1082 |
+
},
|
| 1083 |
+
{
|
| 1084 |
+
"epoch": 0.54,
|
| 1085 |
+
"learning_rate": 0.00026918446601941745,
|
| 1086 |
+
"loss": 0.4743,
|
| 1087 |
+
"step": 1690
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 0.55,
|
| 1091 |
+
"learning_rate": 0.0002689902912621359,
|
| 1092 |
+
"loss": 0.4764,
|
| 1093 |
+
"step": 1700
|
| 1094 |
+
},
|
| 1095 |
+
{
|
| 1096 |
+
"epoch": 0.55,
|
| 1097 |
+
"learning_rate": 0.00026879611650485434,
|
| 1098 |
+
"loss": 0.4837,
|
| 1099 |
+
"step": 1710
|
| 1100 |
+
},
|
| 1101 |
+
{
|
| 1102 |
+
"epoch": 0.55,
|
| 1103 |
+
"learning_rate": 0.0002686019417475728,
|
| 1104 |
+
"loss": 0.4765,
|
| 1105 |
+
"step": 1720
|
| 1106 |
+
},
|
| 1107 |
+
{
|
| 1108 |
+
"epoch": 0.56,
|
| 1109 |
+
"learning_rate": 0.00026840776699029123,
|
| 1110 |
+
"loss": 0.4898,
|
| 1111 |
+
"step": 1730
|
| 1112 |
+
},
|
| 1113 |
+
{
|
| 1114 |
+
"epoch": 0.56,
|
| 1115 |
+
"learning_rate": 0.0002682135922330097,
|
| 1116 |
+
"loss": 0.4708,
|
| 1117 |
+
"step": 1740
|
| 1118 |
+
},
|
| 1119 |
+
{
|
| 1120 |
+
"epoch": 0.56,
|
| 1121 |
+
"learning_rate": 0.0002680194174757281,
|
| 1122 |
+
"loss": 0.5113,
|
| 1123 |
+
"step": 1750
|
| 1124 |
+
},
|
| 1125 |
+
{
|
| 1126 |
+
"epoch": 0.57,
|
| 1127 |
+
"learning_rate": 0.00026782524271844654,
|
| 1128 |
+
"loss": 0.4852,
|
| 1129 |
+
"step": 1760
|
| 1130 |
+
},
|
| 1131 |
+
{
|
| 1132 |
+
"epoch": 0.57,
|
| 1133 |
+
"learning_rate": 0.000267631067961165,
|
| 1134 |
+
"loss": 0.4643,
|
| 1135 |
+
"step": 1770
|
| 1136 |
+
},
|
| 1137 |
+
{
|
| 1138 |
+
"epoch": 0.57,
|
| 1139 |
+
"learning_rate": 0.0002674368932038835,
|
| 1140 |
+
"loss": 0.4666,
|
| 1141 |
+
"step": 1780
|
| 1142 |
+
},
|
| 1143 |
+
{
|
| 1144 |
+
"epoch": 0.58,
|
| 1145 |
+
"learning_rate": 0.0002672427184466019,
|
| 1146 |
+
"loss": 0.4927,
|
| 1147 |
+
"step": 1790
|
| 1148 |
+
},
|
| 1149 |
+
{
|
| 1150 |
+
"epoch": 0.58,
|
| 1151 |
+
"learning_rate": 0.0002670485436893204,
|
| 1152 |
+
"loss": 0.4778,
|
| 1153 |
+
"step": 1800
|
| 1154 |
+
},
|
| 1155 |
+
{
|
| 1156 |
+
"epoch": 0.58,
|
| 1157 |
+
"eval_loss": 0.4814455807209015,
|
| 1158 |
+
"eval_runtime": 528.5641,
|
| 1159 |
+
"eval_samples_per_second": 3.784,
|
| 1160 |
+
"eval_steps_per_second": 0.473,
|
| 1161 |
+
"step": 1800
|
| 1162 |
+
},
|
| 1163 |
+
{
|
| 1164 |
+
"epoch": 0.58,
|
| 1165 |
+
"learning_rate": 0.0002668543689320388,
|
| 1166 |
+
"loss": 0.4695,
|
| 1167 |
+
"step": 1810
|
| 1168 |
+
},
|
| 1169 |
+
{
|
| 1170 |
+
"epoch": 0.59,
|
| 1171 |
+
"learning_rate": 0.00026666019417475727,
|
| 1172 |
+
"loss": 0.468,
|
| 1173 |
+
"step": 1820
|
| 1174 |
+
},
|
| 1175 |
+
{
|
| 1176 |
+
"epoch": 0.59,
|
| 1177 |
+
"learning_rate": 0.0002664660194174757,
|
| 1178 |
+
"loss": 0.4492,
|
| 1179 |
+
"step": 1830
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 0.59,
|
| 1183 |
+
"learning_rate": 0.00026627184466019416,
|
| 1184 |
+
"loss": 0.4654,
|
| 1185 |
+
"step": 1840
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 0.59,
|
| 1189 |
+
"learning_rate": 0.0002660776699029126,
|
| 1190 |
+
"loss": 0.4909,
|
| 1191 |
+
"step": 1850
|
| 1192 |
+
},
|
| 1193 |
+
{
|
| 1194 |
+
"epoch": 0.6,
|
| 1195 |
+
"learning_rate": 0.00026588349514563106,
|
| 1196 |
+
"loss": 0.4786,
|
| 1197 |
+
"step": 1860
|
| 1198 |
+
},
|
| 1199 |
+
{
|
| 1200 |
+
"epoch": 0.6,
|
| 1201 |
+
"learning_rate": 0.00026568932038834953,
|
| 1202 |
+
"loss": 0.4633,
|
| 1203 |
+
"step": 1870
|
| 1204 |
+
},
|
| 1205 |
+
{
|
| 1206 |
+
"epoch": 0.6,
|
| 1207 |
+
"learning_rate": 0.00026549514563106795,
|
| 1208 |
+
"loss": 0.4826,
|
| 1209 |
+
"step": 1880
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"epoch": 0.61,
|
| 1213 |
+
"learning_rate": 0.0002653009708737864,
|
| 1214 |
+
"loss": 0.4644,
|
| 1215 |
+
"step": 1890
|
| 1216 |
+
},
|
| 1217 |
+
{
|
| 1218 |
+
"epoch": 0.61,
|
| 1219 |
+
"learning_rate": 0.00026510679611650484,
|
| 1220 |
+
"loss": 0.4573,
|
| 1221 |
+
"step": 1900
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 0.61,
|
| 1225 |
+
"learning_rate": 0.00026491262135922326,
|
| 1226 |
+
"loss": 0.4402,
|
| 1227 |
+
"step": 1910
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 0.62,
|
| 1231 |
+
"learning_rate": 0.00026471844660194173,
|
| 1232 |
+
"loss": 0.5006,
|
| 1233 |
+
"step": 1920
|
| 1234 |
+
},
|
| 1235 |
+
{
|
| 1236 |
+
"epoch": 0.62,
|
| 1237 |
+
"learning_rate": 0.00026452427184466015,
|
| 1238 |
+
"loss": 0.5296,
|
| 1239 |
+
"step": 1930
|
| 1240 |
+
},
|
| 1241 |
+
{
|
| 1242 |
+
"epoch": 0.62,
|
| 1243 |
+
"learning_rate": 0.0002643300970873786,
|
| 1244 |
+
"loss": 0.4689,
|
| 1245 |
+
"step": 1940
|
| 1246 |
+
},
|
| 1247 |
+
{
|
| 1248 |
+
"epoch": 0.63,
|
| 1249 |
+
"learning_rate": 0.00026413592233009704,
|
| 1250 |
+
"loss": 0.4728,
|
| 1251 |
+
"step": 1950
|
| 1252 |
+
},
|
| 1253 |
+
{
|
| 1254 |
+
"epoch": 0.63,
|
| 1255 |
+
"learning_rate": 0.0002639417475728155,
|
| 1256 |
+
"loss": 0.476,
|
| 1257 |
+
"step": 1960
|
| 1258 |
+
},
|
| 1259 |
+
{
|
| 1260 |
+
"epoch": 0.63,
|
| 1261 |
+
"learning_rate": 0.00026374757281553393,
|
| 1262 |
+
"loss": 0.4545,
|
| 1263 |
+
"step": 1970
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 0.64,
|
| 1267 |
+
"learning_rate": 0.0002635533980582524,
|
| 1268 |
+
"loss": 0.47,
|
| 1269 |
+
"step": 1980
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 0.64,
|
| 1273 |
+
"learning_rate": 0.0002633592233009709,
|
| 1274 |
+
"loss": 0.5014,
|
| 1275 |
+
"step": 1990
|
| 1276 |
+
},
|
| 1277 |
+
{
|
| 1278 |
+
"epoch": 0.64,
|
| 1279 |
+
"learning_rate": 0.0002631650485436893,
|
| 1280 |
+
"loss": 0.4726,
|
| 1281 |
+
"step": 2000
|
| 1282 |
+
},
|
| 1283 |
+
{
|
| 1284 |
+
"epoch": 0.64,
|
| 1285 |
+
"eval_loss": 0.4796612560749054,
|
| 1286 |
+
"eval_runtime": 521.7692,
|
| 1287 |
+
"eval_samples_per_second": 3.833,
|
| 1288 |
+
"eval_steps_per_second": 0.479,
|
| 1289 |
+
"step": 2000
|
| 1290 |
+
},
|
| 1291 |
+
{
|
| 1292 |
+
"epoch": 0.65,
|
| 1293 |
+
"learning_rate": 0.00026297087378640777,
|
| 1294 |
+
"loss": 0.4673,
|
| 1295 |
+
"step": 2010
|
| 1296 |
+
},
|
| 1297 |
+
{
|
| 1298 |
+
"epoch": 0.65,
|
| 1299 |
+
"learning_rate": 0.0002627766990291262,
|
| 1300 |
+
"loss": 0.4892,
|
| 1301 |
+
"step": 2020
|
| 1302 |
+
},
|
| 1303 |
+
{
|
| 1304 |
+
"epoch": 0.65,
|
| 1305 |
+
"learning_rate": 0.00026258252427184466,
|
| 1306 |
+
"loss": 0.47,
|
| 1307 |
+
"step": 2030
|
| 1308 |
+
},
|
| 1309 |
+
{
|
| 1310 |
+
"epoch": 0.66,
|
| 1311 |
+
"learning_rate": 0.0002623883495145631,
|
| 1312 |
+
"loss": 0.4532,
|
| 1313 |
+
"step": 2040
|
| 1314 |
+
},
|
| 1315 |
+
{
|
| 1316 |
+
"epoch": 0.66,
|
| 1317 |
+
"learning_rate": 0.0002621941747572815,
|
| 1318 |
+
"loss": 0.4949,
|
| 1319 |
+
"step": 2050
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 0.66,
|
| 1323 |
+
"learning_rate": 0.00026199999999999997,
|
| 1324 |
+
"loss": 0.493,
|
| 1325 |
+
"step": 2060
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 0.67,
|
| 1329 |
+
"learning_rate": 0.0002618058252427184,
|
| 1330 |
+
"loss": 0.4609,
|
| 1331 |
+
"step": 2070
|
| 1332 |
+
},
|
| 1333 |
+
{
|
| 1334 |
+
"epoch": 0.67,
|
| 1335 |
+
"learning_rate": 0.00026161165048543686,
|
| 1336 |
+
"loss": 0.5041,
|
| 1337 |
+
"step": 2080
|
| 1338 |
+
},
|
| 1339 |
+
{
|
| 1340 |
+
"epoch": 0.67,
|
| 1341 |
+
"learning_rate": 0.00026141747572815534,
|
| 1342 |
+
"loss": 0.4871,
|
| 1343 |
+
"step": 2090
|
| 1344 |
+
},
|
| 1345 |
+
{
|
| 1346 |
+
"epoch": 0.68,
|
| 1347 |
+
"learning_rate": 0.00026122330097087375,
|
| 1348 |
+
"loss": 0.4476,
|
| 1349 |
+
"step": 2100
|
| 1350 |
+
},
|
| 1351 |
+
{
|
| 1352 |
+
"epoch": 0.68,
|
| 1353 |
+
"learning_rate": 0.00026102912621359223,
|
| 1354 |
+
"loss": 0.4874,
|
| 1355 |
+
"step": 2110
|
| 1356 |
+
},
|
| 1357 |
+
{
|
| 1358 |
+
"epoch": 0.68,
|
| 1359 |
+
"learning_rate": 0.00026083495145631065,
|
| 1360 |
+
"loss": 0.478,
|
| 1361 |
+
"step": 2120
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 0.68,
|
| 1365 |
+
"learning_rate": 0.0002606407766990291,
|
| 1366 |
+
"loss": 0.5009,
|
| 1367 |
+
"step": 2130
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 0.69,
|
| 1371 |
+
"learning_rate": 0.00026044660194174754,
|
| 1372 |
+
"loss": 0.4905,
|
| 1373 |
+
"step": 2140
|
| 1374 |
+
},
|
| 1375 |
+
{
|
| 1376 |
+
"epoch": 0.69,
|
| 1377 |
+
"learning_rate": 0.000260252427184466,
|
| 1378 |
+
"loss": 0.4664,
|
| 1379 |
+
"step": 2150
|
| 1380 |
+
},
|
| 1381 |
+
{
|
| 1382 |
+
"epoch": 0.69,
|
| 1383 |
+
"learning_rate": 0.00026005825242718443,
|
| 1384 |
+
"loss": 0.4743,
|
| 1385 |
+
"step": 2160
|
| 1386 |
+
},
|
| 1387 |
+
{
|
| 1388 |
+
"epoch": 0.7,
|
| 1389 |
+
"learning_rate": 0.0002598640776699029,
|
| 1390 |
+
"loss": 0.4867,
|
| 1391 |
+
"step": 2170
|
| 1392 |
+
},
|
| 1393 |
+
{
|
| 1394 |
+
"epoch": 0.7,
|
| 1395 |
+
"learning_rate": 0.0002596699029126214,
|
| 1396 |
+
"loss": 0.4851,
|
| 1397 |
+
"step": 2180
|
| 1398 |
+
},
|
| 1399 |
+
{
|
| 1400 |
+
"epoch": 0.7,
|
| 1401 |
+
"learning_rate": 0.0002594757281553398,
|
| 1402 |
+
"loss": 0.4743,
|
| 1403 |
+
"step": 2190
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 0.71,
|
| 1407 |
+
"learning_rate": 0.0002592815533980582,
|
| 1408 |
+
"loss": 0.5097,
|
| 1409 |
+
"step": 2200
|
| 1410 |
+
},
|
| 1411 |
+
{
|
| 1412 |
+
"epoch": 0.71,
|
| 1413 |
+
"eval_loss": 0.47833874821662903,
|
| 1414 |
+
"eval_runtime": 519.0074,
|
| 1415 |
+
"eval_samples_per_second": 3.854,
|
| 1416 |
+
"eval_steps_per_second": 0.482,
|
| 1417 |
+
"step": 2200
|
| 1418 |
+
},
|
| 1419 |
+
{
|
| 1420 |
+
"epoch": 0.71,
|
| 1421 |
+
"learning_rate": 0.0002590873786407767,
|
| 1422 |
+
"loss": 0.4953,
|
| 1423 |
+
"step": 2210
|
| 1424 |
+
},
|
| 1425 |
+
{
|
| 1426 |
+
"epoch": 0.71,
|
| 1427 |
+
"learning_rate": 0.0002588932038834951,
|
| 1428 |
+
"loss": 0.4722,
|
| 1429 |
+
"step": 2220
|
| 1430 |
+
},
|
| 1431 |
+
{
|
| 1432 |
+
"epoch": 0.72,
|
| 1433 |
+
"learning_rate": 0.0002586990291262136,
|
| 1434 |
+
"loss": 0.4592,
|
| 1435 |
+
"step": 2230
|
| 1436 |
+
},
|
| 1437 |
+
{
|
| 1438 |
+
"epoch": 0.72,
|
| 1439 |
+
"learning_rate": 0.000258504854368932,
|
| 1440 |
+
"loss": 0.4625,
|
| 1441 |
+
"step": 2240
|
| 1442 |
+
},
|
| 1443 |
+
{
|
| 1444 |
+
"epoch": 0.72,
|
| 1445 |
+
"learning_rate": 0.00025831067961165047,
|
| 1446 |
+
"loss": 0.4712,
|
| 1447 |
+
"step": 2250
|
| 1448 |
+
},
|
| 1449 |
+
{
|
| 1450 |
+
"epoch": 0.73,
|
| 1451 |
+
"learning_rate": 0.0002581165048543689,
|
| 1452 |
+
"loss": 0.4728,
|
| 1453 |
+
"step": 2260
|
| 1454 |
+
},
|
| 1455 |
+
{
|
| 1456 |
+
"epoch": 0.73,
|
| 1457 |
+
"learning_rate": 0.00025792233009708736,
|
| 1458 |
+
"loss": 0.4849,
|
| 1459 |
+
"step": 2270
|
| 1460 |
+
},
|
| 1461 |
+
{
|
| 1462 |
+
"epoch": 0.73,
|
| 1463 |
+
"learning_rate": 0.00025772815533980583,
|
| 1464 |
+
"loss": 0.4907,
|
| 1465 |
+
"step": 2280
|
| 1466 |
+
},
|
| 1467 |
+
{
|
| 1468 |
+
"epoch": 0.74,
|
| 1469 |
+
"learning_rate": 0.00025753398058252425,
|
| 1470 |
+
"loss": 0.4499,
|
| 1471 |
+
"step": 2290
|
| 1472 |
+
},
|
| 1473 |
+
{
|
| 1474 |
+
"epoch": 0.74,
|
| 1475 |
+
"learning_rate": 0.0002573398058252427,
|
| 1476 |
+
"loss": 0.4648,
|
| 1477 |
+
"step": 2300
|
| 1478 |
+
},
|
| 1479 |
+
{
|
| 1480 |
+
"epoch": 0.74,
|
| 1481 |
+
"learning_rate": 0.00025714563106796114,
|
| 1482 |
+
"loss": 0.4616,
|
| 1483 |
+
"step": 2310
|
| 1484 |
+
},
|
| 1485 |
+
{
|
| 1486 |
+
"epoch": 0.75,
|
| 1487 |
+
"learning_rate": 0.0002569514563106796,
|
| 1488 |
+
"loss": 0.4872,
|
| 1489 |
+
"step": 2320
|
| 1490 |
+
},
|
| 1491 |
+
{
|
| 1492 |
+
"epoch": 0.75,
|
| 1493 |
+
"learning_rate": 0.00025675728155339804,
|
| 1494 |
+
"loss": 0.4986,
|
| 1495 |
+
"step": 2330
|
| 1496 |
+
},
|
| 1497 |
+
{
|
| 1498 |
+
"epoch": 0.75,
|
| 1499 |
+
"learning_rate": 0.00025656310679611645,
|
| 1500 |
+
"loss": 0.4899,
|
| 1501 |
+
"step": 2340
|
| 1502 |
+
},
|
| 1503 |
+
{
|
| 1504 |
+
"epoch": 0.76,
|
| 1505 |
+
"learning_rate": 0.00025636893203883493,
|
| 1506 |
+
"loss": 0.4848,
|
| 1507 |
+
"step": 2350
|
| 1508 |
+
},
|
| 1509 |
+
{
|
| 1510 |
+
"epoch": 0.76,
|
| 1511 |
+
"learning_rate": 0.00025617475728155335,
|
| 1512 |
+
"loss": 0.4586,
|
| 1513 |
+
"step": 2360
|
| 1514 |
+
},
|
| 1515 |
+
{
|
| 1516 |
+
"epoch": 0.76,
|
| 1517 |
+
"learning_rate": 0.0002559805825242718,
|
| 1518 |
+
"loss": 0.4641,
|
| 1519 |
+
"step": 2370
|
| 1520 |
+
},
|
| 1521 |
+
{
|
| 1522 |
+
"epoch": 0.77,
|
| 1523 |
+
"learning_rate": 0.00025578640776699024,
|
| 1524 |
+
"loss": 0.4726,
|
| 1525 |
+
"step": 2380
|
| 1526 |
+
},
|
| 1527 |
+
{
|
| 1528 |
+
"epoch": 0.77,
|
| 1529 |
+
"learning_rate": 0.0002555922330097087,
|
| 1530 |
+
"loss": 0.4858,
|
| 1531 |
+
"step": 2390
|
| 1532 |
+
},
|
| 1533 |
+
{
|
| 1534 |
+
"epoch": 0.77,
|
| 1535 |
+
"learning_rate": 0.0002553980582524272,
|
| 1536 |
+
"loss": 0.4631,
|
| 1537 |
+
"step": 2400
|
| 1538 |
+
},
|
| 1539 |
+
{
|
| 1540 |
+
"epoch": 0.77,
|
| 1541 |
+
"eval_loss": 0.47692033648490906,
|
| 1542 |
+
"eval_runtime": 523.0687,
|
| 1543 |
+
"eval_samples_per_second": 3.824,
|
| 1544 |
+
"eval_steps_per_second": 0.478,
|
| 1545 |
+
"step": 2400
|
| 1546 |
+
},
|
| 1547 |
+
{
|
| 1548 |
+
"epoch": 0.77,
|
| 1549 |
+
"learning_rate": 0.0002552038834951456,
|
| 1550 |
+
"loss": 0.4537,
|
| 1551 |
+
"step": 2410
|
| 1552 |
+
},
|
| 1553 |
+
{
|
| 1554 |
+
"epoch": 0.78,
|
| 1555 |
+
"learning_rate": 0.0002550097087378641,
|
| 1556 |
+
"loss": 0.5043,
|
| 1557 |
+
"step": 2420
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 0.78,
|
| 1561 |
+
"learning_rate": 0.0002548155339805825,
|
| 1562 |
+
"loss": 0.4889,
|
| 1563 |
+
"step": 2430
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 0.78,
|
| 1567 |
+
"learning_rate": 0.00025462135922330097,
|
| 1568 |
+
"loss": 0.4713,
|
| 1569 |
+
"step": 2440
|
| 1570 |
+
},
|
| 1571 |
+
{
|
| 1572 |
+
"epoch": 0.79,
|
| 1573 |
+
"learning_rate": 0.0002544271844660194,
|
| 1574 |
+
"loss": 0.4593,
|
| 1575 |
+
"step": 2450
|
| 1576 |
+
},
|
| 1577 |
+
{
|
| 1578 |
+
"epoch": 0.79,
|
| 1579 |
+
"learning_rate": 0.00025423300970873786,
|
| 1580 |
+
"loss": 0.4617,
|
| 1581 |
+
"step": 2460
|
| 1582 |
+
},
|
| 1583 |
+
{
|
| 1584 |
+
"epoch": 0.79,
|
| 1585 |
+
"learning_rate": 0.0002540388349514563,
|
| 1586 |
+
"loss": 0.4587,
|
| 1587 |
+
"step": 2470
|
| 1588 |
+
},
|
| 1589 |
+
{
|
| 1590 |
+
"epoch": 0.8,
|
| 1591 |
+
"learning_rate": 0.0002538446601941747,
|
| 1592 |
+
"loss": 0.4471,
|
| 1593 |
+
"step": 2480
|
| 1594 |
+
},
|
| 1595 |
+
{
|
| 1596 |
+
"epoch": 0.8,
|
| 1597 |
+
"learning_rate": 0.00025365048543689317,
|
| 1598 |
+
"loss": 0.4732,
|
| 1599 |
+
"step": 2490
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 0.8,
|
| 1603 |
+
"learning_rate": 0.00025345631067961164,
|
| 1604 |
+
"loss": 0.4999,
|
| 1605 |
+
"step": 2500
|
| 1606 |
+
},
|
| 1607 |
+
{
|
| 1608 |
+
"epoch": 0.81,
|
| 1609 |
+
"learning_rate": 0.00025326213592233006,
|
| 1610 |
+
"loss": 0.4736,
|
| 1611 |
+
"step": 2510
|
| 1612 |
+
},
|
| 1613 |
+
{
|
| 1614 |
+
"epoch": 0.81,
|
| 1615 |
+
"learning_rate": 0.00025306796116504853,
|
| 1616 |
+
"loss": 0.4893,
|
| 1617 |
+
"step": 2520
|
| 1618 |
+
},
|
| 1619 |
+
{
|
| 1620 |
+
"epoch": 0.81,
|
| 1621 |
+
"learning_rate": 0.00025287378640776695,
|
| 1622 |
+
"loss": 0.476,
|
| 1623 |
+
"step": 2530
|
| 1624 |
+
},
|
| 1625 |
+
{
|
| 1626 |
+
"epoch": 0.82,
|
| 1627 |
+
"learning_rate": 0.0002526796116504854,
|
| 1628 |
+
"loss": 0.4713,
|
| 1629 |
+
"step": 2540
|
| 1630 |
+
},
|
| 1631 |
+
{
|
| 1632 |
+
"epoch": 0.82,
|
| 1633 |
+
"learning_rate": 0.00025248543689320384,
|
| 1634 |
+
"loss": 0.4982,
|
| 1635 |
+
"step": 2550
|
| 1636 |
+
},
|
| 1637 |
+
{
|
| 1638 |
+
"epoch": 0.82,
|
| 1639 |
+
"learning_rate": 0.0002522912621359223,
|
| 1640 |
+
"loss": 0.4694,
|
| 1641 |
+
"step": 2560
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 0.83,
|
| 1645 |
+
"learning_rate": 0.00025209708737864074,
|
| 1646 |
+
"loss": 0.4822,
|
| 1647 |
+
"step": 2570
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"epoch": 0.83,
|
| 1651 |
+
"learning_rate": 0.0002519029126213592,
|
| 1652 |
+
"loss": 0.4657,
|
| 1653 |
+
"step": 2580
|
| 1654 |
+
},
|
| 1655 |
+
{
|
| 1656 |
+
"epoch": 0.83,
|
| 1657 |
+
"learning_rate": 0.0002517087378640777,
|
| 1658 |
+
"loss": 0.4796,
|
| 1659 |
+
"step": 2590
|
| 1660 |
+
},
|
| 1661 |
+
{
|
| 1662 |
+
"epoch": 0.84,
|
| 1663 |
+
"learning_rate": 0.0002515145631067961,
|
| 1664 |
+
"loss": 0.4606,
|
| 1665 |
+
"step": 2600
|
| 1666 |
+
},
|
| 1667 |
+
{
|
| 1668 |
+
"epoch": 0.84,
|
| 1669 |
+
"eval_loss": 0.4760785698890686,
|
| 1670 |
+
"eval_runtime": 515.4107,
|
| 1671 |
+
"eval_samples_per_second": 3.88,
|
| 1672 |
+
"eval_steps_per_second": 0.485,
|
| 1673 |
+
"step": 2600
|
| 1674 |
+
},
|
| 1675 |
+
{
|
| 1676 |
+
"epoch": 0.84,
|
| 1677 |
+
"learning_rate": 0.0002513203883495146,
|
| 1678 |
+
"loss": 0.4735,
|
| 1679 |
+
"step": 2610
|
| 1680 |
+
},
|
| 1681 |
+
{
|
| 1682 |
+
"epoch": 0.84,
|
| 1683 |
+
"learning_rate": 0.000251126213592233,
|
| 1684 |
+
"loss": 0.4597,
|
| 1685 |
+
"step": 2620
|
| 1686 |
+
},
|
| 1687 |
+
{
|
| 1688 |
+
"epoch": 0.85,
|
| 1689 |
+
"learning_rate": 0.0002509320388349514,
|
| 1690 |
+
"loss": 0.4722,
|
| 1691 |
+
"step": 2630
|
| 1692 |
+
},
|
| 1693 |
+
{
|
| 1694 |
+
"epoch": 0.85,
|
| 1695 |
+
"learning_rate": 0.0002507378640776699,
|
| 1696 |
+
"loss": 0.4378,
|
| 1697 |
+
"step": 2640
|
| 1698 |
+
},
|
| 1699 |
+
{
|
| 1700 |
+
"epoch": 0.85,
|
| 1701 |
+
"learning_rate": 0.0002505436893203883,
|
| 1702 |
+
"loss": 0.438,
|
| 1703 |
+
"step": 2650
|
| 1704 |
+
},
|
| 1705 |
+
{
|
| 1706 |
+
"epoch": 0.86,
|
| 1707 |
+
"learning_rate": 0.0002503495145631068,
|
| 1708 |
+
"loss": 0.4754,
|
| 1709 |
+
"step": 2660
|
| 1710 |
+
},
|
| 1711 |
+
{
|
| 1712 |
+
"epoch": 0.86,
|
| 1713 |
+
"learning_rate": 0.0002501553398058252,
|
| 1714 |
+
"loss": 0.4753,
|
| 1715 |
+
"step": 2670
|
| 1716 |
+
},
|
| 1717 |
+
{
|
| 1718 |
+
"epoch": 0.86,
|
| 1719 |
+
"learning_rate": 0.00024996116504854367,
|
| 1720 |
+
"loss": 0.4605,
|
| 1721 |
+
"step": 2680
|
| 1722 |
+
},
|
| 1723 |
+
{
|
| 1724 |
+
"epoch": 0.86,
|
| 1725 |
+
"learning_rate": 0.00024976699029126214,
|
| 1726 |
+
"loss": 0.4981,
|
| 1727 |
+
"step": 2690
|
| 1728 |
+
},
|
| 1729 |
+
{
|
| 1730 |
+
"epoch": 0.87,
|
| 1731 |
+
"learning_rate": 0.00024957281553398056,
|
| 1732 |
+
"loss": 0.4848,
|
| 1733 |
+
"step": 2700
|
| 1734 |
+
},
|
| 1735 |
+
{
|
| 1736 |
+
"epoch": 0.87,
|
| 1737 |
+
"learning_rate": 0.00024937864077669903,
|
| 1738 |
+
"loss": 0.4493,
|
| 1739 |
+
"step": 2710
|
| 1740 |
+
},
|
| 1741 |
+
{
|
| 1742 |
+
"epoch": 0.87,
|
| 1743 |
+
"learning_rate": 0.00024918446601941745,
|
| 1744 |
+
"loss": 0.4865,
|
| 1745 |
+
"step": 2720
|
| 1746 |
+
},
|
| 1747 |
+
{
|
| 1748 |
+
"epoch": 0.88,
|
| 1749 |
+
"learning_rate": 0.0002489902912621359,
|
| 1750 |
+
"loss": 0.4728,
|
| 1751 |
+
"step": 2730
|
| 1752 |
+
},
|
| 1753 |
+
{
|
| 1754 |
+
"epoch": 0.88,
|
| 1755 |
+
"learning_rate": 0.00024879611650485434,
|
| 1756 |
+
"loss": 0.4769,
|
| 1757 |
+
"step": 2740
|
| 1758 |
+
},
|
| 1759 |
+
{
|
| 1760 |
+
"epoch": 0.88,
|
| 1761 |
+
"learning_rate": 0.0002486019417475728,
|
| 1762 |
+
"loss": 0.4817,
|
| 1763 |
+
"step": 2750
|
| 1764 |
+
},
|
| 1765 |
+
{
|
| 1766 |
+
"epoch": 0.89,
|
| 1767 |
+
"learning_rate": 0.00024840776699029123,
|
| 1768 |
+
"loss": 0.4685,
|
| 1769 |
+
"step": 2760
|
| 1770 |
+
},
|
| 1771 |
+
{
|
| 1772 |
+
"epoch": 0.89,
|
| 1773 |
+
"learning_rate": 0.00024821359223300965,
|
| 1774 |
+
"loss": 0.4718,
|
| 1775 |
+
"step": 2770
|
| 1776 |
+
},
|
| 1777 |
+
{
|
| 1778 |
+
"epoch": 0.89,
|
| 1779 |
+
"learning_rate": 0.0002480194174757281,
|
| 1780 |
+
"loss": 0.4627,
|
| 1781 |
+
"step": 2780
|
| 1782 |
+
},
|
| 1783 |
+
{
|
| 1784 |
+
"epoch": 0.9,
|
| 1785 |
+
"learning_rate": 0.00024782524271844654,
|
| 1786 |
+
"loss": 0.4779,
|
| 1787 |
+
"step": 2790
|
| 1788 |
+
},
|
| 1789 |
+
{
|
| 1790 |
+
"epoch": 0.9,
|
| 1791 |
+
"learning_rate": 0.000247631067961165,
|
| 1792 |
+
"loss": 0.4673,
|
| 1793 |
+
"step": 2800
|
| 1794 |
+
},
|
| 1795 |
+
{
|
| 1796 |
+
"epoch": 0.9,
|
| 1797 |
+
"eval_loss": 0.47458600997924805,
|
| 1798 |
+
"eval_runtime": 513.2712,
|
| 1799 |
+
"eval_samples_per_second": 3.897,
|
| 1800 |
+
"eval_steps_per_second": 0.487,
|
| 1801 |
+
"step": 2800
|
| 1802 |
+
},
|
| 1803 |
+
{
|
| 1804 |
+
"epoch": 0.9,
|
| 1805 |
+
"learning_rate": 0.0002474368932038835,
|
| 1806 |
+
"loss": 0.4714,
|
| 1807 |
+
"step": 2810
|
| 1808 |
+
},
|
| 1809 |
+
{
|
| 1810 |
+
"epoch": 0.91,
|
| 1811 |
+
"learning_rate": 0.0002472427184466019,
|
| 1812 |
+
"loss": 0.5071,
|
| 1813 |
+
"step": 2820
|
| 1814 |
+
},
|
| 1815 |
+
{
|
| 1816 |
+
"epoch": 0.91,
|
| 1817 |
+
"learning_rate": 0.0002470485436893204,
|
| 1818 |
+
"loss": 0.478,
|
| 1819 |
+
"step": 2830
|
| 1820 |
+
},
|
| 1821 |
+
{
|
| 1822 |
+
"epoch": 0.91,
|
| 1823 |
+
"learning_rate": 0.0002468543689320388,
|
| 1824 |
+
"loss": 0.4769,
|
| 1825 |
+
"step": 2840
|
| 1826 |
+
},
|
| 1827 |
+
{
|
| 1828 |
+
"epoch": 0.92,
|
| 1829 |
+
"learning_rate": 0.0002466601941747573,
|
| 1830 |
+
"loss": 0.4871,
|
| 1831 |
+
"step": 2850
|
| 1832 |
+
},
|
| 1833 |
+
{
|
| 1834 |
+
"epoch": 0.92,
|
| 1835 |
+
"learning_rate": 0.0002464660194174757,
|
| 1836 |
+
"loss": 0.5094,
|
| 1837 |
+
"step": 2860
|
| 1838 |
+
},
|
| 1839 |
+
{
|
| 1840 |
+
"epoch": 0.92,
|
| 1841 |
+
"learning_rate": 0.00024627184466019417,
|
| 1842 |
+
"loss": 0.4551,
|
| 1843 |
+
"step": 2870
|
| 1844 |
+
},
|
| 1845 |
+
{
|
| 1846 |
+
"epoch": 0.93,
|
| 1847 |
+
"learning_rate": 0.0002460776699029126,
|
| 1848 |
+
"loss": 0.4891,
|
| 1849 |
+
"step": 2880
|
| 1850 |
+
},
|
| 1851 |
+
{
|
| 1852 |
+
"epoch": 0.93,
|
| 1853 |
+
"learning_rate": 0.00024588349514563106,
|
| 1854 |
+
"loss": 0.4588,
|
| 1855 |
+
"step": 2890
|
| 1856 |
+
},
|
| 1857 |
+
{
|
| 1858 |
+
"epoch": 0.93,
|
| 1859 |
+
"learning_rate": 0.00024568932038834953,
|
| 1860 |
+
"loss": 0.4609,
|
| 1861 |
+
"step": 2900
|
| 1862 |
+
},
|
| 1863 |
+
{
|
| 1864 |
+
"epoch": 0.94,
|
| 1865 |
+
"learning_rate": 0.00024549514563106795,
|
| 1866 |
+
"loss": 0.4792,
|
| 1867 |
+
"step": 2910
|
| 1868 |
+
},
|
| 1869 |
+
{
|
| 1870 |
+
"epoch": 0.94,
|
| 1871 |
+
"learning_rate": 0.00024530097087378637,
|
| 1872 |
+
"loss": 0.4758,
|
| 1873 |
+
"step": 2920
|
| 1874 |
+
},
|
| 1875 |
+
{
|
| 1876 |
+
"epoch": 0.94,
|
| 1877 |
+
"learning_rate": 0.00024510679611650484,
|
| 1878 |
+
"loss": 0.5096,
|
| 1879 |
+
"step": 2930
|
| 1880 |
+
},
|
| 1881 |
+
{
|
| 1882 |
+
"epoch": 0.95,
|
| 1883 |
+
"learning_rate": 0.00024491262135922326,
|
| 1884 |
+
"loss": 0.4798,
|
| 1885 |
+
"step": 2940
|
| 1886 |
+
},
|
| 1887 |
+
{
|
| 1888 |
+
"epoch": 0.95,
|
| 1889 |
+
"learning_rate": 0.00024471844660194173,
|
| 1890 |
+
"loss": 0.4777,
|
| 1891 |
+
"step": 2950
|
| 1892 |
+
},
|
| 1893 |
+
{
|
| 1894 |
+
"epoch": 0.95,
|
| 1895 |
+
"learning_rate": 0.00024452427184466015,
|
| 1896 |
+
"loss": 0.4827,
|
| 1897 |
+
"step": 2960
|
| 1898 |
+
},
|
| 1899 |
+
{
|
| 1900 |
+
"epoch": 0.95,
|
| 1901 |
+
"learning_rate": 0.0002443300970873786,
|
| 1902 |
+
"loss": 0.4713,
|
| 1903 |
+
"step": 2970
|
| 1904 |
+
},
|
| 1905 |
+
{
|
| 1906 |
+
"epoch": 0.96,
|
| 1907 |
+
"learning_rate": 0.00024413592233009707,
|
| 1908 |
+
"loss": 0.4716,
|
| 1909 |
+
"step": 2980
|
| 1910 |
+
},
|
| 1911 |
+
{
|
| 1912 |
+
"epoch": 0.96,
|
| 1913 |
+
"learning_rate": 0.00024394174757281552,
|
| 1914 |
+
"loss": 0.4644,
|
| 1915 |
+
"step": 2990
|
| 1916 |
+
},
|
| 1917 |
+
{
|
| 1918 |
+
"epoch": 0.96,
|
| 1919 |
+
"learning_rate": 0.00024374757281553396,
|
| 1920 |
+
"loss": 0.4695,
|
| 1921 |
+
"step": 3000
|
| 1922 |
+
},
|
| 1923 |
+
{
|
| 1924 |
+
"epoch": 0.96,
|
| 1925 |
+
"eval_loss": 0.4740288257598877,
|
| 1926 |
+
"eval_runtime": 514.5469,
|
| 1927 |
+
"eval_samples_per_second": 3.887,
|
| 1928 |
+
"eval_steps_per_second": 0.486,
|
| 1929 |
+
"step": 3000
|
| 1930 |
+
},
|
| 1931 |
+
{
|
| 1932 |
