Training in progress, epoch 2, checkpoint
Browse files
last-checkpoint/model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1575259780
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e8ae5ab01ccbb566cd3dd8ee6dc1252a85b1c7f271331643ea03050b0e8ec9d3
|
| 3 |
size 1575259780
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 3150397656
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:acc7dc7d28456c78f8d6080cf32977d4aebc2e514871c0bf819e045e28972103
|
| 3 |
size 3150397656
|
last-checkpoint/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14244
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1e8361756a3864b69a416a28667f28ee22be0ff5d84048a54680e7de8b87f656
|
| 3 |
size 14244
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1064
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:dede00403a11b2b14e170429444fee20dcef7db14bc3c7a358467f448966c579
|
| 3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric": 0.
|
| 3 |
-
"best_model_checkpoint": "/kaggle/tmp/amr-tst-indo/AMRBART-id/fine-tune/../outputs/mbart-en-id-smaller-fted/checkpoint-
|
| 4 |
-
"epoch": 0
|
| 5 |
"eval_steps": 500,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -1181,6 +1181,1174 @@
|
|
| 1181 |
"eval_samples_per_second": 0.864,
|
| 1182 |
"eval_steps_per_second": 0.432,
|
| 1183 |
"step": 3869
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1184 |
}
|
| 1185 |
],
|
| 1186 |
"logging_steps": 20,
|
|
@@ -1200,7 +2368,7 @@
|
|
| 1200 |
"attributes": {}
|
| 1201 |
}
|
| 1202 |
},
|
| 1203 |
-
"total_flos":
|
| 1204 |
"train_batch_size": 2,
|
| 1205 |
"trial_name": null,
|
| 1206 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 0.0747,
|
| 3 |
+
"best_model_checkpoint": "/kaggle/tmp/amr-tst-indo/AMRBART-id/fine-tune/../outputs/mbart-en-id-smaller-fted/checkpoint-7739",
|
| 4 |
+
"epoch": 2.0,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 7739,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 1181 |
"eval_samples_per_second": 0.864,
|
| 1182 |
"eval_steps_per_second": 0.432,
|
| 1183 |
"step": 3869
|
| 1184 |
+
},
|
| 1185 |
+
{
|
| 1186 |
+
"epoch": 1.0027135288797002,
|
| 1187 |
+
"learning_rate": 1.9403604304421105e-07,
|
| 1188 |
+
"loss": 3.0219,
|
| 1189 |
+
"step": 3880
|
| 1190 |
+
},
|
| 1191 |
+
{
|
| 1192 |
+
"epoch": 1.0078821553172244,
|
| 1193 |
+
"learning_rate": 1.9400363023466873e-07,
|
| 1194 |
+
"loss": 3.0378,
|
| 1195 |
+
"step": 3900
|
| 1196 |
+
},
|
| 1197 |
+
{
|
| 1198 |
+
"epoch": 1.0130507817547487,
|
| 1199 |
+
"learning_rate": 1.939712174251264e-07,
|
| 1200 |
+
"loss": 3.058,
|
| 1201 |
+
"step": 3920
|
| 1202 |
+
},
|
| 1203 |
+
{
|
| 1204 |
+
"epoch": 1.0182194081922729,
|
| 1205 |
+
"learning_rate": 1.9393880461558406e-07,
|
| 1206 |
+
"loss": 3.0479,
|
| 1207 |
+
"step": 3940
|
| 1208 |
+
},
|
| 1209 |
+
{
|
| 1210 |
+
"epoch": 1.0233880346297972,
|
| 1211 |
+
"learning_rate": 1.9390639180604175e-07,
|
| 1212 |
+
"loss": 3.0733,
|
| 1213 |
+
"step": 3960
|
| 1214 |
+
},
|
| 1215 |
+
{
|
| 1216 |
+
"epoch": 1.0285566610673214,
|
| 1217 |
+
"learning_rate": 1.938739789964994e-07,
|
| 1218 |
+
"loss": 3.0543,
|
| 1219 |
+
"step": 3980
|
| 1220 |
+
},
|
| 1221 |
+
{
|
| 1222 |
+
"epoch": 1.0337252875048455,
|
| 1223 |
+
"learning_rate": 1.9384156618695707e-07,
|
| 1224 |
+
"loss": 2.9607,
|
| 1225 |
+
"step": 4000
|
| 1226 |
+
},
|
| 1227 |
+
{
|
| 1228 |
+
"epoch": 1.0388939139423699,
|
| 1229 |
+
"learning_rate": 1.9380915337741476e-07,
|
| 1230 |
+
"loss": 2.9848,
|
| 1231 |
+
"step": 4020
|
| 1232 |
+
},
|
| 1233 |
+
{
|
| 1234 |
+
"epoch": 1.044062540379894,
|
| 1235 |
+
"learning_rate": 1.937767405678724e-07,
|
| 1236 |
+
"loss": 3.0765,
|
| 1237 |
+
"step": 4040
|
| 1238 |
+
},
|
| 1239 |
+
{
|
| 1240 |
+
"epoch": 1.0492311668174183,
|
| 1241 |
+
"learning_rate": 1.9374432775833008e-07,
|
| 1242 |
+
"loss": 2.9827,
|
| 1243 |
+
"step": 4060
|
| 1244 |
+
},
|
| 1245 |
+
{
|
| 1246 |
+
"epoch": 1.0543997932549425,
|
| 1247 |
+
"learning_rate": 1.9371191494878777e-07,
|
| 1248 |
+
"loss": 3.0336,
|
| 1249 |
+
"step": 4080
|
| 1250 |
+
},
|
| 1251 |
+
{
|
| 1252 |
+
"epoch": 1.0595684196924666,
|
| 1253 |
+
"learning_rate": 1.936795021392454e-07,
|
| 1254 |
+
"loss": 3.0382,
|
| 1255 |
+
"step": 4100
|
| 1256 |
+
},
|
| 1257 |
+
{
|
| 1258 |
+
"epoch": 1.064737046129991,
|
| 1259 |
+
"learning_rate": 1.936470893297031e-07,
|
| 1260 |
+
"loss": 2.9706,
|
| 1261 |
+
"step": 4120
|
| 1262 |
+
},
|
| 1263 |
+
{
|
| 1264 |
+
"epoch": 1.0699056725675151,
|
| 1265 |
+
"learning_rate": 1.9361467652016076e-07,
|
| 1266 |
+
"loss": 2.9261,
|
| 1267 |
+
"step": 4140
|
| 1268 |
+
},
|
| 1269 |
+
{
|
| 1270 |
+
"epoch": 1.0750742990050395,
|
| 1271 |
+
"learning_rate": 1.9358226371061842e-07,
