Upload folder using huggingface_hub
Browse files- checkpoint-100/config.json +56 -0
- checkpoint-100/generation_config.json +16 -0
- checkpoint-100/model.safetensors +3 -0
- checkpoint-100/optimizer.pt +3 -0
- checkpoint-100/rng_state.pth +3 -0
- checkpoint-100/scheduler.pt +3 -0
- checkpoint-100/trainer_state.json +1033 -0
- checkpoint-100/training_args.bin +3 -0
- checkpoint-2/config.json +56 -0
- checkpoint-2/generation_config.json +16 -0
- checkpoint-2/model.safetensors +3 -0
- checkpoint-2/optimizer.pt +3 -0
- checkpoint-2/rng_state.pth +3 -0
- checkpoint-2/scheduler.pt +3 -0
- checkpoint-2/trainer_state.json +53 -0
- checkpoint-2/training_args.bin +3 -0
- checkpoint-200/config.json +56 -0
- checkpoint-200/generation_config.json +16 -0
- checkpoint-200/model.safetensors +3 -0
- checkpoint-200/optimizer.pt +3 -0
- checkpoint-200/rng_state.pth +3 -0
- checkpoint-200/scheduler.pt +3 -0
- checkpoint-200/trainer_state.json +2023 -0
- checkpoint-200/training_args.bin +3 -0
- config.json +56 -0
- generation_config.json +16 -0
- model.safetensors +3 -0
- runs/Dec03_08-07-34_e9eef5465588/events.out.tfevents.1733213606.e9eef5465588.150.0 +3 -0
- runs/Dec03_08-07-34_e9eef5465588/events.out.tfevents.1733213681.e9eef5465588.150.1 +3 -0
- runs/Dec03_08-07-34_e9eef5465588/events.out.tfevents.1733213721.e9eef5465588.150.2 +3 -0
- runs/Dec03_08-19-26_e9eef5465588/events.out.tfevents.1733214041.e9eef5465588.150.3 +3 -0
- runs/Dec03_08-19-26_e9eef5465588/events.out.tfevents.1733214098.e9eef5465588.150.4 +3 -0
- runs/Dec03_08-48-25_e9eef5465588/events.out.tfevents.1733215721.e9eef5465588.150.5 +3 -0
- runs/Dec03_08-52-34_e9eef5465588/events.out.tfevents.1733215972.e9eef5465588.150.6 +3 -0
- source.spm +0 -0
- special_tokens_map.json +5 -0
- target.spm +0 -0
- tokenizer_config.json +38 -0
- vocab.json +0 -0
checkpoint-100/config.json
ADDED
|
@@ -0,0 +1,56 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "Helsinki-NLP/opus-mt-zh-en",
|
| 3 |
+
"activation_dropout": 0.0,
|
| 4 |
+
"activation_function": "swish",
|
| 5 |
+
"add_bias_logits": false,
|
| 6 |
+
"add_final_layer_norm": false,
|
| 7 |
+
"architectures": [
|
| 8 |
+
"MarianMTModel"
|
| 9 |
+
],
|
| 10 |
+
"attention_dropout": 0.0,
|
| 11 |
+
"bos_token_id": 0,
|
| 12 |
+
"classif_dropout": 0.0,
|
| 13 |
+
"classifier_dropout": 0.0,
|
| 14 |
+
"d_model": 512,
|
| 15 |
+
"decoder_attention_heads": 8,
|
| 16 |
+
"decoder_ffn_dim": 2048,
|
| 17 |
+
"decoder_layerdrop": 0.0,
|
| 18 |
+
"decoder_layers": 6,
|
| 19 |
+
"decoder_start_token_id": 65000,
|
| 20 |
+
"decoder_vocab_size": 65001,
|
| 21 |
+
"dropout": 0.1,
|
| 22 |
+
"encoder_attention_heads": 8,
|
| 23 |
+
"encoder_ffn_dim": 2048,
|
| 24 |
+
"encoder_layerdrop": 0.0,
|
| 25 |
+
"encoder_layers": 6,
|
| 26 |
+
"eos_token_id": 0,
|
| 27 |
+
"extra_pos_embeddings": 65001,
|
| 28 |
+
"forced_eos_token_id": 0,
|
| 29 |
+
"id2label": {
|
| 30 |
+
"0": "LABEL_0",
|
| 31 |
+
"1": "LABEL_1",
|
| 32 |
+
"2": "LABEL_2"
|
| 33 |
+
},
|
| 34 |
+
"init_std": 0.02,
|
| 35 |
+
"is_encoder_decoder": true,
|
| 36 |
+
"label2id": {
|
| 37 |
+
"LABEL_0": 0,
|
| 38 |
+
"LABEL_1": 1,
|
| 39 |
+
"LABEL_2": 2
|
| 40 |
+
},
|
| 41 |
+
"max_length": null,
|
| 42 |
+
"max_position_embeddings": 512,
|
| 43 |
+
"model_type": "marian",
|
| 44 |
+
"normalize_before": false,
|
| 45 |
+
"normalize_embedding": false,
|
| 46 |
+
"num_beams": null,
|
| 47 |
+
"num_hidden_layers": 6,
|
| 48 |
+
"pad_token_id": 65000,
|
| 49 |
+
"scale_embedding": true,
|
| 50 |
+
"share_encoder_decoder_embeddings": true,
|
| 51 |
+
"static_position_embeddings": true,
|
| 52 |
+
"torch_dtype": "float32",
|
| 53 |
+
"transformers_version": "4.46.2",
|
| 54 |
+
"use_cache": true,
|
| 55 |
+
"vocab_size": 65001
|
| 56 |
+
}
|
checkpoint-100/generation_config.json
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bad_words_ids": [
|
| 3 |
+
[
|
| 4 |
+
65000
|
| 5 |
+
]
|
| 6 |
+
],
|
| 7 |
+
"bos_token_id": 0,
|
| 8 |
+
"decoder_start_token_id": 65000,
|
| 9 |
+
"eos_token_id": 0,
|
| 10 |
+
"forced_eos_token_id": 0,
|
| 11 |
+
"max_length": 512,
|
| 12 |
+
"num_beams": 6,
|
| 13 |
+
"pad_token_id": 65000,
|
| 14 |
+
"renormalize_logits": true,
|
| 15 |
+
"transformers_version": "4.46.2"
|
| 16 |
+
}
|
checkpoint-100/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:68953ab2ff5e623b0db0c4745d06c47a40ad25b77641b99e8c9cb666afdb7726
|
| 3 |
+
size 309965092
|
checkpoint-100/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8b55001a294670a5ef85b3c539c37e3928daa41b1716e920b1218e311f09e773
|
| 3 |
+
size 619556538
|
checkpoint-100/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2d0c94ab556094703aa6c8e21234c41e1b8704b3a8d16bdd6b271aa58581fe74
|
| 3 |
+
size 13990
|
checkpoint-100/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fcd60da393ff67ad2f08779377eeaa08189a6bf520f752442788825d9bce1449
|
| 3 |
+
size 1064
|
checkpoint-100/trainer_state.json
ADDED
|
@@ -0,0 +1,1033 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 3.4803478717803955,
|
| 3 |
+
"best_model_checkpoint": "opus-mt-zh-en-finetuned-zh-to-en/checkpoint-2",
|
| 4 |
+
"epoch": 100.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 100,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 1.0,
|
| 13 |
+
"eval_bleu": 5.6918,
|
| 14 |
+
"eval_gen_len": 57.3333,
|
| 15 |
+
"eval_loss": 3.5217063426971436,
|
| 16 |
+
"eval_runtime": 10.9763,
|
| 17 |
+
"eval_samples_per_second": 0.273,
|
| 18 |
+
"eval_steps_per_second": 0.091,
|
| 19 |
+
"step": 1
|
| 20 |
+
},
|
| 21 |
+
{
|
| 22 |
+
"epoch": 2.0,
|
| 23 |
+
"eval_bleu": 7.6304,
|
| 24 |
+
"eval_gen_len": 55.3333,
|
| 25 |
+
"eval_loss": 3.4803478717803955,
|
| 26 |
+
"eval_runtime": 8.3555,
|
| 27 |
+
"eval_samples_per_second": 0.359,
|
| 28 |
+
"eval_steps_per_second": 0.12,
|
| 29 |
+
"step": 2
|
| 30 |
+
},
|
| 31 |
+
{
|
| 32 |
+
"epoch": 3.0,
|
| 33 |
+
"eval_bleu": 7.9453,
|
| 34 |
+
"eval_gen_len": 56.0,
|
| 35 |
+
"eval_loss": 3.491027593612671,
|
| 36 |
+
"eval_runtime": 9.8597,
|
| 37 |
+
"eval_samples_per_second": 0.304,
|
| 38 |
+
"eval_steps_per_second": 0.101,
|
| 39 |
+
"step": 3
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 4.0,
|
| 43 |
+
"eval_bleu": 6.5495,
|
| 44 |
+
"eval_gen_len": 79.0,
|
| 45 |
+
"eval_loss": 3.5166053771972656,
|
| 46 |
+
"eval_runtime": 13.007,
|
| 47 |
+
"eval_samples_per_second": 0.231,
|
| 48 |
+
"eval_steps_per_second": 0.077,
|
| 49 |
+
"step": 4
|
| 50 |
+
},
|
| 51 |
+
{
|
| 52 |
+
"epoch": 5.0,
|
| 53 |
+
"eval_bleu": 6.6932,
|
| 54 |
+
"eval_gen_len": 80.0,
|
| 55 |
+
"eval_loss": 3.5517735481262207,
|
| 56 |
+
"eval_runtime": 14.9524,
|
| 57 |
+
"eval_samples_per_second": 0.201,
|
| 58 |
+
"eval_steps_per_second": 0.067,
|
| 59 |
+
"step": 5
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 6.0,
|
| 63 |
+
"eval_bleu": 5.9333,
|
| 64 |
+
"eval_gen_len": 137.0,
|
| 65 |
+
"eval_loss": 3.5805959701538086,
|
| 66 |
+
"eval_runtime": 61.8901,
|
| 67 |
+
"eval_samples_per_second": 0.048,
|
| 68 |
+