+
"epoch": 0.97,
|
| 1933 |
+
"learning_rate": 0.0002435533980582524,
|
| 1934 |
+
"loss": 0.4763,
|
| 1935 |
+
"step": 3010
|
| 1936 |
+
},
|
| 1937 |
+
{
|
| 1938 |
+
"epoch": 0.97,
|
| 1939 |
+
"learning_rate": 0.00024335922330097085,
|
| 1940 |
+
"loss": 0.4761,
|
| 1941 |
+
"step": 3020
|
| 1942 |
+
},
|
| 1943 |
+
{
|
| 1944 |
+
"epoch": 0.97,
|
| 1945 |
+
"learning_rate": 0.0002431650485436893,
|
| 1946 |
+
"loss": 0.4926,
|
| 1947 |
+
"step": 3030
|
| 1948 |
+
},
|
| 1949 |
+
{
|
| 1950 |
+
"epoch": 0.98,
|
| 1951 |
+
"learning_rate": 0.00024297087378640777,
|
| 1952 |
+
"loss": 0.4571,
|
| 1953 |
+
"step": 3040
|
| 1954 |
+
},
|
| 1955 |
+
{
|
| 1956 |
+
"epoch": 0.98,
|
| 1957 |
+
"learning_rate": 0.00024277669902912616,
|
| 1958 |
+
"loss": 0.4774,
|
| 1959 |
+
"step": 3050
|
| 1960 |
+
},
|
| 1961 |
+
{
|
| 1962 |
+
"epoch": 0.98,
|
| 1963 |
+
"learning_rate": 0.00024258252427184464,
|
| 1964 |
+
"loss": 0.4651,
|
| 1965 |
+
"step": 3060
|
| 1966 |
+
},
|
| 1967 |
+
{
|
| 1968 |
+
"epoch": 0.99,
|
| 1969 |
+
"learning_rate": 0.00024238834951456308,
|
| 1970 |
+
"loss": 0.4816,
|
| 1971 |
+
"step": 3070
|
| 1972 |
+
},
|
| 1973 |
+
{
|
| 1974 |
+
"epoch": 0.99,
|
| 1975 |
+
"learning_rate": 0.00024219417475728153,
|
| 1976 |
+
"loss": 0.4713,
|
| 1977 |
+
"step": 3080
|
| 1978 |
+
},
|
| 1979 |
+
{
|
| 1980 |
+
"epoch": 0.99,
|
| 1981 |
+
"learning_rate": 0.00024199999999999997,
|
| 1982 |
+
"loss": 0.503,
|
| 1983 |
+
"step": 3090
|
| 1984 |
+
},
|
| 1985 |
+
{
|
| 1986 |
+
"epoch": 1.0,
|
| 1987 |
+
"learning_rate": 0.00024180582524271842,
|
| 1988 |
+
"loss": 0.5096,
|
| 1989 |
+
"step": 3100
|
| 1990 |
+
},
|
| 1991 |
+
{
|
| 1992 |
+
"epoch": 1.0,
|
| 1993 |
+
"learning_rate": 0.00024161165048543687,
|
| 1994 |
+
"loss": 0.4786,
|
| 1995 |
+
"step": 3110
|
| 1996 |
+
},
|
| 1997 |
+
{
|
| 1998 |
+
"epoch": 1.0,
|
| 1999 |
+
"learning_rate": 0.0002414174757281553,
|
| 2000 |
+
"loss": 0.4194,
|
| 2001 |
+
"step": 3120
|
| 2002 |
+
},
|
| 2003 |
+
{
|
| 2004 |
+
"epoch": 1.01,
|
| 2005 |
+
"learning_rate": 0.00024122330097087376,
|
| 2006 |
+
"loss": 0.4373,
|
| 2007 |
+
"step": 3130
|
| 2008 |
+
},
|
| 2009 |
+
{
|
| 2010 |
+
"epoch": 1.01,
|
| 2011 |
+
"learning_rate": 0.00024102912621359223,
|
| 2012 |
+
"loss": 0.4531,
|
| 2013 |
+
"step": 3140
|
| 2014 |
+
},
|
| 2015 |
+
{
|
| 2016 |
+
"epoch": 1.01,
|
| 2017 |
+
"learning_rate": 0.00024083495145631068,
|
| 2018 |
+
"loss": 0.4706,
|
| 2019 |
+
"step": 3150
|
| 2020 |
+
},
|
| 2021 |
+
{
|
| 2022 |
+
"epoch": 1.02,
|
| 2023 |
+
"learning_rate": 0.00024064077669902912,
|
| 2024 |
+
"loss": 0.4283,
|
| 2025 |
+
"step": 3160
|
| 2026 |
+
},
|
| 2027 |
+
{
|
| 2028 |
+
"epoch": 1.02,
|
| 2029 |
+
"learning_rate": 0.00024044660194174757,
|
| 2030 |
+
"loss": 0.4589,
|
| 2031 |
+
"step": 3170
|
| 2032 |
+
},
|
| 2033 |
+
{
|
| 2034 |
+
"epoch": 1.02,
|
| 2035 |
+
"learning_rate": 0.000240252427184466,
|
| 2036 |
+
"loss": 0.4506,
|
| 2037 |
+
"step": 3180
|
| 2038 |
+
},
|
| 2039 |
+
{
|
| 2040 |
+
"epoch": 1.03,
|
| 2041 |
+
"learning_rate": 0.00024005825242718446,
|
| 2042 |
+
"loss": 0.4305,
|
| 2043 |
+
"step": 3190
|
| 2044 |
+
},
|
| 2045 |
+
{
|
| 2046 |
+
"epoch": 1.03,
|
| 2047 |
+
"learning_rate": 0.00023986407766990288,
|
| 2048 |
+
"loss": 0.451,
|
| 2049 |
+
"step": 3200
|
| 2050 |
+
},
|
| 2051 |
+
{
|
| 2052 |
+
"epoch": 1.03,
|
| 2053 |
+
"eval_loss": 0.4739780128002167,
|
| 2054 |
+
"eval_runtime": 520.0007,
|
| 2055 |
+
"eval_samples_per_second": 3.846,
|
| 2056 |
+
"eval_steps_per_second": 0.481,
|
| 2057 |
+
"step": 3200
|
| 2058 |
+
},
|
| 2059 |
+
{
|
| 2060 |
+
"epoch": 1.03,
|
| 2061 |
+
"learning_rate": 0.00023966990291262132,
|
| 2062 |
+
"loss": 0.4354,
|
| 2063 |
+
"step": 3210
|
| 2064 |
+
},
|
| 2065 |
+
{
|
| 2066 |
+
"epoch": 1.04,
|
| 2067 |
+
"learning_rate": 0.00023947572815533977,
|
| 2068 |
+
"loss": 0.4396,
|
| 2069 |
+
"step": 3220
|
| 2070 |
+
},
|
| 2071 |
+
{
|
| 2072 |
+
"epoch": 1.04,
|
| 2073 |
+
"learning_rate": 0.00023928155339805821,
|
| 2074 |
+
"loss": 0.423,
|
| 2075 |
+
"step": 3230
|
| 2076 |
+
},
|
| 2077 |
+
{
|
| 2078 |
+
"epoch": 1.04,
|
| 2079 |
+
"learning_rate": 0.00023908737864077666,
|
| 2080 |
+
"loss": 0.4737,
|
| 2081 |
+
"step": 3240
|
| 2082 |
+
},
|
| 2083 |
+
{
|
| 2084 |
+
"epoch": 1.05,
|
| 2085 |
+
"learning_rate": 0.00023889320388349513,
|
| 2086 |
+
"loss": 0.4466,
|
| 2087 |
+
"step": 3250
|
| 2088 |
+
},
|
| 2089 |
+
{
|
| 2090 |
+
"epoch": 1.05,
|
| 2091 |
+
"learning_rate": 0.00023869902912621358,
|
| 2092 |
+
"loss": 0.4624,
|
| 2093 |
+
"step": 3260
|
| 2094 |
+
},
|
| 2095 |
+
{
|
| 2096 |
+
"epoch": 1.05,
|
| 2097 |
+
"learning_rate": 0.00023850485436893203,
|
| 2098 |
+
"loss": 0.455,
|
| 2099 |
+
"step": 3270
|
| 2100 |
+
},
|
| 2101 |
+
{
|
| 2102 |
+
"epoch": 1.05,
|
| 2103 |
+
"learning_rate": 0.00023831067961165047,
|
| 2104 |
+
"loss": 0.4699,
|
| 2105 |
+
"step": 3280
|
| 2106 |
+
},
|
| 2107 |
+
{
|
| 2108 |
+
"epoch": 1.06,
|
| 2109 |
+
"learning_rate": 0.00023811650485436892,
|
| 2110 |
+
"loss": 0.4218,
|
| 2111 |
+
"step": 3290
|
| 2112 |
+
},
|
| 2113 |
+
{
|
| 2114 |
+
"epoch": 1.06,
|
| 2115 |
+
"learning_rate": 0.00023792233009708736,
|
| 2116 |
+
"loss": 0.4495,
|
| 2117 |
+
"step": 3300
|
| 2118 |
+
},
|
| 2119 |
+
{
|
| 2120 |
+
"epoch": 1.06,
|
| 2121 |
+
"learning_rate": 0.0002377281553398058,
|
| 2122 |
+
"loss": 0.4738,
|
| 2123 |
+
"step": 3310
|
| 2124 |
+
},
|
| 2125 |
+
{
|
| 2126 |
+
"epoch": 1.07,
|
| 2127 |
+
"learning_rate": 0.00023753398058252425,
|
| 2128 |
+
"loss": 0.4574,
|
| 2129 |
+
"step": 3320
|
| 2130 |
+
},
|
| 2131 |
+
{
|
| 2132 |
+
"epoch": 1.07,
|
| 2133 |
+
"learning_rate": 0.0002373398058252427,
|
| 2134 |
+
"loss": 0.4699,
|
| 2135 |
+
"step": 3330
|
| 2136 |
+
},
|
| 2137 |
+
{
|
| 2138 |
+
"epoch": 1.07,
|
| 2139 |
+
"learning_rate": 0.00023714563106796112,
|
| 2140 |
+
"loss": 0.454,
|
| 2141 |
+
"step": 3340
|
| 2142 |
+
},
|
| 2143 |
+
{
|
| 2144 |
+
"epoch": 1.08,
|
| 2145 |
+
"learning_rate": 0.00023695145631067956,
|
| 2146 |
+
"loss": 0.463,
|
| 2147 |
+
"step": 3350
|
| 2148 |
+
},
|
| 2149 |
+
{
|
| 2150 |
+
"epoch": 1.08,
|
| 2151 |
+
"learning_rate": 0.00023675728155339804,
|
| 2152 |
+
"loss": 0.4533,
|
| 2153 |
+
"step": 3360
|
| 2154 |
+
},
|
| 2155 |
+
{
|
| 2156 |
+
"epoch": 1.08,
|
| 2157 |
+
"learning_rate": 0.00023656310679611648,
|
| 2158 |
+
"loss": 0.4525,
|
| 2159 |
+
"step": 3370
|
| 2160 |
+
},
|
| 2161 |
+
{
|
| 2162 |
+
"epoch": 1.09,
|
| 2163 |
+
"learning_rate": 0.00023636893203883493,
|
| 2164 |
+
"loss": 0.4391,
|
| 2165 |
+
"step": 3380
|
| 2166 |
+
},
|
| 2167 |
+
{
|
| 2168 |
+
"epoch": 1.09,
|
| 2169 |
+
"learning_rate": 0.00023617475728155338,
|
| 2170 |
+
"loss": 0.4542,
|
| 2171 |
+
"step": 3390
|
| 2172 |
+
},
|
| 2173 |
+
{
|
| 2174 |
+
"epoch": 1.09,
|
| 2175 |
+
"learning_rate": 0.00023598058252427182,
|
| 2176 |
+
"loss": 0.4255,
|
| 2177 |
+
"step": 3400
|
| 2178 |
+
},
|
| 2179 |
+
{
|
| 2180 |
+
"epoch": 1.09,
|
| 2181 |
+
"eval_loss": 0.47360363602638245,
|
| 2182 |
+
"eval_runtime": 513.4848,
|
| 2183 |
+
"eval_samples_per_second": 3.895,
|
| 2184 |
+
"eval_steps_per_second": 0.487,
|
| 2185 |
+
"step": 3400
|
| 2186 |
+
},
|
| 2187 |
+
{
|
| 2188 |
+
"epoch": 1.1,
|
| 2189 |
+
"learning_rate": 0.00023578640776699027,
|
| 2190 |
+
"loss": 0.4341,
|
| 2191 |
+
"step": 3410
|
| 2192 |
+
},
|
| 2193 |
+
{
|
| 2194 |
+
"epoch": 1.1,
|
| 2195 |
+
"learning_rate": 0.0002355922330097087,
|
| 2196 |
+
"loss": 0.4439,
|
| 2197 |
+
"step": 3420
|
| 2198 |
+
},
|
| 2199 |
+
{
|
| 2200 |
+
"epoch": 1.1,
|
| 2201 |
+
"learning_rate": 0.00023539805825242716,
|
| 2202 |
+
"loss": 0.4512,
|
| 2203 |
+
"step": 3430
|
| 2204 |
+
},
|
| 2205 |
+
{
|
| 2206 |
+
"epoch": 1.11,
|
| 2207 |
+
"learning_rate": 0.0002352038834951456,
|
| 2208 |
+
"loss": 0.4454,
|
| 2209 |
+
"step": 3440
|
| 2210 |
+
},
|
| 2211 |
+
{
|
| 2212 |
+
"epoch": 1.11,
|
| 2213 |
+
"learning_rate": 0.00023500970873786408,
|
| 2214 |
+
"loss": 0.4621,
|
| 2215 |
+
"step": 3450
|
| 2216 |
+
},
|
| 2217 |
+
{
|
| 2218 |
+
"epoch": 1.11,
|
| 2219 |
+
"learning_rate": 0.00023481553398058252,
|
| 2220 |
+
"loss": 0.4597,
|
| 2221 |
+
"step": 3460
|
| 2222 |
+
},
|
| 2223 |
+
{
|
| 2224 |
+
"epoch": 1.12,
|
| 2225 |
+
"learning_rate": 0.00023462135922330097,
|
| 2226 |
+
"loss": 0.4546,
|
| 2227 |
+
"step": 3470
|
| 2228 |
+
},
|
| 2229 |
+
{
|
| 2230 |
+
"epoch": 1.12,
|
| 2231 |
+
"learning_rate": 0.00023442718446601941,
|
| 2232 |
+
"loss": 0.4488,
|
| 2233 |
+
"step": 3480
|
| 2234 |
+
},
|
| 2235 |
+
{
|
| 2236 |
+
"epoch": 1.12,
|
| 2237 |
+
"learning_rate": 0.00023423300970873783,
|
| 2238 |
+
"loss": 0.4553,
|
| 2239 |
+
"step": 3490
|
| 2240 |
+
},
|
| 2241 |
+
{
|
| 2242 |
+
"epoch": 1.13,
|
| 2243 |
+
"learning_rate": 0.00023403883495145628,
|
| 2244 |
+
"loss": 0.4334,
|
| 2245 |
+
"step": 3500
|
| 2246 |
+
},
|
| 2247 |
+
{
|
| 2248 |
+
"epoch": 1.13,
|
| 2249 |
+
"learning_rate": 0.00023384466019417473,
|
| 2250 |
+
"loss": 0.4609,
|
| 2251 |
+
"step": 3510
|
| 2252 |
+
},
|
| 2253 |
+
{
|
| 2254 |
+
"epoch": 1.13,
|
| 2255 |
+
"learning_rate": 0.00023365048543689317,
|
| 2256 |
+
"loss": 0.4517,
|
| 2257 |
+
"step": 3520
|
| 2258 |
+
},
|
| 2259 |
+
{
|
| 2260 |
+
"epoch": 1.14,
|
| 2261 |
+
"learning_rate": 0.00023345631067961162,
|
| 2262 |
+
"loss": 0.4627,
|
| 2263 |
+
"step": 3530
|
| 2264 |
+
},
|
| 2265 |
+
{
|
| 2266 |
+
"epoch": 1.14,
|
| 2267 |
+
"learning_rate": 0.00023326213592233006,
|
| 2268 |
+
"loss": 0.4355,
|
| 2269 |
+
"step": 3540
|
| 2270 |
+
},
|
| 2271 |
+
{
|
| 2272 |
+
"epoch": 1.14,
|
| 2273 |
+
"learning_rate": 0.00023306796116504854,
|
| 2274 |
+
"loss": 0.4451,
|
| 2275 |
+
"step": 3550
|
| 2276 |
+
},
|
| 2277 |
+
{
|
| 2278 |
+
"epoch": 1.14,
|
| 2279 |
+
"learning_rate": 0.00023287378640776698,
|
| 2280 |
+
"loss": 0.4565,
|
| 2281 |
+
"step": 3560
|
| 2282 |
+
},
|
| 2283 |
+
{
|
| 2284 |
+
"epoch": 1.15,
|
| 2285 |
+
"learning_rate": 0.00023267961165048543,
|
| 2286 |
+
"loss": 0.4652,
|
| 2287 |
+
"step": 3570
|
| 2288 |
+
},
|
| 2289 |
+
{
|
| 2290 |
+
"epoch": 1.15,
|
| 2291 |
+
"learning_rate": 0.00023248543689320387,
|
| 2292 |
+
"loss": 0.4422,
|
| 2293 |
+
"step": 3580
|
| 2294 |
+
},
|
| 2295 |
+
{
|
| 2296 |
+
"epoch": 1.15,
|
| 2297 |
+
"learning_rate": 0.00023229126213592232,
|
| 2298 |
+
"loss": 0.4361,
|
| 2299 |
+
"step": 3590
|
| 2300 |
+
},
|
| 2301 |
+
{
|
| 2302 |
+
"epoch": 1.16,
|
| 2303 |
+
"learning_rate": 0.00023209708737864076,
|
| 2304 |
+
"loss": 0.4279,
|
| 2305 |
+
"step": 3600
|
| 2306 |
+
},
|
| 2307 |
+
{
|
| 2308 |
+
"epoch": 1.16,
|
| 2309 |
+
"eval_loss": 0.4739069640636444,
|
| 2310 |
+
"eval_runtime": 517.674,
|
| 2311 |
+
"eval_samples_per_second": 3.863,
|
| 2312 |
+
"eval_steps_per_second": 0.483,
|
| 2313 |
+
"step": 3600
|
| 2314 |
+
},
|
| 2315 |
+
{
|
| 2316 |
+
"epoch": 1.16,
|
| 2317 |
+
"learning_rate": 0.0002319029126213592,
|
| 2318 |
+
"loss": 0.4432,
|
| 2319 |
+
"step": 3610
|
| 2320 |
+
},
|
| 2321 |
+
{
|
| 2322 |
+
"epoch": 1.16,
|
| 2323 |
+
"learning_rate": 0.00023170873786407766,
|
| 2324 |
+
"loss": 0.4719,
|
| 2325 |
+
"step": 3620
|
| 2326 |
+
},
|
| 2327 |
+
{
|
| 2328 |
+
"epoch": 1.17,
|
| 2329 |
+
"learning_rate": 0.00023151456310679608,
|
| 2330 |
+
"loss": 0.4603,
|
| 2331 |
+
"step": 3630
|
| 2332 |
+
},
|
| 2333 |
+
{
|
| 2334 |
+
"epoch": 1.17,
|
| 2335 |
+
"learning_rate": 0.00023132038834951452,
|
| 2336 |
+
"loss": 0.4878,
|
| 2337 |
+
"step": 3640
|
| 2338 |
+
},
|
| 2339 |
+
{
|
| 2340 |
+
"epoch": 1.17,
|
| 2341 |
+
"learning_rate": 0.00023112621359223297,
|
| 2342 |
+
"loss": 0.4439,
|
| 2343 |
+
"step": 3650
|
| 2344 |
+
},
|
| 2345 |
+
{
|
| 2346 |
+
"epoch": 1.18,
|
| 2347 |
+
"learning_rate": 0.00023093203883495144,
|
| 2348 |
+
"loss": 0.4301,
|
| 2349 |
+
"step": 3660
|
| 2350 |
+
},
|
| 2351 |
+
{
|
| 2352 |
+
"epoch": 1.18,
|
| 2353 |
+
"learning_rate": 0.00023073786407766989,
|
| 2354 |
+
"loss": 0.4362,
|
| 2355 |
+
"step": 3670
|
| 2356 |
+
},
|
| 2357 |
+
{
|
| 2358 |
+
"epoch": 1.18,
|
| 2359 |
+
"learning_rate": 0.00023054368932038833,
|
| 2360 |
+
"loss": 0.4612,
|
| 2361 |
+
"step": 3680
|
| 2362 |
+
},
|
| 2363 |
+
{
|
| 2364 |
+
"epoch": 1.19,
|
| 2365 |
+
"learning_rate": 0.00023034951456310678,
|
| 2366 |
+
"loss": 0.4681,
|
| 2367 |
+
"step": 3690
|
| 2368 |
+
},
|
| 2369 |
+
{
|
| 2370 |
+
"epoch": 1.19,
|
| 2371 |
+
"learning_rate": 0.00023015533980582522,
|
| 2372 |
+
"loss": 0.4303,
|
| 2373 |
+
"step": 3700
|
| 2374 |
+
},
|
| 2375 |
+
{
|
| 2376 |
+
"epoch": 1.19,
|
| 2377 |
+
"learning_rate": 0.00022996116504854367,
|
| 2378 |
+
"loss": 0.4663,
|
| 2379 |
+
"step": 3710
|
| 2380 |
+
},
|
| 2381 |
+
{
|
| 2382 |
+
"epoch": 1.2,
|
| 2383 |
+
"learning_rate": 0.00022976699029126211,
|
| 2384 |
+
"loss": 0.4523,
|
| 2385 |
+
"step": 3720
|
| 2386 |
+
},
|
| 2387 |
+
{
|
| 2388 |
+
"epoch": 1.2,
|
| 2389 |
+
"learning_rate": 0.00022957281553398056,
|
| 2390 |
+
"loss": 0.4478,
|
| 2391 |
+
"step": 3730
|
| 2392 |
+
},
|
| 2393 |
+
{
|
| 2394 |
+
"epoch": 1.2,
|
| 2395 |
+
"learning_rate": 0.000229378640776699,
|
| 2396 |
+
"loss": 0.4428,
|
| 2397 |
+
"step": 3740
|
| 2398 |
+
},
|
| 2399 |
+
{
|
| 2400 |
+
"epoch": 1.21,
|
| 2401 |
+
"learning_rate": 0.00022918446601941748,
|
| 2402 |
+
"loss": 0.4427,
|
| 2403 |
+
"step": 3750
|
| 2404 |
+
},
|
| 2405 |
+
{
|
| 2406 |
+
"epoch": 1.21,
|
| 2407 |
+
"learning_rate": 0.00022899029126213593,
|
| 2408 |
+
"loss": 0.4247,
|
| 2409 |
+
"step": 3760
|
| 2410 |
+
},
|
| 2411 |
+
{
|
| 2412 |
+
"epoch": 1.21,
|
| 2413 |
+
"learning_rate": 0.00022879611650485437,
|
| 2414 |
+
"loss": 0.4431,
|
| 2415 |
+
"step": 3770
|
| 2416 |
+
},
|
| 2417 |
+
{
|
| 2418 |
+
"epoch": 1.22,
|
| 2419 |
+
"learning_rate": 0.0002286019417475728,
|
| 2420 |
+
"loss": 0.466,
|
| 2421 |
+
"step": 3780
|
| 2422 |
+
},
|
| 2423 |
+
{
|
| 2424 |
+
"epoch": 1.22,
|
| 2425 |
+
"learning_rate": 0.00022840776699029124,
|
| 2426 |
+
"loss": 0.4704,
|
| 2427 |
+
"step": 3790
|
| 2428 |
+
},
|
| 2429 |
+
{
|
| 2430 |
+
"epoch": 1.22,
|
| 2431 |
+
"learning_rate": 0.00022821359223300968,
|
| 2432 |
+
"loss": 0.4698,
|
| 2433 |
+
"step": 3800
|
| 2434 |
+
},
|
| 2435 |
+
{
|
| 2436 |
+
"epoch": 1.22,
|
| 2437 |
+
"eval_loss": 0.4726303517818451,
|
| 2438 |
+
"eval_runtime": 519.452,
|
| 2439 |
+
"eval_samples_per_second": 3.85,
|
| 2440 |
+
"eval_steps_per_second": 0.481,
|
| 2441 |
+
"step": 3800
|
| 2442 |
+
},
|
| 2443 |
+
{
|
| 2444 |
+
"epoch": 1.23,
|
| 2445 |
+
"learning_rate": 0.00022801941747572813,
|
| 2446 |
+
"loss": 0.4643,
|
| 2447 |
+
"step": 3810
|
| 2448 |
+
},
|
| 2449 |
+
{
|
| 2450 |
+
"epoch": 1.23,
|
| 2451 |
+
"learning_rate": 0.00022782524271844657,
|
| 2452 |
+
"loss": 0.4357,
|
| 2453 |
+
"step": 3820
|
| 2454 |
+
},
|
| 2455 |
+
{
|
| 2456 |
+
"epoch": 1.23,
|
| 2457 |
+
"learning_rate": 0.00022763106796116502,
|
| 2458 |
+
"loss": 0.4749,
|
| 2459 |
+
"step": 3830
|
| 2460 |
+
},
|
| 2461 |
+
{
|
| 2462 |
+
"epoch": 1.23,
|
| 2463 |
+
"learning_rate": 0.00022743689320388346,
|
| 2464 |
+
"loss": 0.428,
|
| 2465 |
+
"step": 3840
|
| 2466 |
+
},
|
| 2467 |
+
{
|
| 2468 |
+
"epoch": 1.24,
|
| 2469 |
+
"learning_rate": 0.0002272427184466019,
|
| 2470 |
+
"loss": 0.4705,
|
| 2471 |
+
"step": 3850
|
| 2472 |
+
},
|
| 2473 |
+
{
|
| 2474 |
+
"epoch": 1.24,
|
| 2475 |
+
"learning_rate": 0.00022704854368932038,
|
| 2476 |
+
"loss": 0.4448,
|
| 2477 |
+
"step": 3860
|
| 2478 |
+
},
|
| 2479 |
+
{
|
| 2480 |
+
"epoch": 1.24,
|
| 2481 |
+
"learning_rate": 0.00022685436893203883,
|
| 2482 |
+
"loss": 0.4613,
|
| 2483 |
+
"step": 3870
|
| 2484 |
+
},
|
| 2485 |
+
{
|
| 2486 |
+
"epoch": 1.25,
|
| 2487 |
+
"learning_rate": 0.00022666019417475728,
|
| 2488 |
+
"loss": 0.4643,
|
| 2489 |
+
"step": 3880
|
| 2490 |
+
},
|
| 2491 |
+
{
|
| 2492 |
+
"epoch": 1.25,
|
| 2493 |
+
"learning_rate": 0.00022646601941747572,
|
| 2494 |
+
"loss": 0.4416,
|
| 2495 |
+
"step": 3890
|
| 2496 |
+
},
|
| 2497 |
+
{
|
| 2498 |
+
"epoch": 1.25,
|
| 2499 |
+
"learning_rate": 0.00022627184466019417,
|
| 2500 |
+
"loss": 0.4494,
|
| 2501 |
+
"step": 3900
|
| 2502 |
+
},
|
| 2503 |
+
{
|
| 2504 |
+
"epoch": 1.26,
|
| 2505 |
+
"learning_rate": 0.0002260776699029126,
|
| 2506 |
+
"loss": 0.4343,
|
| 2507 |
+
"step": 3910
|
| 2508 |
+
},
|
| 2509 |
+
{
|
| 2510 |
+
"epoch": 1.26,
|
| 2511 |
+
"learning_rate": 0.00022588349514563103,
|
| 2512 |
+
"loss": 0.4322,
|
| 2513 |
+
"step": 3920
|
| 2514 |
+
},
|
| 2515 |
+
{
|
| 2516 |
+
"epoch": 1.26,
|
| 2517 |
+
"learning_rate": 0.00022568932038834948,
|
| 2518 |
+
"loss": 0.454,
|
| 2519 |
+
"step": 3930
|
| 2520 |
+
},
|
| 2521 |
+
{
|
| 2522 |
+
"epoch": 1.27,
|
| 2523 |
+
"learning_rate": 0.00022549514563106792,
|
| 2524 |
+
"loss": 0.4432,
|
| 2525 |
+
"step": 3940
|
| 2526 |
+
},
|
| 2527 |
+
{
|
| 2528 |
+
"epoch": 1.27,
|
| 2529 |
+
"learning_rate": 0.00022530097087378637,
|
| 2530 |
+
"loss": 0.4393,
|
| 2531 |
+
"step": 3950
|
| 2532 |
+
},
|
| 2533 |
+
{
|
| 2534 |
+
"epoch": 1.27,
|
| 2535 |
+
"learning_rate": 0.00022510679611650484,
|
| 2536 |
+
"loss": 0.4438,
|
| 2537 |
+
"step": 3960
|
| 2538 |
+
},
|
| 2539 |
+
{
|
| 2540 |
+
"epoch": 1.28,
|
| 2541 |
+
"learning_rate": 0.0002249126213592233,
|
| 2542 |
+
"loss": 0.4144,
|
| 2543 |
+
"step": 3970
|
| 2544 |
+
},
|
| 2545 |
+
{
|
| 2546 |
+
"epoch": 1.28,
|
| 2547 |
+
"learning_rate": 0.00022471844660194173,
|
| 2548 |
+
"loss": 0.4304,
|
| 2549 |
+
"step": 3980
|
| 2550 |
+
},
|
| 2551 |
+
{
|
| 2552 |
+
"epoch": 1.28,
|
| 2553 |
+
"learning_rate": 0.00022452427184466018,
|
| 2554 |
+
"loss": 0.454,
|
| 2555 |
+
"step": 3990
|
| 2556 |
+
},
|
| 2557 |
+
{
|
| 2558 |
+
"epoch": 1.29,
|
| 2559 |
+
"learning_rate": 0.00022433009708737863,
|
| 2560 |
+
"loss": 0.4663,
|
| 2561 |
+
"step": 4000
|
| 2562 |
+
},
|
| 2563 |
+
{
|
| 2564 |
+
"epoch": 1.29,
|
| 2565 |
+
"eval_loss": 0.4728265702724457,
|
| 2566 |
+
"eval_runtime": 520.4674,
|
| 2567 |
+
"eval_samples_per_second": 3.843,
|
| 2568 |
+
"eval_steps_per_second": 0.48,
|
| 2569 |
+
"step": 4000
|
| 2570 |
+
},
|
| 2571 |
+
{
|
| 2572 |
+
"epoch": 1.29,
|
| 2573 |
+
"learning_rate": 0.00022413592233009707,
|
| 2574 |
+
"loss": 0.4803,
|
| 2575 |
+
"step": 4010
|
| 2576 |
+
},
|
| 2577 |
+
{
|
| 2578 |
+
"epoch": 1.29,
|
| 2579 |
+
"learning_rate": 0.00022394174757281552,
|
| 2580 |
+
"loss": 0.4528,
|
| 2581 |
+
"step": 4020
|
| 2582 |
+
},
|
| 2583 |
+
{
|
| 2584 |
+
"epoch": 1.3,
|
| 2585 |
+
"learning_rate": 0.00022374757281553396,
|
| 2586 |
+
"loss": 0.4333,
|
| 2587 |
+
"step": 4030
|
| 2588 |
+
},
|
| 2589 |
+
{
|
| 2590 |
+
"epoch": 1.3,
|
| 2591 |
+
"learning_rate": 0.0002235533980582524,
|
| 2592 |
+
"loss": 0.4548,
|
| 2593 |
+
"step": 4040
|
| 2594 |
+
},
|
| 2595 |
+
{
|
| 2596 |
+
"epoch": 1.3,
|
| 2597 |
+
"learning_rate": 0.00022335922330097088,
|
| 2598 |
+
"loss": 0.4533,
|
| 2599 |
+
"step": 4050
|
| 2600 |
+
},
|
| 2601 |
+
{
|
| 2602 |
+
"epoch": 1.31,
|
| 2603 |
+
"learning_rate": 0.00022316504854368927,
|
| 2604 |
+
"loss": 0.4269,
|
| 2605 |
+
"step": 4060
|
| 2606 |
+
},
|
| 2607 |
+
{
|
| 2608 |
+
"epoch": 1.31,
|
| 2609 |
+
"learning_rate": 0.00022297087378640775,
|
| 2610 |
+
"loss": 0.4489,
|
| 2611 |
+
"step": 4070
|
| 2612 |
+
},
|
| 2613 |
+
{
|
| 2614 |
+
"epoch": 1.31,
|
| 2615 |
+
"learning_rate": 0.0002227766990291262,
|
| 2616 |
+
"loss": 0.4581,
|
| 2617 |
+
"step": 4080
|
| 2618 |
+
},
|
| 2619 |
+
{
|
| 2620 |
+
"epoch": 1.32,
|
| 2621 |
+
"learning_rate": 0.00022258252427184464,
|
| 2622 |
+
"loss": 0.4474,
|
| 2623 |
+
"step": 4090
|
| 2624 |
+
},
|
| 2625 |
+
{
|
| 2626 |
+
"epoch": 1.32,
|
| 2627 |
+
"learning_rate": 0.00022238834951456308,
|
| 2628 |
+
"loss": 0.4498,
|
| 2629 |
+
"step": 4100
|
| 2630 |
+
},
|
| 2631 |
+
{
|
| 2632 |
+
"epoch": 1.32,
|
| 2633 |
+
"learning_rate": 0.00022219417475728153,
|
| 2634 |
+
"loss": 0.4737,
|
| 2635 |
+
"step": 4110
|
| 2636 |
+
},
|
| 2637 |
+
{
|
| 2638 |
+
"epoch": 1.32,
|
| 2639 |
+
"learning_rate": 0.00022199999999999998,
|
| 2640 |
+
"loss": 0.4412,
|
| 2641 |
+
"step": 4120
|
| 2642 |
+
},
|
| 2643 |
+
{
|
| 2644 |
+
"epoch": 1.33,
|
| 2645 |
+
"learning_rate": 0.00022180582524271842,
|
| 2646 |
+
"loss": 0.4365,
|
| 2647 |
+
"step": 4130
|
| 2648 |
+
},
|
| 2649 |
+
{
|
| 2650 |
+
"epoch": 1.33,
|
| 2651 |
+
"learning_rate": 0.00022161165048543687,
|
| 2652 |
+
"loss": 0.4266,
|
| 2653 |
+
"step": 4140
|
| 2654 |
+
},
|
| 2655 |
+
{
|
| 2656 |
+
"epoch": 1.33,
|
| 2657 |
+
"learning_rate": 0.0002214174757281553,
|
| 2658 |
+
"loss": 0.4492,
|
| 2659 |
+
"step": 4150
|
| 2660 |
+
},
|
| 2661 |
+
{
|
| 2662 |
+
"epoch": 1.34,
|
| 2663 |
+
"learning_rate": 0.00022122330097087379,
|
| 2664 |
+
"loss": 0.4472,
|
| 2665 |
+
"step": 4160
|
| 2666 |
+
},
|
| 2667 |
+
{
|
| 2668 |
+
"epoch": 1.34,
|
| 2669 |
+
"learning_rate": 0.00022102912621359223,
|
| 2670 |
+
"loss": 0.4137,
|
| 2671 |
+
"step": 4170
|
| 2672 |
+
},
|
| 2673 |
+
{
|
| 2674 |
+
"epoch": 1.34,
|
| 2675 |
+
"learning_rate": 0.00022083495145631068,
|
| 2676 |
+
"loss": 0.4497,
|
| 2677 |
+
"step": 4180
|
| 2678 |
+
},
|
| 2679 |
+
{
|
| 2680 |
+
"epoch": 1.35,
|
| 2681 |
+
"learning_rate": 0.00022064077669902912,
|
| 2682 |
+
"loss": 0.4324,
|
| 2683 |
+
"step": 4190
|
| 2684 |
+
},
|
| 2685 |
+
{
|
| 2686 |
+
"epoch": 1.35,
|
| 2687 |
+
"learning_rate": 0.00022044660194174757,
|
| 2688 |
+
"loss": 0.4332,
|
| 2689 |
+
"step": 4200
|
| 2690 |
+
},
|
| 2691 |
+
{
|
| 2692 |
+
"epoch": 1.35,
|
| 2693 |
+
"eval_loss": 0.471797376871109,
|
| 2694 |
+
"eval_runtime": 518.6631,
|
| 2695 |
+
"eval_samples_per_second": 3.856,
|
| 2696 |
+
"eval_steps_per_second": 0.482,
|
| 2697 |
+
"step": 4200
|
| 2698 |
+
},
|
| 2699 |
+
{
|
| 2700 |
+
"epoch": 1.35,
|
| 2701 |
+
"learning_rate": 0.000220252427184466,
|
| 2702 |
+
"loss": 0.452,
|
| 2703 |
+
"step": 4210
|
| 2704 |
+
},
|
| 2705 |
+
{
|
| 2706 |
+
"epoch": 1.36,
|
| 2707 |
+
"learning_rate": 0.00022005825242718443,
|
| 2708 |
+
"loss": 0.4372,
|
| 2709 |
+
"step": 4220
|
| 2710 |
+
},
|
| 2711 |
+
{
|
| 2712 |
+
"epoch": 1.36,
|
| 2713 |
+
"learning_rate": 0.00021986407766990288,
|
| 2714 |
+
"loss": 0.4346,
|
| 2715 |
+
"step": 4230
|
| 2716 |
+
},
|
| 2717 |
+
{
|
| 2718 |
+
"epoch": 1.36,
|
| 2719 |
+
"learning_rate": 0.00021966990291262133,
|
| 2720 |
+
"loss": 0.4511,
|
| 2721 |
+
"step": 4240
|
| 2722 |
+
},
|
| 2723 |
+
{
|
| 2724 |
+
"epoch": 1.37,
|
| 2725 |
+
"learning_rate": 0.00021947572815533977,
|
| 2726 |
+
"loss": 0.4594,
|
| 2727 |
+
"step": 4250
|
| 2728 |
+
},
|
| 2729 |
+
{
|
| 2730 |
+
"epoch": 1.37,
|
| 2731 |
+
"learning_rate": 0.00021928155339805822,
|
| 2732 |
+
"loss": 0.4655,
|
| 2733 |
+
"step": 4260
|
| 2734 |
+
},
|
| 2735 |
+
{
|
| 2736 |
+
"epoch": 1.37,
|
| 2737 |
+
"learning_rate": 0.0002190873786407767,
|
| 2738 |
+
"loss": 0.4613,
|
| 2739 |
+
"step": 4270
|
| 2740 |
+
},
|
| 2741 |
+
{
|
| 2742 |
+
"epoch": 1.38,
|
| 2743 |
+
"learning_rate": 0.00021889320388349514,
|
| 2744 |
+
"loss": 0.4436,
|
| 2745 |
+
"step": 4280
|
| 2746 |
+
},
|
| 2747 |
+
{
|
| 2748 |
+
"epoch": 1.38,
|
| 2749 |
+
"learning_rate": 0.00021869902912621358,
|
| 2750 |
+
"loss": 0.4434,
|
| 2751 |
+
"step": 4290
|
| 2752 |
+
},
|
| 2753 |
+
{
|
| 2754 |
+
"epoch": 1.38,
|
| 2755 |
+
"learning_rate": 0.00021850485436893203,
|
| 2756 |
+
"loss": 0.4554,
|
| 2757 |
+
"step": 4300
|
| 2758 |
+
},
|
| 2759 |
+
{
|
| 2760 |
+
"epoch": 1.39,
|
| 2761 |
+
"learning_rate": 0.00021831067961165047,
|
| 2762 |
+
"loss": 0.4838,
|
| 2763 |
+
"step": 4310
|
| 2764 |
+
},
|
| 2765 |
+
{
|
| 2766 |
+
"epoch": 1.39,
|
| 2767 |
+
"learning_rate": 0.00021811650485436892,
|
| 2768 |
+
"loss": 0.4451,
|
| 2769 |
+
"step": 4320
|
| 2770 |
+
},
|
| 2771 |
+
{
|
| 2772 |
+
"epoch": 1.39,
|
| 2773 |
+
"learning_rate": 0.00021792233009708736,
|
| 2774 |
+
"loss": 0.4606,
|
| 2775 |
+
"step": 4330
|
| 2776 |
+
},
|
| 2777 |
+
{
|
| 2778 |
+
"epoch": 1.4,
|
| 2779 |
+
"learning_rate": 0.0002177281553398058,
|
| 2780 |
+
"loss": 0.4239,
|
| 2781 |
+