|
| 1272 |
+
"loss": 3.0358,
|
| 1273 |
+
"step": 4160
|
| 1274 |
+
},
|
| 1275 |
+
{
|
| 1276 |
+
"epoch": 1.0802429254425636,
|
| 1277 |
+
"learning_rate": 1.935498509010761e-07,
|
| 1278 |
+
"loss": 3.0442,
|
| 1279 |
+
"step": 4180
|
| 1280 |
+
},
|
| 1281 |
+
{
|
| 1282 |
+
"epoch": 1.085411551880088,
|
| 1283 |
+
"learning_rate": 1.9351743809153377e-07,
|
| 1284 |
+
"loss": 2.9493,
|
| 1285 |
+
"step": 4200
|
| 1286 |
+
},
|
| 1287 |
+
{
|
| 1288 |
+
"epoch": 1.0905801783176121,
|
| 1289 |
+
"learning_rate": 1.9348502528199143e-07,
|
| 1290 |
+
"loss": 2.9963,
|
| 1291 |
+
"step": 4220
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 1.0957488047551363,
|
| 1295 |
+
"learning_rate": 1.9345261247244912e-07,
|
| 1296 |
+
"loss": 3.0102,
|
| 1297 |
+
"step": 4240
|
| 1298 |
+
},
|
| 1299 |
+
{
|
| 1300 |
+
"epoch": 1.1009174311926606,
|
| 1301 |
+
"learning_rate": 1.9342019966290675e-07,
|
| 1302 |
+
"loss": 2.9861,
|
| 1303 |
+
"step": 4260
|
| 1304 |
+
},
|
| 1305 |
+
{
|
| 1306 |
+
"epoch": 1.1060860576301847,
|
| 1307 |
+
"learning_rate": 1.9338778685336444e-07,
|
| 1308 |
+
"loss": 3.0209,
|
| 1309 |
+
"step": 4280
|
| 1310 |
+
},
|
| 1311 |
+
{
|
| 1312 |
+
"epoch": 1.111254684067709,
|
| 1313 |
+
"learning_rate": 1.933553740438221e-07,
|
| 1314 |
+
"loss": 2.9864,
|
| 1315 |
+
"step": 4300
|
| 1316 |
+
},
|
| 1317 |
+
{
|
| 1318 |
+
"epoch": 1.1164233105052332,
|
| 1319 |
+
"learning_rate": 1.9332296123427977e-07,
|
| 1320 |
+
"loss": 2.9766,
|
| 1321 |
+
"step": 4320
|
| 1322 |
+
},
|
| 1323 |
+
{
|
| 1324 |
+
"epoch": 1.1215919369427574,
|
| 1325 |
+
"learning_rate": 1.9329054842473745e-07,
|
| 1326 |
+
"loss": 3.0403,
|
| 1327 |
+
"step": 4340
|
| 1328 |
+
},
|
| 1329 |
+
{
|
| 1330 |
+
"epoch": 1.1267605633802817,
|
| 1331 |
+
"learning_rate": 1.9325813561519512e-07,
|
| 1332 |
+
"loss": 2.9861,
|
| 1333 |
+
"step": 4360
|
| 1334 |
+
},
|
| 1335 |
+
{
|
| 1336 |
+
"epoch": 1.1319291898178059,
|
| 1337 |
+
"learning_rate": 1.9322572280565278e-07,
|
| 1338 |
+
"loss": 3.0244,
|
| 1339 |
+
"step": 4380
|
| 1340 |
+
},
|
| 1341 |
+
{
|
| 1342 |
+
"epoch": 1.1370978162553302,
|
| 1343 |
+
"learning_rate": 1.9319330999611047e-07,
|
| 1344 |
+
"loss": 2.9827,
|
| 1345 |
+
"step": 4400
|
| 1346 |
+
},
|
| 1347 |
+
{
|
| 1348 |
+
"epoch": 1.1422664426928544,
|
| 1349 |
+
"learning_rate": 1.931608971865681e-07,
|
| 1350 |
+
"loss": 3.0343,
|
| 1351 |
+
"step": 4420
|
| 1352 |
+
},
|
| 1353 |
+
{
|
| 1354 |
+
"epoch": 1.1474350691303785,
|
| 1355 |
+
"learning_rate": 1.931284843770258e-07,
|
| 1356 |
+
"loss": 2.9305,
|
| 1357 |
+
"step": 4440
|
| 1358 |
+
},
|
| 1359 |
+
{
|
| 1360 |
+
"epoch": 1.1526036955679029,
|
| 1361 |
+
"learning_rate": 1.9309607156748348e-07,
|
| 1362 |
+
"loss": 2.9797,
|
| 1363 |
+
"step": 4460
|
| 1364 |
+
},
|
| 1365 |
+
{
|
| 1366 |
+
"epoch": 1.157772322005427,
|
| 1367 |
+
"learning_rate": 1.930636587579411e-07,
|
| 1368 |
+
"loss": 2.9846,
|
| 1369 |
+
"step": 4480
|
| 1370 |
+
},
|
| 1371 |
+
{
|
| 1372 |
+
"epoch": 1.1629409484429514,
|
| 1373 |
+
"learning_rate": 1.930312459483988e-07,
|
| 1374 |
+
"loss": 3.0176,
|
| 1375 |
+
"step": 4500
|
| 1376 |
+
},
|
| 1377 |
+
{
|
| 1378 |
+
"epoch": 1.1681095748804755,
|
| 1379 |
+
"learning_rate": 1.9299883313885646e-07,
|
| 1380 |
+
"loss": 2.9897,
|
| 1381 |
+
"step": 4520
|
| 1382 |
+
},
|
| 1383 |
+
{
|
| 1384 |
+
"epoch": 1.1732782013179996,
|
| 1385 |
+
"learning_rate": 1.9296642032931413e-07,
|
| 1386 |
+
"loss": 3.0404,
|
| 1387 |
+
"step": 4540
|
| 1388 |
+
},
|
| 1389 |
+
{
|
| 1390 |
+
"epoch": 1.178446827755524,
|
| 1391 |
+
"learning_rate": 1.9293400751977181e-07,
|
| 1392 |
+
"loss": 2.9724,
|
| 1393 |
+
"step": 4560
|
| 1394 |
+
},
|
| 1395 |
+
{
|
| 1396 |
+
"epoch": 1.1836154541930481,
|
| 1397 |
+
"learning_rate": 1.9290159471022948e-07,
|
| 1398 |
+
"loss": 3.019,
|
| 1399 |
+
"step": 4580
|
| 1400 |
+
},
|
| 1401 |
+
{
|
| 1402 |
+
"epoch": 1.1887840806305725,
|
| 1403 |
+
"learning_rate": 1.9286918190068714e-07,
|
| 1404 |
+
"loss": 2.9637,
|
| 1405 |
+
"step": 4600
|
| 1406 |
+
},
|
| 1407 |
+
{
|
| 1408 |
+
"epoch": 1.1939527070680966,
|
| 1409 |
+
"learning_rate": 1.9283676909114483e-07,
|
| 1410 |
+
"loss": 3.0281,
|
| 1411 |
+
"step": 4620
|
| 1412 |
+
},
|
| 1413 |
+
{
|
| 1414 |
+
"epoch": 1.1991213335056208,
|
| 1415 |
+
"learning_rate": 1.9280435628160246e-07,
|
| 1416 |
+
"loss": 2.9964,
|
| 1417 |
+
"step": 4640
|
| 1418 |
+
},
|
| 1419 |
+
{
|
| 1420 |
+
"epoch": 1.2042899599431451,
|
| 1421 |
+
"learning_rate": 1.9277194347206015e-07,
|
| 1422 |
+
"loss": 3.0762,
|
| 1423 |
+
"step": 4660
|
| 1424 |
+
},
|
| 1425 |
+
{
|
| 1426 |
+
"epoch": 1.2094585863806693,
|
| 1427 |
+