"eval_steps_per_second": 0.016,
|
| 69 |
+
"step": 6
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 7.0,
|
| 73 |
+
"eval_bleu": 5.6605,
|
| 74 |
+
"eval_gen_len": 133.6667,
|
| 75 |
+
"eval_loss": 3.6102888584136963,
|
| 76 |
+
"eval_runtime": 55.757,
|
| 77 |
+
"eval_samples_per_second": 0.054,
|
| 78 |
+
"eval_steps_per_second": 0.018,
|
| 79 |
+
"step": 7
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 8.0,
|
| 83 |
+
"eval_bleu": 7.1686,
|
| 84 |
+
"eval_gen_len": 135.3333,
|
| 85 |
+
"eval_loss": 3.645458936691284,
|
| 86 |
+
"eval_runtime": 52.3901,
|
| 87 |
+
"eval_samples_per_second": 0.057,
|
| 88 |
+
"eval_steps_per_second": 0.019,
|
| 89 |
+
"step": 8
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"epoch": 9.0,
|
| 93 |
+
"eval_bleu": 9.8037,
|
| 94 |
+
"eval_gen_len": 146.0,
|
| 95 |
+
"eval_loss": 3.6737775802612305,
|
| 96 |
+
"eval_runtime": 50.7208,
|
| 97 |
+
"eval_samples_per_second": 0.059,
|
| 98 |
+
"eval_steps_per_second": 0.02,
|
| 99 |
+
"step": 9
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 10.0,
|
| 103 |
+
"eval_bleu": 9.9552,
|
| 104 |
+
"eval_gen_len": 139.0,
|
| 105 |
+
"eval_loss": 3.707125186920166,
|
| 106 |
+
"eval_runtime": 49.5408,
|
| 107 |
+
"eval_samples_per_second": 0.061,
|
| 108 |
+
"eval_steps_per_second": 0.02,
|
| 109 |
+
"step": 10
|
| 110 |
+
},
|
| 111 |
+
{
|
| 112 |
+
"epoch": 11.0,
|
| 113 |
+
"eval_bleu": 10.311,
|
| 114 |
+
"eval_gen_len": 134.0,
|
| 115 |
+
"eval_loss": 3.7424490451812744,
|
| 116 |
+
"eval_runtime": 45.4229,
|
| 117 |
+
"eval_samples_per_second": 0.066,
|
| 118 |
+
"eval_steps_per_second": 0.022,
|
| 119 |
+
"step": 11
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 12.0,
|
| 123 |
+
"eval_bleu": 10.1929,
|
| 124 |
+
"eval_gen_len": 131.6667,
|
| 125 |
+
"eval_loss": 3.786301612854004,
|
| 126 |
+
"eval_runtime": 45.6583,
|
| 127 |
+
"eval_samples_per_second": 0.066,
|
| 128 |
+
"eval_steps_per_second": 0.022,
|
| 129 |
+
"step": 12
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 13.0,
|
| 133 |
+
"eval_bleu": 9.8671,
|
| 134 |
+
"eval_gen_len": 119.3333,
|
| 135 |
+
"eval_loss": 3.825416088104248,
|
| 136 |
+
"eval_runtime": 44.3396,
|
| 137 |
+
"eval_samples_per_second": 0.068,
|
| 138 |
+
"eval_steps_per_second": 0.023,
|
| 139 |
+
"step": 13
|
| 140 |
+
},
|
| 141 |
+
{
|
| 142 |
+
"epoch": 14.0,
|
| 143 |
+
"eval_bleu": 11.4259,
|
| 144 |
+
"eval_gen_len": 107.3333,
|
| 145 |
+
"eval_loss": 3.866556167602539,
|
| 146 |
+
"eval_runtime": 60.118,
|
| 147 |
+
"eval_samples_per_second": 0.05,
|
| 148 |
+
"eval_steps_per_second": 0.017,
|
| 149 |
+
"step": 14
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 15.0,
|
| 153 |
+
"eval_bleu": 11.4172,
|
| 154 |
+
"eval_gen_len": 100.6667,
|
| 155 |
+
"eval_loss": 3.900843620300293,
|
| 156 |
+
"eval_runtime": 56.7929,
|
| 157 |
+
"eval_samples_per_second": 0.053,
|
| 158 |
+
"eval_steps_per_second": 0.018,
|
| 159 |
+
"step": 15
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 16.0,
|
| 163 |
+
"eval_bleu": 10.8738,
|
| 164 |
+
"eval_gen_len": 117.6667,
|
| 165 |
+
"eval_loss": 3.935436964035034,
|
| 166 |
+
"eval_runtime": 38.5973,
|
| 167 |
+
"eval_samples_per_second": 0.078,
|
| 168 |
+
"eval_steps_per_second": 0.026,
|
| 169 |
+
"step": 16
|
| 170 |
+
},
|
| 171 |
+
{
|
| 172 |
+
"epoch": 17.0,
|
| 173 |
+
"eval_bleu": 10.7717,
|
| 174 |
+
"eval_gen_len": 117.0,
|
| 175 |
+
"eval_loss": 3.9768829345703125,
|
| 176 |
+
"eval_runtime": 36.4507,
|
| 177 |
+
"eval_samples_per_second": 0.082,
|
| 178 |
+
"eval_steps_per_second": 0.027,
|
| 179 |
+
"step": 17
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 18.0,
|
| 183 |
+
"eval_bleu": 9.9034,
|
| 184 |
+
"eval_gen_len": 118.6667,
|
| 185 |
+
"eval_loss": 4.017246246337891,
|
| 186 |
+
"eval_runtime": 36.6426,
|
| 187 |
+
"eval_samples_per_second": 0.082,
|
| 188 |
+
"eval_steps_per_second": 0.027,
|
| 189 |
+
"step": 18
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 19.0,
|
| 193 |
+
"eval_bleu": 10.0376,
|
| 194 |
+
"eval_gen_len": 121.6667,
|
| 195 |
+
"eval_loss": 4.05275821685791,
|
| 196 |
+
"eval_runtime": 34.9881,
|
| 197 |
+
"eval_samples_per_second": 0.086,
|
| 198 |
+
"eval_steps_per_second": 0.029,
|
| 199 |
+
"step": 19
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 20.0,
|
| 203 |
+
"eval_bleu": 8.6054,
|
| 204 |
+
"eval_gen_len": 61.0,
|
| 205 |
+
"eval_loss": 4.080726623535156,
|
| 206 |
+
"eval_runtime": 9.9523,
|
| 207 |
+
"eval_samples_per_second": 0.301,
|
| 208 |
+
"eval_steps_per_second": 0.1,
|
| 209 |
+
"step": 20
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 21.0,
|
| 213 |
+
"eval_bleu": 10.9496,
|
| 214 |
+
"eval_gen_len": 67.0,
|
| 215 |
+
"eval_loss": 4.109140396118164,
|
| 216 |
+
"eval_runtime": 11.0517,
|
| 217 |
+
"eval_samples_per_second": 0.271,
|
| 218 |
+
"eval_steps_per_second": 0.09,
|
| 219 |
+
"step": 21
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 22.0,
|
| 223 |
+
"eval_bleu": 10.8268,
|
| 224 |
+
"eval_gen_len": 65.6667,
|
| 225 |
+
"eval_loss": 4.133188247680664,
|
| 226 |
+
"eval_runtime": 10.5163,
|
| 227 |
+
"eval_samples_per_second": 0.285,
|
| 228 |
+
"eval_steps_per_second": 0.095,
|
| 229 |
+
"step": 22
|
| 230 |
+
},
|
| 231 |
+
{
|
| 232 |
+
"epoch": 23.0,
|
| 233 |
+
"eval_bleu": 10.5556,
|
| 234 |
+
"eval_gen_len": 63.6667,
|
| 235 |
+
"eval_loss": 4.158883571624756,
|
| 236 |
+
"eval_runtime": 9.4741,
|
| 237 |
+
"eval_samples_per_second": 0.317,
|
| 238 |
+
"eval_steps_per_second": 0.106,
|
| 239 |
+
"step": 23
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 24.0,
|
| 243 |
+
"eval_bleu": 10.8678,
|
| 244 |
+
"eval_gen_len": 62.0,
|
| 245 |
+
"eval_loss": 4.181970119476318,
|
| 246 |
+
"eval_runtime": 10.5606,
|
| 247 |
+
"eval_samples_per_second": 0.284,
|
| 248 |
+
"eval_steps_per_second": 0.095,
|
| 249 |
+
"step": 24
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 25.0,
|
| 253 |
+
"eval_bleu": 10.8678,
|
| 254 |
+
"eval_gen_len": 61.6667,
|
| 255 |
+
"eval_loss": 4.196631908416748,
|
| 256 |
+
"eval_runtime": 9.7369,
|
| 257 |
+
"eval_samples_per_second": 0.308,
|
| 258 |
+
"eval_steps_per_second": 0.103,
|
| 259 |
+
"step": 25
|
| 260 |
+
},
|
| 261 |
+
{
|
| 262 |
+
"epoch": 26.0,
|
| 263 |
+
"eval_bleu": 10.27,
|
| 264 |
+
"eval_gen_len": 61.6667,
|
| 265 |
+
"eval_loss": 4.209479808807373,
|
| 266 |
+
"eval_runtime": 9.9007,
|
| 267 |
+
"eval_samples_per_second": 0.303,
|
| 268 |
+
"eval_steps_per_second": 0.101,
|
| 269 |
+
"step": 26
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 27.0,
|
| 273 |
+
"eval_bleu": 10.6577,
|
| 274 |
+
"eval_gen_len": 60.0,
|
| 275 |
+
"eval_loss": 4.220566272735596,
|
| 276 |
+
"eval_runtime": 9.9798,
|
| 277 |
+
"eval_samples_per_second": 0.301,
|
| 278 |
+
"eval_steps_per_second": 0.1,
|
| 279 |
+
"step": 27
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 28.0,
|
| 283 |
+
"eval_bleu": 10.2137,
|
| 284 |
+
"eval_gen_len": 60.6667,
|
| 285 |
+
"eval_loss": 4.2330474853515625,