"step": 4340
|
| 2782 |
+
},
|
| 2783 |
+
{
|
| 2784 |
+
"epoch": 1.4,
|
| 2785 |
+
"learning_rate": 0.00021753398058252423,
|
| 2786 |
+
"loss": 0.4328,
|
| 2787 |
+
"step": 4350
|
| 2788 |
+
},
|
| 2789 |
+
{
|
| 2790 |
+
"epoch": 1.4,
|
| 2791 |
+
"learning_rate": 0.00021733980582524268,
|
| 2792 |
+
"loss": 0.4712,
|
| 2793 |
+
"step": 4360
|
| 2794 |
+
},
|
| 2795 |
+
{
|
| 2796 |
+
"epoch": 1.41,
|
| 2797 |
+
"learning_rate": 0.00021714563106796112,
|
| 2798 |
+
"loss": 0.4412,
|
| 2799 |
+
"step": 4370
|
| 2800 |
+
},
|
| 2801 |
+
{
|
| 2802 |
+
"epoch": 1.41,
|
| 2803 |
+
"learning_rate": 0.0002169514563106796,
|
| 2804 |
+
"loss": 0.4629,
|
| 2805 |
+
"step": 4380
|
| 2806 |
+
},
|
| 2807 |
+
{
|
| 2808 |
+
"epoch": 1.41,
|
| 2809 |
+
"learning_rate": 0.00021675728155339804,
|
| 2810 |
+
"loss": 0.4536,
|
| 2811 |
+
"step": 4390
|
| 2812 |
+
},
|
| 2813 |
+
{
|
| 2814 |
+
"epoch": 1.41,
|
| 2815 |
+
"learning_rate": 0.00021656310679611649,
|
| 2816 |
+
"loss": 0.4208,
|
| 2817 |
+
"step": 4400
|
| 2818 |
+
},
|
| 2819 |
+
{
|
| 2820 |
+
"epoch": 1.41,
|
| 2821 |
+
"eval_loss": 0.4719524681568146,
|
| 2822 |
+
"eval_runtime": 524.6738,
|
| 2823 |
+
"eval_samples_per_second": 3.812,
|
| 2824 |
+
"eval_steps_per_second": 0.476,
|
| 2825 |
+
"step": 4400
|
| 2826 |
+
},
|
| 2827 |
+
{
|
| 2828 |
+
"epoch": 1.42,
|
| 2829 |
+
"learning_rate": 0.00021636893203883493,
|
| 2830 |
+
"loss": 0.438,
|
| 2831 |
+
"step": 4410
|
| 2832 |
+
},
|
| 2833 |
+
{
|
| 2834 |
+
"epoch": 1.42,
|
| 2835 |
+
"learning_rate": 0.00021617475728155338,
|
| 2836 |
+
"loss": 0.4399,
|
| 2837 |
+
"step": 4420
|
| 2838 |
+
},
|
| 2839 |
+
{
|
| 2840 |
+
"epoch": 1.42,
|
| 2841 |
+
"learning_rate": 0.00021598058252427182,
|
| 2842 |
+
"loss": 0.4568,
|
| 2843 |
+
"step": 4430
|
| 2844 |
+
},
|
| 2845 |
+
{
|
| 2846 |
+
"epoch": 1.43,
|
| 2847 |
+
"learning_rate": 0.00021578640776699027,
|
| 2848 |
+
"loss": 0.4695,
|
| 2849 |
+
"step": 4440
|
| 2850 |
+
},
|
| 2851 |
+
{
|
| 2852 |
+
"epoch": 1.43,
|
| 2853 |
+
"learning_rate": 0.00021559223300970871,
|
| 2854 |
+
"loss": 0.4387,
|
| 2855 |
+
"step": 4450
|
| 2856 |
+
},
|
| 2857 |
+
{
|
| 2858 |
+
"epoch": 1.43,
|
| 2859 |
+
"learning_rate": 0.0002153980582524272,
|
| 2860 |
+
"loss": 0.4466,
|
| 2861 |
+
"step": 4460
|
| 2862 |
+
},
|
| 2863 |
+
{
|
| 2864 |
+
"epoch": 1.44,
|
| 2865 |
+
"learning_rate": 0.00021520388349514563,
|
| 2866 |
+
"loss": 0.4652,
|
| 2867 |
+
"step": 4470
|
| 2868 |
+
},
|
| 2869 |
+
{
|
| 2870 |
+
"epoch": 1.44,
|
| 2871 |
+
"learning_rate": 0.00021500970873786408,
|
| 2872 |
+
"loss": 0.4155,
|
| 2873 |
+
"step": 4480
|
| 2874 |
+
},
|
| 2875 |
+
{
|
| 2876 |
+
"epoch": 1.44,
|
| 2877 |
+
"learning_rate": 0.00021481553398058253,
|
| 2878 |
+
"loss": 0.4457,
|
| 2879 |
+
"step": 4490
|
| 2880 |
+
},
|
| 2881 |
+
{
|
| 2882 |
+
"epoch": 1.45,
|
| 2883 |
+
"learning_rate": 0.00021462135922330094,
|
| 2884 |
+
"loss": 0.4349,
|
| 2885 |
+
"step": 4500
|
| 2886 |
+
},
|
| 2887 |
+
{
|
| 2888 |
+
"epoch": 1.45,
|
| 2889 |
+
"learning_rate": 0.0002144271844660194,
|
| 2890 |
+
"loss": 0.4304,
|
| 2891 |
+
"step": 4510
|
| 2892 |
+
},
|
| 2893 |
+
{
|
| 2894 |
+
"epoch": 1.45,
|
| 2895 |
+
"learning_rate": 0.00021423300970873784,
|
| 2896 |
+
"loss": 0.4222,
|
| 2897 |
+
"step": 4520
|
| 2898 |
+
},
|
| 2899 |
+
{
|
| 2900 |
+
"epoch": 1.46,
|
| 2901 |
+
"learning_rate": 0.00021403883495145628,
|
| 2902 |
+
"loss": 0.4373,
|
| 2903 |
+
"step": 4530
|
| 2904 |
+
},
|
| 2905 |
+
{
|
| 2906 |
+
"epoch": 1.46,
|
| 2907 |
+
"learning_rate": 0.00021384466019417473,
|
| 2908 |
+
"loss": 0.4526,
|
| 2909 |
+
"step": 4540
|
| 2910 |
+
},
|
| 2911 |
+
{
|
| 2912 |
+
"epoch": 1.46,
|
| 2913 |
+
"learning_rate": 0.00021365048543689317,
|
| 2914 |
+
"loss": 0.4804,
|
| 2915 |
+
"step": 4550
|
| 2916 |
+
},
|
| 2917 |
+
{
|
| 2918 |
+
"epoch": 1.47,
|
| 2919 |
+
"learning_rate": 0.00021345631067961162,
|
| 2920 |
+
"loss": 0.468,
|
| 2921 |
+
"step": 4560
|
| 2922 |
+
},
|
| 2923 |
+
{
|
| 2924 |
+
"epoch": 1.47,
|
| 2925 |
+
"learning_rate": 0.0002132621359223301,
|
| 2926 |
+
"loss": 0.4664,
|
| 2927 |
+
"step": 4570
|
| 2928 |
+
},
|
| 2929 |
+
{
|
| 2930 |
+
"epoch": 1.47,
|
| 2931 |
+
"learning_rate": 0.00021306796116504854,
|
| 2932 |
+
"loss": 0.4688,
|
| 2933 |
+
"step": 4580
|
| 2934 |
+
},
|
| 2935 |
+
{
|
| 2936 |
+
"epoch": 1.48,
|
| 2937 |
+
"learning_rate": 0.00021287378640776698,
|
| 2938 |
+
"loss": 0.4262,
|
| 2939 |
+
"step": 4590
|
| 2940 |
+
},
|
| 2941 |
+
{
|
| 2942 |
+
"epoch": 1.48,
|
| 2943 |
+
"learning_rate": 0.00021267961165048543,
|
| 2944 |
+
"loss": 0.4431,
|
| 2945 |
+
"step": 4600
|
| 2946 |
+
},
|
| 2947 |
+
{
|
| 2948 |
+
"epoch": 1.48,
|
| 2949 |
+
"eval_loss": 0.47054779529571533,
|
| 2950 |
+
"eval_runtime": 526.5834,
|
| 2951 |
+
"eval_samples_per_second": 3.798,
|
| 2952 |
+
"eval_steps_per_second": 0.475,
|
| 2953 |
+
"step": 4600
|
| 2954 |
+
},
|
| 2955 |
+
{
|
| 2956 |
+
"epoch": 1.48,
|
| 2957 |
+
"learning_rate": 0.00021248543689320387,
|
| 2958 |
+
"loss": 0.4371,
|
| 2959 |
+
"step": 4610
|
| 2960 |
+
},
|
| 2961 |
+
{
|
| 2962 |
+
"epoch": 1.49,
|
| 2963 |
+
"learning_rate": 0.00021229126213592232,
|
| 2964 |
+
"loss": 0.4555,
|
| 2965 |
+
"step": 4620
|
| 2966 |
+
},
|
| 2967 |
+
{
|
| 2968 |
+
"epoch": 1.49,
|
| 2969 |
+
"learning_rate": 0.00021209708737864077,
|
| 2970 |
+
"loss": 0.4538,
|
| 2971 |
+
"step": 4630
|
| 2972 |
+
},
|
| 2973 |
+
{
|
| 2974 |
+
"epoch": 1.49,
|
| 2975 |
+
"learning_rate": 0.00021190291262135919,
|
| 2976 |
+
"loss": 0.4501,
|
| 2977 |
+
"step": 4640
|
| 2978 |
+
},
|
| 2979 |
+
{
|
| 2980 |
+
"epoch": 1.5,
|
| 2981 |
+
"learning_rate": 0.00021170873786407763,
|
| 2982 |
+
"loss": 0.4201,
|
| 2983 |
+
"step": 4650
|
| 2984 |
+
},
|
| 2985 |
+
{
|
| 2986 |
+
"epoch": 1.5,
|
| 2987 |
+
"learning_rate": 0.00021151456310679608,
|
| 2988 |
+
"loss": 0.4346,
|
| 2989 |
+
"step": 4660
|
| 2990 |
+
},
|
| 2991 |
+
{
|
| 2992 |
+
"epoch": 1.5,
|
| 2993 |
+
"learning_rate": 0.00021132038834951452,
|
| 2994 |
+
"loss": 0.447,
|
| 2995 |
+
"step": 4670
|
| 2996 |
+
},
|
| 2997 |
+
{
|
| 2998 |
+
"epoch": 1.5,
|
| 2999 |
+
"learning_rate": 0.000211126213592233,
|
| 3000 |
+
"loss": 0.451,
|
| 3001 |
+
"step": 4680
|
| 3002 |
+
},
|
| 3003 |
+
{
|
| 3004 |
+
"epoch": 1.51,
|
| 3005 |
+
"learning_rate": 0.00021093203883495144,
|
| 3006 |
+
"loss": 0.4548,
|
| 3007 |
+
"step": 4690
|
| 3008 |
+
},
|
| 3009 |
+
{
|
| 3010 |
+
"epoch": 1.51,
|
| 3011 |
+
"learning_rate": 0.0002107378640776699,
|
| 3012 |
+
"loss": 0.4386,
|
| 3013 |
+
"step": 4700
|
| 3014 |
+
},
|
| 3015 |
+
{
|
| 3016 |
+
"epoch": 1.51,
|
| 3017 |
+
"learning_rate": 0.00021054368932038833,
|
| 3018 |
+
"loss": 0.4408,
|
| 3019 |
+
"step": 4710
|
| 3020 |
+
},
|
| 3021 |
+
{
|
| 3022 |
+
"epoch": 1.52,
|
| 3023 |
+
"learning_rate": 0.00021034951456310678,
|
| 3024 |
+
"loss": 0.467,
|
| 3025 |
+
"step": 4720
|
| 3026 |
+
},
|
| 3027 |
+
{
|
| 3028 |
+
"epoch": 1.52,
|
| 3029 |
+
"learning_rate": 0.00021015533980582522,
|
| 3030 |
+
"loss": 0.4447,
|
| 3031 |
+
"step": 4730
|
| 3032 |
+
},
|
| 3033 |
+
{
|
| 3034 |
+
"epoch": 1.52,
|
| 3035 |
+
"learning_rate": 0.00020996116504854367,
|
| 3036 |
+
"loss": 0.4566,
|
| 3037 |
+
"step": 4740
|
| 3038 |
+
},
|
| 3039 |
+
{
|
| 3040 |
+
"epoch": 1.53,
|
| 3041 |
+
"learning_rate": 0.00020976699029126212,
|
| 3042 |
+
"loss": 0.4391,
|
| 3043 |
+
"step": 4750
|
| 3044 |
+
},
|
| 3045 |
+
{
|
| 3046 |
+
"epoch": 1.53,
|
| 3047 |
+
"learning_rate": 0.00020957281553398056,
|
| 3048 |
+
"loss": 0.4547,
|
| 3049 |
+
"step": 4760
|
| 3050 |
+
},
|
| 3051 |
+
{
|
| 3052 |
+
"epoch": 1.53,
|
| 3053 |
+
"learning_rate": 0.00020937864077669904,
|
| 3054 |
+
"loss": 0.4466,
|
| 3055 |
+
"step": 4770
|
| 3056 |
+
},
|
| 3057 |
+
{
|
| 3058 |
+
"epoch": 1.54,
|
| 3059 |
+
"learning_rate": 0.00020918446601941748,
|
| 3060 |
+
"loss": 0.4368,
|
| 3061 |
+
"step": 4780
|
| 3062 |
+
},
|
| 3063 |
+
{
|
| 3064 |
+
"epoch": 1.54,
|
| 3065 |
+
"learning_rate": 0.0002089902912621359,
|
| 3066 |
+
"loss": 0.446,
|
| 3067 |
+
"step": 4790
|
| 3068 |
+
},
|
| 3069 |
+
{
|
| 3070 |
+
"epoch": 1.54,
|
| 3071 |
+
"learning_rate": 0.00020879611650485435,
|
| 3072 |
+
"loss": 0.4228,
|
| 3073 |
+
"step": 4800
|
| 3074 |
+
},
|
| 3075 |
+
{
|
| 3076 |
+
"epoch": 1.54,
|
| 3077 |
+
"eval_loss": 0.4704694151878357,
|
| 3078 |
+
"eval_runtime": 522.9734,
|
| 3079 |
+
"eval_samples_per_second": 3.824,
|
| 3080 |
+
"eval_steps_per_second": 0.478,
|
| 3081 |
+
"step": 4800
|
| 3082 |
+
},
|
| 3083 |
+
{
|
| 3084 |
+
"epoch": 1.55,
|
| 3085 |
+
"learning_rate": 0.0002086019417475728,
|
| 3086 |
+
"loss": 0.4241,
|
| 3087 |
+
"step": 4810
|
| 3088 |
+
},
|
| 3089 |
+
{
|
| 3090 |
+
"epoch": 1.55,
|
| 3091 |
+
"learning_rate": 0.00020840776699029124,
|
| 3092 |
+
"loss": 0.4635,
|
| 3093 |
+
"step": 4820
|
| 3094 |
+
},
|
| 3095 |
+
{
|
| 3096 |
+
"epoch": 1.55,
|
| 3097 |
+
"learning_rate": 0.00020821359223300968,
|
| 3098 |
+
"loss": 0.4618,
|
| 3099 |
+
"step": 4830
|
| 3100 |
+
},
|
| 3101 |
+
{
|
| 3102 |
+
"epoch": 1.56,
|
| 3103 |
+
"learning_rate": 0.00020801941747572813,
|
| 3104 |
+
"loss": 0.4364,
|
| 3105 |
+
"step": 4840
|
| 3106 |
+
},
|
| 3107 |
+
{
|
| 3108 |
+
"epoch": 1.56,
|
| 3109 |
+
"learning_rate": 0.00020782524271844657,
|
| 3110 |
+
"loss": 0.4532,
|
| 3111 |
+
"step": 4850
|
| 3112 |
+
},
|
| 3113 |
+
{
|
| 3114 |
+
"epoch": 1.56,
|
| 3115 |
+
"learning_rate": 0.00020763106796116502,
|
| 3116 |
+
"loss": 0.4522,
|
| 3117 |
+
"step": 4860
|
| 3118 |
+
},
|
| 3119 |
+
{
|
| 3120 |
+
"epoch": 1.57,
|
| 3121 |
+
"learning_rate": 0.0002074368932038835,
|
| 3122 |
+
"loss": 0.4459,
|
| 3123 |
+
"step": 4870
|
| 3124 |
+
},
|
| 3125 |
+
{
|
| 3126 |
+
"epoch": 1.57,
|
| 3127 |
+
"learning_rate": 0.00020724271844660194,
|
| 3128 |
+
"loss": 0.4599,
|
| 3129 |
+
"step": 4880
|
| 3130 |
+
},
|
| 3131 |
+
{
|
| 3132 |
+
"epoch": 1.57,
|
| 3133 |
+
"learning_rate": 0.00020704854368932039,
|
| 3134 |
+
"loss": 0.4526,
|
| 3135 |
+
"step": 4890
|
| 3136 |
+
},
|
| 3137 |
+
{
|
| 3138 |
+
"epoch": 1.58,
|
| 3139 |
+
"learning_rate": 0.00020685436893203883,
|
| 3140 |
+
"loss": 0.4497,
|
| 3141 |
+
"step": 4900
|
| 3142 |
+
},
|
| 3143 |
+
{
|
| 3144 |
+
"epoch": 1.58,
|
| 3145 |
+
"learning_rate": 0.00020666019417475728,
|
| 3146 |
+
"loss": 0.4357,
|
| 3147 |
+
"step": 4910
|
| 3148 |
+
},
|
| 3149 |
+
{
|
| 3150 |
+
"epoch": 1.58,
|
| 3151 |
+
"learning_rate": 0.00020646601941747572,
|
| 3152 |
+
"loss": 0.4353,
|
| 3153 |
+
"step": 4920
|
| 3154 |
+
},
|
| 3155 |
+
{
|
| 3156 |
+
"epoch": 1.59,
|
| 3157 |
+
"learning_rate": 0.00020627184466019414,
|
| 3158 |
+
"loss": 0.4383,
|
| 3159 |
+
"step": 4930
|
| 3160 |
+
},
|
| 3161 |
+
{
|
| 3162 |
+
"epoch": 1.59,
|
| 3163 |
+
"learning_rate": 0.0002060776699029126,
|
| 3164 |
+
"loss": 0.4654,
|
| 3165 |
+
"step": 4940
|
| 3166 |
+
},
|
| 3167 |
+
{
|
| 3168 |
+
"epoch": 1.59,
|
| 3169 |
+
"learning_rate": 0.00020588349514563103,
|
| 3170 |
+
"loss": 0.455,
|
| 3171 |
+
"step": 4950
|
| 3172 |
+
},
|
| 3173 |
+
{
|
| 3174 |
+
"epoch": 1.59,
|
| 3175 |
+
"learning_rate": 0.00020568932038834948,
|
| 3176 |
+
"loss": 0.4544,
|
| 3177 |
+
"step": 4960
|
| 3178 |
+
},
|
| 3179 |
+
{
|
| 3180 |
+
"epoch": 1.6,
|
| 3181 |
+
"learning_rate": 0.00020549514563106792,
|
| 3182 |
+
"loss": 0.4531,
|
| 3183 |
+
"step": 4970
|
| 3184 |
+
},
|
| 3185 |
+
{
|
| 3186 |
+
"epoch": 1.6,
|
| 3187 |
+
"learning_rate": 0.0002053009708737864,
|
| 3188 |
+
"loss": 0.4586,
|
| 3189 |
+
"step": 4980
|
| 3190 |
+
},
|
| 3191 |
+
{
|
| 3192 |
+
"epoch": 1.6,
|
| 3193 |
+
"learning_rate": 0.00020510679611650484,
|
| 3194 |
+
"loss": 0.4648,
|
| 3195 |
+
"step": 4990
|
| 3196 |
+
},
|
| 3197 |
+
{
|
| 3198 |
+
"epoch": 1.61,
|
| 3199 |
+
"learning_rate": 0.0002049126213592233,
|
| 3200 |
+
"loss": 0.4527,
|
| 3201 |
+
"step": 5000
|
| 3202 |
+
},
|
| 3203 |
+
{
|
| 3204 |
+
"epoch": 1.61,
|
| 3205 |
+
"eval_loss": 0.47035008668899536,
|
| 3206 |
+
"eval_runtime": 520.9438,
|
| 3207 |
+
"eval_samples_per_second": 3.839,
|
| 3208 |
+
"eval_steps_per_second": 0.48,
|
| 3209 |
+
"step": 5000
|
| 3210 |
+
},
|
| 3211 |
+
{
|
| 3212 |
+
"epoch": 1.61,
|
| 3213 |
+
"learning_rate": 0.00020471844660194174,
|
| 3214 |
+
"loss": 0.4497,
|
| 3215 |
+
"step": 5010
|
| 3216 |
+
},
|
| 3217 |
+
{
|
| 3218 |
+
"epoch": 1.61,
|
| 3219 |
+
"learning_rate": 0.00020452427184466018,
|
| 3220 |
+
"loss": 0.4321,
|
| 3221 |
+
"step": 5020
|
| 3222 |
+
},
|
| 3223 |
+
{
|
| 3224 |
+
"epoch": 1.62,
|
| 3225 |
+
"learning_rate": 0.00020433009708737863,
|
| 3226 |
+
"loss": 0.4548,
|
| 3227 |
+
"step": 5030
|
| 3228 |
+
},
|
| 3229 |
+
{
|
| 3230 |
+
"epoch": 1.62,
|
| 3231 |
+
"learning_rate": 0.00020413592233009707,
|
| 3232 |
+
"loss": 0.4687,
|
| 3233 |
+
"step": 5040
|
| 3234 |
+
},
|
| 3235 |
+
{
|
| 3236 |
+
"epoch": 1.62,
|
| 3237 |
+
"learning_rate": 0.00020394174757281552,
|
| 3238 |
+
"loss": 0.4773,
|
| 3239 |
+
"step": 5050
|
| 3240 |
+
},
|
| 3241 |
+
{
|
| 3242 |
+
"epoch": 1.63,
|
| 3243 |
+
"learning_rate": 0.00020374757281553396,
|
| 3244 |
+
"loss": 0.4692,
|
| 3245 |
+
"step": 5060
|
| 3246 |
+
},
|
| 3247 |
+
{
|
| 3248 |
+
"epoch": 1.63,
|
| 3249 |
+
"learning_rate": 0.00020355339805825244,
|
| 3250 |
+
"loss": 0.4443,
|
| 3251 |
+
"step": 5070
|
| 3252 |
+
},
|
| 3253 |
+
{
|
| 3254 |
+
"epoch": 1.63,
|
| 3255 |
+
"learning_rate": 0.00020335922330097083,
|
| 3256 |
+
"loss": 0.4751,
|
| 3257 |
+
"step": 5080
|
| 3258 |
+
},
|
| 3259 |
+
{
|
| 3260 |
+
"epoch": 1.64,
|
| 3261 |
+
"learning_rate": 0.0002031650485436893,
|
| 3262 |
+
"loss": 0.4473,
|
| 3263 |
+
"step": 5090
|
| 3264 |
+
},
|
| 3265 |
+
{
|
| 3266 |
+
"epoch": 1.64,
|
| 3267 |
+
"learning_rate": 0.00020297087378640775,
|
| 3268 |
+
"loss": 0.4551,
|
| 3269 |
+
"step": 5100
|
| 3270 |
+
},
|
| 3271 |
+
{
|
| 3272 |
+
"epoch": 1.64,
|
| 3273 |
+
"learning_rate": 0.0002027766990291262,
|
| 3274 |
+
"loss": 0.4753,
|
| 3275 |
+
"step": 5110
|
| 3276 |
+
},
|
| 3277 |
+
{
|
| 3278 |
+
"epoch": 1.65,
|
| 3279 |
+
"learning_rate": 0.00020258252427184464,
|
| 3280 |
+
"loss": 0.4325,
|
| 3281 |
+
"step": 5120
|
| 3282 |
+
},
|
| 3283 |
+
{
|
| 3284 |
+
"epoch": 1.65,
|
| 3285 |
+
"learning_rate": 0.00020238834951456309,
|
| 3286 |
+
"loss": 0.4563,
|
| 3287 |
+
"step": 5130
|
| 3288 |
+
},
|
| 3289 |
+
{
|
| 3290 |
+
"epoch": 1.65,
|
| 3291 |
+
"learning_rate": 0.00020219417475728153,
|
| 3292 |
+
"loss": 0.456,
|
| 3293 |
+
"step": 5140
|
| 3294 |
+
},
|
| 3295 |
+
{
|
| 3296 |
+
"epoch": 1.66,
|
| 3297 |
+
"learning_rate": 0.00020199999999999998,
|
| 3298 |
+
"loss": 0.4598,
|
| 3299 |
+
"step": 5150
|
| 3300 |
+
},
|
| 3301 |
+
{
|
| 3302 |
+
"epoch": 1.66,
|
| 3303 |
+
"learning_rate": 0.00020180582524271842,
|
| 3304 |
+
"loss": 0.4407,
|
| 3305 |
+
"step": 5160
|
| 3306 |
+
},
|
| 3307 |
+
{
|
| 3308 |
+
"epoch": 1.66,
|
| 3309 |
+
"learning_rate": 0.00020161165048543687,
|
| 3310 |
+
"loss": 0.4424,
|
| 3311 |
+
"step": 5170
|
| 3312 |
+
},
|
| 3313 |
+
{
|
| 3314 |
+
"epoch": 1.67,
|
| 3315 |
+
"learning_rate": 0.00020141747572815534,
|
| 3316 |
+
"loss": 0.4291,
|
| 3317 |
+
"step": 5180
|
| 3318 |
+
},
|
| 3319 |
+
{
|
| 3320 |
+
"epoch": 1.67,
|
| 3321 |
+
"learning_rate": 0.0002012233009708738,
|
| 3322 |
+
"loss": 0.4544,
|
| 3323 |
+
"step": 5190
|
| 3324 |
+
},
|
| 3325 |
+
{
|
| 3326 |
+
"epoch": 1.67,
|
| 3327 |
+
"learning_rate": 0.00020102912621359223,
|
| 3328 |
+
"loss": 0.4351,
|
| 3329 |
+
"step": 5200
|
| 3330 |
+
},
|
| 3331 |
+
{
|
| 3332 |
+
"epoch": 1.67,
|
| 3333 |
+
"eval_loss": 0.46933242678642273,
|
| 3334 |
+
"eval_runtime": 521.6481,
|
| 3335 |
+
"eval_samples_per_second": 3.834,
|
| 3336 |
+
"eval_steps_per_second": 0.479,
|
| 3337 |
+
"step": 5200
|
| 3338 |
+
},
|
| 3339 |
+
{
|
| 3340 |
+
"epoch": 1.68,
|
| 3341 |
+
"learning_rate": 0.00020083495145631068,
|
| 3342 |
+
"loss": 0.4503,
|
| 3343 |
+
"step": 5210
|
| 3344 |
+
},
|
| 3345 |
+
{
|
| 3346 |
+
"epoch": 1.68,
|
| 3347 |
+
"learning_rate": 0.0002006407766990291,
|
| 3348 |
+
"loss": 0.4352,
|
| 3349 |
+
"step": 5220
|
| 3350 |
+
},
|
| 3351 |
+
{
|
| 3352 |
+
"epoch": 1.68,
|
| 3353 |
+
"learning_rate": 0.00020044660194174754,
|
| 3354 |
+
"loss": 0.4509,
|
| 3355 |
+
"step": 5230
|
| 3356 |
+
},
|
| 3357 |
+
{
|
| 3358 |
+
"epoch": 1.68,
|
| 3359 |
+
"learning_rate": 0.000200252427184466,
|
| 3360 |
+
"loss": 0.4656,
|
| 3361 |
+
"step": 5240
|
| 3362 |
+
},
|
| 3363 |
+
{
|
| 3364 |
+
"epoch": 1.69,
|
| 3365 |
+
"learning_rate": 0.00020005825242718444,
|
| 3366 |
+
"loss": 0.4578,
|
| 3367 |
+
"step": 5250
|
| 3368 |
+
},
|
| 3369 |
+
{
|
| 3370 |
+
"epoch": 1.69,
|
| 3371 |
+
"learning_rate": 0.00019986407766990288,
|
| 3372 |
+
"loss": 0.4338,
|
| 3373 |
+
"step": 5260
|
| 3374 |
+
},
|
| 3375 |
+
{
|
| 3376 |
+
"epoch": 1.69,
|
| 3377 |
+
"learning_rate": 0.00019966990291262133,
|
| 3378 |
+
"loss": 0.4285,
|
| 3379 |
+
"step": 5270
|
| 3380 |
+
},
|
| 3381 |
+
{
|
| 3382 |
+
"epoch": 1.7,
|
| 3383 |
+
"learning_rate": 0.0001994757281553398,
|
| 3384 |
+
"loss": 0.4219,
|
| 3385 |
+
"step": 5280
|
| 3386 |
+
},
|
| 3387 |
+
{
|
| 3388 |
+
"epoch": 1.7,
|
| 3389 |
+
"learning_rate": 0.00019928155339805825,
|
| 3390 |
+
"loss": 0.4709,
|
| 3391 |
+
"step": 5290
|
| 3392 |
+
},
|
| 3393 |
+
{
|
| 3394 |
+
"epoch": 1.7,
|
| 3395 |
+
"learning_rate": 0.0001990873786407767,
|
| 3396 |
+
"loss": 0.4335,
|
| 3397 |
+
"step": 5300
|
| 3398 |
+
},
|
| 3399 |
+
{
|
| 3400 |
+
"epoch": 1.71,
|
| 3401 |
+
"learning_rate": 0.00019889320388349514,
|
| 3402 |
+
"loss": 0.421,
|
| 3403 |
+
"step": 5310
|
| 3404 |
+
},
|
| 3405 |
+
{
|
| 3406 |
+
"epoch": 1.71,
|
| 3407 |
+
"learning_rate": 0.00019869902912621358,
|
| 3408 |
+
"loss": 0.4594,
|
| 3409 |
+
"step": 5320
|
| 3410 |
+
},
|
| 3411 |
+
{
|
| 3412 |
+
"epoch": 1.71,
|
| 3413 |
+
"learning_rate": 0.00019850485436893203,
|
| 3414 |
+
"loss": 0.4361,
|
| 3415 |
+
"step": 5330
|
| 3416 |
+
},
|
| 3417 |
+
{
|
| 3418 |
+
"epoch": 1.72,
|
| 3419 |
+
"learning_rate": 0.00019831067961165047,
|
| 3420 |
+
"loss": 0.4673,
|
| 3421 |
+
"step": 5340
|
| 3422 |
+
},
|
| 3423 |
+
{
|
| 3424 |
+
"epoch": 1.72,
|
| 3425 |
+
"learning_rate": 0.00019811650485436892,
|
| 3426 |
+
"loss": 0.4499,
|
| 3427 |
+
"step": 5350
|
| 3428 |
+
},
|
| 3429 |
+
{
|
| 3430 |
+
"epoch": 1.72,
|
| 3431 |
+
"learning_rate": 0.00019792233009708734,
|
| 3432 |
+
"loss": 0.4634,
|
| 3433 |
+
"step": 5360
|
| 3434 |
+
},
|
| 3435 |
+
{
|
| 3436 |
+
"epoch": 1.73,
|
| 3437 |
+
"learning_rate": 0.00019772815533980579,
|
| 3438 |
+
"loss": 0.4267,
|
| 3439 |
+
"step": 5370
|
| 3440 |
+
},
|
| 3441 |
+
{
|
| 3442 |
+
"epoch": 1.73,
|
| 3443 |
+
"learning_rate": 0.00019753398058252423,
|
| 3444 |
+
"loss": 0.453,
|
| 3445 |
+
"step": 5380
|
| 3446 |
+
},
|
| 3447 |
+
{
|
| 3448 |
+
"epoch": 1.73,
|
| 3449 |
+
"learning_rate": 0.0001973398058252427,
|
| 3450 |
+
"loss": 0.4406,
|
| 3451 |
+
"step": 5390
|
| 3452 |
+
},
|
| 3453 |
+
{
|
| 3454 |
+
"epoch": 1.74,
|
| 3455 |
+
"learning_rate": 0.00019714563106796115,
|
| 3456 |
+
"loss": 0.4516,
|
| 3457 |
+
"step": 5400
|
| 3458 |
+
},
|
| 3459 |
+
{
|
| 3460 |
+
"epoch": 1.74,
|
| 3461 |
+
"eval_loss": 0.4691580832004547,
|
| 3462 |
+
"eval_runtime": 518.7603,
|
| 3463 |
+
"eval_samples_per_second": 3.855,
|
| 3464 |
+
"eval_steps_per_second": 0.482,
|
| 3465 |
+
"step": 5400
|
| 3466 |
+
},
|
| 3467 |
+
{
|
| 3468 |
+
"epoch": 1.74,
|
| 3469 |
+
"learning_rate": 0.0001969514563106796,
|
| 3470 |
+
"loss": 0.4744,
|
| 3471 |
+
"step": 5410
|
| 3472 |
+
},
|
| 3473 |
+
{
|
| 3474 |
+
"epoch": 1.74,
|
| 3475 |
+
"learning_rate": 0.00019675728155339804,
|
| 3476 |
+
"loss": 0.4744,
|
| 3477 |
+
"step": 5420
|
| 3478 |
+
},
|
| 3479 |
+
{
|
| 3480 |
+
"epoch": 1.75,
|
| 3481 |
+
"learning_rate": 0.0001965631067961165,
|
| 3482 |
+
"loss": 0.4546,
|
| 3483 |
+
"step": 5430
|
| 3484 |
+
},
|
| 3485 |
+
{
|
| 3486 |
+
"epoch": 1.75,
|
| 3487 |
+
"learning_rate": 0.00019636893203883493,
|
| 3488 |
+
"loss": 0.4553,
|
| 3489 |
+
"step": 5440
|
| 3490 |
+
},
|
| 3491 |
+
{
|
| 3492 |
+
"epoch": 1.75,
|
| 3493 |
+
"learning_rate": 0.00019617475728155338,
|
| 3494 |
+
"loss": 0.4562,
|
| 3495 |
+
"step": 5450
|
| 3496 |
+
},
|
| 3497 |
+
{
|
| 3498 |
+
"epoch": 1.76,
|
| 3499 |
+
"learning_rate": 0.00019598058252427182,
|
| 3500 |
+
"loss": 0.4426,
|
| 3501 |
+
"step": 5460
|
| 3502 |
+
},
|
| 3503 |
+
{
|
| 3504 |
+
"epoch": 1.76,
|
| 3505 |
+
"learning_rate": 0.00019578640776699027,
|
| 3506 |
+
"loss": 0.4425,
|
| 3507 |
+
"step": 5470
|
| 3508 |
+
},
|
| 3509 |
+
{
|
| 3510 |
+
"epoch": 1.76,
|
| 3511 |
+
"learning_rate": 0.00019559223300970874,
|
| 3512 |
+
"loss": 0.4288,
|
| 3513 |
+
"step": 5480
|
| 3514 |
+
},
|
| 3515 |
+
{
|
| 3516 |
+
"epoch": 1.77,
|
| 3517 |
+
"learning_rate": 0.0001953980582524272,
|
| 3518 |
+
"loss": 0.4413,
|
| 3519 |
+
"step": 5490
|
| 3520 |
+
},
|
| 3521 |
+
{
|
| 3522 |
+
"epoch": 1.77,
|
| 3523 |
+
"learning_rate": 0.00019520388349514564,
|
| 3524 |
+
"loss": 0.4488,
|
| 3525 |
+
"step": 5500
|
| 3526 |
+
},
|
| 3527 |
+
{
|
| 3528 |
+
"epoch": 1.77,
|
| 3529 |
+
"learning_rate": 0.00019500970873786405,
|
| 3530 |
+
"loss": 0.4461,
|
| 3531 |
+
"step": 5510
|
| 3532 |
+
},
|
| 3533 |
+
{
|
| 3534 |
+
"epoch": 1.77,
|
| 3535 |
+
"learning_rate": 0.0001948155339805825,
|
| 3536 |
+
"loss": 0.4361,
|
| 3537 |
+
"step": 5520
|
| 3538 |
+
},
|
| 3539 |
+
{
|
| 3540 |
+
"epoch": 1.78,
|
| 3541 |
+
"learning_rate": 0.00019462135922330095,
|
| 3542 |
+
"loss": 0.4146,
|
| 3543 |
+
"step": 5530
|
| 3544 |
+
},
|
| 3545 |
+
{
|
| 3546 |
+
"epoch": 1.78,
|
| 3547 |
+
"learning_rate": 0.0001944271844660194,
|
| 3548 |
+
"loss": 0.4378,
|
| 3549 |
+
"step": 5540
|
| 3550 |
+
},
|
| 3551 |
+
{
|
| 3552 |
+
"epoch": 1.78,
|
| 3553 |
+
"learning_rate": 0.00019423300970873784,
|
| 3554 |
+
"loss": 0.4375,
|
| 3555 |
+
"step": 5550
|
| 3556 |
+
},
|
| 3557 |
+
{
|
| 3558 |
+
"epoch": 1.79,
|
| 3559 |
+
"learning_rate": 0.00019403883495145628,
|
| 3560 |
+
"loss": 0.4541,
|
| 3561 |
+
"step": 5560
|
| 3562 |
+
},
|
| 3563 |
+
{
|
| 3564 |
+
"epoch": 1.79,
|
| 3565 |
+
"learning_rate": 0.00019384466019417473,
|
| 3566 |
+
"loss": 0.4228,
|
| 3567 |
+
"step": 5570
|
| 3568 |
+
},
|
| 3569 |
+
{
|
| 3570 |
+
"epoch": 1.79,
|
| 3571 |
+
"learning_rate": 0.00019365048543689317,
|
| 3572 |
+
"loss": 0.4349,
|
| 3573 |
+
"step": 5580
|
| 3574 |
+
},
|
| 3575 |
+
{
|
| 3576 |
+
"epoch": 1.8,
|
| 3577 |
+
"learning_rate": 0.00019345631067961165,
|
| 3578 |
+
"loss": 0.4401,
|
| 3579 |
+
"step": 5590
|
| 3580 |
+
},
|
| 3581 |
+
{
|
| 3582 |
+
"epoch": 1.8,
|
| 3583 |
+
"learning_rate": 0.0001932621359223301,
|
| 3584 |
+
"loss": 0.4494,
|
| 3585 |
+
"step": 5600
|
| 3586 |
+
},
|
| 3587 |
+
{
|
| 3588 |
+
"epoch": 1.8,
|
| 3589 |
+
"eval_loss": 0.46821147203445435,
|
| 3590 |
+
"eval_runtime": 516.2558,
|
| 3591 |
+
"eval_samples_per_second": 3.874,
|
| 3592 |
+
"eval_steps_per_second": 0.484,
|
| 3593 |
+
"step": 5600
|
| 3594 |
+
},
|
| 3595 |
+
{
|
| 3596 |
+
"epoch": 1.8,
|
| 3597 |
+
"learning_rate": 0.00019306796116504854,
|
| 3598 |
+
"loss": 0.4666,
|
| 3599 |
+
"step": 5610
|
| 3600 |
+
},
|
| 3601 |
+
{
|
| 3602 |
+
"epoch": 1.81,
|
| 3603 |
+
"learning_rate": 0.00019287378640776699,
|
| 3604 |
+
"loss": 0.4385,
|
| 3605 |
+
"step": 5620
|
| 3606 |
+
},
|
| 3607 |
+
{
|
| 3608 |
+
"epoch": 1.81,
|
| 3609 |
+
"learning_rate": 0.00019267961165048543,
|
| 3610 |
+
"loss": 0.4202,
|
| 3611 |
+
"step": 5630
|
| 3612 |
+
},
|
| 3613 |
+
{
|
| 3614 |
+
"epoch": 1.81,
|
| 3615 |
+
"learning_rate": 0.00019248543689320388,
|
| 3616 |
+
"loss": 0.4564,
|
| 3617 |
+
"step": 5640
|
| 3618 |
+
},
|
| 3619 |
+
{
|
| 3620 |
+
"epoch": 1.82,
|
| 3621 |
+
"learning_rate": 0.0001922912621359223,
|
| 3622 |
+
"loss": 0.4315,
|
| 3623 |
+
"step": 5650
|
| 3624 |
+
},
|
| 3625 |
+
{
|
| 3626 |
+
"epoch": 1.82,
|
| 3627 |
+
"learning_rate": 0.00019209708737864074,
|
| 3628 |
+
"loss": 0.4549,
|
| 3629 |
+
"step": 5660
|
| 3630 |
+
},
|