"learning_rate": 1.9273953066251784e-07,
|
| 1428 |
+
"loss": 3.035,
|
| 1429 |
+
"step": 4680
|
| 1430 |
+
},
|
| 1431 |
+
{
|
| 1432 |
+
"epoch": 1.2146272128181936,
|
| 1433 |
+
"learning_rate": 1.9270711785297547e-07,
|
| 1434 |
+
"loss": 2.9578,
|
| 1435 |
+
"step": 4700
|
| 1436 |
+
},
|
| 1437 |
+
{
|
| 1438 |
+
"epoch": 1.2197958392557178,
|
| 1439 |
+
"learning_rate": 1.9267470504343316e-07,
|
| 1440 |
+
"loss": 2.9568,
|
| 1441 |
+
"step": 4720
|
| 1442 |
+
},
|
| 1443 |
+
{
|
| 1444 |
+
"epoch": 1.2249644656932421,
|
| 1445 |
+
"learning_rate": 1.9264229223389082e-07,
|
| 1446 |
+
"loss": 3.0495,
|
| 1447 |
+
"step": 4740
|
| 1448 |
+
},
|
| 1449 |
+
{
|
| 1450 |
+
"epoch": 1.2301330921307663,
|
| 1451 |
+
"learning_rate": 1.9260987942434848e-07,
|
| 1452 |
+
"loss": 3.0054,
|
| 1453 |
+
"step": 4760
|
| 1454 |
+
},
|
| 1455 |
+
{
|
| 1456 |
+
"epoch": 1.2353017185682904,
|
| 1457 |
+
"learning_rate": 1.9257746661480617e-07,
|
| 1458 |
+
"loss": 3.0101,
|
| 1459 |
+
"step": 4780
|
| 1460 |
+
},
|
| 1461 |
+
{
|
| 1462 |
+
"epoch": 1.2404703450058148,
|
| 1463 |
+
"learning_rate": 1.9254505380526384e-07,
|
| 1464 |
+
"loss": 2.9621,
|
| 1465 |
+
"step": 4800
|
| 1466 |
+
},
|
| 1467 |
+
{
|
| 1468 |
+
"epoch": 1.245638971443339,
|
| 1469 |
+
"learning_rate": 1.925126409957215e-07,
|
| 1470 |
+
"loss": 3.0161,
|
| 1471 |
+
"step": 4820
|
| 1472 |
+
},
|
| 1473 |
+
{
|
| 1474 |
+
"epoch": 1.250807597880863,
|
| 1475 |
+
"learning_rate": 1.9248022818617919e-07,
|
| 1476 |
+
"loss": 2.9672,
|
| 1477 |
+
"step": 4840
|
| 1478 |
+
},
|
| 1479 |
+
{
|
| 1480 |
+
"epoch": 1.2559762243183874,
|
| 1481 |
+
"learning_rate": 1.9244781537663682e-07,
|
| 1482 |
+
"loss": 2.9749,
|
| 1483 |
+
"step": 4860
|
| 1484 |
+
},
|
| 1485 |
+
{
|
| 1486 |
+
"epoch": 1.2611448507559115,
|
| 1487 |
+
"learning_rate": 1.924154025670945e-07,
|
| 1488 |
+
"loss": 2.9898,
|
| 1489 |
+
"step": 4880
|
| 1490 |
+
},
|
| 1491 |
+
{
|
| 1492 |
+
"epoch": 1.266313477193436,
|
| 1493 |
+
"learning_rate": 1.9238298975755217e-07,
|
| 1494 |
+
"loss": 2.9917,
|
| 1495 |
+
"step": 4900
|
| 1496 |
+
},
|
| 1497 |
+
{
|
| 1498 |
+
"epoch": 1.27148210363096,
|
| 1499 |
+
"learning_rate": 1.9235057694800983e-07,
|
| 1500 |
+
"loss": 2.9675,
|
| 1501 |
+
"step": 4920
|
| 1502 |
+
},
|
| 1503 |
+
{
|
| 1504 |
+
"epoch": 1.2766507300684844,
|
| 1505 |
+
"learning_rate": 1.9231816413846752e-07,
|
| 1506 |
+
"loss": 2.9495,
|
| 1507 |
+
"step": 4940
|
| 1508 |
+
},
|
| 1509 |
+
{
|
| 1510 |
+
"epoch": 1.2818193565060085,
|
| 1511 |
+
"learning_rate": 1.9228575132892518e-07,
|
| 1512 |
+
"loss": 3.0443,
|
| 1513 |
+
"step": 4960
|
| 1514 |
+
},
|
| 1515 |
+
{
|
| 1516 |
+
"epoch": 1.2869879829435327,
|
| 1517 |
+
"learning_rate": 1.9225333851938284e-07,
|
| 1518 |
+
"loss": 2.9756,
|
| 1519 |
+
"step": 4980
|
| 1520 |
+
},
|
| 1521 |
+
{
|
| 1522 |
+
"epoch": 1.292156609381057,
|
| 1523 |
+
"learning_rate": 1.9222092570984053e-07,
|
| 1524 |
+
"loss": 2.9824,
|
| 1525 |
+
"step": 5000
|
| 1526 |
+
},
|
| 1527 |
+
{
|
| 1528 |
+
"epoch": 1.2973252358185812,
|
| 1529 |
+
"learning_rate": 1.9218851290029817e-07,
|
| 1530 |
+
"loss": 3.0607,
|
| 1531 |
+
"step": 5020
|
| 1532 |
+
},
|
| 1533 |
+
{
|
| 1534 |
+
"epoch": 1.3024938622561053,
|
| 1535 |
+
"learning_rate": 1.9215610009075586e-07,
|
| 1536 |
+
"loss": 3.0266,
|
| 1537 |
+
"step": 5040
|
| 1538 |
+
},
|
| 1539 |
+
{
|
| 1540 |
+
"epoch": 1.3076624886936297,
|
| 1541 |
+
"learning_rate": 1.9212368728121355e-07,
|
| 1542 |
+
"loss": 2.9212,
|
| 1543 |
+
"step": 5060
|
| 1544 |
+
},
|
| 1545 |
+
{
|
| 1546 |
+
"epoch": 1.312831115131154,
|
| 1547 |
+
"learning_rate": 1.9209127447167118e-07,
|
| 1548 |
+
"loss": 3.0226,
|
| 1549 |
+
"step": 5080
|
| 1550 |
+
},
|
| 1551 |
+
{
|
| 1552 |
+
"epoch": 1.3179997415686782,
|
| 1553 |
+
"learning_rate": 1.9205886166212887e-07,
|
| 1554 |
+
"loss": 2.9282,
|
| 1555 |
+
"step": 5100
|
| 1556 |
+
},
|
| 1557 |
+
{
|
| 1558 |
+
"epoch": 1.3231683680062023,
|
| 1559 |
+
"learning_rate": 1.9202644885258653e-07,
|
| 1560 |
+
"loss": 2.9725,
|
| 1561 |
+
"step": 5120
|
| 1562 |
+
},
|
| 1563 |
+
{
|
| 1564 |
+
"epoch": 1.3283369944437267,
|
| 1565 |
+
"learning_rate": 1.919940360430442e-07,
|
| 1566 |
+
"loss": 2.9777,
|
| 1567 |
+
"step": 5140
|
| 1568 |
+
},
|
| 1569 |
+
{
|
| 1570 |
+
"epoch": 1.3335056208812508,
|
| 1571 |
+
"learning_rate": 1.9196162323350188e-07,
|
| 1572 |
+
"loss": 2.9426,
|
| 1573 |
+
"step": 5160
|
| 1574 |
+
},
|
| 1575 |
+
{
|
| 1576 |
+
"epoch": 1.338674247318775,
|
| 1577 |
+
"learning_rate": 1.9192921042395954e-07,
|
| 1578 |
+
"loss": 2.9584,
|
| 1579 |
+
"step": 5180
|
| 1580 |
+
},
|
| 1581 |
+
{
|
| 1582 |
+
"epoch": 1.3438428737562993,
|
| 1583 |
+