|
| 286 |
+
"eval_runtime": 11.0027,
|
| 287 |
+
"eval_samples_per_second": 0.273,
|
| 288 |
+
"eval_steps_per_second": 0.091,
|
| 289 |
+
"step": 28
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 29.0,
|
| 293 |
+
"eval_bleu": 10.0108,
|
| 294 |
+
"eval_gen_len": 62.0,
|
| 295 |
+
"eval_loss": 4.245401382446289,
|
| 296 |
+
"eval_runtime": 10.7321,
|
| 297 |
+
"eval_samples_per_second": 0.28,
|
| 298 |
+
"eval_steps_per_second": 0.093,
|
| 299 |
+
"step": 29
|
| 300 |
+
},
|
| 301 |
+
{
|
| 302 |
+
"epoch": 30.0,
|
| 303 |
+
"eval_bleu": 10.6984,
|
| 304 |
+
"eval_gen_len": 60.0,
|
| 305 |
+
"eval_loss": 4.257299423217773,
|
| 306 |
+
"eval_runtime": 9.8765,
|
| 307 |
+
"eval_samples_per_second": 0.304,
|
| 308 |
+
"eval_steps_per_second": 0.101,
|
| 309 |
+
"step": 30
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 31.0,
|
| 313 |
+
"eval_bleu": 9.6494,
|
| 314 |
+
"eval_gen_len": 83.0,
|
| 315 |
+
"eval_loss": 4.2664995193481445,
|
| 316 |
+
"eval_runtime": 13.2171,
|
| 317 |
+
"eval_samples_per_second": 0.227,
|
| 318 |
+
"eval_steps_per_second": 0.076,
|
| 319 |
+
"step": 31
|
| 320 |
+
},
|
| 321 |
+
{
|
| 322 |
+
"epoch": 32.0,
|
| 323 |
+
"eval_bleu": 9.7503,
|
| 324 |
+
"eval_gen_len": 84.3333,
|
| 325 |
+
"eval_loss": 4.276342868804932,
|
| 326 |
+
"eval_runtime": 12.9373,
|
| 327 |
+
"eval_samples_per_second": 0.232,
|
| 328 |
+
"eval_steps_per_second": 0.077,
|
| 329 |
+
"step": 32
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 33.0,
|
| 333 |
+
"eval_bleu": 9.7503,
|
| 334 |
+
"eval_gen_len": 84.3333,
|
| 335 |
+
"eval_loss": 4.287381649017334,
|
| 336 |
+
"eval_runtime": 12.9473,
|
| 337 |
+
"eval_samples_per_second": 0.232,
|
| 338 |
+
"eval_steps_per_second": 0.077,
|
| 339 |
+
"step": 33
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 34.0,
|
| 343 |
+
"eval_bleu": 9.7503,
|
| 344 |
+
"eval_gen_len": 84.3333,
|
| 345 |
+
"eval_loss": 4.295342445373535,
|
| 346 |
+
"eval_runtime": 13.0533,
|
| 347 |
+
"eval_samples_per_second": 0.23,
|
| 348 |
+
"eval_steps_per_second": 0.077,
|
| 349 |
+
"step": 34
|
| 350 |
+
},
|
| 351 |
+
{
|
| 352 |
+
"epoch": 35.0,
|
| 353 |
+
"eval_bleu": 9.7503,
|
| 354 |
+
"eval_gen_len": 84.3333,
|
| 355 |
+
"eval_loss": 4.301455020904541,
|
| 356 |
+
"eval_runtime": 13.2102,
|
| 357 |
+
"eval_samples_per_second": 0.227,
|
| 358 |
+
"eval_steps_per_second": 0.076,
|
| 359 |
+
"step": 35
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 36.0,
|
| 363 |
+
"eval_bleu": 9.4923,
|
| 364 |
+
"eval_gen_len": 70.0,
|
| 365 |
+
"eval_loss": 4.303869247436523,
|
| 366 |
+
"eval_runtime": 105.6574,
|
| 367 |
+
"eval_samples_per_second": 0.028,
|
| 368 |
+
"eval_steps_per_second": 0.009,
|
| 369 |
+
"step": 36
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 37.0,
|
| 373 |
+
"eval_bleu": 9.7956,
|
| 374 |
+
"eval_gen_len": 64.6667,
|
| 375 |
+
"eval_loss": 4.308076858520508,
|
| 376 |
+
"eval_runtime": 86.8335,
|
| 377 |
+
"eval_samples_per_second": 0.035,
|
| 378 |
+
"eval_steps_per_second": 0.012,
|
| 379 |
+
"step": 37
|
| 380 |
+
},
|
| 381 |
+
{
|
| 382 |
+
"epoch": 38.0,
|
| 383 |
+
"eval_bleu": 10.3592,
|
| 384 |
+
"eval_gen_len": 70.6667,
|
| 385 |
+
"eval_loss": 4.311928749084473,
|
| 386 |
+
"eval_runtime": 10.4025,
|
| 387 |
+
"eval_samples_per_second": 0.288,
|
| 388 |
+
"eval_steps_per_second": 0.096,
|
| 389 |
+
"step": 38
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 39.0,
|
| 393 |
+
"eval_bleu": 11.5296,
|
| 394 |
+
"eval_gen_len": 86.0,
|
| 395 |
+
"eval_loss": 4.315915584564209,
|
| 396 |
+
"eval_runtime": 13.3622,
|
| 397 |
+
"eval_samples_per_second": 0.225,
|
| 398 |
+
"eval_steps_per_second": 0.075,
|
| 399 |
+
"step": 39
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 40.0,
|
| 403 |
+
"eval_bleu": 11.1681,
|
| 404 |
+
"eval_gen_len": 78.0,
|
| 405 |
+
"eval_loss": 4.319802284240723,
|
| 406 |
+
"eval_runtime": 13.5231,
|
| 407 |
+
"eval_samples_per_second": 0.222,
|
| 408 |
+
"eval_steps_per_second": 0.074,
|
| 409 |
+
"step": 40
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 41.0,
|
| 413 |
+
"eval_bleu": 10.2925,
|
| 414 |
+
"eval_gen_len": 82.6667,
|
| 415 |
+
"eval_loss": 4.323220252990723,
|
| 416 |
+
"eval_runtime": 13.3207,
|
| 417 |
+
"eval_samples_per_second": 0.225,
|
| 418 |
+
"eval_steps_per_second": 0.075,
|
| 419 |
+
"step": 41
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 42.0,
|
| 423 |
+
"eval_bleu": 11.2059,
|
| 424 |
+
"eval_gen_len": 82.3333,
|
| 425 |
+
"eval_loss": 4.325761795043945,
|
| 426 |
+
"eval_runtime": 16.8934,
|
| 427 |
+
"eval_samples_per_second": 0.178,
|
| 428 |
+
"eval_steps_per_second": 0.059,
|
| 429 |
+
"step": 42
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 43.0,
|
| 433 |
+
"eval_bleu": 11.1664,
|
| 434 |
+
"eval_gen_len": 85.3333,
|
| 435 |
+
"eval_loss": 4.327906131744385,
|
| 436 |
+
"eval_runtime": 14.4307,
|
| 437 |
+
"eval_samples_per_second": 0.208,
|
| 438 |
+
"eval_steps_per_second": 0.069,
|
| 439 |
+
"step": 43
|
| 440 |
+
},
|
| 441 |
+
{
|
| 442 |
+
"epoch": 44.0,
|
| 443 |
+
"eval_bleu": 11.2609,
|
| 444 |
+
"eval_gen_len": 81.6667,
|
| 445 |
+
"eval_loss": 4.329981803894043,
|
| 446 |
+
"eval_runtime": 14.847,
|
| 447 |
+
"eval_samples_per_second": 0.202,
|
| 448 |
+
"eval_steps_per_second": 0.067,
|
| 449 |
+
"step": 44
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 45.0,
|
| 453 |
+
"eval_bleu": 11.2609,
|
| 454 |
+
"eval_gen_len": 81.6667,
|
| 455 |
+
"eval_loss": 4.332195281982422,
|
| 456 |
+
"eval_runtime": 15.2747,
|
| 457 |
+
"eval_samples_per_second": 0.196,
|
| 458 |
+
"eval_steps_per_second": 0.065,
|
| 459 |
+
"step": 45
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 46.0,
|
| 463 |
+
"eval_bleu": 11.0203,
|
| 464 |
+
"eval_gen_len": 90.3333,
|
| 465 |
+
"eval_loss": 4.3332905769348145,
|
| 466 |
+
"eval_runtime": 15.7496,
|
| 467 |
+
"eval_samples_per_second": 0.19,
|
| 468 |
+
"eval_steps_per_second": 0.063,
|
| 469 |
+
"step": 46
|
| 470 |
+
},
|
| 471 |
+
{
|
| 472 |
+
"epoch": 47.0,
|
| 473 |
+
"eval_bleu": 11.1664,
|
| 474 |
+
"eval_gen_len": 85.3333,
|
| 475 |
+
"eval_loss": 4.3369364738464355,
|
| 476 |
+
"eval_runtime": 14.8004,
|
| 477 |
+
"eval_samples_per_second": 0.203,
|
| 478 |
+
"eval_steps_per_second": 0.068,
|
| 479 |
+
"step": 47
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 48.0,
|
| 483 |
+
"eval_bleu": 11.3295,
|
| 484 |
+
"eval_gen_len": 82.0,
|
| 485 |
+
"eval_loss": 4.340444564819336,
|
| 486 |
+
"eval_runtime": 13.1824,
|
| 487 |
+
"eval_samples_per_second": 0.228,
|
| 488 |
+
"eval_steps_per_second": 0.076,
|
| 489 |
+
"step": 48
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 49.0,
|
| 493 |
+
"eval_bleu": 11.1364,
|
| 494 |
+
"eval_gen_len": 68.3333,
|
| 495 |
+
"eval_loss": 4.343842506408691,
|
| 496 |
+
"eval_runtime": 12.8795,
|
| 497 |
+
"eval_samples_per_second": 0.233,
|
| 498 |
+
"eval_steps_per_second": 0.078,
|
| 499 |
+
"step": 49
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 50.0,