| 3631 |
+
{
|
| 3632 |
+
"epoch": 1.82,
|
| 3633 |
+
"learning_rate": 0.0001919029126213592,
|
| 3634 |
+
"loss": 0.4515,
|
| 3635 |
+
"step": 5670
|
| 3636 |
+
},
|
| 3637 |
+
{
|
| 3638 |
+
"epoch": 1.83,
|
| 3639 |
+
"learning_rate": 0.00019170873786407763,
|
| 3640 |
+
"loss": 0.4478,
|
| 3641 |
+
"step": 5680
|
| 3642 |
+
},
|
| 3643 |
+
{
|
| 3644 |
+
"epoch": 1.83,
|
| 3645 |
+
"learning_rate": 0.0001915145631067961,
|
| 3646 |
+
"loss": 0.4541,
|
| 3647 |
+
"step": 5690
|
| 3648 |
+
},
|
| 3649 |
+
{
|
| 3650 |
+
"epoch": 1.83,
|
| 3651 |
+
"learning_rate": 0.00019132038834951455,
|
| 3652 |
+
"loss": 0.4282,
|
| 3653 |
+
"step": 5700
|
| 3654 |
+
},
|
| 3655 |
+
{
|
| 3656 |
+
"epoch": 1.84,
|
| 3657 |
+
"learning_rate": 0.000191126213592233,
|
| 3658 |
+
"loss": 0.4333,
|
| 3659 |
+
"step": 5710
|
| 3660 |
+
},
|
| 3661 |
+
{
|
| 3662 |
+
"epoch": 1.84,
|
| 3663 |
+
"learning_rate": 0.00019093203883495144,
|
| 3664 |
+
"loss": 0.4587,
|
| 3665 |
+
"step": 5720
|
| 3666 |
+
},
|
| 3667 |
+
{
|
| 3668 |
+
"epoch": 1.84,
|
| 3669 |
+
"learning_rate": 0.0001907378640776699,
|
| 3670 |
+
"loss": 0.4671,
|
| 3671 |
+
"step": 5730
|
| 3672 |
+
},
|
| 3673 |
+
{
|
| 3674 |
+
"epoch": 1.85,
|
| 3675 |
+
"learning_rate": 0.00019054368932038834,
|
| 3676 |
+
"loss": 0.4618,
|
| 3677 |
+
"step": 5740
|
| 3678 |
+
},
|
| 3679 |
+
{
|
| 3680 |
+
"epoch": 1.85,
|
| 3681 |
+
"learning_rate": 0.00019034951456310678,
|
| 3682 |
+
"loss": 0.4576,
|
| 3683 |
+
"step": 5750
|
| 3684 |
+
},
|
| 3685 |
+
{
|
| 3686 |
+
"epoch": 1.85,
|
| 3687 |
+
"learning_rate": 0.00019015533980582523,
|
| 3688 |
+
"loss": 0.4435,
|
| 3689 |
+
"step": 5760
|
| 3690 |
+
},
|
| 3691 |
+
{
|
| 3692 |
+
"epoch": 1.86,
|
| 3693 |
+
"learning_rate": 0.00018996116504854367,
|
| 3694 |
+
"loss": 0.4311,
|
| 3695 |
+
"step": 5770
|
| 3696 |
+
},
|
| 3697 |
+
{
|
| 3698 |
+
"epoch": 1.86,
|
| 3699 |
+
"learning_rate": 0.00018976699029126215,
|
| 3700 |
+
"loss": 0.4414,
|
| 3701 |
+
"step": 5780
|
| 3702 |
+
},
|
| 3703 |
+
{
|
| 3704 |
+
"epoch": 1.86,
|
| 3705 |
+
"learning_rate": 0.0001895728155339806,
|
| 3706 |
+
"loss": 0.455,
|
| 3707 |
+
"step": 5790
|
| 3708 |
+
},
|
| 3709 |
+
{
|
| 3710 |
+
"epoch": 1.86,
|
| 3711 |
+
"learning_rate": 0.000189378640776699,
|
| 3712 |
+
"loss": 0.47,
|
| 3713 |
+
"step": 5800
|
| 3714 |
+
},
|
| 3715 |
+
{
|
| 3716 |
+
"epoch": 1.86,
|
| 3717 |
+
"eval_loss": 0.46795403957366943,
|
| 3718 |
+
"eval_runtime": 528.6661,
|
| 3719 |
+
"eval_samples_per_second": 3.783,
|
| 3720 |
+
"eval_steps_per_second": 0.473,
|
| 3721 |
+
"step": 5800
|
| 3722 |
+
},
|
| 3723 |
+
{
|
| 3724 |
+
"epoch": 1.87,
|
| 3725 |
+
"learning_rate": 0.00018918446601941746,
|
| 3726 |
+
"loss": 0.4624,
|
| 3727 |
+
"step": 5810
|
| 3728 |
+
},
|
| 3729 |
+
{
|
| 3730 |
+
"epoch": 1.87,
|
| 3731 |
+
"learning_rate": 0.0001889902912621359,
|
| 3732 |
+
"loss": 0.4606,
|
| 3733 |
+
"step": 5820
|
| 3734 |
+
},
|
| 3735 |
+
{
|
| 3736 |
+
"epoch": 1.87,
|
| 3737 |
+
"learning_rate": 0.00018879611650485435,
|
| 3738 |
+
"loss": 0.4533,
|
| 3739 |
+
"step": 5830
|
| 3740 |
+
},
|
| 3741 |
+
{
|
| 3742 |
+
"epoch": 1.88,
|
| 3743 |
+
"learning_rate": 0.0001886019417475728,
|
| 3744 |
+
"loss": 0.4458,
|
| 3745 |
+
"step": 5840
|
| 3746 |
+
},
|
| 3747 |
+
{
|
| 3748 |
+
"epoch": 1.88,
|
| 3749 |
+
"learning_rate": 0.00018840776699029124,
|
| 3750 |
+
"loss": 0.4455,
|
| 3751 |
+
"step": 5850
|
| 3752 |
+
},
|
| 3753 |
+
{
|
| 3754 |
+
"epoch": 1.88,
|
| 3755 |
+
"learning_rate": 0.00018821359223300968,
|
| 3756 |
+
"loss": 0.4416,
|
| 3757 |
+
"step": 5860
|
| 3758 |
+
},
|
| 3759 |
+
{
|
| 3760 |
+
"epoch": 1.89,
|
| 3761 |
+
"learning_rate": 0.00018801941747572813,
|
| 3762 |
+
"loss": 0.4364,
|
| 3763 |
+
"step": 5870
|
| 3764 |
+
},
|
| 3765 |
+
{
|
| 3766 |
+
"epoch": 1.89,
|
| 3767 |
+
"learning_rate": 0.00018782524271844658,
|
| 3768 |
+
"loss": 0.4537,
|
| 3769 |
+
"step": 5880
|
| 3770 |
+
},
|
| 3771 |
+
{
|
| 3772 |
+
"epoch": 1.89,
|
| 3773 |
+
"learning_rate": 0.00018763106796116505,
|
| 3774 |
+
"loss": 0.4368,
|
| 3775 |
+
"step": 5890
|
| 3776 |
+
},
|
| 3777 |
+
{
|
| 3778 |
+
"epoch": 1.9,
|
| 3779 |
+
"learning_rate": 0.0001874368932038835,
|
| 3780 |
+
"loss": 0.4618,
|
| 3781 |
+
"step": 5900
|
| 3782 |
+
},
|
| 3783 |
+
{
|
| 3784 |
+
"epoch": 1.9,
|
| 3785 |
+
"learning_rate": 0.00018724271844660194,
|
| 3786 |
+
"loss": 0.4295,
|
| 3787 |
+
"step": 5910
|
| 3788 |
+
},
|
| 3789 |
+
{
|
| 3790 |
+
"epoch": 1.9,
|
| 3791 |
+
"learning_rate": 0.0001870485436893204,
|
| 3792 |
+
"loss": 0.437,
|
| 3793 |
+
"step": 5920
|
| 3794 |
+
},
|
| 3795 |
+
{
|
| 3796 |
+
"epoch": 1.91,
|
| 3797 |
+
"learning_rate": 0.00018685436893203883,
|
| 3798 |
+
"loss": 0.4455,
|
| 3799 |
+
"step": 5930
|
| 3800 |
+
},
|
| 3801 |
+
{
|
| 3802 |
+
"epoch": 1.91,
|
| 3803 |
+
"learning_rate": 0.00018666019417475725,
|
| 3804 |
+
"loss": 0.4536,
|
| 3805 |
+
"step": 5940
|
| 3806 |
+
},
|
| 3807 |
+
{
|
| 3808 |
+
"epoch": 1.91,
|
| 3809 |
+
"learning_rate": 0.0001864660194174757,
|
| 3810 |
+
"loss": 0.4268,
|
| 3811 |
+
"step": 5950
|
| 3812 |
+
},
|
| 3813 |
+
{
|
| 3814 |
+
"epoch": 1.92,
|
| 3815 |
+
"learning_rate": 0.00018627184466019414,
|
| 3816 |
+
"loss": 0.4446,
|
| 3817 |
+
"step": 5960
|
| 3818 |
+
},
|
| 3819 |
+
{
|
| 3820 |
+
"epoch": 1.92,
|
| 3821 |
+
"learning_rate": 0.0001860776699029126,
|
| 3822 |
+
"loss": 0.4402,
|
| 3823 |
+
"step": 5970
|
| 3824 |
+
},
|
| 3825 |
+
{
|
| 3826 |
+
"epoch": 1.92,
|
| 3827 |
+
"learning_rate": 0.00018588349514563103,
|
| 3828 |
+
"loss": 0.4461,
|
| 3829 |
+
"step": 5980
|
| 3830 |
+
},
|
| 3831 |
+
{
|
| 3832 |
+
"epoch": 1.93,
|
| 3833 |
+
"learning_rate": 0.00018568932038834948,
|
| 3834 |
+
"loss": 0.4428,
|
| 3835 |
+
"step": 5990
|
| 3836 |
+
},
|
| 3837 |
+
{
|
| 3838 |
+
"epoch": 1.93,
|
| 3839 |
+
"learning_rate": 0.00018549514563106795,
|
| 3840 |
+
"loss": 0.4511,
|
| 3841 |
+
"step": 6000
|
| 3842 |
+
},
|
| 3843 |
+
{
|
| 3844 |
+
"epoch": 1.93,
|
| 3845 |
+
"eval_loss": 0.46734848618507385,
|
| 3846 |
+
"eval_runtime": 541.2095,
|
| 3847 |
+
"eval_samples_per_second": 3.695,
|
| 3848 |
+
"eval_steps_per_second": 0.462,
|
| 3849 |
+
"step": 6000
|
| 3850 |
+
},
|
| 3851 |
+
{
|
| 3852 |
+
"epoch": 1.93,
|
| 3853 |
+
"learning_rate": 0.0001853009708737864,
|
| 3854 |
+
"loss": 0.4351,
|
| 3855 |
+
"step": 6010
|
| 3856 |
+
},
|
| 3857 |
+
{
|
| 3858 |
+
"epoch": 1.94,
|
| 3859 |
+
"learning_rate": 0.00018510679611650485,
|
| 3860 |
+
"loss": 0.4829,
|
| 3861 |
+
"step": 6020
|
| 3862 |
+
},
|
| 3863 |
+
{
|
| 3864 |
+
"epoch": 1.94,
|
| 3865 |
+
"learning_rate": 0.0001849126213592233,
|
| 3866 |
+
"loss": 0.4493,
|
| 3867 |
+
"step": 6030
|
| 3868 |
+
},
|
| 3869 |
+
{
|
| 3870 |
+
"epoch": 1.94,
|
| 3871 |
+
"learning_rate": 0.00018471844660194174,
|
| 3872 |
+
"loss": 0.4435,
|
| 3873 |
+
"step": 6040
|
| 3874 |
+
},
|
| 3875 |
+
{
|
| 3876 |
+
"epoch": 1.95,
|
| 3877 |
+
"learning_rate": 0.00018452427184466018,
|
| 3878 |
+
"loss": 0.4295,
|
| 3879 |
+
"step": 6050
|
| 3880 |
+
},
|
| 3881 |
+
{
|
| 3882 |
+
"epoch": 1.95,
|
| 3883 |
+
"learning_rate": 0.00018433009708737863,
|
| 3884 |
+
"loss": 0.4416,
|
| 3885 |
+
"step": 6060
|
| 3886 |
+
},
|
| 3887 |
+
{
|
| 3888 |
+
"epoch": 1.95,
|
| 3889 |
+
"learning_rate": 0.00018413592233009707,
|
| 3890 |
+
"loss": 0.4233,
|
| 3891 |
+
"step": 6070
|
| 3892 |
+
},
|
| 3893 |
+
{
|
| 3894 |
+
"epoch": 1.95,
|
| 3895 |
+
"learning_rate": 0.00018394174757281555,
|
| 3896 |
+
"loss": 0.4569,
|
| 3897 |
+
"step": 6080
|
| 3898 |
+
},
|
| 3899 |
+
{
|
| 3900 |
+
"epoch": 1.96,
|
| 3901 |
+
"learning_rate": 0.00018374757281553394,
|
| 3902 |
+
"loss": 0.4402,
|
| 3903 |
+
"step": 6090
|
| 3904 |
+
},
|
| 3905 |
+
{
|
| 3906 |
+
"epoch": 1.96,
|
| 3907 |
+
"learning_rate": 0.0001835533980582524,
|
| 3908 |
+
"loss": 0.4358,
|
| 3909 |
+
"step": 6100
|
| 3910 |
+
},
|
| 3911 |
+
{
|
| 3912 |
+
"epoch": 1.96,
|
| 3913 |
+
"learning_rate": 0.00018335922330097086,
|
| 3914 |
+
"loss": 0.4552,
|
| 3915 |
+
"step": 6110
|
| 3916 |
+
},
|
| 3917 |
+
{
|
| 3918 |
+
"epoch": 1.97,
|
| 3919 |
+
"learning_rate": 0.0001831650485436893,
|
| 3920 |
+
"loss": 0.4475,
|
| 3921 |
+
"step": 6120
|
| 3922 |
+
},
|
| 3923 |
+
{
|
| 3924 |
+
"epoch": 1.97,
|
| 3925 |
+
"learning_rate": 0.00018297087378640775,
|
| 3926 |
+
"loss": 0.4479,
|
| 3927 |
+
"step": 6130
|
| 3928 |
+
},
|
| 3929 |
+
{
|
| 3930 |
+
"epoch": 1.97,
|
| 3931 |
+
"learning_rate": 0.0001827766990291262,
|
| 3932 |
+
"loss": 0.4538,
|
| 3933 |
+
"step": 6140
|
| 3934 |
+
},
|
| 3935 |
+
{
|
| 3936 |
+
"epoch": 1.98,
|
| 3937 |
+
"learning_rate": 0.00018258252427184464,
|
| 3938 |
+
"loss": 0.4635,
|
| 3939 |
+
"step": 6150
|
| 3940 |
+
},
|
| 3941 |
+
{
|
| 3942 |
+
"epoch": 1.98,
|
| 3943 |
+
"learning_rate": 0.0001823883495145631,
|
| 3944 |
+
"loss": 0.4366,
|
| 3945 |
+
"step": 6160
|
| 3946 |
+
},
|
| 3947 |
+
{
|
| 3948 |
+
"epoch": 1.98,
|
| 3949 |
+
"learning_rate": 0.00018219417475728153,
|
| 3950 |
+
"loss": 0.4194,
|
| 3951 |
+
"step": 6170
|
| 3952 |
+
},
|
| 3953 |
+
{
|
| 3954 |
+
"epoch": 1.99,
|
| 3955 |
+
"learning_rate": 0.00018199999999999998,
|
| 3956 |
+
"loss": 0.4795,
|
| 3957 |
+
"step": 6180
|
| 3958 |
+
},
|
| 3959 |
+
{
|
| 3960 |
+
"epoch": 1.99,
|
| 3961 |
+
"learning_rate": 0.00018180582524271845,
|
| 3962 |
+
"loss": 0.4484,
|
| 3963 |
+
"step": 6190
|
| 3964 |
+
},
|
| 3965 |
+
{
|
| 3966 |
+
"epoch": 1.99,
|
| 3967 |
+
"learning_rate": 0.0001816116504854369,
|
| 3968 |
+
"loss": 0.4451,
|
| 3969 |
+
"step": 6200
|
| 3970 |
+
},
|
| 3971 |
+
{
|
| 3972 |
+
"epoch": 1.99,
|
| 3973 |
+
"eval_loss": 0.4666987955570221,
|
| 3974 |
+
"eval_runtime": 560.4767,
|
| 3975 |
+
"eval_samples_per_second": 3.568,
|
| 3976 |
+
"eval_steps_per_second": 0.446,
|
| 3977 |
+
"step": 6200
|
| 3978 |
+
}
|
| 3979 |
+
],
|
| 3980 |
+
"logging_steps": 10,
|
| 3981 |
+
"max_steps": 15550,
|
| 3982 |
+
"num_train_epochs": 5,
|
| 3983 |
+
"save_steps": 200,
|
| 3984 |
+
"total_flos": 1.5245364281459343e+18,
|
| 3985 |
+
"trial_name": null,
|
| 3986 |
+
"trial_params": null
|
| 3987 |
+
}
|
1b_52k/best_model/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d0bbfcf35518c9d4954ee618a998a985354ed72127eb8b29c48fd9060ce68685
|
| 3 |
+
size 4027
|
alpaca_text_52K/best_model/adapter_config.json
ADDED
|
@@ -0,0 +1,21 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"base_model_name_or_path": "huggyllama/llama-7b",
|
| 3 |
+
"bias": "none",
|
| 4 |
+
"enable_lora": null,
|
| 5 |
+
"fan_in_fan_out": false,
|
| 6 |
+
"inference_mode": true,
|
| 7 |
+
"init_lora_weights": true,
|
| 8 |
+
"lora_alpha": 16,
|
| 9 |
+
"lora_dropout": 0.05,
|
| 10 |
+
"merge_weights": false,
|
| 11 |
+
"modules_to_save": null,
|
| 12 |
+
"peft_type": "LORA",
|
| 13 |
+
"r": 16,
|
| 14 |
+
"target_modules": [
|
| 15 |
+
"q_proj",
|
| 16 |
+
"k_proj",
|
| 17 |
+
"v_proj",
|
| 18 |
+
"o_proj"
|
| 19 |
+
],
|
| 20 |
+
"task_type": "CAUSAL_LM"
|
| 21 |
+
}
|
alpaca_text_52K/best_model/adapter_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1cf0eee3aa5238d429c69dd5cda61c3a39c10e37c171ba3b7252ab2f82610f83
|
| 3 |
+
size 67201357
|
alpaca_text_52K/best_model/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1061da3884c7f2382c57724180759f46a51091bb905941cf000f9f4083ad40b7
|
| 3 |
+
size 134433093
|
alpaca_text_52K/best_model/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:05dd6bdef1188b917af2cf086637afd93711481154b752237000241493a7d9f7
|
| 3 |
+
size 14575
|
alpaca_text_52K/best_model/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:604dec44d466307737354134b927f551b8e9372dcf7e0c7602123bda7cd9adfc
|
| 3 |
+
size 627
|
alpaca_text_52K/best_model/trainer_state.json
ADDED
|
@@ -0,0 +1,3859 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 1.0564384460449219,
|
| 3 |
+
"best_model_checkpoint": "checkpoints/alpaca_text_52K/checkpoint-6000",
|
| 4 |
+
"epoch": 1.9198464122870171,
|
| 5 |
+
"eval_steps": 200,
|
| 6 |
+
"global_step": 6000,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.0,
|
| 13 |
+
"learning_rate": 1.4999999999999999e-05,
|
| 14 |
+
"loss": 2.2432,
|
| 15 |
+
"step": 10
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"epoch": 0.01,
|
| 19 |
+
"learning_rate": 4.4999999999999996e-05,
|
| 20 |
+
"loss": 2.184,
|
| 21 |
+
"step": 20
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"epoch": 0.01,
|
| 25 |
+
"learning_rate": 7.5e-05,
|
| 26 |
+
"loss": 2.1422,
|
| 27 |
+
"step": 30
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.01,
|
| 31 |
+
"learning_rate": 0.000102,
|
| 32 |
+
"loss": 1.7323,
|
| 33 |
+
"step": 40
|
| 34 |
+
},
|
| 35 |
+
{
|
| 36 |
+
"epoch": 0.02,
|
| 37 |
+
"learning_rate": 0.00013199999999999998,
|
| 38 |
+
"loss": 1.5829,
|
| 39 |
+
"step": 50
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.02,
|
| 43 |
+
"learning_rate": 0.000162,
|
| 44 |
+
"loss": 1.5178,
|
| 45 |
+
"step": 60
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.02,
|
| 49 |
+
"learning_rate": 0.00019199999999999998,
|
| 50 |
+
"loss": 1.2123,
|
| 51 |
+
"step": 70
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.03,
|
| 55 |
+
"learning_rate": 0.00022199999999999998,
|
| 56 |
+
"loss": 1.1177,
|
| 57 |
+
"step": 80
|
| 58 |
+
},
|
| 59 |
+
{
|
| 60 |
+
"epoch": 0.03,
|
| 61 |
+
"learning_rate": 0.00025199999999999995,
|
| 62 |
+
"loss": 1.1426,
|
| 63 |
+
"step": 90
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.03,
|
| 67 |
+
"learning_rate": 0.00028199999999999997,
|
| 68 |
+
"loss": 1.1881,
|
| 69 |
+
"step": 100
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.04,
|
| 73 |
+
"learning_rate": 0.0002999227053140096,
|
| 74 |
+
"loss": 1.1073,
|
| 75 |
+
"step": 110
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"epoch": 0.04,
|
| 79 |
+
"learning_rate": 0.0002997294685990338,
|
| 80 |
+
"loss": 1.1338,
|
| 81 |
+
"step": 120
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.04,
|
| 85 |
+
"learning_rate": 0.00029953623188405795,
|
| 86 |
+
"loss": 1.0979,
|
| 87 |
+
"step": 130
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.04,
|
| 91 |
+
"learning_rate": 0.0002993429951690821,
|
| 92 |
+
"loss": 1.1325,
|
| 93 |
+
"step": 140
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.05,
|
| 97 |
+
"learning_rate": 0.00029914975845410623,
|
| 98 |
+
"loss": 1.1361,
|
| 99 |
+
"step": 150
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.05,
|
| 103 |
+
"learning_rate": 0.0002989565217391304,
|
| 104 |
+
"loss": 1.1412,
|
| 105 |
+
"step": 160
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 0.05,
|
| 109 |
+
"learning_rate": 0.00029876328502415457,
|
| 110 |
+
"loss": 1.0977,
|
| 111 |
+
"step": 170
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"epoch": 0.06,
|
| 115 |
+
"learning_rate": 0.0002985700483091787,
|
| 116 |
+
"loss": 1.1252,
|
| 117 |
+
"step": 180
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 0.06,
|
| 121 |
+
"learning_rate": 0.00029837681159420286,
|
| 122 |
+
"loss": 1.1185,
|
| 123 |
+
"step": 190
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"epoch": 0.06,
|
| 127 |
+
"learning_rate": 0.000298183574879227,
|
| 128 |
+
"loss": 1.0904,
|
| 129 |
+
"step": 200
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.06,
|
| 133 |
+
"eval_loss": 1.1103301048278809,
|
| 134 |
+
"eval_runtime": 299.3056,
|
| 135 |
+
"eval_samples_per_second": 6.682,
|
| 136 |
+
"eval_steps_per_second": 0.835,
|
| 137 |
+
"step": 200
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 0.07,
|
| 141 |
+
"learning_rate": 0.0002979903381642512,
|
| 142 |
+
"loss": 1.1097,
|
| 143 |
+
"step": 210
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.07,
|
| 147 |
+
"learning_rate": 0.00029779710144927534,
|
| 148 |
+
"loss": 1.1154,
|
| 149 |
+
"step": 220
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.07,
|
| 153 |
+
"learning_rate": 0.0002976038647342995,
|
| 154 |
+
"loss": 1.1053,
|
| 155 |
+
"step": 230
|
| 156 |
+
},
|
| 157 |
+
{
|
| 158 |
+
"epoch": 0.08,
|
| 159 |
+
"learning_rate": 0.0002974106280193236,
|
| 160 |
+
"loss": 1.1247,
|
| 161 |
+
"step": 240
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"epoch": 0.08,
|
| 165 |
+
"learning_rate": 0.00029721739130434777,
|
| 166 |
+
"loss": 1.1455,
|
| 167 |
+
"step": 250
|
| 168 |
+
},
|
| 169 |
+
{
|
| 170 |
+
"epoch": 0.08,
|
| 171 |
+
"learning_rate": 0.00029702415458937196,
|
| 172 |
+
"loss": 1.1066,
|
| 173 |
+
"step": 260
|
| 174 |
+
},
|
| 175 |
+
{
|
| 176 |
+
"epoch": 0.09,
|
| 177 |
+
"learning_rate": 0.0002968309178743961,
|
| 178 |
+
"loss": 1.0765,
|
| 179 |
+
"step": 270
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 0.09,
|
| 183 |
+
"learning_rate": 0.00029663768115942025,
|
| 184 |
+
"loss": 1.0976,
|
| 185 |
+
"step": 280
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.09,
|
| 189 |
+
"learning_rate": 0.0002964444444444444,
|
| 190 |
+
"loss": 1.0866,
|
| 191 |
+
"step": 290
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.1,
|
| 195 |
+
"learning_rate": 0.0002962512077294686,
|
| 196 |
+
"loss": 1.0844,
|
| 197 |
+
"step": 300
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 0.1,
|
| 201 |
+
"learning_rate": 0.00029605797101449273,
|
| 202 |
+
"loss": 1.1352,
|
| 203 |
+
"step": 310
|
| 204 |
+
},
|
| 205 |
+
{
|
| 206 |
+
"epoch": 0.1,
|
| 207 |
+
"learning_rate": 0.0002958647342995169,
|
| 208 |
+
"loss": 1.0878,
|
| 209 |
+
"step": 320
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 0.11,
|
| 213 |
+
"learning_rate": 0.000295671497584541,
|
| 214 |
+
"loss": 1.0814,
|
| 215 |
+
"step": 330
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 0.11,
|
| 219 |
+
"learning_rate": 0.00029547826086956516,
|
| 220 |
+
"loss": 1.0676,
|
| 221 |
+
"step": 340
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 0.11,
|
| 225 |
+
"learning_rate": 0.00029528502415458936,
|
| 226 |
+
"loss": 1.043,
|
| 227 |
+
"step": 350
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 0.12,
|
| 231 |
+
"learning_rate": 0.0002950917874396135,
|
| 232 |
+
"loss": 1.0888,
|
| 233 |
+
"step": 360
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.12,
|
| 237 |
+
"learning_rate": 0.00029489855072463764,
|
| 238 |
+
"loss": 1.0948,
|
| 239 |
+
"step": 370
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 0.12,
|
| 243 |
+
"learning_rate": 0.0002947053140096618,
|
| 244 |
+
"loss": 1.0595,
|
| 245 |
+
"step": 380
|
| 246 |
+
},
|
| 247 |
+
{
|
| 248 |
+
"epoch": 0.12,
|
| 249 |
+
"learning_rate": 0.000294512077294686,
|
| 250 |
+
"loss": 1.1063,
|
| 251 |
+
"step": 390
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 0.13,
|
| 255 |
+
"learning_rate": 0.0002943188405797101,
|
| 256 |
+
"loss": 1.0437,
|
| 257 |
+
"step": 400
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 0.13,
|
| 261 |
+
"eval_loss": 1.0938334465026855,
|
| 262 |
+
"eval_runtime": 298.9118,
|
| 263 |
+
"eval_samples_per_second": 6.691,
|
| 264 |
+
"eval_steps_per_second": 0.836,
|
| 265 |
+
"step": 400
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 0.13,
|
| 269 |
+
"learning_rate": 0.00029412560386473427,
|
| 270 |
+
"loss": 1.107,
|
| 271 |
+
"step": 410
|
| 272 |
+
},
|
| 273 |
+
{
|
| 274 |
+
"epoch": 0.13,
|
| 275 |
+
"learning_rate": 0.0002939323671497584,
|
| 276 |
+
"loss": 1.1255,
|
| 277 |
+
"step": 420
|
| 278 |
+
},
|
| 279 |
+
{
|
| 280 |
+
"epoch": 0.14,
|
| 281 |
+
"learning_rate": 0.0002937391304347826,
|
| 282 |
+
"loss": 1.1168,
|
| 283 |
+
"step": 430
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 0.14,
|
| 287 |
+
"learning_rate": 0.00029354589371980675,
|
| 288 |
+
"loss": 1.0599,
|
| 289 |
+
"step": 440
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.14,
|
| 293 |
+
"learning_rate": 0.0002933526570048309,
|
| 294 |
+
"loss": 1.0614,
|
| 295 |
+
"step": 450
|
| 296 |
+
},
|
| 297 |
+
{
|
| 298 |
+
"epoch": 0.15,
|
| 299 |
+
"learning_rate": 0.00029315942028985503,
|
| 300 |
+
"loss": 1.1127,
|
| 301 |
+
"step": 460
|
| 302 |
+
},
|
| 303 |
+
{
|
| 304 |
+
"epoch": 0.15,
|
| 305 |
+
"learning_rate": 0.0002929661835748792,
|
| 306 |
+
"loss": 1.0629,
|
| 307 |
+
"step": 470
|
| 308 |
+
},
|
| 309 |
+
{
|
| 310 |
+
"epoch": 0.15,
|
| 311 |
+
"learning_rate": 0.00029277294685990337,
|
| 312 |
+
"loss": 1.0952,
|
| 313 |
+
"step": 480
|
| 314 |
+
},
|
| 315 |
+
{
|
| 316 |
+
"epoch": 0.16,
|
| 317 |
+
"learning_rate": 0.0002925797101449275,
|
| 318 |
+
"loss": 1.1008,
|
| 319 |
+
"step": 490
|
| 320 |
+
},
|
| 321 |
+
{
|
| 322 |
+
"epoch": 0.16,
|
| 323 |
+
"learning_rate": 0.00029238647342995166,
|
| 324 |
+
"loss": 1.0932,
|
| 325 |
+
"step": 500
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.16,
|
| 329 |
+
"learning_rate": 0.0002921932367149758,
|
| 330 |
+
"loss": 1.0604,
|
| 331 |
+
"step": 510
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.17,
|
| 335 |
+
"learning_rate": 0.000292,
|
| 336 |
+
"loss": 1.0925,
|
| 337 |
+
"step": 520
|
| 338 |
+
},
|
| 339 |
+
{
|
| 340 |
+
"epoch": 0.17,
|
| 341 |
+
"learning_rate": 0.00029180676328502414,
|
| 342 |
+
"loss": 1.0489,
|
| 343 |
+
"step": 530
|
| 344 |
+
},
|
| 345 |
+
{
|
| 346 |
+
"epoch": 0.17,
|
| 347 |
+
"learning_rate": 0.0002916135265700483,
|
| 348 |
+
"loss": 1.1185,
|
| 349 |
+
"step": 540
|
| 350 |
+
},
|
| 351 |
+
{
|
| 352 |
+
"epoch": 0.18,
|
| 353 |
+
"learning_rate": 0.0002914202898550724,
|
| 354 |
+
"loss": 1.0939,
|
| 355 |
+
"step": 550
|
| 356 |
+
},
|
| 357 |
+
{
|
| 358 |
+
"epoch": 0.18,
|
| 359 |
+
"learning_rate": 0.00029122705314009657,
|
| 360 |
+
"loss": 1.1288,
|
| 361 |
+
"step": 560
|
| 362 |
+
},
|
| 363 |
+
{
|
| 364 |
+
"epoch": 0.18,
|
| 365 |
+
"learning_rate": 0.00029103381642512076,
|
| 366 |
+
"loss": 1.0368,
|
| 367 |
+
"step": 570
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 0.19,
|
| 371 |
+
"learning_rate": 0.0002908405797101449,
|
| 372 |
+
"loss": 1.0762,
|
| 373 |
+
"step": 580
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 0.19,
|
| 377 |
+
"learning_rate": 0.00029064734299516905,
|
| 378 |
+
"loss": 1.0966,
|
| 379 |
+
"step": 590
|
| 380 |
+
},
|
| 381 |
+
{
|
| 382 |
+
"epoch": 0.19,
|
| 383 |
+
"learning_rate": 0.0002904541062801932,
|
| 384 |
+
"loss": 1.0941,
|
| 385 |
+
"step": 600
|
| 386 |
+
},
|
| 387 |
+
{
|
| 388 |
+
"epoch": 0.19,
|
| 389 |
+
"eval_loss": 1.0855311155319214,
|
| 390 |
+
"eval_runtime": 299.2543,
|
| 391 |
+
"eval_samples_per_second": 6.683,
|
| 392 |
+
"eval_steps_per_second": 0.835,
|
| 393 |
+
"step": 600
|
| 394 |
+
},
|
| 395 |
+
{
|
| 396 |
+
"epoch": 0.2,
|
| 397 |
+
"learning_rate": 0.0002902608695652174,
|
| 398 |
+
"loss": 1.0732,
|
| 399 |
+
"step": 610
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 0.2,
|
| 403 |
+
"learning_rate": 0.00029006763285024153,
|
| 404 |
+
"loss": 1.0497,
|
| 405 |
+
"step": 620
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 0.2,
|
| 409 |
+
"learning_rate": 0.00028987439613526567,
|
| 410 |
+
"loss": 1.0601,
|
| 411 |
+
"step": 630
|
| 412 |
+
},
|
| 413 |
+
{
|
| 414 |
+
"epoch": 0.2,
|
| 415 |
+
"learning_rate": 0.0002896811594202898,
|
| 416 |
+
"loss": 1.0886,
|
| 417 |
+
"step": 640
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 0.21,
|
| 421 |
+
"learning_rate": 0.00028948792270531396,
|
| 422 |
+
"loss": 1.0681,
|
| 423 |
+
"step": 650
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 0.21,
|
| 427 |
+
"learning_rate": 0.00028929468599033815,
|
| 428 |
+
"loss": 1.0559,
|
| 429 |
+
"step": 660
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.21,
|
| 433 |
+
"learning_rate": 0.0002891014492753623,
|
| 434 |
+
"loss": 1.0879,
|
| 435 |
+
"step": 670
|
| 436 |
+
},
|
| 437 |
+
{
|
| 438 |
+
"epoch": 0.22,
|
| 439 |
+
"learning_rate": 0.00028890821256038644,
|
| 440 |
+
"loss": 1.089,
|
| 441 |
+
"step": 680
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 0.22,
|
| 445 |
+
"learning_rate": 0.0002887149758454106,
|
| 446 |
+
"loss": 1.1005,
|
| 447 |
+
"step": 690
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 0.22,
|
| 451 |
+
"learning_rate": 0.0002885217391304348,