"learning_rate": 1.918967976144172e-07,
|
| 1584 |
+
"loss": 2.9244,
|
| 1585 |
+
"step": 5200
|
| 1586 |
+
},
|
| 1587 |
+
{
|
| 1588 |
+
"epoch": 1.3490115001938234,
|
| 1589 |
+
"learning_rate": 1.918643848048749e-07,
|
| 1590 |
+
"loss": 3.0344,
|
| 1591 |
+
"step": 5220
|
| 1592 |
+
},
|
| 1593 |
+
{
|
| 1594 |
+
"epoch": 1.3541801266313478,
|
| 1595 |
+
"learning_rate": 1.9183197199533253e-07,
|
| 1596 |
+
"loss": 2.9877,
|
| 1597 |
+
"step": 5240
|
| 1598 |
+
},
|
| 1599 |
+
{
|
| 1600 |
+
"epoch": 1.359348753068872,
|
| 1601 |
+
"learning_rate": 1.9179955918579022e-07,
|
| 1602 |
+
"loss": 2.9763,
|
| 1603 |
+
"step": 5260
|
| 1604 |
+
},
|
| 1605 |
+
{
|
| 1606 |
+
"epoch": 1.3645173795063963,
|
| 1607 |
+
"learning_rate": 1.917671463762479e-07,
|
| 1608 |
+
"loss": 2.9684,
|
| 1609 |
+
"step": 5280
|
| 1610 |
+
},
|
| 1611 |
+
{
|
| 1612 |
+
"epoch": 1.3696860059439204,
|
| 1613 |
+
"learning_rate": 1.9173473356670554e-07,
|
| 1614 |
+
"loss": 3.0331,
|
| 1615 |
+
"step": 5300
|
| 1616 |
+
},
|
| 1617 |
+
{
|
| 1618 |
+
"epoch": 1.3748546323814446,
|
| 1619 |
+
"learning_rate": 1.9170232075716323e-07,
|
| 1620 |
+
"loss": 3.0052,
|
| 1621 |
+
"step": 5320
|
| 1622 |
+
},
|
| 1623 |
+
{
|
| 1624 |
+
"epoch": 1.380023258818969,
|
| 1625 |
+
"learning_rate": 1.916699079476209e-07,
|
| 1626 |
+
"loss": 2.9511,
|
| 1627 |
+
"step": 5340
|
| 1628 |
+
},
|
| 1629 |
+
{
|
| 1630 |
+
"epoch": 1.385191885256493,
|
| 1631 |
+
"learning_rate": 1.9163749513807855e-07,
|
| 1632 |
+
"loss": 2.9444,
|
| 1633 |
+
"step": 5360
|
| 1634 |
+
},
|
| 1635 |
+
{
|
| 1636 |
+
"epoch": 1.3903605116940172,
|
| 1637 |
+
"learning_rate": 1.9160508232853624e-07,
|
| 1638 |
+
"loss": 2.9757,
|
| 1639 |
+
"step": 5380
|
| 1640 |
+
},
|
| 1641 |
+
{
|
| 1642 |
+
"epoch": 1.3955291381315416,
|
| 1643 |
+
"learning_rate": 1.915726695189939e-07,
|
| 1644 |
+
"loss": 3.0247,
|
| 1645 |
+
"step": 5400
|
| 1646 |
+
},
|
| 1647 |
+
{
|
| 1648 |
+
"epoch": 1.4006977645690657,
|
| 1649 |
+
"learning_rate": 1.9154025670945156e-07,
|
| 1650 |
+
"loss": 3.0173,
|
| 1651 |
+
"step": 5420
|
| 1652 |
+
},
|
| 1653 |
+
{
|
| 1654 |
+
"epoch": 1.40586639100659,
|
| 1655 |
+
"learning_rate": 1.9150784389990925e-07,
|
| 1656 |
+
"loss": 2.9848,
|
| 1657 |
+
"step": 5440
|
| 1658 |
+
},
|
| 1659 |
+
{
|
| 1660 |
+
"epoch": 1.4110350174441142,
|
| 1661 |
+
"learning_rate": 1.914754310903669e-07,
|
| 1662 |
+
"loss": 2.9544,
|
| 1663 |
+
"step": 5460
|
| 1664 |
+
},
|
| 1665 |
+
{
|
| 1666 |
+
"epoch": 1.4162036438816386,
|
| 1667 |
+
"learning_rate": 1.9144301828082458e-07,
|
| 1668 |
+
"loss": 2.9586,
|
| 1669 |
+
"step": 5480
|
| 1670 |
+
},
|
| 1671 |
+
{
|
| 1672 |
+
"epoch": 1.4213722703191627,
|
| 1673 |
+
"learning_rate": 1.9141060547128224e-07,
|
| 1674 |
+
"loss": 2.9663,
|
| 1675 |
+
"step": 5500
|
| 1676 |
+
},
|
| 1677 |
+
{
|
| 1678 |
+
"epoch": 1.4265408967566868,
|
| 1679 |
+
"learning_rate": 1.913781926617399e-07,
|
| 1680 |
+
"loss": 2.9736,
|
| 1681 |
+
"step": 5520
|
| 1682 |
+
},
|
| 1683 |
+
{
|
| 1684 |
+
"epoch": 1.4317095231942112,
|
| 1685 |
+
"learning_rate": 1.913457798521976e-07,
|
| 1686 |
+
"loss": 2.9392,
|
| 1687 |
+
"step": 5540
|
| 1688 |
+
},
|
| 1689 |
+
{
|
| 1690 |
+
"epoch": 1.4368781496317353,
|
| 1691 |
+
"learning_rate": 1.9131336704265525e-07,
|
| 1692 |
+
"loss": 2.9724,
|
| 1693 |
+
"step": 5560
|
| 1694 |
+
},
|
| 1695 |
+
{
|
| 1696 |
+
"epoch": 1.4420467760692595,
|
| 1697 |
+
"learning_rate": 1.912809542331129e-07,
|
| 1698 |
+
"loss": 2.9819,
|
| 1699 |
+
"step": 5580
|
| 1700 |
+
},
|
| 1701 |
+
{
|
| 1702 |
+
"epoch": 1.4472154025067838,
|
| 1703 |
+
"learning_rate": 1.912485414235706e-07,
|
| 1704 |
+
"loss": 2.8861,
|
| 1705 |
+
"step": 5600
|
| 1706 |
+
},
|
| 1707 |
+
{
|
| 1708 |
+
"epoch": 1.4523840289443082,
|
| 1709 |
+
"learning_rate": 1.9121612861402824e-07,
|
| 1710 |
+
"loss": 2.9408,
|
| 1711 |
+
"step": 5620
|
| 1712 |
+
},
|
| 1713 |
+
{
|
| 1714 |
+
"epoch": 1.4575526553818323,
|
| 1715 |
+
"learning_rate": 1.9118371580448592e-07,
|
| 1716 |
+
"loss": 3.0236,
|
| 1717 |
+
"step": 5640
|
| 1718 |
+
},
|
| 1719 |
+
{
|
| 1720 |
+
"epoch": 1.4627212818193565,
|
| 1721 |
+
"learning_rate": 1.911513029949436e-07,
|
| 1722 |
+
"loss": 2.9481,
|
| 1723 |
+
"step": 5660
|
| 1724 |
+
},
|
| 1725 |
+
{
|
| 1726 |
+
"epoch": 1.4678899082568808,
|
| 1727 |
+
"learning_rate": 1.9111889018540125e-07,
|
| 1728 |
+
"loss": 2.9892,
|
| 1729 |
+
"step": 5680
|
| 1730 |
+
},
|
| 1731 |
+
{
|
| 1732 |
+
"epoch": 1.473058534694405,
|
| 1733 |
+
"learning_rate": 1.9108647737585894e-07,
|
| 1734 |
+
"loss": 2.9723,
|
| 1735 |
+
"step": 5700
|
| 1736 |
+
},
|
| 1737 |
+
{
|
| 1738 |
+
"epoch": 1.478227161131929,
|
| 1739 |
+