|
| 503 |
+
"eval_bleu": 11.1364,
|
| 504 |
+
"eval_gen_len": 68.3333,
|
| 505 |
+
"eval_loss": 4.347080230712891,
|
| 506 |
+
"eval_runtime": 13.1248,
|
| 507 |
+
"eval_samples_per_second": 0.229,
|
| 508 |
+
"eval_steps_per_second": 0.076,
|
| 509 |
+
"step": 50
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 51.0,
|
| 513 |
+
"eval_bleu": 11.1208,
|
| 514 |
+
"eval_gen_len": 65.6667,
|
| 515 |
+
"eval_loss": 4.349867820739746,
|
| 516 |
+
"eval_runtime": 11.6147,
|
| 517 |
+
"eval_samples_per_second": 0.258,
|
| 518 |
+
"eval_steps_per_second": 0.086,
|
| 519 |
+
"step": 51
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 52.0,
|
| 523 |
+
"eval_bleu": 11.1208,
|
| 524 |
+
"eval_gen_len": 65.6667,
|
| 525 |
+
"eval_loss": 4.352280616760254,
|
| 526 |
+
"eval_runtime": 11.3187,
|
| 527 |
+
"eval_samples_per_second": 0.265,
|
| 528 |
+
"eval_steps_per_second": 0.088,
|
| 529 |
+
"step": 52
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 53.0,
|
| 533 |
+
"eval_bleu": 11.1208,
|
| 534 |
+
"eval_gen_len": 65.6667,
|
| 535 |
+
"eval_loss": 4.353754997253418,
|
| 536 |
+
"eval_runtime": 12.1023,
|
| 537 |
+
"eval_samples_per_second": 0.248,
|
| 538 |
+
"eval_steps_per_second": 0.083,
|
| 539 |
+
"step": 53
|
| 540 |
+
},
|
| 541 |
+
{
|
| 542 |
+
"epoch": 54.0,
|
| 543 |
+
"eval_bleu": 10.9215,
|
| 544 |
+
"eval_gen_len": 66.0,
|
| 545 |
+
"eval_loss": 4.355079174041748,
|
| 546 |
+
"eval_runtime": 10.6524,
|
| 547 |
+
"eval_samples_per_second": 0.282,
|
| 548 |
+
"eval_steps_per_second": 0.094,
|
| 549 |
+
"step": 54
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 55.0,
|
| 553 |
+
"eval_bleu": 11.0309,
|
| 554 |
+
"eval_gen_len": 72.0,
|
| 555 |
+
"eval_loss": 4.356247901916504,
|
| 556 |
+
"eval_runtime": 14.1128,
|
| 557 |
+
"eval_samples_per_second": 0.213,
|
| 558 |
+
"eval_steps_per_second": 0.071,
|
| 559 |
+
"step": 55
|
| 560 |
+
},
|
| 561 |
+
{
|
| 562 |
+
"epoch": 56.0,
|
| 563 |
+
"eval_bleu": 11.0309,
|
| 564 |
+
"eval_gen_len": 72.0,
|
| 565 |
+
"eval_loss": 4.357237815856934,
|
| 566 |
+
"eval_runtime": 14.0892,
|
| 567 |
+
"eval_samples_per_second": 0.213,
|
| 568 |
+
"eval_steps_per_second": 0.071,
|
| 569 |
+
"step": 56
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 57.0,
|
| 573 |
+
"eval_bleu": 10.678,
|
| 574 |
+
"eval_gen_len": 71.3333,
|
| 575 |
+
"eval_loss": 4.358083248138428,
|
| 576 |
+
"eval_runtime": 12.7806,
|
| 577 |
+
"eval_samples_per_second": 0.235,
|
| 578 |
+
"eval_steps_per_second": 0.078,
|
| 579 |
+
"step": 57
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 58.0,
|
| 583 |
+
"eval_bleu": 10.678,
|
| 584 |
+
"eval_gen_len": 71.3333,
|
| 585 |
+
"eval_loss": 4.358962059020996,
|
| 586 |
+
"eval_runtime": 12.8538,
|
| 587 |
+
"eval_samples_per_second": 0.233,
|
| 588 |
+
"eval_steps_per_second": 0.078,
|
| 589 |
+
"step": 58
|
| 590 |
+
},
|
| 591 |
+
{
|
| 592 |
+
"epoch": 59.0,
|
| 593 |
+
"eval_bleu": 10.678,
|
| 594 |
+
"eval_gen_len": 71.3333,
|
| 595 |
+
"eval_loss": 4.359888076782227,
|
| 596 |
+
"eval_runtime": 12.3355,
|
| 597 |
+
"eval_samples_per_second": 0.243,
|
| 598 |
+
"eval_steps_per_second": 0.081,
|
| 599 |
+
"step": 59
|
| 600 |
+
},
|
| 601 |
+
{
|
| 602 |
+
"epoch": 60.0,
|
| 603 |
+
"eval_bleu": 10.678,
|
| 604 |
+
"eval_gen_len": 71.3333,
|
| 605 |
+
"eval_loss": 4.360787391662598,
|
| 606 |
+
"eval_runtime": 12.7009,
|
| 607 |
+
"eval_samples_per_second": 0.236,
|
| 608 |
+
"eval_steps_per_second": 0.079,
|
| 609 |
+
"step": 60
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 61.0,
|
| 613 |
+
"eval_bleu": 10.4416,
|
| 614 |
+
"eval_gen_len": 86.3333,
|
| 615 |
+
"eval_loss": 4.36198091506958,
|
| 616 |
+
"eval_runtime": 12.4279,
|
| 617 |
+
"eval_samples_per_second": 0.241,
|
| 618 |
+
"eval_steps_per_second": 0.08,
|
| 619 |
+
"step": 61
|
| 620 |
+
},
|
| 621 |
+
{
|
| 622 |
+
"epoch": 62.0,
|
| 623 |
+
"eval_bleu": 10.678,
|
| 624 |
+
"eval_gen_len": 71.3333,
|
| 625 |
+
"eval_loss": 4.362994194030762,
|
| 626 |
+
"eval_runtime": 12.6907,
|
| 627 |
+
"eval_samples_per_second": 0.236,
|
| 628 |
+
"eval_steps_per_second": 0.079,
|
| 629 |
+
"step": 62
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 63.0,
|
| 633 |
+
"eval_bleu": 10.678,
|
| 634 |
+
"eval_gen_len": 71.3333,
|
| 635 |
+
"eval_loss": 4.366813659667969,
|
| 636 |
+
"eval_runtime": 12.6482,
|
| 637 |
+
"eval_samples_per_second": 0.237,
|
| 638 |
+
"eval_steps_per_second": 0.079,
|
| 639 |
+
"step": 63
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 64.0,
|
| 643 |
+
"eval_bleu": 11.1601,
|
| 644 |
+
"eval_gen_len": 74.6667,
|
| 645 |
+
"eval_loss": 4.370214462280273,
|
| 646 |
+
"eval_runtime": 11.7414,
|
| 647 |
+
"eval_samples_per_second": 0.256,
|
| 648 |
+
"eval_steps_per_second": 0.085,
|
| 649 |
+
"step": 64
|
| 650 |
+
},
|
| 651 |
+
{
|
| 652 |
+
"epoch": 65.0,
|
| 653 |
+
"eval_bleu": 11.137,
|
| 654 |
+
"eval_gen_len": 78.0,
|
| 655 |
+
"eval_loss": 4.37328577041626,
|
| 656 |
+
"eval_runtime": 13.9564,
|
| 657 |
+
"eval_samples_per_second": 0.215,
|
| 658 |
+
"eval_steps_per_second": 0.072,
|
| 659 |
+
"step": 65
|
| 660 |
+
},
|
| 661 |
+
{
|
| 662 |
+
"epoch": 66.0,
|
| 663 |
+
"eval_bleu": 11.137,
|
| 664 |
+
"eval_gen_len": 78.0,
|
| 665 |
+
"eval_loss": 4.3749165534973145,
|
| 666 |
+
"eval_runtime": 13.8294,
|
| 667 |
+
"eval_samples_per_second": 0.217,
|
| 668 |
+
"eval_steps_per_second": 0.072,
|
| 669 |
+
"step": 66
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 67.0,
|
| 673 |
+
"eval_bleu": 11.137,
|
| 674 |
+
"eval_gen_len": 78.0,
|
| 675 |
+
"eval_loss": 4.3767194747924805,
|
| 676 |
+
"eval_runtime": 13.7066,
|
| 677 |
+
"eval_samples_per_second": 0.219,
|
| 678 |
+
"eval_steps_per_second": 0.073,
|
| 679 |
+
"step": 67
|
| 680 |
+
},
|
| 681 |
+
{
|
| 682 |
+
"epoch": 68.0,
|
| 683 |
+
"eval_bleu": 10.8916,
|
| 684 |
+
"eval_gen_len": 88.0,
|
| 685 |
+
"eval_loss": 4.3782854080200195,
|
| 686 |
+
"eval_runtime": 16.3165,
|
| 687 |
+
"eval_samples_per_second": 0.184,
|
| 688 |
+
"eval_steps_per_second": 0.061,
|
| 689 |
+
"step": 68
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 69.0,
|
| 693 |
+
"eval_bleu": 10.7733,
|
| 694 |
+
"eval_gen_len": 67.3333,
|
| 695 |
+
"eval_loss": 4.379751682281494,
|
| 696 |
+
"eval_runtime": 11.7272,
|
| 697 |
+
"eval_samples_per_second": 0.256,
|
| 698 |
+
"eval_steps_per_second": 0.085,
|
| 699 |
+
"step": 69
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 70.0,
|
| 703 |
+
"eval_bleu": 11.5509,
|
| 704 |
+
"eval_gen_len": 66.3333,
|
| 705 |
+
"eval_loss": 4.381042957305908,
|
| 706 |
+
"eval_runtime": 11.8594,
|
| 707 |
+
"eval_samples_per_second": 0.253,
|
| 708 |
+
"eval_steps_per_second": 0.084,
|
| 709 |
+
"step": 70
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"epoch": 71.0,
|
| 713 |
+
"eval_bleu": 11.4262,
|
| 714 |
+
"eval_gen_len": 70.0,
|
| 715 |
+
"eval_loss": 4.382328510284424,
|
| 716 |
+
"eval_runtime": 13.0297,
|
| 717 |
+
"eval_samples_per_second": 0.23,