|
| 452 |
+
"loss": 1.1494,
|
| 453 |
+
"step": 700
|
| 454 |
+
},
|
| 455 |
+
{
|
| 456 |
+
"epoch": 0.23,
|
| 457 |
+
"learning_rate": 0.0002883285024154589,
|
| 458 |
+
"loss": 1.0847,
|
| 459 |
+
"step": 710
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 0.23,
|
| 463 |
+
"learning_rate": 0.00028813526570048306,
|
| 464 |
+
"loss": 1.0514,
|
| 465 |
+
"step": 720
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 0.23,
|
| 469 |
+
"learning_rate": 0.0002879420289855072,
|
| 470 |
+
"loss": 1.1038,
|
| 471 |
+
"step": 730
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 0.24,
|
| 475 |
+
"learning_rate": 0.0002877487922705314,
|
| 476 |
+
"loss": 1.0171,
|
| 477 |
+
"step": 740
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 0.24,
|
| 481 |
+
"learning_rate": 0.00028755555555555555,
|
| 482 |
+
"loss": 1.1281,
|
| 483 |
+
"step": 750
|
| 484 |
+
},
|
| 485 |
+
{
|
| 486 |
+
"epoch": 0.24,
|
| 487 |
+
"learning_rate": 0.0002873623188405797,
|
| 488 |
+
"loss": 1.0677,
|
| 489 |
+
"step": 760
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 0.25,
|
| 493 |
+
"learning_rate": 0.00028716908212560383,
|
| 494 |
+
"loss": 1.0963,
|
| 495 |
+
"step": 770
|
| 496 |
+
},
|
| 497 |
+
{
|
| 498 |
+
"epoch": 0.25,
|
| 499 |
+
"learning_rate": 0.000286975845410628,
|
| 500 |
+
"loss": 1.1041,
|
| 501 |
+
"step": 780
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 0.25,
|
| 505 |
+
"learning_rate": 0.00028678260869565217,
|
| 506 |
+
"loss": 1.0847,
|
| 507 |
+
"step": 790
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 0.26,
|
| 511 |
+
"learning_rate": 0.0002865893719806763,
|
| 512 |
+
"loss": 1.0871,
|
| 513 |
+
"step": 800
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 0.26,
|
| 517 |
+
"eval_loss": 1.0817850828170776,
|
| 518 |
+
"eval_runtime": 298.1366,
|
| 519 |
+
"eval_samples_per_second": 6.708,
|
| 520 |
+
"eval_steps_per_second": 0.839,
|
| 521 |
+
"step": 800
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 0.26,
|
| 525 |
+
"learning_rate": 0.00028639613526570046,
|
| 526 |
+
"loss": 1.0593,
|
| 527 |
+
"step": 810
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 0.26,
|
| 531 |
+
"learning_rate": 0.0002862028985507246,
|
| 532 |
+
"loss": 1.0254,
|
| 533 |
+
"step": 820
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 0.27,
|
| 537 |
+
"learning_rate": 0.0002860096618357488,
|
| 538 |
+
"loss": 1.0503,
|
| 539 |
+
"step": 830
|
| 540 |
+
},
|
| 541 |
+
{
|
| 542 |
+
"epoch": 0.27,
|
| 543 |
+
"learning_rate": 0.00028581642512077294,
|
| 544 |
+
"loss": 1.1017,
|
| 545 |
+
"step": 840
|
| 546 |
+
},
|
| 547 |
+
{
|
| 548 |
+
"epoch": 0.27,
|
| 549 |
+
"learning_rate": 0.0002856231884057971,
|
| 550 |
+
"loss": 1.0658,
|
| 551 |
+
"step": 850
|
| 552 |
+
},
|
| 553 |
+
{
|
| 554 |
+
"epoch": 0.28,
|
| 555 |
+
"learning_rate": 0.0002854299516908212,
|
| 556 |
+
"loss": 1.1069,
|
| 557 |
+
"step": 860
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 0.28,
|
| 561 |
+
"learning_rate": 0.00028523671497584537,
|
| 562 |
+
"loss": 1.0603,
|
| 563 |
+
"step": 870
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 0.28,
|
| 567 |
+
"learning_rate": 0.00028504347826086956,
|
| 568 |
+
"loss": 1.082,
|
| 569 |
+
"step": 880
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 0.28,
|
| 573 |
+
"learning_rate": 0.0002848502415458937,
|
| 574 |
+
"loss": 1.1248,
|
| 575 |
+
"step": 890
|
| 576 |
+
},
|
| 577 |
+
{
|
| 578 |
+
"epoch": 0.29,
|
| 579 |
+
"learning_rate": 0.00028465700483091785,
|
| 580 |
+
"loss": 1.1223,
|
| 581 |
+
"step": 900
|
| 582 |
+
},
|
| 583 |
+
{
|
| 584 |
+
"epoch": 0.29,
|
| 585 |
+
"learning_rate": 0.000284463768115942,
|
| 586 |
+
"loss": 1.0672,
|
| 587 |
+
"step": 910
|
| 588 |
+
},
|
| 589 |
+
{
|
| 590 |
+
"epoch": 0.29,
|
| 591 |
+
"learning_rate": 0.0002842705314009662,
|
| 592 |
+
"loss": 1.1095,
|
| 593 |
+
"step": 920
|
| 594 |
+
},
|
| 595 |
+
{
|
| 596 |
+
"epoch": 0.3,
|
| 597 |
+
"learning_rate": 0.00028407729468599033,
|
| 598 |
+
"loss": 1.0227,
|
| 599 |
+
"step": 930
|
| 600 |
+
},
|
| 601 |
+
{
|
| 602 |
+
"epoch": 0.3,
|
| 603 |
+
"learning_rate": 0.00028388405797101447,
|
| 604 |
+
"loss": 1.0969,
|
| 605 |
+
"step": 940
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 0.3,
|
| 609 |
+
"learning_rate": 0.0002836908212560386,
|
| 610 |
+
"loss": 1.0638,
|
| 611 |
+
"step": 950
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 0.31,
|
| 615 |
+
"learning_rate": 0.00028349758454106276,
|
| 616 |
+
"loss": 1.0892,
|
| 617 |
+
"step": 960
|
| 618 |
+
},
|
| 619 |
+
{
|
| 620 |
+
"epoch": 0.31,
|
| 621 |
+
"learning_rate": 0.00028330434782608695,
|
| 622 |
+
"loss": 1.0565,
|
| 623 |
+
"step": 970
|
| 624 |
+
},
|
| 625 |
+
{
|
| 626 |
+
"epoch": 0.31,
|
| 627 |
+
"learning_rate": 0.0002831111111111111,
|
| 628 |
+
"loss": 1.1078,
|
| 629 |
+
"step": 980
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 0.32,
|
| 633 |
+
"learning_rate": 0.00028291787439613524,
|
| 634 |
+
"loss": 1.1106,
|
| 635 |
+
"step": 990
|
| 636 |
+
},
|
| 637 |
+
{
|
| 638 |
+
"epoch": 0.32,
|
| 639 |
+
"learning_rate": 0.0002827246376811594,
|
| 640 |
+
"loss": 1.066,
|
| 641 |
+
"step": 1000
|
| 642 |
+
},
|
| 643 |
+
{
|
| 644 |
+
"epoch": 0.32,
|
| 645 |
+
"eval_loss": 1.0781841278076172,
|
| 646 |
+
"eval_runtime": 298.6342,
|
| 647 |
+
"eval_samples_per_second": 6.697,
|
| 648 |
+
"eval_steps_per_second": 0.837,
|
| 649 |
+
"step": 1000
|
| 650 |
+
},
|
| 651 |
+
{
|
| 652 |
+
"epoch": 0.32,
|
| 653 |
+
"learning_rate": 0.0002825314009661836,
|
| 654 |
+
"loss": 1.107,
|
| 655 |
+
"step": 1010
|
| 656 |
+
},
|
| 657 |
+
{
|
| 658 |
+
"epoch": 0.33,
|
| 659 |
+
"learning_rate": 0.0002823381642512077,
|
| 660 |
+
"loss": 1.0654,
|
| 661 |
+
"step": 1020
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 0.33,
|
| 665 |
+
"learning_rate": 0.00028214492753623186,
|
| 666 |
+
"loss": 1.0342,
|
| 667 |
+
"step": 1030
|
| 668 |
+
},
|
| 669 |
+
{
|
| 670 |
+
"epoch": 0.33,
|
| 671 |
+
"learning_rate": 0.000281951690821256,
|
| 672 |
+
"loss": 1.088,
|
| 673 |
+
"step": 1040
|
| 674 |
+
},
|
| 675 |
+
{
|
| 676 |
+
"epoch": 0.34,
|
| 677 |
+
"learning_rate": 0.0002817584541062802,
|
| 678 |
+
"loss": 1.0871,
|
| 679 |
+
"step": 1050
|
| 680 |
+
},
|
| 681 |
+
{
|
| 682 |
+
"epoch": 0.34,
|
| 683 |
+
"learning_rate": 0.00028156521739130434,
|
| 684 |
+
"loss": 1.1038,
|
| 685 |
+
"step": 1060
|
| 686 |
+
},
|
| 687 |
+
{
|
| 688 |
+
"epoch": 0.34,
|
| 689 |
+
"learning_rate": 0.0002813719806763285,
|
| 690 |
+
"loss": 1.0912,
|
| 691 |
+
"step": 1070
|
| 692 |
+
},
|
| 693 |
+
{
|
| 694 |
+
"epoch": 0.35,
|
| 695 |
+
"learning_rate": 0.00028117874396135263,
|
| 696 |
+
"loss": 1.0836,
|
| 697 |
+
"step": 1080
|
| 698 |
+
},
|
| 699 |
+
{
|
| 700 |
+
"epoch": 0.35,
|
| 701 |
+
"learning_rate": 0.00028098550724637677,
|
| 702 |
+
"loss": 1.0852,
|
| 703 |
+
"step": 1090
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 0.35,
|
| 707 |
+
"learning_rate": 0.00028079227053140097,
|
| 708 |
+
"loss": 1.0757,
|
| 709 |
+
"step": 1100
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"epoch": 0.36,
|
| 713 |
+
"learning_rate": 0.0002805990338164251,
|
| 714 |
+
"loss": 1.0274,
|
| 715 |
+
"step": 1110
|
| 716 |
+
},
|
| 717 |
+
{
|
| 718 |
+
"epoch": 0.36,
|
| 719 |
+
"learning_rate": 0.00028040579710144925,
|
| 720 |
+
"loss": 1.0815,
|
| 721 |
+
"step": 1120
|
| 722 |
+
},
|
| 723 |
+
{
|
| 724 |
+
"epoch": 0.36,
|
| 725 |
+
"learning_rate": 0.0002802125603864734,
|
| 726 |
+
"loss": 1.0917,
|
| 727 |
+
"step": 1130
|
| 728 |
+
},
|
| 729 |
+
{
|
| 730 |
+
"epoch": 0.36,
|
| 731 |
+
"learning_rate": 0.0002800193236714976,
|
| 732 |
+
"loss": 1.0986,
|
| 733 |
+
"step": 1140
|
| 734 |
+
},
|
| 735 |
+
{
|
| 736 |
+
"epoch": 0.37,
|
| 737 |
+
"learning_rate": 0.00027982608695652174,
|
| 738 |
+
"loss": 1.0712,
|
| 739 |
+
"step": 1150
|
| 740 |
+
},
|
| 741 |
+
{
|
| 742 |
+
"epoch": 0.37,
|
| 743 |
+
"learning_rate": 0.0002796328502415459,
|
| 744 |
+
"loss": 1.0725,
|
| 745 |
+
"step": 1160
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 0.37,
|
| 749 |
+
"learning_rate": 0.00027943961352657,
|
| 750 |
+
"loss": 1.0358,
|
| 751 |
+
"step": 1170
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 0.38,
|
| 755 |
+
"learning_rate": 0.00027924637681159416,
|
| 756 |
+
"loss": 1.0775,
|
| 757 |
+
"step": 1180
|
| 758 |
+
},
|
| 759 |
+
{
|
| 760 |
+
"epoch": 0.38,
|
| 761 |
+
"learning_rate": 0.00027905314009661836,
|
| 762 |
+
"loss": 1.083,
|
| 763 |
+
"step": 1190
|
| 764 |
+
},
|
| 765 |
+
{
|
| 766 |
+
"epoch": 0.38,
|
| 767 |
+
"learning_rate": 0.0002788599033816425,
|
| 768 |
+
"loss": 1.0618,
|
| 769 |
+
"step": 1200
|
| 770 |
+
},
|
| 771 |
+
{
|
| 772 |
+
"epoch": 0.38,
|
| 773 |
+
"eval_loss": 1.0745840072631836,
|
| 774 |
+
"eval_runtime": 300.795,
|
| 775 |
+
"eval_samples_per_second": 6.649,
|
| 776 |
+
"eval_steps_per_second": 0.831,
|
| 777 |
+
"step": 1200
|
| 778 |
+
},
|
| 779 |
+
{
|
| 780 |
+
"epoch": 0.39,
|
| 781 |
+
"learning_rate": 0.00027866666666666665,
|
| 782 |
+
"loss": 1.0851,
|
| 783 |
+
"step": 1210
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 0.39,
|
| 787 |
+
"learning_rate": 0.0002784734299516908,
|
| 788 |
+
"loss": 1.1221,
|
| 789 |
+
"step": 1220
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 0.39,
|
| 793 |
+
"learning_rate": 0.000278280193236715,
|
| 794 |
+
"loss": 1.1222,
|
| 795 |
+
"step": 1230
|
| 796 |
+
},
|
| 797 |
+
{
|
| 798 |
+
"epoch": 0.4,
|
| 799 |
+
"learning_rate": 0.00027808695652173913,
|
| 800 |
+
"loss": 1.0941,
|
| 801 |
+
"step": 1240
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 0.4,
|
| 805 |
+
"learning_rate": 0.00027789371980676327,
|
| 806 |
+
"loss": 1.0994,
|
| 807 |
+
"step": 1250
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 0.4,
|
| 811 |
+
"learning_rate": 0.0002777004830917874,
|
| 812 |
+
"loss": 1.0973,
|
| 813 |
+
"step": 1260
|
| 814 |
+
},
|
| 815 |
+
{
|
| 816 |
+
"epoch": 0.41,
|
| 817 |
+
"learning_rate": 0.00027750724637681156,
|
| 818 |
+
"loss": 0.9984,
|
| 819 |
+
"step": 1270
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 0.41,
|
| 823 |
+
"learning_rate": 0.00027731400966183575,
|
| 824 |
+
"loss": 1.0531,
|
| 825 |
+
"step": 1280
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 0.41,
|
| 829 |
+
"learning_rate": 0.0002771207729468599,
|
| 830 |
+
"loss": 1.0872,
|
| 831 |
+
"step": 1290
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 0.42,
|
| 835 |
+
"learning_rate": 0.00027692753623188404,
|
| 836 |
+
"loss": 1.0575,
|
| 837 |
+
"step": 1300
|
| 838 |
+
},
|
| 839 |
+
{
|
| 840 |
+
"epoch": 0.42,
|
| 841 |
+
"learning_rate": 0.0002767342995169082,
|
| 842 |
+
"loss": 1.0515,
|
| 843 |
+
"step": 1310
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 0.42,
|
| 847 |
+
"learning_rate": 0.0002765410628019324,
|
| 848 |
+
"loss": 1.0888,
|
| 849 |
+
"step": 1320
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 0.43,
|
| 853 |
+
"learning_rate": 0.0002763478260869565,
|
| 854 |
+
"loss": 1.058,
|
| 855 |
+
"step": 1330
|
| 856 |
+
},
|
| 857 |
+
{
|
| 858 |
+
"epoch": 0.43,
|
| 859 |
+
"learning_rate": 0.00027615458937198066,
|
| 860 |
+
"loss": 1.0355,
|
| 861 |
+
"step": 1340
|
| 862 |
+
},
|
| 863 |
+
{
|
| 864 |
+
"epoch": 0.43,
|
| 865 |
+
"learning_rate": 0.0002759613526570048,
|
| 866 |
+
"loss": 1.0612,
|
| 867 |
+
"step": 1350
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 0.44,
|
| 871 |
+
"learning_rate": 0.000275768115942029,
|
| 872 |
+
"loss": 1.1059,
|
| 873 |
+
"step": 1360
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 0.44,
|
| 877 |
+
"learning_rate": 0.00027557487922705314,
|
| 878 |
+
"loss": 1.0836,
|
| 879 |
+
"step": 1370
|
| 880 |
+
},
|
| 881 |
+
{
|
| 882 |
+
"epoch": 0.44,
|
| 883 |
+
"learning_rate": 0.0002753816425120773,
|
| 884 |
+
"loss": 1.0433,
|
| 885 |
+
"step": 1380
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 0.44,
|
| 889 |
+
"learning_rate": 0.00027518840579710143,
|
| 890 |
+
"loss": 1.1255,
|
| 891 |
+
"step": 1390
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 0.45,
|
| 895 |
+
"learning_rate": 0.00027499516908212557,
|
| 896 |
+
"loss": 1.0494,
|
| 897 |
+
"step": 1400
|
| 898 |
+
},
|
| 899 |
+
{
|
| 900 |
+
"epoch": 0.45,
|
| 901 |
+
"eval_loss": 1.0709148645401,
|
| 902 |
+
"eval_runtime": 299.6045,
|
| 903 |
+
"eval_samples_per_second": 6.675,
|
| 904 |
+
"eval_steps_per_second": 0.834,
|
| 905 |
+
"step": 1400
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 0.45,
|
| 909 |
+
"learning_rate": 0.00027480193236714977,
|
| 910 |
+
"loss": 1.0646,
|
| 911 |
+
"step": 1410
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 0.45,
|
| 915 |
+
"learning_rate": 0.0002746086956521739,
|
| 916 |
+
"loss": 1.0492,
|
| 917 |
+
"step": 1420
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"epoch": 0.46,
|
| 921 |
+
"learning_rate": 0.00027441545893719805,
|
| 922 |
+
"loss": 1.1201,
|
| 923 |
+
"step": 1430
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 0.46,
|
| 927 |
+
"learning_rate": 0.0002742222222222222,
|
| 928 |
+
"loss": 1.1151,
|
| 929 |
+
"step": 1440
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 0.46,
|
| 933 |
+
"learning_rate": 0.0002740289855072464,
|
| 934 |
+
"loss": 1.0785,
|
| 935 |
+
"step": 1450
|
| 936 |
+
},
|
| 937 |
+
{
|
| 938 |
+
"epoch": 0.47,
|
| 939 |
+
"learning_rate": 0.00027383574879227054,
|
| 940 |
+
"loss": 1.1167,
|
| 941 |
+
"step": 1460
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 0.47,
|
| 945 |
+
"learning_rate": 0.0002736425120772947,
|
| 946 |
+
"loss": 1.0623,
|
| 947 |
+
"step": 1470
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 0.47,
|
| 951 |
+
"learning_rate": 0.0002734492753623188,
|
| 952 |
+
"loss": 1.084,
|
| 953 |
+
"step": 1480
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 0.48,
|
| 957 |
+
"learning_rate": 0.00027325603864734296,
|
| 958 |
+
"loss": 1.0931,
|
| 959 |
+
"step": 1490
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 0.48,
|
| 963 |
+
"learning_rate": 0.00027306280193236716,
|
| 964 |
+
"loss": 1.0783,
|
| 965 |
+
"step": 1500
|
| 966 |
+
},
|
| 967 |
+
{
|
| 968 |
+
"epoch": 0.48,
|
| 969 |
+
"learning_rate": 0.0002728695652173913,
|
| 970 |
+
"loss": 1.0801,
|
| 971 |
+
"step": 1510
|
| 972 |
+
},
|
| 973 |
+
{
|
| 974 |
+
"epoch": 0.49,
|
| 975 |
+
"learning_rate": 0.00027267632850241544,
|
| 976 |
+
"loss": 1.0362,
|
| 977 |
+
"step": 1520
|
| 978 |
+
},
|
| 979 |
+
{
|
| 980 |
+
"epoch": 0.49,
|
| 981 |
+
"learning_rate": 0.0002724830917874396,
|
| 982 |
+
"loss": 1.1147,
|
| 983 |
+
"step": 1530
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 0.49,
|
| 987 |
+
"learning_rate": 0.0002722898550724638,
|
| 988 |
+
"loss": 1.0659,
|
| 989 |
+
"step": 1540
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 0.5,
|
| 993 |
+
"learning_rate": 0.0002720966183574879,
|
| 994 |
+
"loss": 1.0629,
|
| 995 |
+
"step": 1550
|
| 996 |
+
},
|
| 997 |
+
{
|
| 998 |
+
"epoch": 0.5,
|
| 999 |
+
"learning_rate": 0.00027190338164251207,
|
| 1000 |
+
"loss": 1.0563,
|
| 1001 |
+
"step": 1560
|
| 1002 |
+
},
|
| 1003 |
+
{
|
| 1004 |
+
"epoch": 0.5,
|
| 1005 |
+
"learning_rate": 0.0002717101449275362,
|
| 1006 |
+
"loss": 1.0661,
|
| 1007 |
+
"step": 1570
|
| 1008 |
+
},
|
| 1009 |
+
{
|
| 1010 |
+
"epoch": 0.51,
|
| 1011 |
+
"learning_rate": 0.00027151690821256035,
|
| 1012 |
+
"loss": 1.0722,
|
| 1013 |
+
"step": 1580
|
| 1014 |
+
},
|
| 1015 |
+
{
|
| 1016 |
+
"epoch": 0.51,
|
| 1017 |
+
"learning_rate": 0.00027132367149758455,
|
| 1018 |
+
"loss": 1.0175,
|
| 1019 |
+
"step": 1590
|
| 1020 |
+
},
|
| 1021 |
+
{
|
| 1022 |
+
"epoch": 0.51,
|
| 1023 |
+
"learning_rate": 0.0002711304347826087,
|
| 1024 |
+
"loss": 1.0829,
|
| 1025 |
+
"step": 1600
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 0.51,
|
| 1029 |
+
"eval_loss": 1.0700575113296509,
|
| 1030 |
+
"eval_runtime": 301.3359,
|
| 1031 |
+
"eval_samples_per_second": 6.637,
|
| 1032 |
+
"eval_steps_per_second": 0.83,
|
| 1033 |
+
"step": 1600
|
| 1034 |
+
},
|
| 1035 |
+
{
|
| 1036 |
+
"epoch": 0.52,
|
| 1037 |
+
"learning_rate": 0.00027093719806763284,
|
| 1038 |
+
"loss": 1.0935,
|
| 1039 |
+
"step": 1610
|
| 1040 |
+
},
|
| 1041 |
+
{
|
| 1042 |
+
"epoch": 0.52,
|
| 1043 |
+
"learning_rate": 0.000270743961352657,
|
| 1044 |
+
"loss": 1.094,
|
| 1045 |
+
"step": 1620
|
| 1046 |
+
},
|
| 1047 |
+
{
|
| 1048 |
+
"epoch": 0.52,
|
| 1049 |
+
"learning_rate": 0.0002705507246376812,
|
| 1050 |
+
"loss": 1.0603,
|
| 1051 |
+
"step": 1630
|
| 1052 |
+
},
|
| 1053 |
+
{
|
| 1054 |
+
"epoch": 0.52,
|
| 1055 |
+
"learning_rate": 0.0002703574879227053,
|
| 1056 |
+
"loss": 1.0813,
|
| 1057 |
+
"step": 1640
|
| 1058 |
+
},
|
| 1059 |
+
{
|
| 1060 |
+
"epoch": 0.53,
|
| 1061 |
+
"learning_rate": 0.00027016425120772946,
|
| 1062 |
+
"loss": 1.0836,
|
| 1063 |
+
"step": 1650
|
| 1064 |
+
},
|
| 1065 |
+
{
|
| 1066 |
+
"epoch": 0.53,
|
| 1067 |
+
"learning_rate": 0.0002699710144927536,
|
| 1068 |
+
"loss": 1.0538,
|
| 1069 |
+
"step": 1660
|
| 1070 |
+
},
|
| 1071 |
+
{
|
| 1072 |
+
"epoch": 0.53,
|
| 1073 |
+
"learning_rate": 0.00026977777777777775,
|
| 1074 |
+
"loss": 1.0892,
|
| 1075 |
+
"step": 1670
|
| 1076 |
+
},
|
| 1077 |
+
{
|
| 1078 |
+
"epoch": 0.54,
|
| 1079 |
+
"learning_rate": 0.00026958454106280194,
|
| 1080 |
+
"loss": 1.0814,
|
| 1081 |
+
"step": 1680
|
| 1082 |
+
},
|
| 1083 |
+
{
|
| 1084 |
+
"epoch": 0.54,
|
| 1085 |
+
"learning_rate": 0.0002693913043478261,
|
| 1086 |
+
"loss": 1.0407,
|
| 1087 |
+
"step": 1690
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 0.54,
|
| 1091 |
+
"learning_rate": 0.00026919806763285023,
|
| 1092 |
+
"loss": 1.0628,
|
| 1093 |
+
"step": 1700
|
| 1094 |
+
},
|
| 1095 |
+
{
|
| 1096 |
+
"epoch": 0.55,
|
| 1097 |
+
"learning_rate": 0.00026900483091787437,
|
| 1098 |
+
"loss": 1.0678,
|
| 1099 |
+
"step": 1710
|
| 1100 |
+
},
|
| 1101 |
+
{
|
| 1102 |
+
"epoch": 0.55,
|
| 1103 |
+
"learning_rate": 0.0002688115942028985,
|
| 1104 |
+
"loss": 1.0832,
|
| 1105 |
+
"step": 1720
|
| 1106 |
+
},
|
| 1107 |
+
{
|
| 1108 |
+
"epoch": 0.55,
|
| 1109 |
+
"learning_rate": 0.0002686183574879227,
|
| 1110 |
+
"loss": 1.1199,
|
| 1111 |
+
"step": 1730
|
| 1112 |
+
},
|
| 1113 |
+
{
|
| 1114 |
+
"epoch": 0.56,
|
| 1115 |
+
"learning_rate": 0.00026842512077294685,
|
| 1116 |
+
"loss": 1.0674,
|
| 1117 |
+
"step": 1740
|
| 1118 |
+
},
|
| 1119 |
+
{
|
| 1120 |
+
"epoch": 0.56,
|
| 1121 |
+
"learning_rate": 0.000268231884057971,
|
| 1122 |
+
"loss": 1.0719,
|
| 1123 |
+
"step": 1750
|
| 1124 |
+
},
|
| 1125 |
+
{
|
| 1126 |
+
"epoch": 0.56,
|
| 1127 |
+
"learning_rate": 0.00026803864734299514,
|
| 1128 |
+
"loss": 1.0754,
|
| 1129 |
+
"step": 1760
|
| 1130 |
+
},
|
| 1131 |
+
{
|
| 1132 |
+
"epoch": 0.57,
|
| 1133 |
+
"learning_rate": 0.00026784541062801933,
|
| 1134 |
+
"loss": 1.0248,
|
| 1135 |
+
"step": 1770
|
| 1136 |
+
},
|
| 1137 |
+
{
|
| 1138 |
+
"epoch": 0.57,
|
| 1139 |
+
"learning_rate": 0.0002676521739130435,
|
| 1140 |
+
"loss": 1.0659,
|
| 1141 |
+
"step": 1780
|
| 1142 |
+
},
|
| 1143 |
+
{
|
| 1144 |
+
"epoch": 0.57,
|
| 1145 |
+
"learning_rate": 0.0002674589371980676,
|
| 1146 |
+
"loss": 1.0341,
|
| 1147 |
+
"step": 1790
|
| 1148 |
+
},
|
| 1149 |
+
{
|
| 1150 |
+
"epoch": 0.58,
|
| 1151 |
+
"learning_rate": 0.00026726570048309176,
|
| 1152 |
+
"loss": 1.1131,
|
| 1153 |
+
"step": 1800
|
| 1154 |
+
},
|
| 1155 |
+
{
|
| 1156 |
+
"epoch": 0.58,
|
| 1157 |
+
"eval_loss": 1.0687577724456787,
|
| 1158 |
+
"eval_runtime": 298.631,
|
| 1159 |
+
"eval_samples_per_second": 6.697,
|
| 1160 |
+
"eval_steps_per_second": 0.837,
|
| 1161 |
+
"step": 1800
|
| 1162 |
+
},
|
| 1163 |
+
{
|
| 1164 |
+
"epoch": 0.58,
|
| 1165 |
+
"learning_rate": 0.0002670724637681159,
|
| 1166 |
+
"loss": 1.0585,
|
| 1167 |
+
"step": 1810
|
| 1168 |
+
},
|
| 1169 |
+
{
|
| 1170 |
+
"epoch": 0.58,
|
| 1171 |
+
"learning_rate": 0.0002668792270531401,
|
| 1172 |
+
"loss": 1.0657,
|
| 1173 |
+
"step": 1820
|
| 1174 |
+
},
|
| 1175 |
+
{
|
| 1176 |
+
"epoch": 0.59,
|
| 1177 |
+
"learning_rate": 0.00026668599033816424,
|
| 1178 |
+
"loss": 1.0435,
|
| 1179 |
+
"step": 1830
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 0.59,
|
| 1183 |
+
"learning_rate": 0.0002664927536231884,
|
| 1184 |
+
"loss": 1.0697,
|
| 1185 |
+
"step": 1840
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 0.59,
|
| 1189 |
+
"learning_rate": 0.00026629951690821253,
|
| 1190 |
+
"loss": 1.086,
|
| 1191 |
+
"step": 1850
|
| 1192 |
+
},
|
| 1193 |
+
{
|
| 1194 |
+
"epoch": 0.6,
|
| 1195 |
+
"learning_rate": 0.0002661062801932367,
|
| 1196 |
+
"loss": 1.0853,
|
| 1197 |
+
"step": 1860
|
| 1198 |
+
},
|
| 1199 |
+
{
|
| 1200 |
+
"epoch": 0.6,
|
| 1201 |
+
"learning_rate": 0.00026591304347826087,
|
| 1202 |
+
"loss": 1.0214,
|
| 1203 |
+
"step": 1870
|
| 1204 |
+
},
|
| 1205 |
+
{
|
| 1206 |
+
"epoch": 0.6,
|
| 1207 |
+
"learning_rate": 0.000265719806763285,
|
| 1208 |
+
"loss": 1.069,
|
| 1209 |
+
"step": 1880
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"epoch": 0.6,
|
| 1213 |
+
"learning_rate": 0.00026552657004830915,
|
| 1214 |
+
"loss": 1.0793,
|
| 1215 |
+
"step": 1890
|
| 1216 |
+
},
|
| 1217 |
+
{
|
| 1218 |
+
"epoch": 0.61,
|
| 1219 |
+
"learning_rate": 0.0002653333333333333,
|
| 1220 |
+
"loss": 1.0631,
|
| 1221 |
+
"step": 1900
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 0.61,
|
| 1225 |
+
"learning_rate": 0.0002651400966183575,
|
| 1226 |
+
"loss": 1.0605,
|
| 1227 |
+
"step": 1910
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 0.61,
|
| 1231 |
+
"learning_rate": 0.00026494685990338163,
|
| 1232 |
+
"loss": 1.0322,
|
| 1233 |
+
"step": 1920
|
| 1234 |
+
},
|
| 1235 |
+
{
|
| 1236 |
+
"epoch": 0.62,
|
| 1237 |
+
"learning_rate": 0.0002647536231884058,
|
| 1238 |
+
"loss": 1.0644,
|
| 1239 |
+
"step": 1930
|
| 1240 |
+
},
|
| 1241 |
+
{
|
| 1242 |
+
"epoch": 0.62,
|
| 1243 |
+
"learning_rate": 0.0002645603864734299,
|
| 1244 |
+
"loss": 1.0665,
|
| 1245 |
+
"step": 1940
|
| 1246 |
+
},
|
| 1247 |
+
{
|
| 1248 |
+
"epoch": 0.62,
|
| 1249 |
+
"learning_rate": 0.00026436714975845406,
|
| 1250 |
+
"loss": 1.1168,
|
| 1251 |
+
"step": 1950
|
| 1252 |
+
},
|
| 1253 |
+
{
|
| 1254 |
+
"epoch": 0.63,
|
| 1255 |
+
"learning_rate": 0.00026417391304347826,
|
| 1256 |
+
"loss": 1.0518,
|
| 1257 |
+
"step": 1960
|
| 1258 |
+
},
|
| 1259 |
+
{
|
| 1260 |
+
"epoch": 0.63,
|
| 1261 |
+
"learning_rate": 0.0002639806763285024,
|
| 1262 |
+
"loss": 1.0634,
|
| 1263 |
+
"step": 1970
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 0.63,
|
| 1267 |
+
"learning_rate": 0.00026378743961352654,
|
| 1268 |
+
"loss": 1.0483,
|
| 1269 |
+
"step": 1980
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 0.64,
|
| 1273 |
+
"learning_rate": 0.0002635942028985507,
|
| 1274 |
+
"loss": 1.0808,
|
| 1275 |
+
"step": 1990
|
| 1276 |
+
},
|
| 1277 |
+
{
|
| 1278 |
+
"epoch": 0.64,
|
| 1279 |
+
"learning_rate": 0.0002634009661835749,
|
| 1280 |
+
"loss": 1.0519,
|
| 1281 |
+
"step": 2000
|
| 1282 |
+
},
|
| 1283 |
+
{
|
| 1284 |
+
"epoch": 0.64,
|
| 1285 |
+
"eval_loss": 1.0664184093475342,
|
| 1286 |
+
"eval_runtime": 297.6732,
|
| 1287 |
+
"eval_samples_per_second": 6.719,
|
| 1288 |
+
"eval_steps_per_second": 0.84,
|
| 1289 |
+
"step": 2000
|
| 1290 |
+
},
|
| 1291 |
+
{
|
| 1292 |
+
"epoch": 0.64,
|
| 1293 |
+
"learning_rate": 0.000263207729468599,
|
| 1294 |
+
"loss": 1.0351,
|
| 1295 |
+
"step": 2010
|
| 1296 |
+
},
|
| 1297 |
+
{
|
| 1298 |
+
"epoch": 0.65,
|
| 1299 |
+
"learning_rate": 0.00026301449275362317,
|
| 1300 |
+
"loss": 1.0342,
|
| 1301 |
+
"step": 2020
|
| 1302 |
+
},
|
| 1303 |
+
{
|
| 1304 |
+
"epoch": 0.65,
|
| 1305 |
+
"learning_rate": 0.0002628212560386473,
|
| 1306 |
+
"loss": 1.029,
|
| 1307 |
+
"step": 2030
|
| 1308 |
+
},
|
| 1309 |
+
{
|
| 1310 |
+
"epoch": 0.65,
|
| 1311 |
+
"learning_rate": 0.00026262801932367145,
|
| 1312 |
+
"loss": 1.0726,
|
| 1313 |
+
"step": 2040
|
| 1314 |
+
},
|
| 1315 |
+
{
|
| 1316 |
+
"epoch": 0.66,