"learning_rate": 1.910540645663166e-07,
|
| 1740 |
+
"loss": 2.8791,
|
| 1741 |
+
"step": 5720
|
| 1742 |
+
},
|
| 1743 |
+
{
|
| 1744 |
+
"epoch": 1.4833957875694535,
|
| 1745 |
+
"learning_rate": 1.9102165175677426e-07,
|
| 1746 |
+
"loss": 2.9381,
|
| 1747 |
+
"step": 5740
|
| 1748 |
+
},
|
| 1749 |
+
{
|
| 1750 |
+
"epoch": 1.4885644140069776,
|
| 1751 |
+
"learning_rate": 1.9098923894723195e-07,
|
| 1752 |
+
"loss": 2.8873,
|
| 1753 |
+
"step": 5760
|
| 1754 |
+
},
|
| 1755 |
+
{
|
| 1756 |
+
"epoch": 1.493733040444502,
|
| 1757 |
+
"learning_rate": 1.909568261376896e-07,
|
| 1758 |
+
"loss": 2.8586,
|
| 1759 |
+
"step": 5780
|
| 1760 |
+
},
|
| 1761 |
+
{
|
| 1762 |
+
"epoch": 1.498901666882026,
|
| 1763 |
+
"learning_rate": 1.9092441332814727e-07,
|
| 1764 |
+
"loss": 2.9308,
|
| 1765 |
+
"step": 5800
|
| 1766 |
+
},
|
| 1767 |
+
{
|
| 1768 |
+
"epoch": 1.5040702933195504,
|
| 1769 |
+
"learning_rate": 1.9089200051860496e-07,
|
| 1770 |
+
"loss": 2.9222,
|
| 1771 |
+
"step": 5820
|
| 1772 |
+
},
|
| 1773 |
+
{
|
| 1774 |
+
"epoch": 1.5092389197570746,
|
| 1775 |
+
"learning_rate": 1.908595877090626e-07,
|
| 1776 |
+
"loss": 2.9787,
|
| 1777 |
+
"step": 5840
|
| 1778 |
+
},
|
| 1779 |
+
{
|
| 1780 |
+
"epoch": 1.5144075461945987,
|
| 1781 |
+
"learning_rate": 1.9082717489952028e-07,
|
| 1782 |
+
"loss": 2.9207,
|
| 1783 |
+
"step": 5860
|
| 1784 |
+
},
|
| 1785 |
+
{
|
| 1786 |
+
"epoch": 1.519576172632123,
|
| 1787 |
+
"learning_rate": 1.9079476208997797e-07,
|
| 1788 |
+
"loss": 2.9304,
|
| 1789 |
+
"step": 5880
|
| 1790 |
+
},
|
| 1791 |
+
{
|
| 1792 |
+
"epoch": 1.5247447990696472,
|
| 1793 |
+
"learning_rate": 1.907623492804356e-07,
|
| 1794 |
+
"loss": 2.942,
|
| 1795 |
+
"step": 5900
|
| 1796 |
+
},
|
| 1797 |
+
{
|
| 1798 |
+
"epoch": 1.5299134255071714,
|
| 1799 |
+
"learning_rate": 1.907299364708933e-07,
|
| 1800 |
+
"loss": 2.9672,
|
| 1801 |
+
"step": 5920
|
| 1802 |
+
},
|
| 1803 |
+
{
|
| 1804 |
+
"epoch": 1.5350820519446957,
|
| 1805 |
+
"learning_rate": 1.9069752366135096e-07,
|
| 1806 |
+
"loss": 2.9461,
|
| 1807 |
+
"step": 5940
|
| 1808 |
+
},
|
| 1809 |
+
{
|
| 1810 |
+
"epoch": 1.54025067838222,
|
| 1811 |
+
"learning_rate": 1.9066511085180862e-07,
|
| 1812 |
+
"loss": 2.9206,
|
| 1813 |
+
"step": 5960
|
| 1814 |
+
},
|
| 1815 |
+
{
|
| 1816 |
+
"epoch": 1.545419304819744,
|
| 1817 |
+
"learning_rate": 1.906326980422663e-07,
|
| 1818 |
+
"loss": 3.0094,
|
| 1819 |
+
"step": 5980
|
| 1820 |
+
},
|
| 1821 |
+
{
|
| 1822 |
+
"epoch": 1.5505879312572683,
|
| 1823 |
+
"learning_rate": 1.9060028523272397e-07,
|
| 1824 |
+
"loss": 2.9333,
|
| 1825 |
+
"step": 6000
|
| 1826 |
+
},
|
| 1827 |
+
{
|
| 1828 |
+
"epoch": 1.5557565576947927,
|
| 1829 |
+
"learning_rate": 1.9056787242318163e-07,
|
| 1830 |
+
"loss": 2.8769,
|
| 1831 |
+
"step": 6020
|
| 1832 |
+
},
|
| 1833 |
+
{
|
| 1834 |
+
"epoch": 1.5609251841323168,
|
| 1835 |
+
"learning_rate": 1.9053545961363932e-07,
|
| 1836 |
+
"loss": 2.8884,
|
| 1837 |
+
"step": 6040
|
| 1838 |
+
},
|
| 1839 |
+
{
|
| 1840 |
+
"epoch": 1.566093810569841,
|
| 1841 |
+
"learning_rate": 1.9050304680409696e-07,
|
| 1842 |
+
"loss": 2.9742,
|
| 1843 |
+
"step": 6060
|
| 1844 |
+
},
|
| 1845 |
+
{
|
| 1846 |
+
"epoch": 1.5712624370073653,
|
| 1847 |
+
"learning_rate": 1.9047063399455464e-07,
|
| 1848 |
+
"loss": 3.0092,
|
| 1849 |
+
"step": 6080
|
| 1850 |
+
},
|
| 1851 |
+
{
|
| 1852 |
+
"epoch": 1.5764310634448895,
|
| 1853 |
+
"learning_rate": 1.904382211850123e-07,
|
| 1854 |
+
"loss": 2.9553,
|
| 1855 |
+
"step": 6100
|
| 1856 |
+
},
|
| 1857 |
+
{
|
| 1858 |
+
"epoch": 1.5815996898824136,
|
| 1859 |
+
"learning_rate": 1.9040580837546997e-07,
|
| 1860 |
+
"loss": 2.8705,
|
| 1861 |
+
"step": 6120
|
| 1862 |
+
},
|
| 1863 |
+
{
|
| 1864 |
+
"epoch": 1.586768316319938,
|
| 1865 |
+
"learning_rate": 1.9037339556592766e-07,
|
| 1866 |
+
"loss": 2.9877,
|
| 1867 |
+
"step": 6140
|
| 1868 |
+
},
|
| 1869 |
+
{
|
| 1870 |
+
"epoch": 1.5919369427574623,
|
| 1871 |
+
"learning_rate": 1.9034098275638532e-07,
|
| 1872 |
+
"loss": 2.966,
|
| 1873 |
+
"step": 6160
|
| 1874 |
+
},
|
| 1875 |
+
{
|
| 1876 |
+
"epoch": 1.5971055691949865,
|
| 1877 |
+
"learning_rate": 1.9030856994684298e-07,
|
| 1878 |
+
"loss": 2.9414,
|
| 1879 |
+
"step": 6180
|
| 1880 |
+
},
|
| 1881 |
+
{
|
| 1882 |
+
"epoch": 1.6022741956325106,
|
| 1883 |
+
"learning_rate": 1.9027615713730067e-07,
|
| 1884 |
+
"loss": 2.9618,
|
| 1885 |
+
"step": 6200
|
| 1886 |
+
},
|
| 1887 |
+
{
|
| 1888 |
+
"epoch": 1.607442822070035,
|
| 1889 |
+
"learning_rate": 1.902437443277583e-07,
|
| 1890 |
+
"loss": 2.9328,
|
| 1891 |
+
"step": 6220
|
| 1892 |
+
},
|
| 1893 |
+
{
|
| 1894 |
+
"epoch": 1.612611448507559,
|
| 1895 |