|
| 718 |
+
"eval_steps_per_second": 0.077,
|
| 719 |
+
"step": 71
|
| 720 |
+
},
|
| 721 |
+
{
|
| 722 |
+
"epoch": 72.0,
|
| 723 |
+
"eval_bleu": 11.4262,
|
| 724 |
+
"eval_gen_len": 70.0,
|
| 725 |
+
"eval_loss": 4.383648872375488,
|
| 726 |
+
"eval_runtime": 14.0598,
|
| 727 |
+
"eval_samples_per_second": 0.213,
|
| 728 |
+
"eval_steps_per_second": 0.071,
|
| 729 |
+
"step": 72
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 73.0,
|
| 733 |
+
"eval_bleu": 11.195,
|
| 734 |
+
"eval_gen_len": 66.3333,
|
| 735 |
+
"eval_loss": 4.384763717651367,
|
| 736 |
+
"eval_runtime": 11.4527,
|
| 737 |
+
"eval_samples_per_second": 0.262,
|
| 738 |
+
"eval_steps_per_second": 0.087,
|
| 739 |
+
"step": 73
|
| 740 |
+
},
|
| 741 |
+
{
|
| 742 |
+
"epoch": 74.0,
|
| 743 |
+
"eval_bleu": 11.1115,
|
| 744 |
+
"eval_gen_len": 67.0,
|
| 745 |
+
"eval_loss": 4.384941577911377,
|
| 746 |
+
"eval_runtime": 12.3777,
|
| 747 |
+
"eval_samples_per_second": 0.242,
|
| 748 |
+
"eval_steps_per_second": 0.081,
|
| 749 |
+
"step": 74
|
| 750 |
+
},
|
| 751 |
+
{
|
| 752 |
+
"epoch": 75.0,
|
| 753 |
+
"eval_bleu": 11.2741,
|
| 754 |
+
"eval_gen_len": 67.6667,
|
| 755 |
+
"eval_loss": 4.385134696960449,
|
| 756 |
+
"eval_runtime": 12.3325,
|
| 757 |
+
"eval_samples_per_second": 0.243,
|
| 758 |
+
"eval_steps_per_second": 0.081,
|
| 759 |
+
"step": 75
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 76.0,
|
| 763 |
+
"eval_bleu": 11.6137,
|
| 764 |
+
"eval_gen_len": 71.0,
|
| 765 |
+
"eval_loss": 4.38496208190918,
|
| 766 |
+
"eval_runtime": 12.6988,
|
| 767 |
+
"eval_samples_per_second": 0.236,
|
| 768 |
+
"eval_steps_per_second": 0.079,
|
| 769 |
+
"step": 76
|
| 770 |
+
},
|
| 771 |
+
{
|
| 772 |
+
"epoch": 77.0,
|
| 773 |
+
"eval_bleu": 11.2741,
|
| 774 |
+
"eval_gen_len": 67.6667,
|
| 775 |
+
"eval_loss": 4.3848419189453125,
|
| 776 |
+
"eval_runtime": 11.4524,
|
| 777 |
+
"eval_samples_per_second": 0.262,
|
| 778 |
+
"eval_steps_per_second": 0.087,
|
| 779 |
+
"step": 77
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 78.0,
|
| 783 |
+
"eval_bleu": 11.1454,
|
| 784 |
+
"eval_gen_len": 62.0,
|
| 785 |
+
"eval_loss": 4.38457727432251,
|
| 786 |
+
"eval_runtime": 10.4375,
|
| 787 |
+
"eval_samples_per_second": 0.287,
|
| 788 |
+
"eval_steps_per_second": 0.096,
|
| 789 |
+
"step": 78
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 79.0,
|
| 793 |
+
"eval_bleu": 11.1454,
|
| 794 |
+
"eval_gen_len": 62.0,
|
| 795 |
+
"eval_loss": 4.3842973709106445,
|
| 796 |
+
"eval_runtime": 10.4146,
|
| 797 |
+
"eval_samples_per_second": 0.288,
|
| 798 |
+
"eval_steps_per_second": 0.096,
|
| 799 |
+
"step": 79
|
| 800 |
+
},
|
| 801 |
+
{
|
| 802 |
+
"epoch": 80.0,
|
| 803 |
+
"eval_bleu": 11.267,
|
| 804 |
+
"eval_gen_len": 68.6667,
|
| 805 |
+
"eval_loss": 4.383995532989502,
|
| 806 |
+
"eval_runtime": 12.6896,
|
| 807 |
+
"eval_samples_per_second": 0.236,
|
| 808 |
+
"eval_steps_per_second": 0.079,
|
| 809 |
+
"step": 80
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 81.0,
|
| 813 |
+
"eval_bleu": 11.267,
|
| 814 |
+
"eval_gen_len": 68.6667,
|
| 815 |
+
"eval_loss": 4.383782386779785,
|
| 816 |
+
"eval_runtime": 12.4982,
|
| 817 |
+
"eval_samples_per_second": 0.24,
|
| 818 |
+
"eval_steps_per_second": 0.08,
|
| 819 |
+
"step": 81
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 82.0,
|
| 823 |
+
"eval_bleu": 11.267,
|
| 824 |
+
"eval_gen_len": 68.6667,
|
| 825 |
+
"eval_loss": 4.383605480194092,
|
| 826 |
+
"eval_runtime": 12.2559,
|
| 827 |
+
"eval_samples_per_second": 0.245,
|
| 828 |
+
"eval_steps_per_second": 0.082,
|
| 829 |
+
"step": 82
|
| 830 |
+
},
|
| 831 |
+
{
|
| 832 |
+
"epoch": 83.0,
|
| 833 |
+
"eval_bleu": 11.267,
|
| 834 |
+
"eval_gen_len": 68.6667,
|
| 835 |
+
"eval_loss": 4.383455753326416,
|
| 836 |
+
"eval_runtime": 12.7951,
|
| 837 |
+
"eval_samples_per_second": 0.234,
|
| 838 |
+
"eval_steps_per_second": 0.078,
|
| 839 |
+
"step": 83
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 84.0,
|
| 843 |
+
"eval_bleu": 11.267,
|
| 844 |
+
"eval_gen_len": 68.6667,
|
| 845 |
+
"eval_loss": 4.38338565826416,
|
| 846 |
+
"eval_runtime": 12.5193,
|
| 847 |
+
"eval_samples_per_second": 0.24,
|
| 848 |
+
"eval_steps_per_second": 0.08,
|
| 849 |
+
"step": 84
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 85.0,
|
| 853 |
+
"eval_bleu": 11.4092,
|
| 854 |
+
"eval_gen_len": 69.0,
|
| 855 |
+
"eval_loss": 4.383331298828125,
|
| 856 |
+
"eval_runtime": 12.8027,
|
| 857 |
+
"eval_samples_per_second": 0.234,
|
| 858 |
+
"eval_steps_per_second": 0.078,
|
| 859 |
+
"step": 85
|
| 860 |
+
},
|
| 861 |
+
{
|
| 862 |
+
"epoch": 86.0,
|
| 863 |
+
"eval_bleu": 11.4092,
|
| 864 |
+
"eval_gen_len": 69.0,
|
| 865 |
+
"eval_loss": 4.38341760635376,
|
| 866 |
+
"eval_runtime": 12.849,
|
| 867 |
+
"eval_samples_per_second": 0.233,
|
| 868 |
+
"eval_steps_per_second": 0.078,
|
| 869 |
+
"step": 86
|
| 870 |
+
},
|
| 871 |
+
{
|
| 872 |
+
"epoch": 87.0,
|
| 873 |
+
"eval_bleu": 11.4092,
|
| 874 |
+
"eval_gen_len": 69.0,
|
| 875 |
+
"eval_loss": 4.383518695831299,
|
| 876 |
+
"eval_runtime": 12.3575,
|
| 877 |
+
"eval_samples_per_second": 0.243,
|
| 878 |
+
"eval_steps_per_second": 0.081,
|
| 879 |
+
"step": 87
|
| 880 |
+
},
|
| 881 |
+
{
|
| 882 |
+
"epoch": 88.0,
|
| 883 |
+
"eval_bleu": 11.4092,
|
| 884 |
+
"eval_gen_len": 69.0,
|
| 885 |
+
"eval_loss": 4.3836445808410645,
|
| 886 |
+
"eval_runtime": 12.8174,
|
| 887 |
+
"eval_samples_per_second": 0.234,
|
| 888 |
+
"eval_steps_per_second": 0.078,
|
| 889 |
+
"step": 88
|
| 890 |
+
},
|
| 891 |
+
{
|
| 892 |
+
"epoch": 89.0,
|
| 893 |
+
"eval_bleu": 11.4092,
|
| 894 |
+
"eval_gen_len": 69.0,
|
| 895 |
+
"eval_loss": 4.383749961853027,
|
| 896 |
+
"eval_runtime": 12.8695,
|
| 897 |
+
"eval_samples_per_second": 0.233,
|
| 898 |
+
"eval_steps_per_second": 0.078,
|
| 899 |
+
"step": 89
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 90.0,
|
| 903 |
+
"eval_bleu": 11.4092,
|
| 904 |
+
"eval_gen_len": 69.0,
|
| 905 |
+
"eval_loss": 4.383831977844238,
|
| 906 |
+
"eval_runtime": 12.8527,
|
| 907 |
+
"eval_samples_per_second": 0.233,
|
| 908 |
+
"eval_steps_per_second": 0.078,
|
| 909 |
+
"step": 90
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 91.0,
|
| 913 |
+
"eval_bleu": 11.4092,
|
| 914 |
+
"eval_gen_len": 69.0,
|
| 915 |
+
"eval_loss": 4.383894920349121,
|
| 916 |
+
"eval_runtime": 12.7272,
|
| 917 |
+
"eval_samples_per_second": 0.236,
|
| 918 |
+
"eval_steps_per_second": 0.079,
|
| 919 |
+
"step": 91
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 92.0,
|
| 923 |
+
"eval_bleu": 11.4092,
|
| 924 |
+
"eval_gen_len": 69.0,
|
| 925 |
+
"eval_loss": 4.383920669555664,
|
| 926 |
+
"eval_runtime": 14.3034,
|
| 927 |
+
"eval_samples_per_second": 0.21,
|
| 928 |
+
"eval_steps_per_second": 0.07,
|
| 929 |
+
"step": 92
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 93.0,
|
| 933 |
+
"eval_bleu": 11.4092,
|
| 934 |
+
"eval_gen_len": 69.0,
|
| 935 |
+
"eval_loss": 4.38393497467041,