|
| 1317 |
+
"learning_rate": 0.00026243478260869565,
|
| 1318 |
+
"loss": 1.0187,
|
| 1319 |
+
"step": 2050
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 0.66,
|
| 1323 |
+
"learning_rate": 0.0002622415458937198,
|
| 1324 |
+
"loss": 1.0743,
|
| 1325 |
+
"step": 2060
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 0.66,
|
| 1329 |
+
"learning_rate": 0.00026204830917874394,
|
| 1330 |
+
"loss": 1.0544,
|
| 1331 |
+
"step": 2070
|
| 1332 |
+
},
|
| 1333 |
+
{
|
| 1334 |
+
"epoch": 0.67,
|
| 1335 |
+
"learning_rate": 0.0002618550724637681,
|
| 1336 |
+
"loss": 1.0619,
|
| 1337 |
+
"step": 2080
|
| 1338 |
+
},
|
| 1339 |
+
{
|
| 1340 |
+
"epoch": 0.67,
|
| 1341 |
+
"learning_rate": 0.0002616618357487922,
|
| 1342 |
+
"loss": 1.0642,
|
| 1343 |
+
"step": 2090
|
| 1344 |
+
},
|
| 1345 |
+
{
|
| 1346 |
+
"epoch": 0.67,
|
| 1347 |
+
"learning_rate": 0.0002614685990338164,
|
| 1348 |
+
"loss": 1.0353,
|
| 1349 |
+
"step": 2100
|
| 1350 |
+
},
|
| 1351 |
+
{
|
| 1352 |
+
"epoch": 0.68,
|
| 1353 |
+
"learning_rate": 0.00026127536231884056,
|
| 1354 |
+
"loss": 1.0435,
|
| 1355 |
+
"step": 2110
|
| 1356 |
+
},
|
| 1357 |
+
{
|
| 1358 |
+
"epoch": 0.68,
|
| 1359 |
+
"learning_rate": 0.0002610821256038647,
|
| 1360 |
+
"loss": 1.0759,
|
| 1361 |
+
"step": 2120
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 0.68,
|
| 1365 |
+
"learning_rate": 0.00026088888888888885,
|
| 1366 |
+
"loss": 1.1161,
|
| 1367 |
+
"step": 2130
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 0.68,
|
| 1371 |
+
"learning_rate": 0.00026069565217391304,
|
| 1372 |
+
"loss": 1.0283,
|
| 1373 |
+
"step": 2140
|
| 1374 |
+
},
|
| 1375 |
+
{
|
| 1376 |
+
"epoch": 0.69,
|
| 1377 |
+
"learning_rate": 0.0002605024154589372,
|
| 1378 |
+
"loss": 1.0889,
|
| 1379 |
+
"step": 2150
|
| 1380 |
+
},
|
| 1381 |
+
{
|
| 1382 |
+
"epoch": 0.69,
|
| 1383 |
+
"learning_rate": 0.00026030917874396133,
|
| 1384 |
+
"loss": 1.0649,
|
| 1385 |
+
"step": 2160
|
| 1386 |
+
},
|
| 1387 |
+
{
|
| 1388 |
+
"epoch": 0.69,
|
| 1389 |
+
"learning_rate": 0.00026011594202898547,
|
| 1390 |
+
"loss": 1.0541,
|
| 1391 |
+
"step": 2170
|
| 1392 |
+
},
|
| 1393 |
+
{
|
| 1394 |
+
"epoch": 0.7,
|
| 1395 |
+
"learning_rate": 0.0002599227053140096,
|
| 1396 |
+
"loss": 1.086,
|
| 1397 |
+
"step": 2180
|
| 1398 |
+
},
|
| 1399 |
+
{
|
| 1400 |
+
"epoch": 0.7,
|
| 1401 |
+
"learning_rate": 0.0002597294685990338,
|
| 1402 |
+
"loss": 1.0836,
|
| 1403 |
+
"step": 2190
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 0.7,
|
| 1407 |
+
"learning_rate": 0.00025953623188405795,
|
| 1408 |
+
"loss": 1.0128,
|
| 1409 |
+
"step": 2200
|
| 1410 |
+
},
|
| 1411 |
+
{
|
| 1412 |
+
"epoch": 0.7,
|
| 1413 |
+
"eval_loss": 1.0654979944229126,
|
| 1414 |
+
"eval_runtime": 298.984,
|
| 1415 |
+
"eval_samples_per_second": 6.689,
|
| 1416 |
+
"eval_steps_per_second": 0.836,
|
| 1417 |
+
"step": 2200
|
| 1418 |
+
},
|
| 1419 |
+
{
|
| 1420 |
+
"epoch": 0.71,
|
| 1421 |
+
"learning_rate": 0.0002593429951690821,
|
| 1422 |
+
"loss": 1.0502,
|
| 1423 |
+
"step": 2210
|
| 1424 |
+
},
|
| 1425 |
+
{
|
| 1426 |
+
"epoch": 0.71,
|
| 1427 |
+
"learning_rate": 0.00025914975845410624,
|
| 1428 |
+
"loss": 1.0275,
|
| 1429 |
+
"step": 2220
|
| 1430 |
+
},
|
| 1431 |
+
{
|
| 1432 |
+
"epoch": 0.71,
|
| 1433 |
+
"learning_rate": 0.00025895652173913043,
|
| 1434 |
+
"loss": 1.0567,
|
| 1435 |
+
"step": 2230
|
| 1436 |
+
},
|
| 1437 |
+
{
|
| 1438 |
+
"epoch": 0.72,
|
| 1439 |
+
"learning_rate": 0.0002587632850241546,
|
| 1440 |
+
"loss": 1.1105,
|
| 1441 |
+
"step": 2240
|
| 1442 |
+
},
|
| 1443 |
+
{
|
| 1444 |
+
"epoch": 0.72,
|
| 1445 |
+
"learning_rate": 0.0002585700483091787,
|
| 1446 |
+
"loss": 1.0271,
|
| 1447 |
+
"step": 2250
|
| 1448 |
+
},
|
| 1449 |
+
{
|
| 1450 |
+
"epoch": 0.72,
|
| 1451 |
+
"learning_rate": 0.00025837681159420286,
|
| 1452 |
+
"loss": 1.0471,
|
| 1453 |
+
"step": 2260
|
| 1454 |
+
},
|
| 1455 |
+
{
|
| 1456 |
+
"epoch": 0.73,
|
| 1457 |
+
"learning_rate": 0.000258183574879227,
|
| 1458 |
+
"loss": 1.0862,
|
| 1459 |
+
"step": 2270
|
| 1460 |
+
},
|
| 1461 |
+
{
|
| 1462 |
+
"epoch": 0.73,
|
| 1463 |
+
"learning_rate": 0.0002579903381642512,
|
| 1464 |
+
"loss": 1.0829,
|
| 1465 |
+
"step": 2280
|
| 1466 |
+
},
|
| 1467 |
+
{
|
| 1468 |
+
"epoch": 0.73,
|
| 1469 |
+
"learning_rate": 0.00025779710144927534,
|
| 1470 |
+
"loss": 1.0542,
|
| 1471 |
+
"step": 2290
|
| 1472 |
+
},
|
| 1473 |
+
{
|
| 1474 |
+
"epoch": 0.74,
|
| 1475 |
+
"learning_rate": 0.0002576038647342995,
|
| 1476 |
+
"loss": 1.0632,
|
| 1477 |
+
"step": 2300
|
| 1478 |
+
},
|
| 1479 |
+
{
|
| 1480 |
+
"epoch": 0.74,
|
| 1481 |
+
"learning_rate": 0.00025741062801932363,
|
| 1482 |
+
"loss": 1.0714,
|
| 1483 |
+
"step": 2310
|
| 1484 |
+
},
|
| 1485 |
+
{
|
| 1486 |
+
"epoch": 0.74,
|
| 1487 |
+
"learning_rate": 0.00025721739130434777,
|
| 1488 |
+
"loss": 0.9918,
|
| 1489 |
+
"step": 2320
|
| 1490 |
+
},
|
| 1491 |
+
{
|
| 1492 |
+
"epoch": 0.75,
|
| 1493 |
+
"learning_rate": 0.00025702415458937197,
|
| 1494 |
+
"loss": 1.0651,
|
| 1495 |
+
"step": 2330
|
| 1496 |
+
},
|
| 1497 |
+
{
|
| 1498 |
+
"epoch": 0.75,
|
| 1499 |
+
"learning_rate": 0.0002568309178743961,
|
| 1500 |
+
"loss": 1.0521,
|
| 1501 |
+
"step": 2340
|
| 1502 |
+
},
|
| 1503 |
+
{
|
| 1504 |
+
"epoch": 0.75,
|
| 1505 |
+
"learning_rate": 0.00025663768115942025,
|
| 1506 |
+
"loss": 1.057,
|
| 1507 |
+
"step": 2350
|
| 1508 |
+
},
|
| 1509 |
+
{
|
| 1510 |
+
"epoch": 0.76,
|
| 1511 |
+
"learning_rate": 0.0002564444444444444,
|
| 1512 |
+
"loss": 1.0675,
|
| 1513 |
+
"step": 2360
|
| 1514 |
+
},
|
| 1515 |
+
{
|
| 1516 |
+
"epoch": 0.76,
|
| 1517 |
+
"learning_rate": 0.0002562512077294686,
|
| 1518 |
+
"loss": 1.1029,
|
| 1519 |
+
"step": 2370
|
| 1520 |
+
},
|
| 1521 |
+
{
|
| 1522 |
+
"epoch": 0.76,
|
| 1523 |
+
"learning_rate": 0.00025605797101449273,
|
| 1524 |
+
"loss": 1.0636,
|
| 1525 |
+
"step": 2380
|
| 1526 |
+
},
|
| 1527 |
+
{
|
| 1528 |
+
"epoch": 0.76,
|
| 1529 |
+
"learning_rate": 0.0002558647342995169,
|
| 1530 |
+
"loss": 1.0586,
|
| 1531 |
+
"step": 2390
|
| 1532 |
+
},
|
| 1533 |
+
{
|
| 1534 |
+
"epoch": 0.77,
|
| 1535 |
+
"learning_rate": 0.000255671497584541,
|
| 1536 |
+
"loss": 1.0517,
|
| 1537 |
+
"step": 2400
|
| 1538 |
+
},
|
| 1539 |
+
{
|
| 1540 |
+
"epoch": 0.77,
|
| 1541 |
+
"eval_loss": 1.0638346672058105,
|
| 1542 |
+
"eval_runtime": 305.5987,
|
| 1543 |
+
"eval_samples_per_second": 6.545,
|
| 1544 |
+
"eval_steps_per_second": 0.818,
|
| 1545 |
+
"step": 2400
|
| 1546 |
+
},
|
| 1547 |
+
{
|
| 1548 |
+
"epoch": 0.77,
|
| 1549 |
+
"learning_rate": 0.00025547826086956516,
|
| 1550 |
+
"loss": 1.0955,
|
| 1551 |
+
"step": 2410
|
| 1552 |
+
},
|
| 1553 |
+
{
|
| 1554 |
+
"epoch": 0.77,
|
| 1555 |
+
"learning_rate": 0.00025528502415458936,
|
| 1556 |
+
"loss": 1.0724,
|
| 1557 |
+
"step": 2420
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 0.78,
|
| 1561 |
+
"learning_rate": 0.0002550917874396135,
|
| 1562 |
+
"loss": 1.0682,
|
| 1563 |
+
"step": 2430
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 0.78,
|
| 1567 |
+
"learning_rate": 0.00025489855072463764,
|
| 1568 |
+
"loss": 1.0906,
|
| 1569 |
+
"step": 2440
|
| 1570 |
+
},
|
| 1571 |
+
{
|
| 1572 |
+
"epoch": 0.78,
|
| 1573 |
+
"learning_rate": 0.0002547053140096618,
|
| 1574 |
+
"loss": 1.0637,
|
| 1575 |
+
"step": 2450
|
| 1576 |
+
},
|
| 1577 |
+
{
|
| 1578 |
+
"epoch": 0.79,
|
| 1579 |
+
"learning_rate": 0.00025451207729468593,
|
| 1580 |
+
"loss": 1.0794,
|
| 1581 |
+
"step": 2460
|
| 1582 |
+
},
|
| 1583 |
+
{
|
| 1584 |
+
"epoch": 0.79,
|
| 1585 |
+
"learning_rate": 0.0002543188405797101,
|
| 1586 |
+
"loss": 1.0554,
|
| 1587 |
+
"step": 2470
|
| 1588 |
+
},
|
| 1589 |
+
{
|
| 1590 |
+
"epoch": 0.79,
|
| 1591 |
+
"learning_rate": 0.00025412560386473427,
|
| 1592 |
+
"loss": 1.1048,
|
| 1593 |
+
"step": 2480
|
| 1594 |
+
},
|
| 1595 |
+
{
|
| 1596 |
+
"epoch": 0.8,
|
| 1597 |
+
"learning_rate": 0.0002539323671497584,
|
| 1598 |
+
"loss": 1.0526,
|
| 1599 |
+
"step": 2490
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 0.8,
|
| 1603 |
+
"learning_rate": 0.00025373913043478255,
|
| 1604 |
+
"loss": 1.1208,
|
| 1605 |
+
"step": 2500
|
| 1606 |
+
},
|
| 1607 |
+
{
|
| 1608 |
+
"epoch": 0.8,
|
| 1609 |
+
"learning_rate": 0.00025354589371980675,
|
| 1610 |
+
"loss": 1.058,
|
| 1611 |
+
"step": 2510
|
| 1612 |
+
},
|
| 1613 |
+
{
|
| 1614 |
+
"epoch": 0.81,
|
| 1615 |
+
"learning_rate": 0.0002533526570048309,
|
| 1616 |
+
"loss": 1.0374,
|
| 1617 |
+
"step": 2520
|
| 1618 |
+
},
|
| 1619 |
+
{
|
| 1620 |
+
"epoch": 0.81,
|
| 1621 |
+
"learning_rate": 0.00025315942028985504,
|
| 1622 |
+
"loss": 1.0768,
|
| 1623 |
+
"step": 2530
|
| 1624 |
+
},
|
| 1625 |
+
{
|
| 1626 |
+
"epoch": 0.81,
|
| 1627 |
+
"learning_rate": 0.0002529661835748792,
|
| 1628 |
+
"loss": 1.0313,
|
| 1629 |
+
"step": 2540
|
| 1630 |
+
},
|
| 1631 |
+
{
|
| 1632 |
+
"epoch": 0.82,
|
| 1633 |
+
"learning_rate": 0.0002527729468599033,
|
| 1634 |
+
"loss": 1.0322,
|
| 1635 |
+
"step": 2550
|
| 1636 |
+
},
|
| 1637 |
+
{
|
| 1638 |
+
"epoch": 0.82,
|
| 1639 |
+
"learning_rate": 0.0002525797101449275,
|
| 1640 |
+
"loss": 1.0753,
|
| 1641 |
+
"step": 2560
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 0.82,
|
| 1645 |
+
"learning_rate": 0.00025238647342995166,
|
| 1646 |
+
"loss": 1.0702,
|
| 1647 |
+
"step": 2570
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"epoch": 0.83,
|
| 1651 |
+
"learning_rate": 0.0002521932367149758,
|
| 1652 |
+
"loss": 1.0889,
|
| 1653 |
+
"step": 2580
|
| 1654 |
+
},
|
| 1655 |
+
{
|
| 1656 |
+
"epoch": 0.83,
|
| 1657 |
+
"learning_rate": 0.00025199999999999995,
|
| 1658 |
+
"loss": 1.0907,
|
| 1659 |
+
"step": 2590
|
| 1660 |
+
},
|
| 1661 |
+
{
|
| 1662 |
+
"epoch": 0.83,
|
| 1663 |
+
"learning_rate": 0.00025180676328502414,
|
| 1664 |
+
"loss": 1.0813,
|
| 1665 |
+
"step": 2600
|
| 1666 |
+
},
|
| 1667 |
+
{
|
| 1668 |
+
"epoch": 0.83,
|
| 1669 |
+
"eval_loss": 1.06317138671875,
|
| 1670 |
+
"eval_runtime": 301.103,
|
| 1671 |
+
"eval_samples_per_second": 6.642,
|
| 1672 |
+
"eval_steps_per_second": 0.83,
|
| 1673 |
+
"step": 2600
|
| 1674 |
+
},
|
| 1675 |
+
{
|
| 1676 |
+
"epoch": 0.84,
|
| 1677 |
+
"learning_rate": 0.0002516135265700483,
|
| 1678 |
+
"loss": 1.0696,
|
| 1679 |
+
"step": 2610
|
| 1680 |
+
},
|
| 1681 |
+
{
|
| 1682 |
+
"epoch": 0.84,
|
| 1683 |
+
"learning_rate": 0.00025142028985507243,
|
| 1684 |
+
"loss": 1.05,
|
| 1685 |
+
"step": 2620
|
| 1686 |
+
},
|
| 1687 |
+
{
|
| 1688 |
+
"epoch": 0.84,
|
| 1689 |
+
"learning_rate": 0.00025122705314009657,
|
| 1690 |
+
"loss": 1.049,
|
| 1691 |
+
"step": 2630
|
| 1692 |
+
},
|
| 1693 |
+
{
|
| 1694 |
+
"epoch": 0.84,
|
| 1695 |
+
"learning_rate": 0.0002510338164251207,
|
| 1696 |
+
"loss": 1.0756,
|
| 1697 |
+
"step": 2640
|
| 1698 |
+
},
|
| 1699 |
+
{
|
| 1700 |
+
"epoch": 0.85,
|
| 1701 |
+
"learning_rate": 0.0002508405797101449,
|
| 1702 |
+
"loss": 1.0656,
|
| 1703 |
+
"step": 2650
|
| 1704 |
+
},
|
| 1705 |
+
{
|
| 1706 |
+
"epoch": 0.85,
|
| 1707 |
+
"learning_rate": 0.00025064734299516905,
|
| 1708 |
+
"loss": 1.0376,
|
| 1709 |
+
"step": 2660
|
| 1710 |
+
},
|
| 1711 |
+
{
|
| 1712 |
+
"epoch": 0.85,
|
| 1713 |
+
"learning_rate": 0.0002504541062801932,
|
| 1714 |
+
"loss": 1.0677,
|
| 1715 |
+
"step": 2670
|
| 1716 |
+
},
|
| 1717 |
+
{
|
| 1718 |
+
"epoch": 0.86,
|
| 1719 |
+
"learning_rate": 0.00025026086956521734,
|
| 1720 |
+
"loss": 1.0488,
|
| 1721 |
+
"step": 2680
|
| 1722 |
+
},
|
| 1723 |
+
{
|
| 1724 |
+
"epoch": 0.86,
|
| 1725 |
+
"learning_rate": 0.00025006763285024153,
|
| 1726 |
+
"loss": 1.0751,
|
| 1727 |
+
"step": 2690
|
| 1728 |
+
},
|
| 1729 |
+
{
|
| 1730 |
+
"epoch": 0.86,
|
| 1731 |
+
"learning_rate": 0.0002498743961352657,
|
| 1732 |
+
"loss": 1.054,
|
| 1733 |
+
"step": 2700
|
| 1734 |
+
},
|
| 1735 |
+
{
|
| 1736 |
+
"epoch": 0.87,
|
| 1737 |
+
"learning_rate": 0.0002496811594202898,
|
| 1738 |
+
"loss": 1.0478,
|
| 1739 |
+
"step": 2710
|
| 1740 |
+
},
|
| 1741 |
+
{
|
| 1742 |
+
"epoch": 0.87,
|
| 1743 |
+
"learning_rate": 0.00024948792270531396,
|
| 1744 |
+
"loss": 1.0872,
|
| 1745 |
+
"step": 2720
|
| 1746 |
+
},
|
| 1747 |
+
{
|
| 1748 |
+
"epoch": 0.87,
|
| 1749 |
+
"learning_rate": 0.0002492946859903381,
|
| 1750 |
+
"loss": 1.0308,
|
| 1751 |
+
"step": 2730
|
| 1752 |
+
},
|
| 1753 |
+
{
|
| 1754 |
+
"epoch": 0.88,
|
| 1755 |
+
"learning_rate": 0.0002491014492753623,
|
| 1756 |
+
"loss": 1.0349,
|
| 1757 |
+
"step": 2740
|
| 1758 |
+
},
|
| 1759 |
+
{
|
| 1760 |
+
"epoch": 0.88,
|
| 1761 |
+
"learning_rate": 0.00024890821256038644,
|
| 1762 |
+
"loss": 1.0775,
|
| 1763 |
+
"step": 2750
|
| 1764 |
+
},
|
| 1765 |
+
{
|
| 1766 |
+
"epoch": 0.88,
|
| 1767 |
+
"learning_rate": 0.0002487149758454106,
|
| 1768 |
+
"loss": 1.1327,
|
| 1769 |
+
"step": 2760
|
| 1770 |
+
},
|
| 1771 |
+
{
|
| 1772 |
+
"epoch": 0.89,
|
| 1773 |
+
"learning_rate": 0.00024852173913043473,
|
| 1774 |
+
"loss": 1.0442,
|
| 1775 |
+
"step": 2770
|
| 1776 |
+
},
|
| 1777 |
+
{
|
| 1778 |
+
"epoch": 0.89,
|
| 1779 |
+
"learning_rate": 0.0002483285024154589,
|
| 1780 |
+
"loss": 1.0428,
|
| 1781 |
+
"step": 2780
|
| 1782 |
+
},
|
| 1783 |
+
{
|
| 1784 |
+
"epoch": 0.89,
|
| 1785 |
+
"learning_rate": 0.00024813526570048307,
|
| 1786 |
+
"loss": 1.0661,
|
| 1787 |
+
"step": 2790
|
| 1788 |
+
},
|
| 1789 |
+
{
|
| 1790 |
+
"epoch": 0.9,
|
| 1791 |
+
"learning_rate": 0.0002479420289855072,
|
| 1792 |
+
"loss": 1.1036,
|
| 1793 |
+
"step": 2800
|
| 1794 |
+
},
|
| 1795 |
+
{
|
| 1796 |
+
"epoch": 0.9,
|
| 1797 |
+
"eval_loss": 1.0611621141433716,
|
| 1798 |
+
"eval_runtime": 303.5211,
|
| 1799 |
+
"eval_samples_per_second": 6.589,
|
| 1800 |
+
"eval_steps_per_second": 0.824,
|
| 1801 |
+
"step": 2800
|
| 1802 |
+
},
|
| 1803 |
+
{
|
| 1804 |
+
"epoch": 0.9,
|
| 1805 |
+
"learning_rate": 0.00024774879227053135,
|
| 1806 |
+
"loss": 1.0631,
|
| 1807 |
+
"step": 2810
|
| 1808 |
+
},
|
| 1809 |
+
{
|
| 1810 |
+
"epoch": 0.9,
|
| 1811 |
+
"learning_rate": 0.0002475555555555555,
|
| 1812 |
+
"loss": 1.0571,
|
| 1813 |
+
"step": 2820
|
| 1814 |
+
},
|
| 1815 |
+
{
|
| 1816 |
+
"epoch": 0.91,
|
| 1817 |
+
"learning_rate": 0.0002473623188405797,
|
| 1818 |
+
"loss": 1.0389,
|
| 1819 |
+
"step": 2830
|
| 1820 |
+
},
|
| 1821 |
+
{
|
| 1822 |
+
"epoch": 0.91,
|
| 1823 |
+
"learning_rate": 0.00024716908212560383,
|
| 1824 |
+
"loss": 1.0354,
|
| 1825 |
+
"step": 2840
|
| 1826 |
+
},
|
| 1827 |
+
{
|
| 1828 |
+
"epoch": 0.91,
|
| 1829 |
+
"learning_rate": 0.000246975845410628,
|
| 1830 |
+
"loss": 1.0385,
|
| 1831 |
+
"step": 2850
|
| 1832 |
+
},
|
| 1833 |
+
{
|
| 1834 |
+
"epoch": 0.92,
|
| 1835 |
+
"learning_rate": 0.0002467826086956521,
|
| 1836 |
+
"loss": 1.0689,
|
| 1837 |
+
"step": 2860
|
| 1838 |
+
},
|
| 1839 |
+
{
|
| 1840 |
+
"epoch": 0.92,
|
| 1841 |
+
"learning_rate": 0.0002465893719806763,
|
| 1842 |
+
"loss": 1.0414,
|
| 1843 |
+
"step": 2870
|
| 1844 |
+
},
|
| 1845 |
+
{
|
| 1846 |
+
"epoch": 0.92,
|
| 1847 |
+
"learning_rate": 0.00024639613526570046,
|
| 1848 |
+
"loss": 1.069,
|
| 1849 |
+
"step": 2880
|
| 1850 |
+
},
|
| 1851 |
+
{
|
| 1852 |
+
"epoch": 0.92,
|
| 1853 |
+
"learning_rate": 0.0002462028985507246,
|
| 1854 |
+
"loss": 1.0096,
|
| 1855 |
+
"step": 2890
|
| 1856 |
+
},
|
| 1857 |
+
{
|
| 1858 |
+
"epoch": 0.93,
|
| 1859 |
+
"learning_rate": 0.00024600966183574874,
|
| 1860 |
+
"loss": 1.0859,
|
| 1861 |
+
"step": 2900
|
| 1862 |
+
},
|
| 1863 |
+
{
|
| 1864 |
+
"epoch": 0.93,
|
| 1865 |
+
"learning_rate": 0.00024581642512077294,
|
| 1866 |
+
"loss": 1.0518,
|
| 1867 |
+
"step": 2910
|
| 1868 |
+
},
|
| 1869 |
+
{
|
| 1870 |
+
"epoch": 0.93,
|
| 1871 |
+
"learning_rate": 0.0002456231884057971,
|
| 1872 |
+
"loss": 1.0696,
|
| 1873 |
+
"step": 2920
|
| 1874 |
+
},
|
| 1875 |
+
{
|
| 1876 |
+
"epoch": 0.94,
|
| 1877 |
+
"learning_rate": 0.0002454299516908212,
|
| 1878 |
+
"loss": 1.0559,
|
| 1879 |
+
"step": 2930
|
| 1880 |
+
},
|
| 1881 |
+
{
|
| 1882 |
+
"epoch": 0.94,
|
| 1883 |
+
"learning_rate": 0.00024523671497584537,
|
| 1884 |
+
"loss": 1.0453,
|
| 1885 |
+
"step": 2940
|
| 1886 |
+
},
|
| 1887 |
+
{
|
| 1888 |
+
"epoch": 0.94,
|
| 1889 |
+
"learning_rate": 0.0002450434782608695,
|
| 1890 |
+
"loss": 1.0325,
|
| 1891 |
+
"step": 2950
|
| 1892 |
+
},
|
| 1893 |
+
{
|
| 1894 |
+
"epoch": 0.95,
|
| 1895 |
+
"learning_rate": 0.0002448502415458937,
|
| 1896 |
+
"loss": 1.1118,
|
| 1897 |
+
"step": 2960
|
| 1898 |
+
},
|
| 1899 |
+
{
|
| 1900 |
+
"epoch": 0.95,
|
| 1901 |
+
"learning_rate": 0.00024465700483091785,
|
| 1902 |
+
"loss": 1.0666,
|
| 1903 |
+
"step": 2970
|
| 1904 |
+
},
|
| 1905 |
+
{
|
| 1906 |
+
"epoch": 0.95,
|
| 1907 |
+
"learning_rate": 0.000244463768115942,
|
| 1908 |
+
"loss": 1.0467,
|
| 1909 |
+
"step": 2980
|
| 1910 |
+
},
|
| 1911 |
+
{
|
| 1912 |
+
"epoch": 0.96,
|
| 1913 |
+
"learning_rate": 0.00024427053140096614,
|
| 1914 |
+
"loss": 1.0605,
|
| 1915 |
+
"step": 2990
|
| 1916 |
+
},
|
| 1917 |
+
{
|
| 1918 |
+
"epoch": 0.96,
|
| 1919 |
+
"learning_rate": 0.00024407729468599033,
|
| 1920 |
+
"loss": 1.0787,
|
| 1921 |
+
"step": 3000
|
| 1922 |
+
},
|
| 1923 |
+
{
|
| 1924 |
+
"epoch": 0.96,
|
| 1925 |
+
"eval_loss": 1.0607348680496216,
|
| 1926 |
+
"eval_runtime": 301.6471,
|
| 1927 |
+
"eval_samples_per_second": 6.63,
|
| 1928 |
+
"eval_steps_per_second": 0.829,
|
| 1929 |
+
"step": 3000
|
| 1930 |
+
},
|
| 1931 |
+
{
|
| 1932 |
+
"epoch": 0.96,
|
| 1933 |
+
"learning_rate": 0.00024388405797101448,
|
| 1934 |
+
"loss": 1.0407,
|
| 1935 |
+
"step": 3010
|
| 1936 |
+
},
|
| 1937 |
+
{
|
| 1938 |
+
"epoch": 0.97,
|
| 1939 |
+
"learning_rate": 0.00024369082125603862,
|
| 1940 |
+
"loss": 1.0713,
|
| 1941 |
+
"step": 3020
|
| 1942 |
+
},
|
| 1943 |
+
{
|
| 1944 |
+
"epoch": 0.97,
|
| 1945 |
+
"learning_rate": 0.00024349758454106276,
|
| 1946 |
+
"loss": 1.0489,
|
| 1947 |
+
"step": 3030
|
| 1948 |
+
},
|
| 1949 |
+
{
|
| 1950 |
+
"epoch": 0.97,
|
| 1951 |
+
"learning_rate": 0.00024330434782608693,
|
| 1952 |
+
"loss": 1.0443,
|
| 1953 |
+
"step": 3040
|
| 1954 |
+
},
|
| 1955 |
+
{
|
| 1956 |
+
"epoch": 0.98,
|
| 1957 |
+
"learning_rate": 0.0002431111111111111,
|
| 1958 |
+
"loss": 1.0606,
|
| 1959 |
+
"step": 3050
|
| 1960 |
+
},
|
| 1961 |
+
{
|
| 1962 |
+
"epoch": 0.98,
|
| 1963 |
+
"learning_rate": 0.00024291787439613524,
|
| 1964 |
+
"loss": 1.0929,
|
| 1965 |
+
"step": 3060
|
| 1966 |
+
},
|
| 1967 |
+
{
|
| 1968 |
+
"epoch": 0.98,
|
| 1969 |
+
"learning_rate": 0.00024272463768115938,
|
| 1970 |
+
"loss": 1.0656,
|
| 1971 |
+
"step": 3070
|
| 1972 |
+
},
|
| 1973 |
+
{
|
| 1974 |
+
"epoch": 0.99,
|
| 1975 |
+
"learning_rate": 0.00024253140096618355,
|
| 1976 |
+
"loss": 1.0528,
|
| 1977 |
+
"step": 3080
|
| 1978 |
+
},
|
| 1979 |
+
{
|
| 1980 |
+
"epoch": 0.99,
|
| 1981 |
+
"learning_rate": 0.00024233816425120772,
|
| 1982 |
+
"loss": 1.1073,
|
| 1983 |
+
"step": 3090
|
| 1984 |
+
},
|
| 1985 |
+
{
|
| 1986 |
+
"epoch": 0.99,
|
| 1987 |
+
"learning_rate": 0.00024214492753623187,
|
| 1988 |
+
"loss": 1.0326,
|
| 1989 |
+
"step": 3100
|
| 1990 |
+
},
|
| 1991 |
+
{
|
| 1992 |
+
"epoch": 1.0,
|
| 1993 |
+
"learning_rate": 0.000241951690821256,
|
| 1994 |
+
"loss": 1.0446,
|
| 1995 |
+
"step": 3110
|
| 1996 |
+
},
|
| 1997 |
+
{
|
| 1998 |
+
"epoch": 1.0,
|
| 1999 |
+
"learning_rate": 0.00024175845410628015,
|
| 2000 |
+
"loss": 1.0253,
|
| 2001 |
+
"step": 3120
|
| 2002 |
+
},
|
| 2003 |
+
{
|
| 2004 |
+
"epoch": 1.0,
|
| 2005 |
+
"learning_rate": 0.00024156521739130432,
|
| 2006 |
+
"loss": 1.0186,
|
| 2007 |
+
"step": 3130
|
| 2008 |
+
},
|
| 2009 |
+
{
|
| 2010 |
+
"epoch": 1.0,
|
| 2011 |
+
"learning_rate": 0.0002413719806763285,
|
| 2012 |
+
"loss": 1.0226,
|
| 2013 |
+
"step": 3140
|
| 2014 |
+
},
|
| 2015 |
+
{
|
| 2016 |
+
"epoch": 1.01,
|
| 2017 |
+
"learning_rate": 0.00024117874396135263,
|
| 2018 |
+
"loss": 0.9921,
|
| 2019 |
+
"step": 3150
|
| 2020 |
+
},
|
| 2021 |
+
{
|
| 2022 |
+
"epoch": 1.01,
|
| 2023 |
+
"learning_rate": 0.00024098550724637678,
|
| 2024 |
+
"loss": 0.9769,
|
| 2025 |
+
"step": 3160
|
| 2026 |
+
},
|
| 2027 |
+
{
|
| 2028 |
+
"epoch": 1.01,
|
| 2029 |
+
"learning_rate": 0.00024079227053140095,
|
| 2030 |
+
"loss": 1.0191,
|
| 2031 |
+
"step": 3170
|
| 2032 |
+
},
|
| 2033 |
+
{
|
| 2034 |
+
"epoch": 1.02,
|
| 2035 |
+
"learning_rate": 0.00024059903381642512,
|
| 2036 |
+
"loss": 1.0193,
|
| 2037 |
+
"step": 3180
|
| 2038 |
+
},
|
| 2039 |
+
{
|
| 2040 |
+
"epoch": 1.02,
|
| 2041 |
+
"learning_rate": 0.00024040579710144926,
|
| 2042 |
+
"loss": 1.0023,
|
| 2043 |
+
"step": 3190
|
| 2044 |
+
},
|
| 2045 |
+
{
|
| 2046 |
+
"epoch": 1.02,
|
| 2047 |
+
"learning_rate": 0.0002402125603864734,
|
| 2048 |
+
"loss": 1.0317,
|
| 2049 |
+
"step": 3200
|
| 2050 |
+
},
|
| 2051 |
+
{
|
| 2052 |
+
"epoch": 1.02,
|
| 2053 |
+
"eval_loss": 1.0621122121810913,
|
| 2054 |
+
"eval_runtime": 301.2241,
|
| 2055 |
+
"eval_samples_per_second": 6.64,
|
| 2056 |
+
"eval_steps_per_second": 0.83,
|
| 2057 |
+
"step": 3200
|
| 2058 |
+
},
|
| 2059 |
+
{
|
| 2060 |
+
"epoch": 1.03,
|
| 2061 |
+
"learning_rate": 0.00024001932367149754,
|
| 2062 |
+
"loss": 1.0001,
|
| 2063 |
+
"step": 3210
|
| 2064 |
+
},
|
| 2065 |
+
{
|
| 2066 |
+
"epoch": 1.03,
|
| 2067 |
+
"learning_rate": 0.00023982608695652174,
|
| 2068 |
+
"loss": 0.9865,
|
| 2069 |
+
"step": 3220
|
| 2070 |
+
},
|
| 2071 |
+
{
|
| 2072 |
+
"epoch": 1.03,
|
| 2073 |
+
"learning_rate": 0.00023963285024154588,
|
| 2074 |
+
"loss": 0.9851,
|
| 2075 |
+
"step": 3230
|
| 2076 |
+
},
|
| 2077 |
+
{
|
| 2078 |
+
"epoch": 1.04,
|
| 2079 |
+
"learning_rate": 0.00023943961352657002,
|
| 2080 |
+
"loss": 1.017,
|
| 2081 |
+
"step": 3240
|
| 2082 |
+
},
|
| 2083 |
+
{
|
| 2084 |
+
"epoch": 1.04,
|
| 2085 |
+
"learning_rate": 0.00023924637681159417,
|
| 2086 |
+
"loss": 1.0149,
|
| 2087 |
+
"step": 3250
|
| 2088 |
+
},
|
| 2089 |
+
{
|
| 2090 |
+
"epoch": 1.04,
|
| 2091 |
+
"learning_rate": 0.0002390531400966183,
|
| 2092 |
+
"loss": 0.9958,
|
| 2093 |
+
"step": 3260
|
| 2094 |
+
},
|
| 2095 |
+
{
|
| 2096 |
+
"epoch": 1.05,
|
| 2097 |
+
"learning_rate": 0.0002388599033816425,
|
| 2098 |
+
"loss": 1.0346,
|
| 2099 |
+
"step": 3270
|
| 2100 |
+
},
|
| 2101 |
+
{
|
| 2102 |
+
"epoch": 1.05,
|
| 2103 |
+
"learning_rate": 0.00023866666666666665,
|
| 2104 |
+
"loss": 1.0612,
|
| 2105 |
+
"step": 3280
|
| 2106 |
+
},
|
| 2107 |
+
{
|
| 2108 |
+
"epoch": 1.05,
|
| 2109 |
+
"learning_rate": 0.0002384734299516908,
|
| 2110 |
+
"loss": 0.9862,
|
| 2111 |
+
"step": 3290
|
| 2112 |
+
},
|
| 2113 |
+
{
|
| 2114 |
+
"epoch": 1.06,
|
| 2115 |
+
"learning_rate": 0.00023828019323671493,
|
| 2116 |
+
"loss": 0.9982,
|
| 2117 |
+
"step": 3300
|
| 2118 |
+
},
|
| 2119 |
+
{
|
| 2120 |
+
"epoch": 1.06,
|
| 2121 |
+
"learning_rate": 0.00023808695652173913,
|
| 2122 |
+
"loss": 1.046,
|
| 2123 |
+
"step": 3310
|
| 2124 |
+
},
|
| 2125 |
+
{
|
| 2126 |
+
"epoch": 1.06,
|
| 2127 |
+
"learning_rate": 0.00023789371980676327,
|
| 2128 |
+
"loss": 1.0176,
|
| 2129 |
+
"step": 3320
|
| 2130 |
+
},
|
| 2131 |
+
{
|
| 2132 |
+
"epoch": 1.07,
|
| 2133 |
+
"learning_rate": 0.00023770048309178742,
|
| 2134 |
+
"loss": 0.9974,
|
| 2135 |
+
"step": 3330
|
| 2136 |
+
},
|
| 2137 |
+
{
|
| 2138 |
+
"epoch": 1.07,
|
| 2139 |
+
"learning_rate": 0.00023750724637681156,
|
| 2140 |
+
"loss": 1.0194,
|
| 2141 |
+
"step": 3340
|
| 2142 |
+
},
|
| 2143 |
+
{
|
| 2144 |
+
"epoch": 1.07,
|
| 2145 |
+
"learning_rate": 0.0002373140096618357,
|
| 2146 |
+
"loss": 1.0272,
|
| 2147 |
+
"step": 3350
|
| 2148 |
+
},
|
| 2149 |
+
{
|
| 2150 |
+
"epoch": 1.08,
|
| 2151 |
+
"learning_rate": 0.0002371207729468599,
|
| 2152 |
+
"loss": 1.0707,
|
| 2153 |
+
"step": 3360
|
| 2154 |
+
},
|
| 2155 |
+
{
|
| 2156 |
+
"epoch": 1.08,
|
| 2157 |
+
"learning_rate": 0.00023692753623188404,
|
| 2158 |
+
"loss": 0.9941,
|
| 2159 |
+
"step": 3370
|
| 2160 |
+
},
|
| 2161 |
+
{
|
| 2162 |
+
"epoch": 1.08,
|
| 2163 |
+
"learning_rate": 0.00023673429951690818,
|
| 2164 |
+
"loss": 1.0298,
|
| 2165 |
+
"step": 3380
|
| 2166 |
+
},
|
| 2167 |
+
{
|