+
"learning_rate": 1.90211331518216e-07,
|
| 1896 |
+
"loss": 2.8855,
|
| 1897 |
+
"step": 6240
|
| 1898 |
+
},
|
| 1899 |
+
{
|
| 1900 |
+
"epoch": 1.6177800749450832,
|
| 1901 |
+
"learning_rate": 1.9017891870867368e-07,
|
| 1902 |
+
"loss": 2.9168,
|
| 1903 |
+
"step": 6260
|
| 1904 |
+
},
|
| 1905 |
+
{
|
| 1906 |
+
"epoch": 1.6229487013826076,
|
| 1907 |
+
"learning_rate": 1.9014650589913132e-07,
|
| 1908 |
+
"loss": 2.9443,
|
| 1909 |
+
"step": 6280
|
| 1910 |
+
},
|
| 1911 |
+
{
|
| 1912 |
+
"epoch": 1.628117327820132,
|
| 1913 |
+
"learning_rate": 1.90114093089589e-07,
|
| 1914 |
+
"loss": 2.9139,
|
| 1915 |
+
"step": 6300
|
| 1916 |
+
},
|
| 1917 |
+
{
|
| 1918 |
+
"epoch": 1.6332859542576559,
|
| 1919 |
+
"learning_rate": 1.9008168028004667e-07,
|
| 1920 |
+
"loss": 2.9343,
|
| 1921 |
+
"step": 6320
|
| 1922 |
+
},
|
| 1923 |
+
{
|
| 1924 |
+
"epoch": 1.6384545806951802,
|
| 1925 |
+
"learning_rate": 1.9004926747050433e-07,
|
| 1926 |
+
"loss": 2.9279,
|
| 1927 |
+
"step": 6340
|
| 1928 |
+
},
|
| 1929 |
+
{
|
| 1930 |
+
"epoch": 1.6436232071327046,
|
| 1931 |
+
"learning_rate": 1.9001685466096202e-07,
|
| 1932 |
+
"loss": 2.8902,
|
| 1933 |
+
"step": 6360
|
| 1934 |
+
},
|
| 1935 |
+
{
|
| 1936 |
+
"epoch": 1.6487918335702287,
|
| 1937 |
+
"learning_rate": 1.8998444185141968e-07,
|
| 1938 |
+
"loss": 2.9188,
|
| 1939 |
+
"step": 6380
|
| 1940 |
+
},
|
| 1941 |
+
{
|
| 1942 |
+
"epoch": 1.6539604600077529,
|
| 1943 |
+
"learning_rate": 1.8995202904187734e-07,
|
| 1944 |
+
"loss": 2.9314,
|
| 1945 |
+
"step": 6400
|
| 1946 |
+
},
|
| 1947 |
+
{
|
| 1948 |
+
"epoch": 1.6591290864452772,
|
| 1949 |
+
"learning_rate": 1.8991961623233503e-07,
|
| 1950 |
+
"loss": 2.9729,
|
| 1951 |
+
"step": 6420
|
| 1952 |
+
},
|
| 1953 |
+
{
|
| 1954 |
+
"epoch": 1.6642977128828014,
|
| 1955 |
+
"learning_rate": 1.8988720342279266e-07,
|
| 1956 |
+
"loss": 2.9336,
|
| 1957 |
+
"step": 6440
|
| 1958 |
+
},
|
| 1959 |
+
{
|
| 1960 |
+
"epoch": 1.6694663393203255,
|
| 1961 |
+
"learning_rate": 1.8985479061325035e-07,
|
| 1962 |
+
"loss": 2.9242,
|
| 1963 |
+
"step": 6460
|
| 1964 |
+
},
|
| 1965 |
+
{
|
| 1966 |
+
"epoch": 1.6746349657578499,
|
| 1967 |
+
"learning_rate": 1.8982237780370804e-07,
|
| 1968 |
+
"loss": 3.0066,
|
| 1969 |
+
"step": 6480
|
| 1970 |
+
},
|
| 1971 |
+
{
|
| 1972 |
+
"epoch": 1.6798035921953742,
|
| 1973 |
+
"learning_rate": 1.8978996499416568e-07,
|
| 1974 |
+
"loss": 2.9272,
|
| 1975 |
+
"step": 6500
|
| 1976 |
+
},
|
| 1977 |
+
{
|
| 1978 |
+
"epoch": 1.6849722186328981,
|
| 1979 |
+
"learning_rate": 1.8975755218462336e-07,
|
| 1980 |
+
"loss": 2.8841,
|
| 1981 |
+
"step": 6520
|
| 1982 |
+
},
|
| 1983 |
+
{
|
| 1984 |
+
"epoch": 1.6901408450704225,
|
| 1985 |
+
"learning_rate": 1.8972513937508103e-07,
|
| 1986 |
+
"loss": 2.9365,
|
| 1987 |
+
"step": 6540
|
| 1988 |
+
},
|
| 1989 |
+
{
|
| 1990 |
+
"epoch": 1.6953094715079469,
|
| 1991 |
+
"learning_rate": 1.896927265655387e-07,
|
| 1992 |
+
"loss": 2.9948,
|
| 1993 |
+
"step": 6560
|
| 1994 |
+
},
|
| 1995 |
+
{
|
| 1996 |
+
"epoch": 1.700478097945471,
|
| 1997 |
+
"learning_rate": 1.8966031375599638e-07,
|
| 1998 |
+
"loss": 2.9,
|
| 1999 |
+
"step": 6580
|
| 2000 |
+
},
|
| 2001 |
+
{
|
| 2002 |
+
"epoch": 1.7056467243829951,
|
| 2003 |
+
"learning_rate": 1.8962790094645404e-07,
|
| 2004 |
+
"loss": 2.9318,
|
| 2005 |
+
"step": 6600
|
| 2006 |
+
},
|
| 2007 |
+
{
|
| 2008 |
+
"epoch": 1.7108153508205195,
|
| 2009 |
+
"learning_rate": 1.895954881369117e-07,
|
| 2010 |
+
"loss": 2.8761,
|
| 2011 |
+
"step": 6620
|
| 2012 |
+
},
|
| 2013 |
+
{
|
| 2014 |
+
"epoch": 1.7159839772580436,
|
| 2015 |
+
"learning_rate": 1.895630753273694e-07,
|
| 2016 |
+
"loss": 2.9628,
|
| 2017 |
+
"step": 6640
|
| 2018 |
+
},
|
| 2019 |
+
{
|
| 2020 |
+
"epoch": 1.7211526036955678,
|
| 2021 |
+
"learning_rate": 1.8953066251782702e-07,
|
| 2022 |
+
"loss": 2.9114,
|
| 2023 |
+
"step": 6660
|
| 2024 |
+
},
|
| 2025 |
+
{
|
| 2026 |
+
"epoch": 1.7263212301330921,
|
| 2027 |
+
"learning_rate": 1.894982497082847e-07,
|
| 2028 |
+
"loss": 2.9338,
|
| 2029 |
+
"step": 6680
|
| 2030 |
+
},
|
| 2031 |
+
{
|
| 2032 |
+
"epoch": 1.7314898565706165,
|
| 2033 |
+
"learning_rate": 1.8946583689874237e-07,
|
| 2034 |
+
"loss": 2.9956,
|
| 2035 |
+
"step": 6700
|
| 2036 |
+
},
|
| 2037 |
+
{
|
| 2038 |
+
"epoch": 1.7366584830081406,
|
| 2039 |
+
"learning_rate": 1.8943342408920004e-07,
|
| 2040 |
+
"loss": 2.9089,
|
| 2041 |
+
"step": 6720
|
| 2042 |
+
},
|
| 2043 |
+
{
|
| 2044 |
+
"epoch": 1.7418271094456648,
|
| 2045 |
+
"learning_rate": 1.8940101127965772e-07,
|
| 2046 |
+
"loss": 2.9213,
|
| 2047 |
+
"step": 6740
|
| 2048 |
+
},
|
| 2049 |
+
{
|
| 2050 |
+
"epoch": 1.7469957358831891,