|
| 936 |
+
"eval_runtime": 12.4454,
|
| 937 |
+
"eval_samples_per_second": 0.241,
|
| 938 |
+
"eval_steps_per_second": 0.08,
|
| 939 |
+
"step": 93
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 94.0,
|
| 943 |
+
"eval_bleu": 11.4092,
|
| 944 |
+
"eval_gen_len": 69.0,
|
| 945 |
+
"eval_loss": 4.384005546569824,
|
| 946 |
+
"eval_runtime": 12.8192,
|
| 947 |
+
"eval_samples_per_second": 0.234,
|
| 948 |
+
"eval_steps_per_second": 0.078,
|
| 949 |
+
"step": 94
|
| 950 |
+
},
|
| 951 |
+
{
|
| 952 |
+
"epoch": 95.0,
|
| 953 |
+
"eval_bleu": 11.4092,
|
| 954 |
+
"eval_gen_len": 69.0,
|
| 955 |
+
"eval_loss": 4.3840460777282715,
|
| 956 |
+
"eval_runtime": 12.4402,
|
| 957 |
+
"eval_samples_per_second": 0.241,
|
| 958 |
+
"eval_steps_per_second": 0.08,
|
| 959 |
+
"step": 95
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 96.0,
|
| 963 |
+
"eval_bleu": 10.7891,
|
| 964 |
+
"eval_gen_len": 70.6667,
|
| 965 |
+
"eval_loss": 4.384109020233154,
|
| 966 |
+
"eval_runtime": 12.8195,
|
| 967 |
+
"eval_samples_per_second": 0.234,
|
| 968 |
+
"eval_steps_per_second": 0.078,
|
| 969 |
+
"step": 96
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 97.0,
|
| 973 |
+
"eval_bleu": 10.7891,
|
| 974 |
+
"eval_gen_len": 70.6667,
|
| 975 |
+
"eval_loss": 4.384162902832031,
|
| 976 |
+
"eval_runtime": 12.3694,
|
| 977 |
+
"eval_samples_per_second": 0.243,
|
| 978 |
+
"eval_steps_per_second": 0.081,
|
| 979 |
+
"step": 97
|
| 980 |
+
},
|
| 981 |
+
{
|
| 982 |
+
"epoch": 98.0,
|
| 983 |
+
"eval_bleu": 10.7891,
|
| 984 |
+
"eval_gen_len": 70.6667,
|
| 985 |
+
"eval_loss": 4.384200096130371,
|
| 986 |
+
"eval_runtime": 12.8132,
|
| 987 |
+
"eval_samples_per_second": 0.234,
|
| 988 |
+
"eval_steps_per_second": 0.078,
|
| 989 |
+
"step": 98
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 99.0,
|
| 993 |
+
"eval_bleu": 10.7891,
|
| 994 |
+
"eval_gen_len": 70.6667,
|
| 995 |
+
"eval_loss": 4.384230613708496,
|
| 996 |
+
"eval_runtime": 12.6103,
|
| 997 |
+
"eval_samples_per_second": 0.238,
|
| 998 |
+
"eval_steps_per_second": 0.079,
|
| 999 |
+
"step": 99
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 100.0,
|
| 1003 |
+
"eval_bleu": 10.7891,
|
| 1004 |
+
"eval_gen_len": 70.6667,
|
| 1005 |
+
"eval_loss": 4.384246349334717,
|
| 1006 |
+
"eval_runtime": 12.8394,
|
| 1007 |
+
"eval_samples_per_second": 0.234,
|
| 1008 |
+
"eval_steps_per_second": 0.078,
|
| 1009 |
+
"step": 100
|
| 1010 |
+
}
|
| 1011 |
+
],
|
| 1012 |
+
"logging_steps": 500,
|
| 1013 |
+
"max_steps": 100,
|
| 1014 |
+
"num_input_tokens_seen": 0,
|
| 1015 |
+
"num_train_epochs": 100,
|
| 1016 |
+
"save_steps": 500,
|
| 1017 |
+
"stateful_callbacks": {
|
| 1018 |
+
"TrainerControl": {
|
| 1019 |
+
"args": {
|
| 1020 |
+
"should_epoch_stop": false,
|
| 1021 |
+
"should_evaluate": false,
|
| 1022 |
+
"should_log": false,
|
| 1023 |
+
"should_save": true,
|
| 1024 |
+
"should_training_stop": true
|
| 1025 |
+
},
|
| 1026 |
+
"attributes": {}
|
| 1027 |
+
}
|
| 1028 |
+
},
|
| 1029 |
+
"total_flos": 7786030694400.0,
|
| 1030 |
+
"train_batch_size": 16,
|
| 1031 |
+
"trial_name": null,
|
| 1032 |
+
"trial_params": null
|
| 1033 |
+
}
|
checkpoint-100/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bf14fcd93191fa971de1127e1842c7d6d380c54dce1ad0fe115f24bc90e7b529
|
| 3 |
+
size 5368
|
checkpoint-2/config.json
ADDED
|
@@ -0,0 +1,56 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "Helsinki-NLP/opus-mt-zh-en",
|
| 3 |
+
"activation_dropout": 0.0,
|
| 4 |
+
"activation_function": "swish",
|
| 5 |
+
"add_bias_logits": false,
|
| 6 |
+
"add_final_layer_norm": false,
|
| 7 |
+
"architectures": [
|
| 8 |
+
"MarianMTModel"
|
| 9 |
+
],
|
| 10 |
+
"attention_dropout": 0.0,
|
| 11 |
+
"bos_token_id": 0,
|
| 12 |
+
"classif_dropout": 0.0,
|
| 13 |
+
"classifier_dropout": 0.0,
|
| 14 |
+
"d_model": 512,
|
| 15 |
+
"decoder_attention_heads": 8,
|
| 16 |
+
"decoder_ffn_dim": 2048,
|
| 17 |
+
"decoder_layerdrop": 0.0,
|
| 18 |
+
"decoder_layers": 6,
|
| 19 |
+
"decoder_start_token_id": 65000,
|
| 20 |
+
"decoder_vocab_size": 65001,
|
| 21 |
+
"dropout": 0.1,
|
| 22 |
+
"encoder_attention_heads": 8,
|
| 23 |
+
"encoder_ffn_dim": 2048,
|
| 24 |
+
"encoder_layerdrop": 0.0,
|
| 25 |
+
"encoder_layers": 6,
|
| 26 |
+
"eos_token_id": 0,
|
| 27 |
+
"extra_pos_embeddings": 65001,
|
| 28 |
+
"forced_eos_token_id": 0,
|
| 29 |
+
"id2label": {
|
| 30 |
+
"0": "LABEL_0",
|
| 31 |
+
"1": "LABEL_1",
|
| 32 |
+
"2": "LABEL_2"
|
| 33 |
+
},
|
| 34 |
+
"init_std": 0.02,
|
| 35 |
+
"is_encoder_decoder": true,
|
| 36 |
+
"label2id": {
|
| 37 |
+
"LABEL_0": 0,
|
| 38 |
+
"LABEL_1": 1,
|
| 39 |
+
"LABEL_2": 2
|
| 40 |
+
},
|
| 41 |
+
"max_length": null,
|
| 42 |
+
"max_position_embeddings": 512,
|
| 43 |
+
"model_type": "marian",
|
| 44 |
+
"normalize_before": false,
|
| 45 |
+
"normalize_embedding": false,
|
| 46 |
+
"num_beams": null,
|
| 47 |
+
"num_hidden_layers": 6,
|
| 48 |
+
"pad_token_id": 65000,
|
| 49 |
+
"scale_embedding": true,
|
| 50 |
+
"share_encoder_decoder_embeddings": true,
|
| 51 |
+
"static_position_embeddings": true,
|
| 52 |
+
"torch_dtype": "float32",
|
| 53 |
+
"transformers_version": "4.46.2",
|
| 54 |
+
"use_cache": true,
|
| 55 |
+
"vocab_size": 65001
|
| 56 |
+
}
|
checkpoint-2/generation_config.json
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bad_words_ids": [
|
| 3 |
+
[
|
| 4 |
+
65000
|
| 5 |
+
]
|
| 6 |
+
],
|
| 7 |
+
"bos_token_id": 0,
|
| 8 |
+
"decoder_start_token_id": 65000,
|
| 9 |
+
"eos_token_id": 0,
|
| 10 |
+
"forced_eos_token_id": 0,
|
| 11 |
+
"max_length": 512,
|
| 12 |
+
"num_beams": 6,
|
| 13 |
+
"pad_token_id": 65000,
|
| 14 |
+
"renormalize_logits": true,
|
| 15 |
+
"transformers_version": "4.46.2"
|
| 16 |
+
}
|
checkpoint-2/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:056c0879a45c67724d9458dbe29f7e80bc0512df9d78bcd528604342602a417e
|
| 3 |
+
size 309965092
|
checkpoint-2/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1087661f74ac45f79f9c59d85e316a57feb5bf3c54d4e412ef6f04e01495a8dd
|
| 3 |
+
size 619556538
|
checkpoint-2/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3044572c68786ace002a23f2b666810b1947f579bf5d8e46512ec0d4bddeb3e4
|
| 3 |
+
size 13990
|
checkpoint-2/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ec73cccea843cf0c1163762a6cbf1856824592516abc4ee6911210e86a93bdac
|
| 3 |
+
size 1064
|
checkpoint-2/trainer_state.json
ADDED
|
@@ -0,0 +1,53 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 3.4803478717803955,
|
| 3 |
+
"best_model_checkpoint": "opus-mt-zh-en-finetuned-zh-to-en/checkpoint-2",
|
| 4 |
+
"epoch": 2.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 2,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 1.0,
|
| 13 |
+
"eval_bleu": 5.6918,
|
| 14 |
+
"eval_gen_len": 57.3333,
|
| 15 |
+
"eval_loss": 3.5217063426971436,
|
| 16 |
+
"eval_runtime": 10.9763,
|
| 17 |
+
"eval_samples_per_second": 0.273,
|
| 18 |
+
"eval_steps_per_second": 0.091,
|
| 19 |
+
"step": 1
|
| 20 |
+
},
|
| 21 |
+
{
|
| 22 |