| 2168 |
+
"epoch": 1.08,
|
| 2169 |
+
"learning_rate": 0.00023654106280193233,
|
| 2170 |
+
"loss": 1.0192,
|
| 2171 |
+
"step": 3390
|
| 2172 |
+
},
|
| 2173 |
+
{
|
| 2174 |
+
"epoch": 1.09,
|
| 2175 |
+
"learning_rate": 0.00023634782608695652,
|
| 2176 |
+
"loss": 0.99,
|
| 2177 |
+
"step": 3400
|
| 2178 |
+
},
|
| 2179 |
+
{
|
| 2180 |
+
"epoch": 1.09,
|
| 2181 |
+
"eval_loss": 1.0633840560913086,
|
| 2182 |
+
"eval_runtime": 301.1449,
|
| 2183 |
+
"eval_samples_per_second": 6.641,
|
| 2184 |
+
"eval_steps_per_second": 0.83,
|
| 2185 |
+
"step": 3400
|
| 2186 |
+
},
|
| 2187 |
+
{
|
| 2188 |
+
"epoch": 1.09,
|
| 2189 |
+
"learning_rate": 0.00023615458937198067,
|
| 2190 |
+
"loss": 1.0316,
|
| 2191 |
+
"step": 3410
|
| 2192 |
+
},
|
| 2193 |
+
{
|
| 2194 |
+
"epoch": 1.09,
|
| 2195 |
+
"learning_rate": 0.0002359613526570048,
|
| 2196 |
+
"loss": 0.9605,
|
| 2197 |
+
"step": 3420
|
| 2198 |
+
},
|
| 2199 |
+
{
|
| 2200 |
+
"epoch": 1.1,
|
| 2201 |
+
"learning_rate": 0.00023576811594202895,
|
| 2202 |
+
"loss": 0.9902,
|
| 2203 |
+
"step": 3430
|
| 2204 |
+
},
|
| 2205 |
+
{
|
| 2206 |
+
"epoch": 1.1,
|
| 2207 |
+
"learning_rate": 0.0002355748792270531,
|
| 2208 |
+
"loss": 0.99,
|
| 2209 |
+
"step": 3440
|
| 2210 |
+
},
|
| 2211 |
+
{
|
| 2212 |
+
"epoch": 1.1,
|
| 2213 |
+
"learning_rate": 0.0002353816425120773,
|
| 2214 |
+
"loss": 0.9923,
|
| 2215 |
+
"step": 3450
|
| 2216 |
+
},
|
| 2217 |
+
{
|
| 2218 |
+
"epoch": 1.11,
|
| 2219 |
+
"learning_rate": 0.00023518840579710143,
|
| 2220 |
+
"loss": 0.993,
|
| 2221 |
+
"step": 3460
|
| 2222 |
+
},
|
| 2223 |
+
{
|
| 2224 |
+
"epoch": 1.11,
|
| 2225 |
+
"learning_rate": 0.00023499516908212557,
|
| 2226 |
+
"loss": 1.0103,
|
| 2227 |
+
"step": 3470
|
| 2228 |
+
},
|
| 2229 |
+
{
|
| 2230 |
+
"epoch": 1.11,
|
| 2231 |
+
"learning_rate": 0.00023480193236714972,
|
| 2232 |
+
"loss": 1.0536,
|
| 2233 |
+
"step": 3480
|
| 2234 |
+
},
|
| 2235 |
+
{
|
| 2236 |
+
"epoch": 1.12,
|
| 2237 |
+
"learning_rate": 0.00023460869565217391,
|
| 2238 |
+
"loss": 0.9961,
|
| 2239 |
+
"step": 3490
|
| 2240 |
+
},
|
| 2241 |
+
{
|
| 2242 |
+
"epoch": 1.12,
|
| 2243 |
+
"learning_rate": 0.00023441545893719806,
|
| 2244 |
+
"loss": 1.0461,
|
| 2245 |
+
"step": 3500
|
| 2246 |
+
},
|
| 2247 |
+
{
|
| 2248 |
+
"epoch": 1.12,
|
| 2249 |
+
"learning_rate": 0.0002342222222222222,
|
| 2250 |
+
"loss": 0.9959,
|
| 2251 |
+
"step": 3510
|
| 2252 |
+
},
|
| 2253 |
+
{
|
| 2254 |
+
"epoch": 1.13,
|
| 2255 |
+
"learning_rate": 0.00023402898550724634,
|
| 2256 |
+
"loss": 0.9932,
|
| 2257 |
+
"step": 3520
|
| 2258 |
+
},
|
| 2259 |
+
{
|
| 2260 |
+
"epoch": 1.13,
|
| 2261 |
+
"learning_rate": 0.00023383574879227054,
|
| 2262 |
+
"loss": 1.0258,
|
| 2263 |
+
"step": 3530
|
| 2264 |
+
},
|
| 2265 |
+
{
|
| 2266 |
+
"epoch": 1.13,
|
| 2267 |
+
"learning_rate": 0.00023364251207729468,
|
| 2268 |
+
"loss": 1.0591,
|
| 2269 |
+
"step": 3540
|
| 2270 |
+
},
|
| 2271 |
+
{
|
| 2272 |
+
"epoch": 1.14,
|
| 2273 |
+
"learning_rate": 0.00023344927536231882,
|
| 2274 |
+
"loss": 1.0672,
|
| 2275 |
+
"step": 3550
|
| 2276 |
+
},
|
| 2277 |
+
{
|
| 2278 |
+
"epoch": 1.14,
|
| 2279 |
+
"learning_rate": 0.00023325603864734297,
|
| 2280 |
+
"loss": 0.9972,
|
| 2281 |
+
"step": 3560
|
| 2282 |
+
},
|
| 2283 |
+
{
|
| 2284 |
+
"epoch": 1.14,
|
| 2285 |
+
"learning_rate": 0.0002330628019323671,
|
| 2286 |
+
"loss": 1.0063,
|
| 2287 |
+
"step": 3570
|
| 2288 |
+
},
|
| 2289 |
+
{
|
| 2290 |
+
"epoch": 1.15,
|
| 2291 |
+
"learning_rate": 0.0002328695652173913,
|
| 2292 |
+
"loss": 1.0153,
|
| 2293 |
+
"step": 3580
|
| 2294 |
+
},
|
| 2295 |
+
{
|
| 2296 |
+
"epoch": 1.15,
|
| 2297 |
+
"learning_rate": 0.00023267632850241545,
|
| 2298 |
+
"loss": 1.0611,
|
| 2299 |
+
"step": 3590
|
| 2300 |
+
},
|
| 2301 |
+
{
|
| 2302 |
+
"epoch": 1.15,
|
| 2303 |
+
"learning_rate": 0.0002324830917874396,
|
| 2304 |
+
"loss": 1.0215,
|
| 2305 |
+
"step": 3600
|
| 2306 |
+
},
|
| 2307 |
+
{
|
| 2308 |
+
"epoch": 1.15,
|
| 2309 |
+
"eval_loss": 1.0626596212387085,
|
| 2310 |
+
"eval_runtime": 302.4484,
|
| 2311 |
+
"eval_samples_per_second": 6.613,
|
| 2312 |
+
"eval_steps_per_second": 0.827,
|
| 2313 |
+
"step": 3600
|
| 2314 |
+
},
|
| 2315 |
+
{
|
| 2316 |
+
"epoch": 1.16,
|
| 2317 |
+
"learning_rate": 0.00023228985507246373,
|
| 2318 |
+
"loss": 1.0118,
|
| 2319 |
+
"step": 3610
|
| 2320 |
+
},
|
| 2321 |
+
{
|
| 2322 |
+
"epoch": 1.16,
|
| 2323 |
+
"learning_rate": 0.00023209661835748793,
|
| 2324 |
+
"loss": 1.0626,
|
| 2325 |
+
"step": 3620
|
| 2326 |
+
},
|
| 2327 |
+
{
|
| 2328 |
+
"epoch": 1.16,
|
| 2329 |
+
"learning_rate": 0.00023190338164251207,
|
| 2330 |
+
"loss": 1.0051,
|
| 2331 |
+
"step": 3630
|
| 2332 |
+
},
|
| 2333 |
+
{
|
| 2334 |
+
"epoch": 1.16,
|
| 2335 |
+
"learning_rate": 0.00023171014492753622,
|
| 2336 |
+
"loss": 0.9824,
|
| 2337 |
+
"step": 3640
|
| 2338 |
+
},
|
| 2339 |
+
{
|
| 2340 |
+
"epoch": 1.17,
|
| 2341 |
+
"learning_rate": 0.00023151690821256036,
|
| 2342 |
+
"loss": 1.0567,
|
| 2343 |
+
"step": 3650
|
| 2344 |
+
},
|
| 2345 |
+
{
|
| 2346 |
+
"epoch": 1.17,
|
| 2347 |
+
"learning_rate": 0.0002313236714975845,
|
| 2348 |
+
"loss": 1.015,
|
| 2349 |
+
"step": 3660
|
| 2350 |
+
},
|
| 2351 |
+
{
|
| 2352 |
+
"epoch": 1.17,
|
| 2353 |
+
"learning_rate": 0.0002311304347826087,
|
| 2354 |
+
"loss": 1.0129,
|
| 2355 |
+
"step": 3670
|
| 2356 |
+
},
|
| 2357 |
+
{
|
| 2358 |
+
"epoch": 1.18,
|
| 2359 |
+
"learning_rate": 0.00023093719806763284,
|
| 2360 |
+
"loss": 0.9826,
|
| 2361 |
+
"step": 3680
|
| 2362 |
+
},
|
| 2363 |
+
{
|
| 2364 |
+
"epoch": 1.18,
|
| 2365 |
+
"learning_rate": 0.00023074396135265698,
|
| 2366 |
+
"loss": 1.0257,
|
| 2367 |
+
"step": 3690
|
| 2368 |
+
},
|
| 2369 |
+
{
|
| 2370 |
+
"epoch": 1.18,
|
| 2371 |
+
"learning_rate": 0.00023055072463768112,
|
| 2372 |
+
"loss": 1.0426,
|
| 2373 |
+
"step": 3700
|
| 2374 |
+
},
|
| 2375 |
+
{
|
| 2376 |
+
"epoch": 1.19,
|
| 2377 |
+
"learning_rate": 0.00023035748792270532,
|
| 2378 |
+
"loss": 1.0054,
|
| 2379 |
+
"step": 3710
|
| 2380 |
+
},
|
| 2381 |
+
{
|
| 2382 |
+
"epoch": 1.19,
|
| 2383 |
+
"learning_rate": 0.00023016425120772946,
|
| 2384 |
+
"loss": 1.0168,
|
| 2385 |
+
"step": 3720
|
| 2386 |
+
},
|
| 2387 |
+
{
|
| 2388 |
+
"epoch": 1.19,
|
| 2389 |
+
"learning_rate": 0.0002299710144927536,
|
| 2390 |
+
"loss": 1.0,
|
| 2391 |
+
"step": 3730
|
| 2392 |
+
},
|
| 2393 |
+
{
|
| 2394 |
+
"epoch": 1.2,
|
| 2395 |
+
"learning_rate": 0.00022977777777777775,
|
| 2396 |
+
"loss": 1.0138,
|
| 2397 |
+
"step": 3740
|
| 2398 |
+
},
|
| 2399 |
+
{
|
| 2400 |
+
"epoch": 1.2,
|
| 2401 |
+
"learning_rate": 0.0002295845410628019,
|
| 2402 |
+
"loss": 1.0339,
|
| 2403 |
+
"step": 3750
|
| 2404 |
+
},
|
| 2405 |
+
{
|
| 2406 |
+
"epoch": 1.2,
|
| 2407 |
+
"learning_rate": 0.0002293913043478261,
|
| 2408 |
+
"loss": 0.9986,
|
| 2409 |
+
"step": 3760
|
| 2410 |
+
},
|
| 2411 |
+
{
|
| 2412 |
+
"epoch": 1.21,
|
| 2413 |
+
"learning_rate": 0.00022919806763285023,
|
| 2414 |
+
"loss": 1.0171,
|
| 2415 |
+
"step": 3770
|
| 2416 |
+
},
|
| 2417 |
+
{
|
| 2418 |
+
"epoch": 1.21,
|
| 2419 |
+
"learning_rate": 0.00022900483091787437,
|
| 2420 |
+
"loss": 0.9846,
|
| 2421 |
+
"step": 3780
|
| 2422 |
+
},
|
| 2423 |
+
{
|
| 2424 |
+
"epoch": 1.21,
|
| 2425 |
+
"learning_rate": 0.00022881159420289852,
|
| 2426 |
+
"loss": 1.0338,
|
| 2427 |
+
"step": 3790
|
| 2428 |
+
},
|
| 2429 |
+
{
|
| 2430 |
+
"epoch": 1.22,
|
| 2431 |
+
"learning_rate": 0.00022861835748792269,
|
| 2432 |
+
"loss": 1.0161,
|
| 2433 |
+
"step": 3800
|
| 2434 |
+
},
|
| 2435 |
+
{
|
| 2436 |
+
"epoch": 1.22,
|
| 2437 |
+
"eval_loss": 1.0625114440917969,
|
| 2438 |
+
"eval_runtime": 301.9731,
|
| 2439 |
+
"eval_samples_per_second": 6.623,
|
| 2440 |
+
"eval_steps_per_second": 0.828,
|
| 2441 |
+
"step": 3800
|
| 2442 |
+
},
|
| 2443 |
+
{
|
| 2444 |
+
"epoch": 1.22,
|
| 2445 |
+
"learning_rate": 0.00022842512077294686,
|
| 2446 |
+
"loss": 1.0253,
|
| 2447 |
+
"step": 3810
|
| 2448 |
+
},
|
| 2449 |
+
{
|
| 2450 |
+
"epoch": 1.22,
|
| 2451 |
+
"learning_rate": 0.000228231884057971,
|
| 2452 |
+
"loss": 1.0187,
|
| 2453 |
+
"step": 3820
|
| 2454 |
+
},
|
| 2455 |
+
{
|
| 2456 |
+
"epoch": 1.23,
|
| 2457 |
+
"learning_rate": 0.00022803864734299514,
|
| 2458 |
+
"loss": 0.9766,
|
| 2459 |
+
"step": 3830
|
| 2460 |
+
},
|
| 2461 |
+
{
|
| 2462 |
+
"epoch": 1.23,
|
| 2463 |
+
"learning_rate": 0.0002278454106280193,
|
| 2464 |
+
"loss": 1.0336,
|
| 2465 |
+
"step": 3840
|
| 2466 |
+
},
|
| 2467 |
+
{
|
| 2468 |
+
"epoch": 1.23,
|
| 2469 |
+
"learning_rate": 0.00022765217391304348,
|
| 2470 |
+
"loss": 1.0084,
|
| 2471 |
+
"step": 3850
|
| 2472 |
+
},
|
| 2473 |
+
{
|
| 2474 |
+
"epoch": 1.24,
|
| 2475 |
+
"learning_rate": 0.00022745893719806762,
|
| 2476 |
+
"loss": 1.0152,
|
| 2477 |
+
"step": 3860
|
| 2478 |
+
},
|
| 2479 |
+
{
|
| 2480 |
+
"epoch": 1.24,
|
| 2481 |
+
"learning_rate": 0.00022726570048309177,
|
| 2482 |
+
"loss": 0.9628,
|
| 2483 |
+
"step": 3870
|
| 2484 |
+
},
|
| 2485 |
+
{
|
| 2486 |
+
"epoch": 1.24,
|
| 2487 |
+
"learning_rate": 0.0002270724637681159,
|
| 2488 |
+
"loss": 0.9494,
|
| 2489 |
+
"step": 3880
|
| 2490 |
+
},
|
| 2491 |
+
{
|
| 2492 |
+
"epoch": 1.24,
|
| 2493 |
+
"learning_rate": 0.00022687922705314008,
|
| 2494 |
+
"loss": 1.0688,
|
| 2495 |
+
"step": 3890
|
| 2496 |
+
},
|
| 2497 |
+
{
|
| 2498 |
+
"epoch": 1.25,
|
| 2499 |
+
"learning_rate": 0.00022668599033816425,
|
| 2500 |
+
"loss": 1.0242,
|
| 2501 |
+
"step": 3900
|
| 2502 |
+
},
|
| 2503 |
+
{
|
| 2504 |
+
"epoch": 1.25,
|
| 2505 |
+
"learning_rate": 0.0002264927536231884,
|
| 2506 |
+
"loss": 1.0518,
|
| 2507 |
+
"step": 3910
|
| 2508 |
+
},
|
| 2509 |
+
{
|
| 2510 |
+
"epoch": 1.25,
|
| 2511 |
+
"learning_rate": 0.00022629951690821253,
|
| 2512 |
+
"loss": 0.9777,
|
| 2513 |
+
"step": 3920
|
| 2514 |
+
},
|
| 2515 |
+
{
|
| 2516 |
+
"epoch": 1.26,
|
| 2517 |
+
"learning_rate": 0.0002261062801932367,
|
| 2518 |
+
"loss": 0.9929,
|
| 2519 |
+
"step": 3930
|
| 2520 |
+
},
|
| 2521 |
+
{
|
| 2522 |
+
"epoch": 1.26,
|
| 2523 |
+
"learning_rate": 0.00022591304347826084,
|
| 2524 |
+
"loss": 1.0192,
|
| 2525 |
+
"step": 3940
|
| 2526 |
+
},
|
| 2527 |
+
{
|
| 2528 |
+
"epoch": 1.26,
|
| 2529 |
+
"learning_rate": 0.00022571980676328501,
|
| 2530 |
+
"loss": 0.9951,
|
| 2531 |
+
"step": 3950
|
| 2532 |
+
},
|
| 2533 |
+
{
|
| 2534 |
+
"epoch": 1.27,
|
| 2535 |
+
"learning_rate": 0.00022552657004830916,
|
| 2536 |
+
"loss": 1.0014,
|
| 2537 |
+
"step": 3960
|
| 2538 |
+
},
|
| 2539 |
+
{
|
| 2540 |
+
"epoch": 1.27,
|
| 2541 |
+
"learning_rate": 0.0002253333333333333,
|
| 2542 |
+
"loss": 1.035,
|
| 2543 |
+
"step": 3970
|
| 2544 |
+
},
|
| 2545 |
+
{
|
| 2546 |
+
"epoch": 1.27,
|
| 2547 |
+
"learning_rate": 0.00022514009661835747,
|
| 2548 |
+
"loss": 0.9909,
|
| 2549 |
+
"step": 3980
|
| 2550 |
+
},
|
| 2551 |
+
{
|
| 2552 |
+
"epoch": 1.28,
|
| 2553 |
+
"learning_rate": 0.00022494685990338164,
|
| 2554 |
+
"loss": 1.0372,
|
| 2555 |
+
"step": 3990
|
| 2556 |
+
},
|
| 2557 |
+
{
|
| 2558 |
+
"epoch": 1.28,
|
| 2559 |
+
"learning_rate": 0.00022475362318840578,
|
| 2560 |
+
"loss": 0.992,
|
| 2561 |
+
"step": 4000
|
| 2562 |
+
},
|
| 2563 |
+
{
|
| 2564 |
+
"epoch": 1.28,
|
| 2565 |
+
"eval_loss": 1.0622905492782593,
|
| 2566 |
+
"eval_runtime": 303.6073,
|
| 2567 |
+
"eval_samples_per_second": 6.587,
|
| 2568 |
+
"eval_steps_per_second": 0.823,
|
| 2569 |
+
"step": 4000
|
| 2570 |
+
},
|
| 2571 |
+
{
|
| 2572 |
+
"epoch": 1.28,
|
| 2573 |
+
"learning_rate": 0.00022456038647342992,
|
| 2574 |
+
"loss": 1.0224,
|
| 2575 |
+
"step": 4010
|
| 2576 |
+
},
|
| 2577 |
+
{
|
| 2578 |
+
"epoch": 1.29,
|
| 2579 |
+
"learning_rate": 0.0002243671497584541,
|
| 2580 |
+
"loss": 1.0159,
|
| 2581 |
+
"step": 4020
|
| 2582 |
+
},
|
| 2583 |
+
{
|
| 2584 |
+
"epoch": 1.29,
|
| 2585 |
+
"learning_rate": 0.00022417391304347824,
|
| 2586 |
+
"loss": 1.0257,
|
| 2587 |
+
"step": 4030
|
| 2588 |
+
},
|
| 2589 |
+
{
|
| 2590 |
+
"epoch": 1.29,
|
| 2591 |
+
"learning_rate": 0.0002239806763285024,
|
| 2592 |
+
"loss": 1.0391,
|
| 2593 |
+
"step": 4040
|
| 2594 |
+
},
|
| 2595 |
+
{
|
| 2596 |
+
"epoch": 1.3,
|
| 2597 |
+
"learning_rate": 0.00022378743961352655,
|
| 2598 |
+
"loss": 0.9996,
|
| 2599 |
+
"step": 4050
|
| 2600 |
+
},
|
| 2601 |
+
{
|
| 2602 |
+
"epoch": 1.3,
|
| 2603 |
+
"learning_rate": 0.0002235942028985507,
|
| 2604 |
+
"loss": 1.0807,
|
| 2605 |
+
"step": 4060
|
| 2606 |
+
},
|
| 2607 |
+
{
|
| 2608 |
+
"epoch": 1.3,
|
| 2609 |
+
"learning_rate": 0.00022340096618357486,
|
| 2610 |
+
"loss": 0.975,
|
| 2611 |
+
"step": 4070
|
| 2612 |
+
},
|
| 2613 |
+
{
|
| 2614 |
+
"epoch": 1.31,
|
| 2615 |
+
"learning_rate": 0.00022320772946859903,
|
| 2616 |
+
"loss": 1.0069,
|
| 2617 |
+
"step": 4080
|
| 2618 |
+
},
|
| 2619 |
+
{
|
| 2620 |
+
"epoch": 1.31,
|
| 2621 |
+
"learning_rate": 0.00022301449275362317,
|
| 2622 |
+
"loss": 0.9844,
|
| 2623 |
+
"step": 4090
|
| 2624 |
+
},
|
| 2625 |
+
{
|
| 2626 |
+
"epoch": 1.31,
|
| 2627 |
+
"learning_rate": 0.00022282125603864732,
|
| 2628 |
+
"loss": 1.0336,
|
| 2629 |
+
"step": 4100
|
| 2630 |
+
},
|
| 2631 |
+
{
|
| 2632 |
+
"epoch": 1.32,
|
| 2633 |
+
"learning_rate": 0.00022262801932367148,
|
| 2634 |
+
"loss": 1.0124,
|
| 2635 |
+
"step": 4110
|
| 2636 |
+
},
|
| 2637 |
+
{
|
| 2638 |
+
"epoch": 1.32,
|
| 2639 |
+
"learning_rate": 0.00022243478260869563,
|
| 2640 |
+
"loss": 0.9849,
|
| 2641 |
+
"step": 4120
|
| 2642 |
+
},
|
| 2643 |
+
{
|
| 2644 |
+
"epoch": 1.32,
|
| 2645 |
+
"learning_rate": 0.0002222415458937198,
|
| 2646 |
+
"loss": 1.0065,
|
| 2647 |
+
"step": 4130
|
| 2648 |
+
},
|
| 2649 |
+
{
|
| 2650 |
+
"epoch": 1.32,
|
| 2651 |
+
"learning_rate": 0.00022204830917874394,
|
| 2652 |
+
"loss": 1.0319,
|
| 2653 |
+
"step": 4140
|
| 2654 |
+
},
|
| 2655 |
+
{
|
| 2656 |
+
"epoch": 1.33,
|
| 2657 |
+
"learning_rate": 0.00022185507246376808,
|
| 2658 |
+
"loss": 1.0118,
|
| 2659 |
+
"step": 4150
|
| 2660 |
+
},
|
| 2661 |
+
{
|
| 2662 |
+
"epoch": 1.33,
|
| 2663 |
+
"learning_rate": 0.00022166183574879225,
|
| 2664 |
+
"loss": 1.0255,
|
| 2665 |
+
"step": 4160
|
| 2666 |
+
},
|
| 2667 |
+
{
|
| 2668 |
+
"epoch": 1.33,
|
| 2669 |
+
"learning_rate": 0.0002214685990338164,
|
| 2670 |
+
"loss": 0.9649,
|
| 2671 |
+
"step": 4170
|
| 2672 |
+
},
|
| 2673 |
+
{
|
| 2674 |
+
"epoch": 1.34,
|
| 2675 |
+
"learning_rate": 0.00022127536231884056,
|
| 2676 |
+
"loss": 0.9756,
|
| 2677 |
+
"step": 4180
|
| 2678 |
+
},
|
| 2679 |
+
{
|
| 2680 |
+
"epoch": 1.34,
|
| 2681 |
+
"learning_rate": 0.0002210821256038647,
|
| 2682 |
+
"loss": 0.9827,
|
| 2683 |
+
"step": 4190
|
| 2684 |
+
},
|
| 2685 |
+
{
|
| 2686 |
+
"epoch": 1.34,
|
| 2687 |
+
"learning_rate": 0.00022088888888888888,
|
| 2688 |
+
"loss": 1.0245,
|
| 2689 |
+
"step": 4200
|
| 2690 |
+
},
|
| 2691 |
+
{
|
| 2692 |
+
"epoch": 1.34,
|
| 2693 |
+
"eval_loss": 1.0623623132705688,
|
| 2694 |
+
"eval_runtime": 303.2092,
|
| 2695 |
+
"eval_samples_per_second": 6.596,
|
| 2696 |
+
"eval_steps_per_second": 0.825,
|
| 2697 |
+
"step": 4200
|
| 2698 |
+
},
|
| 2699 |
+
{
|
| 2700 |
+
"epoch": 1.35,
|
| 2701 |
+
"learning_rate": 0.00022069565217391302,
|
| 2702 |
+
"loss": 1.0284,
|
| 2703 |
+
"step": 4210
|
| 2704 |
+
},
|
| 2705 |
+
{
|
| 2706 |
+
"epoch": 1.35,
|
| 2707 |
+
"learning_rate": 0.0002205024154589372,
|
| 2708 |
+
"loss": 1.0258,
|
| 2709 |
+
"step": 4220
|
| 2710 |
+
},
|
| 2711 |
+
{
|
| 2712 |
+
"epoch": 1.35,
|
| 2713 |
+
"learning_rate": 0.00022030917874396133,
|
| 2714 |
+
"loss": 1.0665,
|
| 2715 |
+
"step": 4230
|
| 2716 |
+
},
|
| 2717 |
+
{
|
| 2718 |
+
"epoch": 1.36,
|
| 2719 |
+
"learning_rate": 0.0002201159420289855,
|
| 2720 |
+
"loss": 0.9961,
|
| 2721 |
+
"step": 4240
|
| 2722 |
+
},
|
| 2723 |
+
{
|
| 2724 |
+
"epoch": 1.36,
|
| 2725 |
+
"learning_rate": 0.00021992270531400964,
|
| 2726 |
+
"loss": 1.0137,
|
| 2727 |
+
"step": 4250
|
| 2728 |
+
},
|
| 2729 |
+
{
|
| 2730 |
+
"epoch": 1.36,
|
| 2731 |
+
"learning_rate": 0.00021972946859903379,
|
| 2732 |
+
"loss": 0.9965,
|
| 2733 |
+
"step": 4260
|
| 2734 |
+
},
|
| 2735 |
+
{
|
| 2736 |
+
"epoch": 1.37,
|
| 2737 |
+
"learning_rate": 0.00021953623188405796,
|
| 2738 |
+
"loss": 1.0181,
|
| 2739 |
+
"step": 4270
|
| 2740 |
+
},
|
| 2741 |
+
{
|
| 2742 |
+
"epoch": 1.37,
|
| 2743 |
+
"learning_rate": 0.0002193429951690821,
|
| 2744 |
+
"loss": 1.0047,
|
| 2745 |
+
"step": 4280
|
| 2746 |
+
},
|
| 2747 |
+
{
|
| 2748 |
+
"epoch": 1.37,
|
| 2749 |
+
"learning_rate": 0.00021914975845410627,
|
| 2750 |
+
"loss": 1.0193,
|
| 2751 |
+
"step": 4290
|
| 2752 |
+
},
|
| 2753 |
+
{
|
| 2754 |
+
"epoch": 1.38,
|
| 2755 |
+
"learning_rate": 0.0002189565217391304,
|
| 2756 |
+
"loss": 1.011,
|
| 2757 |
+
"step": 4300
|
| 2758 |
+
},
|
| 2759 |
+
{
|
| 2760 |
+
"epoch": 1.38,
|
| 2761 |
+
"learning_rate": 0.00021876328502415458,
|
| 2762 |
+
"loss": 1.0327,
|
| 2763 |
+
"step": 4310
|
| 2764 |
+
},
|
| 2765 |
+
{
|
| 2766 |
+
"epoch": 1.38,
|
| 2767 |
+
"learning_rate": 0.00021857004830917872,
|
| 2768 |
+
"loss": 1.0434,
|
| 2769 |
+
"step": 4320
|
| 2770 |
+
},
|
| 2771 |
+
{
|
| 2772 |
+
"epoch": 1.39,
|
| 2773 |
+
"learning_rate": 0.0002183768115942029,
|
| 2774 |
+
"loss": 1.0138,
|
| 2775 |
+
"step": 4330
|
| 2776 |
+
},
|
| 2777 |
+
{
|
| 2778 |
+
"epoch": 1.39,
|
| 2779 |
+
"learning_rate": 0.00021818357487922703,
|
| 2780 |
+
"loss": 0.9737,
|
| 2781 |
+
"step": 4340
|
| 2782 |
+
},
|
| 2783 |
+
{
|
| 2784 |
+
"epoch": 1.39,
|
| 2785 |
+
"learning_rate": 0.00021799033816425118,
|
| 2786 |
+
"loss": 0.9977,
|
| 2787 |
+
"step": 4350
|
| 2788 |
+
},
|
| 2789 |
+
{
|
| 2790 |
+
"epoch": 1.4,
|
| 2791 |
+
"learning_rate": 0.00021779710144927535,
|
| 2792 |
+
"loss": 1.0181,
|
| 2793 |
+
"step": 4360
|
| 2794 |
+
},
|
| 2795 |
+
{
|
| 2796 |
+
"epoch": 1.4,
|
| 2797 |
+
"learning_rate": 0.0002176038647342995,
|
| 2798 |
+
"loss": 1.0434,
|
| 2799 |
+
"step": 4370
|
| 2800 |
+
},
|
| 2801 |
+
{
|
| 2802 |
+
"epoch": 1.4,
|
| 2803 |
+
"learning_rate": 0.00021741062801932366,
|
| 2804 |
+
"loss": 1.0143,
|
| 2805 |
+
"step": 4380
|
| 2806 |
+
},
|
| 2807 |
+
{
|
| 2808 |
+
"epoch": 1.4,
|
| 2809 |
+
"learning_rate": 0.0002172173913043478,
|
| 2810 |
+
"loss": 0.9786,
|
| 2811 |
+
"step": 4390
|
| 2812 |
+
},
|
| 2813 |
+
{
|
| 2814 |
+
"epoch": 1.41,
|
| 2815 |
+
"learning_rate": 0.00021702415458937194,
|
| 2816 |
+
"loss": 0.9861,
|
| 2817 |
+
"step": 4400
|
| 2818 |
+
},
|
| 2819 |
+
{
|
| 2820 |
+
"epoch": 1.41,
|
| 2821 |
+
"eval_loss": 1.0632884502410889,
|
| 2822 |
+
"eval_runtime": 302.3081,
|
| 2823 |
+
"eval_samples_per_second": 6.616,
|
| 2824 |
+
"eval_steps_per_second": 0.827,
|
| 2825 |
+
"step": 4400
|
| 2826 |
+
},
|
| 2827 |
+
{
|
| 2828 |
+
"epoch": 1.41,
|
| 2829 |
+
"learning_rate": 0.00021683091787439611,
|
| 2830 |
+
"loss": 1.0298,
|
| 2831 |
+
"step": 4410
|
| 2832 |
+
},
|
| 2833 |
+
{
|
| 2834 |
+
"epoch": 1.41,
|
| 2835 |
+
"learning_rate": 0.00021663768115942028,
|
| 2836 |
+
"loss": 1.0203,
|
| 2837 |
+
"step": 4420
|
| 2838 |
+
},
|
| 2839 |
+
{
|
| 2840 |
+
"epoch": 1.42,
|
| 2841 |
+
"learning_rate": 0.00021644444444444443,
|
| 2842 |
+
"loss": 1.0476,
|
| 2843 |
+
"step": 4430
|
| 2844 |
+
},
|
| 2845 |
+
{
|
| 2846 |
+
"epoch": 1.42,
|
| 2847 |
+
"learning_rate": 0.00021625120772946857,
|
| 2848 |
+
"loss": 1.0071,
|
| 2849 |
+
"step": 4440
|
| 2850 |
+
},
|
| 2851 |
+
{
|
| 2852 |
+
"epoch": 1.42,
|
| 2853 |
+
"learning_rate": 0.00021605797101449274,
|
| 2854 |
+
"loss": 1.0125,
|
| 2855 |
+
"step": 4450
|
| 2856 |
+
},
|
| 2857 |
+
{
|
| 2858 |
+
"epoch": 1.43,
|
| 2859 |
+
"learning_rate": 0.00021586473429951688,
|
| 2860 |
+
"loss": 1.0706,
|
| 2861 |
+
"step": 4460
|
| 2862 |
+
},
|
| 2863 |
+
{
|
| 2864 |
+
"epoch": 1.43,
|
| 2865 |
+
"learning_rate": 0.00021567149758454105,
|
| 2866 |
+
"loss": 0.9954,
|
| 2867 |
+
"step": 4470
|
| 2868 |
+
},
|
| 2869 |
+
{
|
| 2870 |
+
"epoch": 1.43,
|
| 2871 |
+
"learning_rate": 0.0002154782608695652,
|
| 2872 |
+
"loss": 1.0229,
|
| 2873 |
+
"step": 4480
|
| 2874 |
+
},
|
| 2875 |
+
{
|
| 2876 |
+
"epoch": 1.44,
|
| 2877 |
+
"learning_rate": 0.00021528502415458934,
|
| 2878 |
+
"loss": 0.9828,
|
| 2879 |
+
"step": 4490
|
| 2880 |
+
},
|
| 2881 |
+
{
|
| 2882 |
+
"epoch": 1.44,
|
| 2883 |
+
"learning_rate": 0.0002150917874396135,
|
| 2884 |
+
"loss": 1.0033,
|
| 2885 |
+
"step": 4500
|
| 2886 |
+
},
|
| 2887 |
+
{
|
| 2888 |
+
"epoch": 1.44,
|
| 2889 |
+
"learning_rate": 0.00021489855072463767,
|
| 2890 |
+
"loss": 1.0132,
|
| 2891 |
+
"step": 4510
|
| 2892 |
+
},
|
| 2893 |
+
{
|
| 2894 |
+
"epoch": 1.45,
|
| 2895 |
+
"learning_rate": 0.00021470531400966182,
|
| 2896 |
+
"loss": 1.0256,
|
| 2897 |
+
"step": 4520
|
| 2898 |
+
},
|
| 2899 |
+
{
|
| 2900 |
+
"epoch": 1.45,
|
| 2901 |
+
"learning_rate": 0.00021451207729468596,
|
| 2902 |
+
"loss": 1.0227,
|
| 2903 |
+
"step": 4530
|
| 2904 |
+
},
|
| 2905 |
+
{
|
| 2906 |
+
"epoch": 1.45,
|
| 2907 |
+
"learning_rate": 0.0002143188405797101,
|
| 2908 |
+
"loss": 1.0434,
|
| 2909 |
+
"step": 4540
|
| 2910 |
+
},
|
| 2911 |
+
{
|
| 2912 |
+
"epoch": 1.46,
|
| 2913 |
+
"learning_rate": 0.0002141256038647343,
|
| 2914 |
+
"loss": 1.0596,
|
| 2915 |
+
"step": 4550
|
| 2916 |
+
},
|
| 2917 |
+
{
|
| 2918 |
+
"epoch": 1.46,
|
| 2919 |
+
"learning_rate": 0.00021393236714975844,
|
| 2920 |
+
"loss": 0.9893,
|
| 2921 |
+
"step": 4560
|
| 2922 |
+
},
|
| 2923 |
+
{
|
| 2924 |
+
"epoch": 1.46,
|
| 2925 |
+
"learning_rate": 0.00021373913043478258,
|
| 2926 |
+
"loss": 1.0225,
|
| 2927 |
+
"step": 4570
|
| 2928 |
+
},
|
| 2929 |
+
{
|
| 2930 |
+
"epoch": 1.47,
|
| 2931 |
+
"learning_rate": 0.00021354589371980673,
|
| 2932 |
+
"loss": 0.981,
|
| 2933 |
+
"step": 4580
|
| 2934 |
+
},
|
| 2935 |
+
{
|
| 2936 |
+
"epoch": 1.47,
|
| 2937 |
+
"learning_rate": 0.0002133526570048309,
|
| 2938 |
+
"loss": 1.0348,
|
| 2939 |
+
"step": 4590
|
| 2940 |
+
},
|
| 2941 |
+
{
|
| 2942 |
+
"epoch": 1.47,
|
| 2943 |
+
"learning_rate": 0.00021315942028985507,
|
| 2944 |
+
"loss": 1.0208,
|
| 2945 |
+
"step": 4600
|
| 2946 |
+
},
|
| 2947 |
+
{
|
| 2948 |
+
"epoch": 1.47,
|
| 2949 |
+
"eval_loss": 1.0610480308532715,
|
| 2950 |
+
"eval_runtime": 302.534,
|
| 2951 |
+
"eval_samples_per_second": 6.611,
|
| 2952 |
+
"eval_steps_per_second": 0.826,
|
| 2953 |
+
"step": 4600
|
| 2954 |
+
},
|
| 2955 |
+
{
|
| 2956 |
+
"epoch": 1.48,
|
| 2957 |
+
"learning_rate": 0.0002129661835748792,
|
| 2958 |
+
"loss": 1.0074,
|
| 2959 |
+
"step": 4610
|
| 2960 |
+
},
|
| 2961 |
+
{
|
| 2962 |
+
"epoch": 1.48,
|
| 2963 |
+
"learning_rate": 0.00021277294685990335,
|
| 2964 |
+
"loss": 1.0173,
|
| 2965 |
+
"step": 4620
|
| 2966 |
+
},
|
| 2967 |
+
{
|
| 2968 |
+
"epoch": 1.48,
|
| 2969 |
+
"learning_rate": 0.0002125797101449275,
|
| 2970 |
+
"loss": 0.9835,
|
| 2971 |
+
"step": 4630
|
| 2972 |
+
},
|
| 2973 |
+
{
|
| 2974 |
+
"epoch": 1.48,
|
| 2975 |
+
"learning_rate": 0.0002123864734299517,
|
| 2976 |
+
"loss": 0.9819,
|
| 2977 |
+
"step": 4640
|
| 2978 |
+
},
|
| 2979 |
+
{
|
| 2980 |
+
"epoch": 1.49,
|
| 2981 |
+
"learning_rate": 0.00021219323671497583,
|
| 2982 |
+
"loss": 1.0399,
|
| 2983 |
+
"step": 4650
|
| 2984 |
+
},
|
| 2985 |
+
{
|
| 2986 |
+
"epoch": 1.49,
|
| 2987 |
+
"learning_rate": 0.00021199999999999998,
|
| 2988 |
+
"loss": 1.0076,
|
| 2989 |
+
"step": 4660
|
| 2990 |
+
},
|
| 2991 |
+
{
|
| 2992 |
+
"epoch": 1.49,
|
| 2993 |
+
"learning_rate": 0.00021180676328502412,
|
| 2994 |
+
"loss": 1.0235,
|
| 2995 |
+
"step": 4670
|
| 2996 |
+
},
|
| 2997 |
+
{
|
| 2998 |
+
"epoch": 1.5,
|
| 2999 |
+
"learning_rate": 0.0002116135265700483,
|
| 3000 |
+
"loss": 1.0269,
|
| 3001 |
+
"step": 4680
|
| 3002 |
+
},
|
| 3003 |
+
{
|
| 3004 |
+
"epoch": 1.5,
|
| 3005 |
+
"learning_rate": 0.00021142028985507246,
|
| 3006 |
+
"loss": 1.0428,
|
| 3007 |
+
"step": 4690
|
| 3008 |
+
},
|
| 3009 |
+
{
|
| 3010 |
+
"epoch": 1.5,
|
| 3011 |
+
"learning_rate": 0.0002112270531400966,
|
| 3012 |
+
"loss": 1.0019,
|
| 3013 |
+
"step": 4700
|
| 3014 |
+
},
|
| 3015 |
+
{
|
| 3016 |
+
"epoch": 1.51,
|
| 3017 |
+