|
| 2051 |
+
"learning_rate": 1.8936859847011539e-07,
|
| 2052 |
+
"loss": 2.9461,
|
| 2053 |
+
"step": 6760
|
| 2054 |
+
},
|
| 2055 |
+
{
|
| 2056 |
+
"epoch": 1.7521643623207133,
|
| 2057 |
+
"learning_rate": 1.8933618566057305e-07,
|
| 2058 |
+
"loss": 2.9103,
|
| 2059 |
+
"step": 6780
|
| 2060 |
+
},
|
| 2061 |
+
{
|
| 2062 |
+
"epoch": 1.7573329887582374,
|
| 2063 |
+
"learning_rate": 1.8930377285103074e-07,
|
| 2064 |
+
"loss": 2.9025,
|
| 2065 |
+
"step": 6800
|
| 2066 |
+
},
|
| 2067 |
+
{
|
| 2068 |
+
"epoch": 1.7625016151957618,
|
| 2069 |
+
"learning_rate": 1.8927136004148837e-07,
|
| 2070 |
+
"loss": 2.9723,
|
| 2071 |
+
"step": 6820
|
| 2072 |
+
},
|
| 2073 |
+
{
|
| 2074 |
+
"epoch": 1.767670241633286,
|
| 2075 |
+
"learning_rate": 1.8923894723194606e-07,
|
| 2076 |
+
"loss": 2.9768,
|
| 2077 |
+
"step": 6840
|
| 2078 |
+
},
|
| 2079 |
+
{
|
| 2080 |
+
"epoch": 1.77283886807081,
|
| 2081 |
+
"learning_rate": 1.8920653442240375e-07,
|
| 2082 |
+
"loss": 2.8984,
|
| 2083 |
+
"step": 6860
|
| 2084 |
+
},
|
| 2085 |
+
{
|
| 2086 |
+
"epoch": 1.7780074945083344,
|
| 2087 |
+
"learning_rate": 1.8917412161286138e-07,
|
| 2088 |
+
"loss": 2.9167,
|
| 2089 |
+
"step": 6880
|
| 2090 |
+
},
|
| 2091 |
+
{
|
| 2092 |
+
"epoch": 1.7831761209458588,
|
| 2093 |
+
"learning_rate": 1.8914170880331907e-07,
|
| 2094 |
+
"loss": 2.9561,
|
| 2095 |
+
"step": 6900
|
| 2096 |
+
},
|
| 2097 |
+
{
|
| 2098 |
+
"epoch": 1.788344747383383,
|
| 2099 |
+
"learning_rate": 1.8910929599377673e-07,
|
| 2100 |
+
"loss": 2.8689,
|
| 2101 |
+
"step": 6920
|
| 2102 |
+
},
|
| 2103 |
+
{
|
| 2104 |
+
"epoch": 1.793513373820907,
|
| 2105 |
+
"learning_rate": 1.890768831842344e-07,
|
| 2106 |
+
"loss": 2.9081,
|
| 2107 |
+
"step": 6940
|
| 2108 |
+
},
|
| 2109 |
+
{
|
| 2110 |
+
"epoch": 1.7986820002584314,
|
| 2111 |
+
"learning_rate": 1.8904447037469208e-07,
|
| 2112 |
+
"loss": 2.8818,
|
| 2113 |
+
"step": 6960
|
| 2114 |
+
},
|
| 2115 |
+
{
|
| 2116 |
+
"epoch": 1.8038506266959555,
|
| 2117 |
+
"learning_rate": 1.8901205756514975e-07,
|
| 2118 |
+
"loss": 2.9503,
|
| 2119 |
+
"step": 6980
|
| 2120 |
+
},
|
| 2121 |
+
{
|
| 2122 |
+
"epoch": 1.8090192531334797,
|
| 2123 |
+
"learning_rate": 1.889796447556074e-07,
|
| 2124 |
+
"loss": 2.985,
|
| 2125 |
+
"step": 7000
|
| 2126 |
+
},
|
| 2127 |
+
{
|
| 2128 |
+
"epoch": 1.814187879571004,
|
| 2129 |
+
"learning_rate": 1.889472319460651e-07,
|
| 2130 |
+
"loss": 2.9812,
|
| 2131 |
+
"step": 7020
|
| 2132 |
+
},
|
| 2133 |
+
{
|
| 2134 |
+
"epoch": 1.8193565060085284,
|
| 2135 |
+
"learning_rate": 1.8891481913652273e-07,
|
| 2136 |
+
"loss": 2.9244,
|
| 2137 |
+
"step": 7040
|
| 2138 |
+
},
|
| 2139 |
+
{
|
| 2140 |
+
"epoch": 1.8245251324460523,
|
| 2141 |
+
"learning_rate": 1.8888240632698042e-07,
|
| 2142 |
+
"loss": 2.9199,
|
| 2143 |
+
"step": 7060
|
| 2144 |
+
},
|
| 2145 |
+
{
|
| 2146 |
+
"epoch": 1.8296937588835767,
|
| 2147 |
+
"learning_rate": 1.8884999351743808e-07,
|
| 2148 |
+
"loss": 2.9121,
|
| 2149 |
+
"step": 7080
|
| 2150 |
+
},
|
| 2151 |
+
{
|
| 2152 |
+
"epoch": 1.834862385321101,
|
| 2153 |
+
"learning_rate": 1.8881758070789574e-07,
|
| 2154 |
+
"loss": 2.9476,
|
| 2155 |
+
"step": 7100
|
| 2156 |
+
},
|
| 2157 |
+
{
|
| 2158 |
+
"epoch": 1.8400310117586252,
|
| 2159 |
+
"learning_rate": 1.8878516789835343e-07,
|
| 2160 |
+
"loss": 2.9117,
|
| 2161 |
+
"step": 7120
|
| 2162 |
+
},
|
| 2163 |
+
{
|
| 2164 |
+
"epoch": 1.8451996381961493,
|
| 2165 |
+
"learning_rate": 1.887527550888111e-07,
|
| 2166 |
+
"loss": 2.8593,
|
| 2167 |
+
"step": 7140
|
| 2168 |
+
},
|
| 2169 |
+
{
|
| 2170 |
+
"epoch": 1.8503682646336737,
|
| 2171 |
+
"learning_rate": 1.8872034227926876e-07,
|
| 2172 |
+
"loss": 2.9515,
|
| 2173 |
+
"step": 7160
|
| 2174 |
+
},
|
| 2175 |
+
{
|
| 2176 |
+
"epoch": 1.8555368910711978,
|
| 2177 |
+
"learning_rate": 1.8868792946972644e-07,
|
| 2178 |
+
"loss": 2.8816,
|
| 2179 |
+
"step": 7180
|
| 2180 |
+
},
|
| 2181 |
+
{
|
| 2182 |
+
"epoch": 1.860705517508722,
|
| 2183 |
+
"learning_rate": 1.8865551666018408e-07,
|
| 2184 |
+
"loss": 2.8811,
|
| 2185 |
+
"step": 7200
|
| 2186 |
+
},
|
| 2187 |
+
{
|
| 2188 |
+
"epoch": 1.8658741439462463,
|
| 2189 |
+
"learning_rate": 1.8862310385064177e-07,
|
| 2190 |
+
"loss": 2.8955,
|
| 2191 |
+
"step": 7220
|
| 2192 |
+
},
|
| 2193 |
+
{
|
| 2194 |
+
"epoch": 1.8710427703837706,
|
| 2195 |
+
"learning_rate": 1.8859069104109943e-07,
|
| 2196 |
+
"loss": 2.8895,
|
| 2197 |
+
"step": 7240
|
| 2198 |
+
},
|
| 2199 |
+
{
|
| 2200 |
+
"epoch": 1.8762113968212948,
|
| 2201 |
+
"learning_rate": 1.885582782315571e-07,
|
| 2202 |
+
"loss": 2.8621,
|
| 2203 |
+
"step": 7260
|
| 2204 |
+
},
|
| 2205 |
+
{
|
| 2206 |
+
"epoch": 1.881380023258819,