+
"epoch": 2.0,
|
| 23 |
+
"eval_bleu": 7.6304,
|
| 24 |
+
"eval_gen_len": 55.3333,
|
| 25 |
+
"eval_loss": 3.4803478717803955,
|
| 26 |
+
"eval_runtime": 8.3555,
|
| 27 |
+
"eval_samples_per_second": 0.359,
|
| 28 |
+
"eval_steps_per_second": 0.12,
|
| 29 |
+
"step": 2
|
| 30 |
+
}
|
| 31 |
+
],
|
| 32 |
+
"logging_steps": 500,
|
| 33 |
+
"max_steps": 100,
|
| 34 |
+
"num_input_tokens_seen": 0,
|
| 35 |
+
"num_train_epochs": 100,
|
| 36 |
+
"save_steps": 500,
|
| 37 |
+
"stateful_callbacks": {
|
| 38 |
+
"TrainerControl": {
|
| 39 |
+
"args": {
|
| 40 |
+
"should_epoch_stop": false,
|
| 41 |
+
"should_evaluate": false,
|
| 42 |
+
"should_log": false,
|
| 43 |
+
"should_save": true,
|
| 44 |
+
"should_training_stop": false
|
| 45 |
+
},
|
| 46 |
+
"attributes": {}
|
| 47 |
+
}
|
| 48 |
+
},
|
| 49 |
+
"total_flos": 155720613888.0,
|
| 50 |
+
"train_batch_size": 16,
|
| 51 |
+
"trial_name": null,
|
| 52 |
+
"trial_params": null
|
| 53 |
+
}
|
checkpoint-2/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bf14fcd93191fa971de1127e1842c7d6d380c54dce1ad0fe115f24bc90e7b529
|
| 3 |
+
size 5368
|
checkpoint-200/config.json
ADDED
|
@@ -0,0 +1,56 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "opus-mt-zh-en-finetuned-zh-to-en",
|
| 3 |
+
"activation_dropout": 0.0,
|
| 4 |
+
"activation_function": "swish",
|
| 5 |
+
"add_bias_logits": false,
|
| 6 |
+
"add_final_layer_norm": false,
|
| 7 |
+
"architectures": [
|
| 8 |
+
"MarianMTModel"
|
| 9 |
+
],
|
| 10 |
+
"attention_dropout": 0.0,
|
| 11 |
+
"bos_token_id": 0,
|
| 12 |
+
"classif_dropout": 0.0,
|
| 13 |
+
"classifier_dropout": 0.0,
|
| 14 |
+
"d_model": 512,
|
| 15 |
+
"decoder_attention_heads": 8,
|
| 16 |
+
"decoder_ffn_dim": 2048,
|
| 17 |
+
"decoder_layerdrop": 0.0,
|
| 18 |
+
"decoder_layers": 6,
|
| 19 |
+
"decoder_start_token_id": 65000,
|
| 20 |
+
"decoder_vocab_size": 65001,
|
| 21 |
+
"dropout": 0.1,
|
| 22 |
+
"encoder_attention_heads": 8,
|
| 23 |
+
"encoder_ffn_dim": 2048,
|
| 24 |
+
"encoder_layerdrop": 0.0,
|
| 25 |
+
"encoder_layers": 6,
|
| 26 |
+
"eos_token_id": 0,
|
| 27 |
+
"extra_pos_embeddings": 65001,
|
| 28 |
+
"forced_eos_token_id": 0,
|
| 29 |
+
"id2label": {
|
| 30 |
+
"0": "LABEL_0",
|
| 31 |
+
"1": "LABEL_1",
|
| 32 |
+
"2": "LABEL_2"
|
| 33 |
+
},
|
| 34 |
+
"init_std": 0.02,
|
| 35 |
+
"is_encoder_decoder": true,
|
| 36 |
+
"label2id": {
|
| 37 |
+
"LABEL_0": 0,
|
| 38 |
+
"LABEL_1": 1,
|
| 39 |
+
"LABEL_2": 2
|
| 40 |
+
},
|
| 41 |
+
"max_length": null,
|
| 42 |
+
"max_position_embeddings": 512,
|
| 43 |
+
"model_type": "marian",
|
| 44 |
+
"normalize_before": false,
|
| 45 |
+
"normalize_embedding": false,
|
| 46 |
+
"num_beams": null,
|
| 47 |
+
"num_hidden_layers": 6,
|
| 48 |
+
"pad_token_id": 65000,
|
| 49 |
+
"scale_embedding": true,
|
| 50 |
+
"share_encoder_decoder_embeddings": true,
|
| 51 |
+
"static_position_embeddings": true,
|
| 52 |
+
"torch_dtype": "float32",
|
| 53 |
+
"transformers_version": "4.46.2",
|
| 54 |
+
"use_cache": true,
|
| 55 |
+
"vocab_size": 65001
|
| 56 |
+
}
|
checkpoint-200/generation_config.json
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bad_words_ids": [
|
| 3 |
+
[
|
| 4 |
+
65000
|
| 5 |
+
]
|
| 6 |
+
],
|
| 7 |
+
"bos_token_id": 0,
|
| 8 |
+
"decoder_start_token_id": 65000,
|
| 9 |
+
"eos_token_id": 0,
|
| 10 |
+
"forced_eos_token_id": 0,
|
| 11 |
+
"max_length": 512,
|
| 12 |
+
"num_beams": 6,
|
| 13 |
+
"pad_token_id": 65000,
|
| 14 |
+
"renormalize_logits": true,
|
| 15 |
+
"transformers_version": "4.46.2"
|
| 16 |
+
}
|
checkpoint-200/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:79c37aab6fda592ede95efaba9c896b7b424148afc87f370bd499262e7c1f1d6
|
| 3 |
+
size 309965092
|
checkpoint-200/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:63681206b5d5fd68d24c0e21127e7eb7cdc8641ee413d8855664d6f4788ef645
|
| 3 |
+
size 619556538
|
checkpoint-200/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:83b1c45d5758c23ecf238dcf0c9a2084623357e4678068ae8468d47768a18f71
|
| 3 |
+
size 13990
|
checkpoint-200/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fb148ad6cd6502048dc3eee274a6708e5b4f6b72c0d56ebc272935446c220acd
|
| 3 |
+
size 1064
|
checkpoint-200/trainer_state.json
ADDED
|
@@ -0,0 +1,2023 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 200.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 200,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 1.0,
|
| 13 |
+
"eval_bleu": 7.1535,
|
| 14 |
+
"eval_gen_len": 34.0,
|
| 15 |
+
"eval_loss": 2.992987632751465,
|
| 16 |
+
"eval_runtime": 5.9633,
|
| 17 |
+
"eval_samples_per_second": 0.503,
|
| 18 |
+
"eval_steps_per_second": 0.168,
|
| 19 |
+
"step": 1
|
| 20 |
+
},
|
| 21 |
+
{
|
| 22 |
+
"epoch": 2.0,
|
| 23 |
+
"eval_bleu": 7.1535,
|
| 24 |
+
"eval_gen_len": 34.0,
|
| 25 |
+
"eval_loss": 3.007624864578247,
|
| 26 |
+
"eval_runtime": 7.1487,
|
| 27 |
+
"eval_samples_per_second": 0.42,
|
| 28 |
+
"eval_steps_per_second": 0.14,
|
| 29 |
+
"step": 2
|
| 30 |
+
},
|
| 31 |
+
{
|
| 32 |
+
"epoch": 3.0,
|
| 33 |
+
"eval_bleu": 7.0446,
|
| 34 |
+
"eval_gen_len": 33.6667,
|
| 35 |
+
"eval_loss": 3.0191893577575684,
|
| 36 |
+
"eval_runtime": 7.6034,
|
| 37 |
+
"eval_samples_per_second": 0.395,
|
| 38 |
+
"eval_steps_per_second": 0.132,
|
| 39 |
+
"step": 3
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 4.0,
|
| 43 |
+
"eval_bleu": 7.0446,
|
| 44 |
+
"eval_gen_len": 33.6667,
|
| 45 |
+
"eval_loss": 3.0266923904418945,
|
| 46 |
+
"eval_runtime": 6.2244,
|
| 47 |
+
"eval_samples_per_second": 0.482,
|
| 48 |
+
"eval_steps_per_second": 0.161,
|
| 49 |
+
"step": 4
|
| 50 |
+
},
|
| 51 |
+
{
|
| 52 |
+
"epoch": 5.0,
|
| 53 |
+
"eval_bleu": 7.0446,
|
| 54 |
+
"eval_gen_len": 33.6667,
|
| 55 |
+
"eval_loss": 3.037548542022705,
|
| 56 |
+
"eval_runtime": 5.9999,
|
| 57 |
+
"eval_samples_per_second": 0.5,
|
| 58 |
+
"eval_steps_per_second": 0.167,
|
| 59 |
+
"step": 5
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 6.0,
|
| 63 |
+
"eval_bleu": 7.0446,
|
| 64 |
+
"eval_gen_len": 33.6667,
|
| 65 |
+
"eval_loss": 3.046501398086548,
|
| 66 |
+
"eval_runtime": 6.5759,
|
| 67 |
+
"eval_samples_per_second": 0.456,
|
| 68 |
+
"eval_steps_per_second": 0.152,
|
| 69 |
+
"step": 6
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 7.0,
|
| 73 |
+
"eval_bleu": 7.0446,
|
| 74 |
+
"eval_gen_len": 33.6667,
|
| 75 |
+
"eval_loss": 3.055983781814575,
|
| 76 |
+
"eval_runtime": 7.1686,
|
| 77 |
+
"eval_samples_per_second": 0.418,
|
| 78 |
+
"eval_steps_per_second": 0.139,
|
| 79 |
+
"step": 7
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 8.0,
|
| 83 |
+
"eval_bleu": 7.0491,
|
| 84 |
+
"eval_gen_len": 33.3333,
|
| 85 |
+
"eval_loss": 3.0670735836029053,
|
| 86 |
+
"eval_runtime": 6.2316,
|
| 87 |
+
"eval_samples_per_second": 0.481,
|
| 88 |
+
"eval_steps_per_second": 0.16,
|
| 89 |
+
"step": 8
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"epoch": 9.0,
|
| 93 |
+
"eval_bleu": 7.0446,
|
| 94 |
+
"eval_gen_len": 33.6667,
|
| 95 |
+