"learning_rate": 0.00021103381642512074,
|
| 3018 |
+
"loss": 0.9701,
|
| 3019 |
+
"step": 4710
|
| 3020 |
+
},
|
| 3021 |
+
{
|
| 3022 |
+
"epoch": 1.51,
|
| 3023 |
+
"learning_rate": 0.00021084057971014489,
|
| 3024 |
+
"loss": 0.977,
|
| 3025 |
+
"step": 4720
|
| 3026 |
+
},
|
| 3027 |
+
{
|
| 3028 |
+
"epoch": 1.51,
|
| 3029 |
+
"learning_rate": 0.00021064734299516908,
|
| 3030 |
+
"loss": 0.9753,
|
| 3031 |
+
"step": 4730
|
| 3032 |
+
},
|
| 3033 |
+
{
|
| 3034 |
+
"epoch": 1.52,
|
| 3035 |
+
"learning_rate": 0.00021045410628019322,
|
| 3036 |
+
"loss": 1.0432,
|
| 3037 |
+
"step": 4740
|
| 3038 |
+
},
|
| 3039 |
+
{
|
| 3040 |
+
"epoch": 1.52,
|
| 3041 |
+
"learning_rate": 0.00021026086956521737,
|
| 3042 |
+
"loss": 1.051,
|
| 3043 |
+
"step": 4750
|
| 3044 |
+
},
|
| 3045 |
+
{
|
| 3046 |
+
"epoch": 1.52,
|
| 3047 |
+
"learning_rate": 0.0002100676328502415,
|
| 3048 |
+
"loss": 1.0301,
|
| 3049 |
+
"step": 4760
|
| 3050 |
+
},
|
| 3051 |
+
{
|
| 3052 |
+
"epoch": 1.53,
|
| 3053 |
+
"learning_rate": 0.00020987439613526565,
|
| 3054 |
+
"loss": 0.9793,
|
| 3055 |
+
"step": 4770
|
| 3056 |
+
},
|
| 3057 |
+
{
|
| 3058 |
+
"epoch": 1.53,
|
| 3059 |
+
"learning_rate": 0.00020968115942028985,
|
| 3060 |
+
"loss": 1.0151,
|
| 3061 |
+
"step": 4780
|
| 3062 |
+
},
|
| 3063 |
+
{
|
| 3064 |
+
"epoch": 1.53,
|
| 3065 |
+
"learning_rate": 0.000209487922705314,
|
| 3066 |
+
"loss": 0.9979,
|
| 3067 |
+
"step": 4790
|
| 3068 |
+
},
|
| 3069 |
+
{
|
| 3070 |
+
"epoch": 1.54,
|
| 3071 |
+
"learning_rate": 0.00020929468599033813,
|
| 3072 |
+
"loss": 1.0126,
|
| 3073 |
+
"step": 4800
|
| 3074 |
+
},
|
| 3075 |
+
{
|
| 3076 |
+
"epoch": 1.54,
|
| 3077 |
+
"eval_loss": 1.0597844123840332,
|
| 3078 |
+
"eval_runtime": 303.6517,
|
| 3079 |
+
"eval_samples_per_second": 6.586,
|
| 3080 |
+
"eval_steps_per_second": 0.823,
|
| 3081 |
+
"step": 4800
|
| 3082 |
+
},
|
| 3083 |
+
{
|
| 3084 |
+
"epoch": 1.54,
|
| 3085 |
+
"learning_rate": 0.00020910144927536228,
|
| 3086 |
+
"loss": 1.0428,
|
| 3087 |
+
"step": 4810
|
| 3088 |
+
},
|
| 3089 |
+
{
|
| 3090 |
+
"epoch": 1.54,
|
| 3091 |
+
"learning_rate": 0.00020890821256038647,
|
| 3092 |
+
"loss": 1.0049,
|
| 3093 |
+
"step": 4820
|
| 3094 |
+
},
|
| 3095 |
+
{
|
| 3096 |
+
"epoch": 1.55,
|
| 3097 |
+
"learning_rate": 0.00020871497584541062,
|
| 3098 |
+
"loss": 1.0006,
|
| 3099 |
+
"step": 4830
|
| 3100 |
+
},
|
| 3101 |
+
{
|
| 3102 |
+
"epoch": 1.55,
|
| 3103 |
+
"learning_rate": 0.00020852173913043476,
|
| 3104 |
+
"loss": 1.022,
|
| 3105 |
+
"step": 4840
|
| 3106 |
+
},
|
| 3107 |
+
{
|
| 3108 |
+
"epoch": 1.55,
|
| 3109 |
+
"learning_rate": 0.0002083285024154589,
|
| 3110 |
+
"loss": 1.0342,
|
| 3111 |
+
"step": 4850
|
| 3112 |
+
},
|
| 3113 |
+
{
|
| 3114 |
+
"epoch": 1.56,
|
| 3115 |
+
"learning_rate": 0.0002081352657004831,
|
| 3116 |
+
"loss": 1.0146,
|
| 3117 |
+
"step": 4860
|
| 3118 |
+
},
|
| 3119 |
+
{
|
| 3120 |
+
"epoch": 1.56,
|
| 3121 |
+
"learning_rate": 0.00020794202898550724,
|
| 3122 |
+
"loss": 1.029,
|
| 3123 |
+
"step": 4870
|
| 3124 |
+
},
|
| 3125 |
+
{
|
| 3126 |
+
"epoch": 1.56,
|
| 3127 |
+
"learning_rate": 0.00020774879227053138,
|
| 3128 |
+
"loss": 1.0381,
|
| 3129 |
+
"step": 4880
|
| 3130 |
+
},
|
| 3131 |
+
{
|
| 3132 |
+
"epoch": 1.56,
|
| 3133 |
+
"learning_rate": 0.00020755555555555553,
|
| 3134 |
+
"loss": 1.0372,
|
| 3135 |
+
"step": 4890
|
| 3136 |
+
},
|
| 3137 |
+
{
|
| 3138 |
+
"epoch": 1.57,
|
| 3139 |
+
"learning_rate": 0.00020736231884057967,
|
| 3140 |
+
"loss": 1.0119,
|
| 3141 |
+
"step": 4900
|
| 3142 |
+
},
|
| 3143 |
+
{
|
| 3144 |
+
"epoch": 1.57,
|
| 3145 |
+
"learning_rate": 0.00020716908212560387,
|
| 3146 |
+
"loss": 1.0797,
|
| 3147 |
+
"step": 4910
|
| 3148 |
+
},
|
| 3149 |
+
{
|
| 3150 |
+
"epoch": 1.57,
|
| 3151 |
+
"learning_rate": 0.000206975845410628,
|
| 3152 |
+
"loss": 0.95,
|
| 3153 |
+
"step": 4920
|
| 3154 |
+
},
|
| 3155 |
+
{
|
| 3156 |
+
"epoch": 1.58,
|
| 3157 |
+
"learning_rate": 0.00020678260869565215,
|
| 3158 |
+
"loss": 1.0012,
|
| 3159 |
+
"step": 4930
|
| 3160 |
+
},
|
| 3161 |
+
{
|
| 3162 |
+
"epoch": 1.58,
|
| 3163 |
+
"learning_rate": 0.0002065893719806763,
|
| 3164 |
+
"loss": 1.0194,
|
| 3165 |
+
"step": 4940
|
| 3166 |
+
},
|
| 3167 |
+
{
|
| 3168 |
+
"epoch": 1.58,
|
| 3169 |
+
"learning_rate": 0.0002063961352657005,
|
| 3170 |
+
"loss": 1.0135,
|
| 3171 |
+
"step": 4950
|
| 3172 |
+
},
|
| 3173 |
+
{
|
| 3174 |
+
"epoch": 1.59,
|
| 3175 |
+
"learning_rate": 0.00020620289855072463,
|
| 3176 |
+
"loss": 1.0294,
|
| 3177 |
+
"step": 4960
|
| 3178 |
+
},
|
| 3179 |
+
{
|
| 3180 |
+
"epoch": 1.59,
|
| 3181 |
+
"learning_rate": 0.00020600966183574877,
|
| 3182 |
+
"loss": 0.9954,
|
| 3183 |
+
"step": 4970
|
| 3184 |
+
},
|
| 3185 |
+
{
|
| 3186 |
+
"epoch": 1.59,
|
| 3187 |
+
"learning_rate": 0.00020581642512077292,
|
| 3188 |
+
"loss": 1.0082,
|
| 3189 |
+
"step": 4980
|
| 3190 |
+
},
|
| 3191 |
+
{
|
| 3192 |
+
"epoch": 1.6,
|
| 3193 |
+
"learning_rate": 0.00020562318840579706,
|
| 3194 |
+
"loss": 1.0343,
|
| 3195 |
+
"step": 4990
|
| 3196 |
+
},
|
| 3197 |
+
{
|
| 3198 |
+
"epoch": 1.6,
|
| 3199 |
+
"learning_rate": 0.00020542995169082126,
|
| 3200 |
+
"loss": 0.9967,
|
| 3201 |
+
"step": 5000
|
| 3202 |
+
},
|
| 3203 |
+
{
|
| 3204 |
+
"epoch": 1.6,
|
| 3205 |
+
"eval_loss": 1.060504674911499,
|
| 3206 |
+
"eval_runtime": 303.6571,
|
| 3207 |
+
"eval_samples_per_second": 6.586,
|
| 3208 |
+
"eval_steps_per_second": 0.823,
|
| 3209 |
+
"step": 5000
|
| 3210 |
+
},
|
| 3211 |
+
{
|
| 3212 |
+
"epoch": 1.6,
|
| 3213 |
+
"learning_rate": 0.0002052367149758454,
|
| 3214 |
+
"loss": 1.0436,
|
| 3215 |
+
"step": 5010
|
| 3216 |
+
},
|
| 3217 |
+
{
|
| 3218 |
+
"epoch": 1.61,
|
| 3219 |
+
"learning_rate": 0.00020504347826086954,
|
| 3220 |
+
"loss": 1.037,
|
| 3221 |
+
"step": 5020
|
| 3222 |
+
},
|
| 3223 |
+
{
|
| 3224 |
+
"epoch": 1.61,
|
| 3225 |
+
"learning_rate": 0.00020485024154589368,
|
| 3226 |
+
"loss": 1.0613,
|
| 3227 |
+
"step": 5030
|
| 3228 |
+
},
|
| 3229 |
+
{
|
| 3230 |
+
"epoch": 1.61,
|
| 3231 |
+
"learning_rate": 0.00020465700483091788,
|
| 3232 |
+
"loss": 1.0305,
|
| 3233 |
+
"step": 5040
|
| 3234 |
+
},
|
| 3235 |
+
{
|
| 3236 |
+
"epoch": 1.62,
|
| 3237 |
+
"learning_rate": 0.00020446376811594202,
|
| 3238 |
+
"loss": 1.0585,
|
| 3239 |
+
"step": 5050
|
| 3240 |
+
},
|
| 3241 |
+
{
|
| 3242 |
+
"epoch": 1.62,
|
| 3243 |
+
"learning_rate": 0.00020427053140096617,
|
| 3244 |
+
"loss": 1.0071,
|
| 3245 |
+
"step": 5060
|
| 3246 |
+
},
|
| 3247 |
+
{
|
| 3248 |
+
"epoch": 1.62,
|
| 3249 |
+
"learning_rate": 0.0002040772946859903,
|
| 3250 |
+
"loss": 0.9929,
|
| 3251 |
+
"step": 5070
|
| 3252 |
+
},
|
| 3253 |
+
{
|
| 3254 |
+
"epoch": 1.63,
|
| 3255 |
+
"learning_rate": 0.00020388405797101445,
|
| 3256 |
+
"loss": 1.051,
|
| 3257 |
+
"step": 5080
|
| 3258 |
+
},
|
| 3259 |
+
{
|
| 3260 |
+
"epoch": 1.63,
|
| 3261 |
+
"learning_rate": 0.00020369082125603865,
|
| 3262 |
+
"loss": 0.9953,
|
| 3263 |
+
"step": 5090
|
| 3264 |
+
},
|
| 3265 |
+
{
|
| 3266 |
+
"epoch": 1.63,
|
| 3267 |
+
"learning_rate": 0.0002034975845410628,
|
| 3268 |
+
"loss": 1.0176,
|
| 3269 |
+
"step": 5100
|
| 3270 |
+
},
|
| 3271 |
+
{
|
| 3272 |
+
"epoch": 1.64,
|
| 3273 |
+
"learning_rate": 0.00020330434782608693,
|
| 3274 |
+
"loss": 1.0107,
|
| 3275 |
+
"step": 5110
|
| 3276 |
+
},
|
| 3277 |
+
{
|
| 3278 |
+
"epoch": 1.64,
|
| 3279 |
+
"learning_rate": 0.00020311111111111108,
|
| 3280 |
+
"loss": 1.0618,
|
| 3281 |
+
"step": 5120
|
| 3282 |
+
},
|
| 3283 |
+
{
|
| 3284 |
+
"epoch": 1.64,
|
| 3285 |
+
"learning_rate": 0.00020291787439613527,
|
| 3286 |
+
"loss": 1.0147,
|
| 3287 |
+
"step": 5130
|
| 3288 |
+
},
|
| 3289 |
+
{
|
| 3290 |
+
"epoch": 1.64,
|
| 3291 |
+
"learning_rate": 0.00020272463768115942,
|
| 3292 |
+
"loss": 0.9754,
|
| 3293 |
+
"step": 5140
|
| 3294 |
+
},
|
| 3295 |
+
{
|
| 3296 |
+
"epoch": 1.65,
|
| 3297 |
+
"learning_rate": 0.00020253140096618356,
|
| 3298 |
+
"loss": 1.0381,
|
| 3299 |
+
"step": 5150
|
| 3300 |
+
},
|
| 3301 |
+
{
|
| 3302 |
+
"epoch": 1.65,
|
| 3303 |
+
"learning_rate": 0.0002023381642512077,
|
| 3304 |
+
"loss": 0.9645,
|
| 3305 |
+
"step": 5160
|
| 3306 |
+
},
|
| 3307 |
+
{
|
| 3308 |
+
"epoch": 1.65,
|
| 3309 |
+
"learning_rate": 0.00020214492753623187,
|
| 3310 |
+
"loss": 0.9645,
|
| 3311 |
+
"step": 5170
|
| 3312 |
+
},
|
| 3313 |
+
{
|
| 3314 |
+
"epoch": 1.66,
|
| 3315 |
+
"learning_rate": 0.00020195169082125604,
|
| 3316 |
+
"loss": 1.0371,
|
| 3317 |
+
"step": 5180
|
| 3318 |
+
},
|
| 3319 |
+
{
|
| 3320 |
+
"epoch": 1.66,
|
| 3321 |
+
"learning_rate": 0.00020175845410628018,
|
| 3322 |
+
"loss": 1.0058,
|
| 3323 |
+
"step": 5190
|
| 3324 |
+
},
|
| 3325 |
+
{
|
| 3326 |
+
"epoch": 1.66,
|
| 3327 |
+
"learning_rate": 0.00020156521739130432,
|
| 3328 |
+
"loss": 1.02,
|
| 3329 |
+
"step": 5200
|
| 3330 |
+
},
|
| 3331 |
+
{
|
| 3332 |
+
"epoch": 1.66,
|
| 3333 |
+
"eval_loss": 1.0592375993728638,
|
| 3334 |
+
"eval_runtime": 303.4125,
|
| 3335 |
+
"eval_samples_per_second": 6.592,
|
| 3336 |
+
"eval_steps_per_second": 0.824,
|
| 3337 |
+
"step": 5200
|
| 3338 |
+
},
|
| 3339 |
+
{
|
| 3340 |
+
"epoch": 1.67,
|
| 3341 |
+
"learning_rate": 0.00020137198067632847,
|
| 3342 |
+
"loss": 1.0409,
|
| 3343 |
+
"step": 5210
|
| 3344 |
+
},
|
| 3345 |
+
{
|
| 3346 |
+
"epoch": 1.67,
|
| 3347 |
+
"learning_rate": 0.00020117874396135266,
|
| 3348 |
+
"loss": 1.0193,
|
| 3349 |
+
"step": 5220
|
| 3350 |
+
},
|
| 3351 |
+
{
|
| 3352 |
+
"epoch": 1.67,
|
| 3353 |
+
"learning_rate": 0.0002009855072463768,
|
| 3354 |
+
"loss": 1.0169,
|
| 3355 |
+
"step": 5230
|
| 3356 |
+
},
|
| 3357 |
+
{
|
| 3358 |
+
"epoch": 1.68,
|
| 3359 |
+
"learning_rate": 0.00020079227053140095,
|
| 3360 |
+
"loss": 1.0484,
|
| 3361 |
+
"step": 5240
|
| 3362 |
+
},
|
| 3363 |
+
{
|
| 3364 |
+
"epoch": 1.68,
|
| 3365 |
+
"learning_rate": 0.0002005990338164251,
|
| 3366 |
+
"loss": 0.9895,
|
| 3367 |
+
"step": 5250
|
| 3368 |
+
},
|
| 3369 |
+
{
|
| 3370 |
+
"epoch": 1.68,
|
| 3371 |
+
"learning_rate": 0.00020040579710144926,
|
| 3372 |
+
"loss": 0.9858,
|
| 3373 |
+
"step": 5260
|
| 3374 |
+
},
|
| 3375 |
+
{
|
| 3376 |
+
"epoch": 1.69,
|
| 3377 |
+
"learning_rate": 0.00020021256038647343,
|
| 3378 |
+
"loss": 0.9993,
|
| 3379 |
+
"step": 5270
|
| 3380 |
+
},
|
| 3381 |
+
{
|
| 3382 |
+
"epoch": 1.69,
|
| 3383 |
+
"learning_rate": 0.00020001932367149757,
|
| 3384 |
+
"loss": 0.9755,
|
| 3385 |
+
"step": 5280
|
| 3386 |
+
},
|
| 3387 |
+
{
|
| 3388 |
+
"epoch": 1.69,
|
| 3389 |
+
"learning_rate": 0.00019982608695652172,
|
| 3390 |
+
"loss": 1.0392,
|
| 3391 |
+
"step": 5290
|
| 3392 |
+
},
|
| 3393 |
+
{
|
| 3394 |
+
"epoch": 1.7,
|
| 3395 |
+
"learning_rate": 0.00019963285024154586,
|
| 3396 |
+
"loss": 1.0123,
|
| 3397 |
+
"step": 5300
|
| 3398 |
+
},
|
| 3399 |
+
{
|
| 3400 |
+
"epoch": 1.7,
|
| 3401 |
+
"learning_rate": 0.00019943961352657003,
|
| 3402 |
+
"loss": 1.0386,
|
| 3403 |
+
"step": 5310
|
| 3404 |
+
},
|
| 3405 |
+
{
|
| 3406 |
+
"epoch": 1.7,
|
| 3407 |
+
"learning_rate": 0.0001992463768115942,
|
| 3408 |
+
"loss": 1.0359,
|
| 3409 |
+
"step": 5320
|
| 3410 |
+
},
|
| 3411 |
+
{
|
| 3412 |
+
"epoch": 1.71,
|
| 3413 |
+
"learning_rate": 0.00019905314009661834,
|
| 3414 |
+
"loss": 1.0003,
|
| 3415 |
+
"step": 5330
|
| 3416 |
+
},
|
| 3417 |
+
{
|
| 3418 |
+
"epoch": 1.71,
|
| 3419 |
+
"learning_rate": 0.00019885990338164248,
|
| 3420 |
+
"loss": 0.9989,
|
| 3421 |
+
"step": 5340
|
| 3422 |
+
},
|
| 3423 |
+
{
|
| 3424 |
+
"epoch": 1.71,
|
| 3425 |
+
"learning_rate": 0.00019866666666666665,
|
| 3426 |
+
"loss": 0.9515,
|
| 3427 |
+
"step": 5350
|
| 3428 |
+
},
|
| 3429 |
+
{
|
| 3430 |
+
"epoch": 1.72,
|
| 3431 |
+
"learning_rate": 0.00019847342995169082,
|
| 3432 |
+
"loss": 1.0035,
|
| 3433 |
+
"step": 5360
|
| 3434 |
+
},
|
| 3435 |
+
{
|
| 3436 |
+
"epoch": 1.72,
|
| 3437 |
+
"learning_rate": 0.00019828019323671497,
|
| 3438 |
+
"loss": 1.0219,
|
| 3439 |
+
"step": 5370
|
| 3440 |
+
},
|
| 3441 |
+
{
|
| 3442 |
+
"epoch": 1.72,
|
| 3443 |
+
"learning_rate": 0.0001980869565217391,
|
| 3444 |
+
"loss": 1.0096,
|
| 3445 |
+
"step": 5380
|
| 3446 |
+
},
|
| 3447 |
+
{
|
| 3448 |
+
"epoch": 1.72,
|
| 3449 |
+
"learning_rate": 0.00019789371980676325,
|
| 3450 |
+
"loss": 1.0077,
|
| 3451 |
+
"step": 5390
|
| 3452 |
+
},
|
| 3453 |
+
{
|
| 3454 |
+
"epoch": 1.73,
|
| 3455 |
+
"learning_rate": 0.00019770048309178742,
|
| 3456 |
+
"loss": 1.0298,
|
| 3457 |
+
"step": 5400
|
| 3458 |
+
},
|
| 3459 |
+
{
|
| 3460 |
+
"epoch": 1.73,
|
| 3461 |
+
"eval_loss": 1.0586228370666504,
|
| 3462 |
+
"eval_runtime": 302.8462,
|
| 3463 |
+
"eval_samples_per_second": 6.604,
|
| 3464 |
+
"eval_steps_per_second": 0.826,
|
| 3465 |
+
"step": 5400
|
| 3466 |
+
},
|
| 3467 |
+
{
|
| 3468 |
+
"epoch": 1.73,
|
| 3469 |
+
"learning_rate": 0.0001975072463768116,
|
| 3470 |
+
"loss": 1.0001,
|
| 3471 |
+
"step": 5410
|
| 3472 |
+
},
|
| 3473 |
+
{
|
| 3474 |
+
"epoch": 1.73,
|
| 3475 |
+
"learning_rate": 0.00019731400966183573,
|
| 3476 |
+
"loss": 1.0465,
|
| 3477 |
+
"step": 5420
|
| 3478 |
+
},
|
| 3479 |
+
{
|
| 3480 |
+
"epoch": 1.74,
|
| 3481 |
+
"learning_rate": 0.00019712077294685987,
|
| 3482 |
+
"loss": 1.0265,
|
| 3483 |
+
"step": 5430
|
| 3484 |
+
},
|
| 3485 |
+
{
|
| 3486 |
+
"epoch": 1.74,
|
| 3487 |
+
"learning_rate": 0.00019692753623188404,
|
| 3488 |
+
"loss": 1.0148,
|
| 3489 |
+
"step": 5440
|
| 3490 |
+
},
|
| 3491 |
+
{
|
| 3492 |
+
"epoch": 1.74,
|
| 3493 |
+
"learning_rate": 0.0001967342995169082,
|
| 3494 |
+
"loss": 0.9763,
|
| 3495 |
+
"step": 5450
|
| 3496 |
+
},
|
| 3497 |
+
{
|
| 3498 |
+
"epoch": 1.75,
|
| 3499 |
+
"learning_rate": 0.00019654106280193236,
|
| 3500 |
+
"loss": 0.9861,
|
| 3501 |
+
"step": 5460
|
| 3502 |
+
},
|
| 3503 |
+
{
|
| 3504 |
+
"epoch": 1.75,
|
| 3505 |
+
"learning_rate": 0.0001963478260869565,
|
| 3506 |
+
"loss": 0.9877,
|
| 3507 |
+
"step": 5470
|
| 3508 |
+
},
|
| 3509 |
+
{
|
| 3510 |
+
"epoch": 1.75,
|
| 3511 |
+
"learning_rate": 0.00019615458937198067,
|
| 3512 |
+
"loss": 0.9489,
|
| 3513 |
+
"step": 5480
|
| 3514 |
+
},
|
| 3515 |
+
{
|
| 3516 |
+
"epoch": 1.76,
|
| 3517 |
+
"learning_rate": 0.0001959613526570048,
|
| 3518 |
+
"loss": 1.0694,
|
| 3519 |
+
"step": 5490
|
| 3520 |
+
},
|
| 3521 |
+
{
|
| 3522 |
+
"epoch": 1.76,
|
| 3523 |
+
"learning_rate": 0.00019576811594202898,
|
| 3524 |
+
"loss": 1.0311,
|
| 3525 |
+
"step": 5500
|
| 3526 |
+
},
|
| 3527 |
+
{
|
| 3528 |
+
"epoch": 1.76,
|
| 3529 |
+
"learning_rate": 0.00019557487922705312,
|
| 3530 |
+
"loss": 1.0494,
|
| 3531 |
+
"step": 5510
|
| 3532 |
+
},
|
| 3533 |
+
{
|
| 3534 |
+
"epoch": 1.77,
|
| 3535 |
+
"learning_rate": 0.00019538164251207727,
|
| 3536 |
+
"loss": 0.9967,
|
| 3537 |
+
"step": 5520
|
| 3538 |
+
},
|
| 3539 |
+
{
|
| 3540 |
+
"epoch": 1.77,
|
| 3541 |
+
"learning_rate": 0.00019518840579710144,
|
| 3542 |
+
"loss": 1.0278,
|
| 3543 |
+
"step": 5530
|
| 3544 |
+
},
|
| 3545 |
+
{
|
| 3546 |
+
"epoch": 1.77,
|
| 3547 |
+
"learning_rate": 0.00019499516908212558,
|
| 3548 |
+
"loss": 1.0384,
|
| 3549 |
+
"step": 5540
|
| 3550 |
+
},
|
| 3551 |
+
{
|
| 3552 |
+
"epoch": 1.78,
|
| 3553 |
+
"learning_rate": 0.00019480193236714975,
|
| 3554 |
+
"loss": 1.0051,
|
| 3555 |
+
"step": 5550
|
| 3556 |
+
},
|
| 3557 |
+
{
|
| 3558 |
+
"epoch": 1.78,
|
| 3559 |
+
"learning_rate": 0.0001946086956521739,
|
| 3560 |
+
"loss": 1.0387,
|
| 3561 |
+
"step": 5560
|
| 3562 |
+
},
|
| 3563 |
+
{
|
| 3564 |
+
"epoch": 1.78,
|
| 3565 |
+
"learning_rate": 0.00019441545893719806,
|
| 3566 |
+
"loss": 1.0094,
|
| 3567 |
+
"step": 5570
|
| 3568 |
+
},
|
| 3569 |
+
{
|
| 3570 |
+
"epoch": 1.79,
|
| 3571 |
+
"learning_rate": 0.0001942222222222222,
|
| 3572 |
+
"loss": 1.0155,
|
| 3573 |
+
"step": 5580
|
| 3574 |
+
},
|
| 3575 |
+
{
|
| 3576 |
+
"epoch": 1.79,
|
| 3577 |
+
"learning_rate": 0.00019402898550724637,
|
| 3578 |
+
"loss": 1.0104,
|
| 3579 |
+
"step": 5590
|
| 3580 |
+
},
|
| 3581 |
+
{
|
| 3582 |
+
"epoch": 1.79,
|
| 3583 |
+
"learning_rate": 0.00019383574879227052,
|
| 3584 |
+
"loss": 1.0264,
|
| 3585 |
+
"step": 5600
|
| 3586 |
+
},
|
| 3587 |
+
{
|
| 3588 |
+
"epoch": 1.79,
|
| 3589 |
+
"eval_loss": 1.058441162109375,
|
| 3590 |
+
"eval_runtime": 301.1664,
|
| 3591 |
+
"eval_samples_per_second": 6.641,
|
| 3592 |
+
"eval_steps_per_second": 0.83,
|
| 3593 |
+
"step": 5600
|
| 3594 |
+
},
|
| 3595 |
+
{
|
| 3596 |
+
"epoch": 1.8,
|
| 3597 |
+
"learning_rate": 0.00019364251207729466,
|
| 3598 |
+
"loss": 0.9823,
|
| 3599 |
+
"step": 5610
|
| 3600 |
+
},
|
| 3601 |
+
{
|
| 3602 |
+
"epoch": 1.8,
|
| 3603 |
+
"learning_rate": 0.00019344927536231883,
|
| 3604 |
+
"loss": 0.9871,
|
| 3605 |
+
"step": 5620
|
| 3606 |
+
},
|
| 3607 |
+
{
|
| 3608 |
+
"epoch": 1.8,
|
| 3609 |
+
"learning_rate": 0.00019325603864734297,
|
| 3610 |
+
"loss": 1.022,
|
| 3611 |
+
"step": 5630
|
| 3612 |
+
},
|
| 3613 |
+
{
|
| 3614 |
+
"epoch": 1.8,
|
| 3615 |
+
"learning_rate": 0.00019306280193236714,
|
| 3616 |
+
"loss": 1.0128,
|
| 3617 |
+
"step": 5640
|
| 3618 |
+
},
|
| 3619 |
+
{
|
| 3620 |
+
"epoch": 1.81,
|
| 3621 |
+
"learning_rate": 0.00019286956521739128,
|
| 3622 |
+
"loss": 1.0502,
|
| 3623 |
+
"step": 5650
|
| 3624 |
+
},
|
| 3625 |
+
{
|
| 3626 |
+
"epoch": 1.81,
|
| 3627 |
+
"learning_rate": 0.00019267632850241545,
|
| 3628 |
+
"loss": 1.0427,
|
| 3629 |
+
"step": 5660
|
| 3630 |
+
},
|
| 3631 |
+
{
|
| 3632 |
+
"epoch": 1.81,
|
| 3633 |
+
"learning_rate": 0.0001924830917874396,
|
| 3634 |
+
"loss": 1.0088,
|
| 3635 |
+
"step": 5670
|
| 3636 |
+
},
|
| 3637 |
+
{
|
| 3638 |
+
"epoch": 1.82,
|
| 3639 |
+
"learning_rate": 0.00019228985507246374,
|
| 3640 |
+
"loss": 1.0319,
|
| 3641 |
+
"step": 5680
|
| 3642 |
+
},
|
| 3643 |
+
{
|
| 3644 |
+
"epoch": 1.82,
|
| 3645 |
+
"learning_rate": 0.0001920966183574879,
|
| 3646 |
+
"loss": 1.02,
|
| 3647 |
+
"step": 5690
|
| 3648 |
+
},
|
| 3649 |
+
{
|
| 3650 |
+
"epoch": 1.82,
|
| 3651 |
+
"learning_rate": 0.00019190338164251205,
|
| 3652 |
+
"loss": 1.0376,
|
| 3653 |
+
"step": 5700
|
| 3654 |
+
},
|
| 3655 |
+
{
|
| 3656 |
+
"epoch": 1.83,
|
| 3657 |
+
"learning_rate": 0.00019171014492753622,
|
| 3658 |
+
"loss": 1.0233,
|
| 3659 |
+
"step": 5710
|
| 3660 |
+
},
|
| 3661 |
+
{
|
| 3662 |
+
"epoch": 1.83,
|
| 3663 |
+
"learning_rate": 0.00019151690821256036,
|
| 3664 |
+
"loss": 1.0285,
|
| 3665 |
+
"step": 5720
|
| 3666 |
+
},
|
| 3667 |
+
{
|
| 3668 |
+
"epoch": 1.83,
|
| 3669 |
+
"learning_rate": 0.00019132367149758453,
|
| 3670 |
+
"loss": 1.0579,
|
| 3671 |
+
"step": 5730
|
| 3672 |
+
},
|
| 3673 |
+
{
|
| 3674 |
+
"epoch": 1.84,
|
| 3675 |
+
"learning_rate": 0.00019113043478260867,
|
| 3676 |
+
"loss": 1.0542,
|
| 3677 |
+
"step": 5740
|
| 3678 |
+
},
|
| 3679 |
+
{
|
| 3680 |
+
"epoch": 1.84,
|
| 3681 |
+
"learning_rate": 0.00019093719806763284,
|
| 3682 |
+
"loss": 1.0427,
|
| 3683 |
+
"step": 5750
|
| 3684 |
+
},
|
| 3685 |
+
{
|
| 3686 |
+
"epoch": 1.84,
|
| 3687 |
+
"learning_rate": 0.00019074396135265699,
|
| 3688 |
+
"loss": 0.9983,
|
| 3689 |
+
"step": 5760
|
| 3690 |
+
},
|
| 3691 |
+
{
|
| 3692 |
+
"epoch": 1.85,
|
| 3693 |
+
"learning_rate": 0.00019055072463768113,
|
| 3694 |
+
"loss": 1.0145,
|
| 3695 |
+
"step": 5770
|
| 3696 |
+
},
|
| 3697 |
+
{
|
| 3698 |
+
"epoch": 1.85,
|
| 3699 |
+
"learning_rate": 0.0001903574879227053,
|
| 3700 |
+
"loss": 1.0471,
|
| 3701 |
+
"step": 5780
|
| 3702 |
+
},
|
| 3703 |
+
{
|
| 3704 |
+
"epoch": 1.85,
|
| 3705 |
+
"learning_rate": 0.00019016425120772947,
|
| 3706 |
+
"loss": 1.0436,
|
| 3707 |
+
"step": 5790
|
| 3708 |
+
},
|
| 3709 |
+
{
|
| 3710 |
+
"epoch": 1.86,
|
| 3711 |
+
"learning_rate": 0.0001899710144927536,
|
| 3712 |
+
"loss": 1.0312,
|
| 3713 |
+
"step": 5800
|
| 3714 |
+
},
|
| 3715 |
+
{
|
| 3716 |
+
"epoch": 1.86,
|
| 3717 |
+
"eval_loss": 1.056977391242981,
|
| 3718 |
+
"eval_runtime": 105.8988,
|
| 3719 |
+
"eval_samples_per_second": 18.886,
|
| 3720 |
+
"eval_steps_per_second": 2.361,
|
| 3721 |
+
"step": 5800
|
| 3722 |
+
},
|
| 3723 |
+
{
|
| 3724 |
+
"epoch": 1.86,
|
| 3725 |
+
"learning_rate": 0.00018977777777777775,
|
| 3726 |
+
"loss": 0.9989,
|
| 3727 |
+
"step": 5810
|
| 3728 |
+
},
|
| 3729 |
+
{
|
| 3730 |
+
"epoch": 1.86,
|
| 3731 |
+
"learning_rate": 0.00018958454106280192,
|
| 3732 |
+
"loss": 1.0022,
|
| 3733 |
+
"step": 5820
|
| 3734 |
+
},
|
| 3735 |
+
{
|
| 3736 |
+
"epoch": 1.87,
|
| 3737 |
+
"learning_rate": 0.00018939130434782606,
|
| 3738 |
+
"loss": 1.0295,
|
| 3739 |
+
"step": 5830
|
| 3740 |
+
},
|
| 3741 |
+
{
|
| 3742 |
+
"epoch": 1.87,
|
| 3743 |
+
"learning_rate": 0.00018919806763285023,
|
| 3744 |
+
"loss": 1.0644,
|
| 3745 |
+
"step": 5840
|
| 3746 |
+
},
|
| 3747 |
+
{
|
| 3748 |
+
"epoch": 1.87,
|
| 3749 |
+
"learning_rate": 0.00018900483091787438,
|
| 3750 |
+
"loss": 1.0238,
|
| 3751 |
+
"step": 5850
|
| 3752 |
+
},
|
| 3753 |
+
{
|
| 3754 |
+
"epoch": 1.88,
|
| 3755 |
+
"learning_rate": 0.00018881159420289852,
|
| 3756 |
+
"loss": 1.06,
|
| 3757 |
+
"step": 5860
|
| 3758 |
+
},
|
| 3759 |
+
{
|
| 3760 |
+
"epoch": 1.88,
|
| 3761 |
+
"learning_rate": 0.0001886183574879227,
|
| 3762 |
+
"loss": 0.9779,
|
| 3763 |
+
"step": 5870
|
| 3764 |
+
},
|
| 3765 |
+
{
|
| 3766 |
+
"epoch": 1.88,
|
| 3767 |
+
"learning_rate": 0.00018842512077294686,
|
| 3768 |
+
"loss": 1.0072,
|
| 3769 |
+
"step": 5880
|
| 3770 |
+
},
|
| 3771 |
+
{
|
| 3772 |
+
"epoch": 1.88,
|
| 3773 |
+
"learning_rate": 0.000188231884057971,
|
| 3774 |
+
"loss": 1.0261,
|
| 3775 |
+
"step": 5890
|
| 3776 |
+
},
|
| 3777 |
+
{
|
| 3778 |
+
"epoch": 1.89,
|
| 3779 |
+
"learning_rate": 0.00018803864734299514,
|
| 3780 |
+
"loss": 1.0632,
|
| 3781 |
+
"step": 5900
|
| 3782 |
+
},
|
| 3783 |
+
{
|
| 3784 |
+
"epoch": 1.89,
|
| 3785 |
+
"learning_rate": 0.0001878454106280193,
|
| 3786 |
+
"loss": 0.9929,
|
| 3787 |
+
"step": 5910
|
| 3788 |
+
},
|
| 3789 |
+
{
|
| 3790 |
+
"epoch": 1.89,
|
| 3791 |
+
"learning_rate": 0.00018765217391304346,
|
| 3792 |
+
"loss": 0.9593,
|
| 3793 |
+
"step": 5920
|
| 3794 |
+
},
|
| 3795 |
+
{
|
| 3796 |
+
"epoch": 1.9,
|
| 3797 |
+
"learning_rate": 0.00018745893719806763,
|
| 3798 |
+
"loss": 1.051,
|
| 3799 |
+
"step": 5930
|
| 3800 |
+
},
|
| 3801 |
+
{
|
| 3802 |
+
"epoch": 1.9,
|
| 3803 |
+
"learning_rate": 0.00018726570048309177,
|
| 3804 |
+
"loss": 1.0117,
|
| 3805 |
+
"step": 5940
|
| 3806 |
+
},
|
| 3807 |
+
{
|
| 3808 |
+
"epoch": 1.9,
|
| 3809 |
+
"learning_rate": 0.0001870724637681159,
|
| 3810 |
+
"loss": 1.041,
|
| 3811 |
+
"step": 5950
|
| 3812 |
+
},
|
| 3813 |
+
{
|
| 3814 |
+
"epoch": 1.91,
|
| 3815 |
+
"learning_rate": 0.00018687922705314008,
|
| 3816 |
+
"loss": 0.9918,
|
| 3817 |
+
"step": 5960
|
| 3818 |
+
},
|
| 3819 |
+
{
|
| 3820 |
+
"epoch": 1.91,
|
| 3821 |
+
"learning_rate": 0.00018668599033816425,
|
| 3822 |
+
"loss": 1.0187,
|
| 3823 |
+
"step": 5970
|
| 3824 |
+
},
|
| 3825 |
+
{
|
| 3826 |
+
"epoch": 1.91,
|
| 3827 |
+
"learning_rate": 0.0001864927536231884,
|
| 3828 |
+
"loss": 1.0205,
|
| 3829 |
+
"step": 5980
|
| 3830 |
+
},
|
| 3831 |
+
{
|
| 3832 |
+
"epoch": 1.92,
|
| 3833 |
+
"learning_rate": 0.00018629951690821254,
|
| 3834 |
+
"loss": 1.0267,
|
| 3835 |
+
"step": 5990
|
| 3836 |
+
},
|
| 3837 |
+
{
|
| 3838 |
+
"epoch": 1.92,
|
| 3839 |
+
"learning_rate": 0.00018610628019323668,
|
| 3840 |
+
"loss": 1.0381,
|
| 3841 |
+
"step": 6000
|
| 3842 |
+
},
|
| 3843 |
+
{
|
| 3844 |
+
"epoch": 1.92,
|
| 3845 |
+
"eval_loss": 1.0564384460449219,
|
| 3846 |
+
"eval_runtime": 105.9283,
|
| 3847 |
+
"eval_samples_per_second": 18.881,
|
| 3848 |
+
"eval_steps_per_second": 2.36,
|
| 3849 |
+
"step": 6000
|
| 3850 |
+
}
|
| 3851 |
+
],
|
| 3852 |
+
"logging_steps": 10,
|
| 3853 |
+
"max_steps": 15625,
|
| 3854 |
+
"num_train_epochs": 5,
|
| 3855 |
+
"save_steps": 200,
|
| 3856 |
+
"total_flos": 6.618622281779773e+17,
|
| 3857 |
+
"trial_name": null,
|
| 3858 |
+
"trial_params": null
|
| 3859 |
+
}
|
alpaca_text_52K/best_model/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a09cbbfbe6d6c002252235b4476abb50ae8f0d55947cd46a53991e96578a822e
|
| 3 |
+
size 4027
|