|
| 2207 |
+
"learning_rate": 1.8852586542201478e-07,
|
| 2208 |
+
"loss": 2.8738,
|
| 2209 |
+
"step": 7280
|
| 2210 |
+
},
|
| 2211 |
+
{
|
| 2212 |
+
"epoch": 1.8865486496963433,
|
| 2213 |
+
"learning_rate": 1.8849345261247244e-07,
|
| 2214 |
+
"loss": 2.9246,
|
| 2215 |
+
"step": 7300
|
| 2216 |
+
},
|
| 2217 |
+
{
|
| 2218 |
+
"epoch": 1.8917172761338674,
|
| 2219 |
+
"learning_rate": 1.884610398029301e-07,
|
| 2220 |
+
"loss": 2.898,
|
| 2221 |
+
"step": 7320
|
| 2222 |
+
},
|
| 2223 |
+
{
|
| 2224 |
+
"epoch": 1.8968859025713916,
|
| 2225 |
+
"learning_rate": 1.884286269933878e-07,
|
| 2226 |
+
"loss": 2.9362,
|
| 2227 |
+
"step": 7340
|
| 2228 |
+
},
|
| 2229 |
+
{
|
| 2230 |
+
"epoch": 1.902054529008916,
|
| 2231 |
+
"learning_rate": 1.8839621418384545e-07,
|
| 2232 |
+
"loss": 2.9173,
|
| 2233 |
+
"step": 7360
|
| 2234 |
+
},
|
| 2235 |
+
{
|
| 2236 |
+
"epoch": 1.90722315544644,
|
| 2237 |
+
"learning_rate": 1.8836380137430312e-07,
|
| 2238 |
+
"loss": 2.9209,
|
| 2239 |
+
"step": 7380
|
| 2240 |
+
},
|
| 2241 |
+
{
|
| 2242 |
+
"epoch": 1.9123917818839642,
|
| 2243 |
+
"learning_rate": 1.883313885647608e-07,
|
| 2244 |
+
"loss": 2.9294,
|
| 2245 |
+
"step": 7400
|
| 2246 |
+
},
|
| 2247 |
+
{
|
| 2248 |
+
"epoch": 1.9175604083214886,
|
| 2249 |
+
"learning_rate": 1.8829897575521844e-07,
|
| 2250 |
+
"loss": 2.9321,
|
| 2251 |
+
"step": 7420
|
| 2252 |
+
},
|
| 2253 |
+
{
|
| 2254 |
+
"epoch": 1.922729034759013,
|
| 2255 |
+
"learning_rate": 1.8826656294567613e-07,
|
| 2256 |
+
"loss": 2.8453,
|
| 2257 |
+
"step": 7440
|
| 2258 |
+
},
|
| 2259 |
+
{
|
| 2260 |
+
"epoch": 1.927897661196537,
|
| 2261 |
+
"learning_rate": 1.882341501361338e-07,
|
| 2262 |
+
"loss": 2.8941,
|
| 2263 |
+
"step": 7460
|
| 2264 |
+
},
|
| 2265 |
+
{
|
| 2266 |
+
"epoch": 1.9330662876340612,
|
| 2267 |
+
"learning_rate": 1.8820173732659145e-07,
|
| 2268 |
+
"loss": 2.932,
|
| 2269 |
+
"step": 7480
|
| 2270 |
+
},
|
| 2271 |
+
{
|
| 2272 |
+
"epoch": 1.9382349140715855,
|
| 2273 |
+
"learning_rate": 1.8816932451704914e-07,
|
| 2274 |
+
"loss": 2.9291,
|
| 2275 |
+
"step": 7500
|
| 2276 |
+
},
|
| 2277 |
+
{
|
| 2278 |
+
"epoch": 1.9434035405091097,
|
| 2279 |
+
"learning_rate": 1.881369117075068e-07,
|
| 2280 |
+
"loss": 2.9375,
|
| 2281 |
+
"step": 7520
|
| 2282 |
+
},
|
| 2283 |
+
{
|
| 2284 |
+
"epoch": 1.9485721669466338,
|
| 2285 |
+
"learning_rate": 1.8810449889796446e-07,
|
| 2286 |
+
"loss": 2.9291,
|
| 2287 |
+
"step": 7540
|
| 2288 |
+
},
|
| 2289 |
+
{
|
| 2290 |
+
"epoch": 1.9537407933841582,
|
| 2291 |
+
"learning_rate": 1.8807208608842215e-07,
|
| 2292 |
+
"loss": 2.8834,
|
| 2293 |
+
"step": 7560
|
| 2294 |
+
},
|
| 2295 |
+
{
|
| 2296 |
+
"epoch": 1.9589094198216825,
|
| 2297 |
+
"learning_rate": 1.880396732788798e-07,
|
| 2298 |
+
"loss": 2.9606,
|
| 2299 |
+
"step": 7580
|
| 2300 |
+
},
|
| 2301 |
+
{
|
| 2302 |
+
"epoch": 1.9640780462592065,
|
| 2303 |
+
"learning_rate": 1.8800726046933748e-07,
|
| 2304 |
+
"loss": 2.8769,
|
| 2305 |
+
"step": 7600
|
| 2306 |
+
},
|
| 2307 |
+
{
|
| 2308 |
+
"epoch": 1.9692466726967308,
|
| 2309 |
+
"learning_rate": 1.8797484765979514e-07,
|
| 2310 |
+
"loss": 2.9195,
|
| 2311 |
+
"step": 7620
|
| 2312 |
+
},
|
| 2313 |
+
{
|
| 2314 |
+
"epoch": 1.9744152991342552,
|
| 2315 |
+
"learning_rate": 1.879424348502528e-07,
|
| 2316 |
+
"loss": 2.9,
|
| 2317 |
+
"step": 7640
|
| 2318 |
+
},
|
| 2319 |
+
{
|
| 2320 |
+
"epoch": 1.9795839255717793,
|
| 2321 |
+
"learning_rate": 1.879100220407105e-07,
|
| 2322 |
+
"loss": 2.9407,
|
| 2323 |
+
"step": 7660
|
| 2324 |
+
},
|
| 2325 |
+
{
|
| 2326 |
+
"epoch": 1.9847525520093035,
|
| 2327 |
+
"learning_rate": 1.8787760923116815e-07,
|
| 2328 |
+
"loss": 2.9094,
|
| 2329 |
+
"step": 7680
|
| 2330 |
+
},
|
| 2331 |
+
{
|
| 2332 |
+
"epoch": 1.9899211784468278,
|
| 2333 |
+
"learning_rate": 1.878451964216258e-07,
|
| 2334 |
+
"loss": 2.9574,
|
| 2335 |
+
"step": 7700
|
| 2336 |
+
},
|
| 2337 |
+
{
|
| 2338 |
+
"epoch": 1.995089804884352,
|
| 2339 |
+
"learning_rate": 1.878127836120835e-07,
|
| 2340 |
+
"loss": 2.8978,
|
| 2341 |
+
"step": 7720
|
| 2342 |
+
},
|
| 2343 |
+
{
|
| 2344 |
+
"epoch": 2.0,
|
| 2345 |
+
"eval_bleu": 0.0747,
|
| 2346 |
+
"eval_gen_len": 113.0081,
|
| 2347 |
+
"eval_loss": 2.8610005378723145,
|
| 2348 |
+
"eval_runtime": 1965.2378,
|
| 2349 |
+
"eval_samples_per_second": 0.876,
|
| 2350 |
+
"eval_steps_per_second": 0.438,
|
| 2351 |
+
"step": 7739
|
| 2352 |
}
|
| 2353 |
],
|
| 2354 |
"logging_steps": 20,
|
|
|
|
| 2368 |
"attributes": {}
|
| 2369 |
}
|
| 2370 |
},
|
| 2371 |
+
"total_flos": 5.079387107794944e+16,
|
| 2372 |
"train_batch_size": 2,
|
| 2373 |
"trial_name": null,
|
| 2374 |
"trial_params": null
|