"eval_loss": 3.0760223865509033,
|
| 96 |
+
"eval_runtime": 5.9592,
|
| 97 |
+
"eval_samples_per_second": 0.503,
|
| 98 |
+
"eval_steps_per_second": 0.168,
|
| 99 |
+
"step": 9
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 10.0,
|
| 103 |
+
"eval_bleu": 8.3068,
|
| 104 |
+
"eval_gen_len": 33.3333,
|
| 105 |
+
"eval_loss": 3.086441993713379,
|
| 106 |
+
"eval_runtime": 6.5845,
|
| 107 |
+
"eval_samples_per_second": 0.456,
|
| 108 |
+
"eval_steps_per_second": 0.152,
|
| 109 |
+
"step": 10
|
| 110 |
+
},
|
| 111 |
+
{
|
| 112 |
+
"epoch": 11.0,
|
| 113 |
+
"eval_bleu": 6.8089,
|
| 114 |
+
"eval_gen_len": 33.6667,
|
| 115 |
+
"eval_loss": 3.090127468109131,
|
| 116 |
+
"eval_runtime": 8.4387,
|
| 117 |
+
"eval_samples_per_second": 0.356,
|
| 118 |
+
"eval_steps_per_second": 0.119,
|
| 119 |
+
"step": 11
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 12.0,
|
| 123 |
+
"eval_bleu": 6.4212,
|
| 124 |
+
"eval_gen_len": 30.6667,
|
| 125 |
+
"eval_loss": 3.091043472290039,
|
| 126 |
+
"eval_runtime": 5.076,
|
| 127 |
+
"eval_samples_per_second": 0.591,
|
| 128 |
+
"eval_steps_per_second": 0.197,
|
| 129 |
+
"step": 12
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 13.0,
|
| 133 |
+
"eval_bleu": 6.4212,
|
| 134 |
+
"eval_gen_len": 30.6667,
|
| 135 |
+
"eval_loss": 3.0892398357391357,
|
| 136 |
+
"eval_runtime": 5.1747,
|
| 137 |
+
"eval_samples_per_second": 0.58,
|
| 138 |
+
"eval_steps_per_second": 0.193,
|
| 139 |
+
"step": 13
|
| 140 |
+
},
|
| 141 |
+
{
|
| 142 |
+
"epoch": 14.0,
|
| 143 |
+
"eval_bleu": 6.4212,
|
| 144 |
+
"eval_gen_len": 30.6667,
|
| 145 |
+
"eval_loss": 3.0853846073150635,
|
| 146 |
+
"eval_runtime": 6.1365,
|
| 147 |
+
"eval_samples_per_second": 0.489,
|
| 148 |
+
"eval_steps_per_second": 0.163,
|
| 149 |
+
"step": 14
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 15.0,
|
| 153 |
+
"eval_bleu": 6.4212,
|
| 154 |
+
"eval_gen_len": 30.6667,
|
| 155 |
+
"eval_loss": 3.084298610687256,
|
| 156 |
+
"eval_runtime": 5.0728,
|
| 157 |
+
"eval_samples_per_second": 0.591,
|
| 158 |
+
"eval_steps_per_second": 0.197,
|
| 159 |
+
"step": 15
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 16.0,
|
| 163 |
+
"eval_bleu": 6.4212,
|
| 164 |
+
"eval_gen_len": 30.6667,
|
| 165 |
+
"eval_loss": 3.084357976913452,
|
| 166 |
+
"eval_runtime": 5.1387,
|
| 167 |
+
"eval_samples_per_second": 0.584,
|
| 168 |
+
"eval_steps_per_second": 0.195,
|
| 169 |
+
"step": 16
|
| 170 |
+
},
|
| 171 |
+
{
|
| 172 |
+
"epoch": 17.0,
|
| 173 |
+
"eval_bleu": 8.0332,
|
| 174 |
+
"eval_gen_len": 33.0,
|
| 175 |
+
"eval_loss": 3.0861120223999023,
|
| 176 |
+
"eval_runtime": 6.9674,
|
| 177 |
+
"eval_samples_per_second": 0.431,
|
| 178 |
+
"eval_steps_per_second": 0.144,
|
| 179 |
+
"step": 17
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 18.0,
|
| 183 |
+
"eval_bleu": 8.0332,
|
| 184 |
+
"eval_gen_len": 33.0,
|
| 185 |
+
"eval_loss": 3.089327812194824,
|
| 186 |
+
"eval_runtime": 6.5905,
|
| 187 |
+
"eval_samples_per_second": 0.455,
|
| 188 |
+
"eval_steps_per_second": 0.152,
|
| 189 |
+
"step": 18
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 19.0,
|
| 193 |
+
"eval_bleu": 8.0332,
|
| 194 |
+
"eval_gen_len": 33.0,
|
| 195 |
+
"eval_loss": 3.093491792678833,
|
| 196 |
+
"eval_runtime": 5.8865,
|
| 197 |
+
"eval_samples_per_second": 0.51,
|
| 198 |
+
"eval_steps_per_second": 0.17,
|
| 199 |
+
"step": 19
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 20.0,
|
| 203 |
+
"eval_bleu": 8.0332,
|
| 204 |
+
"eval_gen_len": 33.0,
|
| 205 |
+
"eval_loss": 3.0987062454223633,
|
| 206 |
+
"eval_runtime": 6.0219,
|
| 207 |
+
"eval_samples_per_second": 0.498,
|
| 208 |
+
"eval_steps_per_second": 0.166,
|
| 209 |
+
"step": 20
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 21.0,
|
| 213 |
+
"eval_bleu": 8.0332,
|
| 214 |
+
"eval_gen_len": 33.0,
|
| 215 |
+
"eval_loss": 3.107353925704956,
|
| 216 |
+
"eval_runtime": 6.8091,
|
| 217 |
+
"eval_samples_per_second": 0.441,
|
| 218 |
+
"eval_steps_per_second": 0.147,
|
| 219 |
+
"step": 21
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 22.0,
|
| 223 |
+
"eval_bleu": 8.0332,
|
| 224 |
+
"eval_gen_len": 33.0,
|
| 225 |
+
"eval_loss": 3.1154139041900635,
|
| 226 |
+
"eval_runtime": 6.6995,
|
| 227 |
+
"eval_samples_per_second": 0.448,
|
| 228 |
+
"eval_steps_per_second": 0.149,
|
| 229 |
+
"step": 22
|
| 230 |
+
},
|
| 231 |
+
{
|
| 232 |
+
"epoch": 23.0,
|
| 233 |
+
"eval_bleu": 8.0332,
|
| 234 |
+
"eval_gen_len": 33.0,
|
| 235 |
+
"eval_loss": 3.123610258102417,
|
| 236 |
+
"eval_runtime": 5.8096,
|
| 237 |
+
"eval_samples_per_second": 0.516,
|
| 238 |
+
"eval_steps_per_second": 0.172,
|
| 239 |
+
"step": 23
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 24.0,
|
| 243 |
+
"eval_bleu": 8.0332,
|
| 244 |
+
"eval_gen_len": 33.0,
|
| 245 |
+
"eval_loss": 3.1322247982025146,
|
| 246 |
+
"eval_runtime": 6.0317,
|
| 247 |
+
"eval_samples_per_second": 0.497,
|
| 248 |
+
"eval_steps_per_second": 0.166,
|
| 249 |
+
"step": 24
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 25.0,
|
| 253 |
+
"eval_bleu": 8.1739,
|
| 254 |
+
"eval_gen_len": 34.0,
|
| 255 |
+
"eval_loss": 3.1408698558807373,
|
| 256 |
+
"eval_runtime": 7.3925,
|
| 257 |
+
"eval_samples_per_second": 0.406,
|
| 258 |
+
"eval_steps_per_second": 0.135,
|
| 259 |
+
"step": 25
|
| 260 |
+
},
|
| 261 |
+
{
|
| 262 |
+
"epoch": 26.0,
|
| 263 |
+
"eval_bleu": 8.2412,
|
| 264 |
+
"eval_gen_len": 34.0,
|
| 265 |
+
"eval_loss": 3.1478536128997803,
|
| 266 |
+
"eval_runtime": 7.2011,
|
| 267 |
+
"eval_samples_per_second": 0.417,
|
| 268 |
+
"eval_steps_per_second": 0.139,
|
| 269 |
+
"step": 26
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 27.0,
|
| 273 |
+
"eval_bleu": 8.2412,
|
| 274 |
+
"eval_gen_len": 34.0,
|
| 275 |
+
"eval_loss": 3.154348134994507,
|
| 276 |
+
"eval_runtime": 6.2153,
|
| 277 |
+
"eval_samples_per_second": 0.483,
|
| 278 |
+
"eval_steps_per_second": 0.161,
|
| 279 |
+
"step": 27
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 28.0,
|
| 283 |
+
"eval_bleu": 8.2412,
|
| 284 |
+
"eval_gen_len": 34.0,
|
| 285 |
+
"eval_loss": 3.160304307937622,
|
| 286 |
+
"eval_runtime": 6.2248,
|
| 287 |
+
"eval_samples_per_second": 0.482,
|
| 288 |
+
"eval_steps_per_second": 0.161,
|
| 289 |
+
"step": 28
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 29.0,
|
| 293 |
+
"eval_bleu": 8.2334,
|
| 294 |
+
"eval_gen_len": 33.6667,
|
| 295 |
+
"eval_loss": 3.1661272048950195,
|
| 296 |
+
"eval_runtime": 7.371,
|
| 297 |
+
"eval_samples_per_second": 0.407,
|
| 298 |
+
"eval_steps_per_second": 0.136,
|
| 299 |
+
"step": 29
|
| 300 |
+
},
|
| 301 |
+
{
|
| 302 |
+
"epoch": 30.0,
|
| 303 |
+
"eval_bleu": 8.2334,
|
| 304 |
+
"eval_gen_len": 33.6667,
|
| 305 |
+
"eval_loss": 3.171717882156372,
|
| 306 |
+
"eval_runtime": 7.4227,
|
| 307 |
+
"eval_samples_per_second": 0.404,
|
| 308 |
+
"eval_steps_per_second": 0.135,
|
| 309 |
+
"step": 30
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 31.0,
|
| 313 |
+
"eval_bleu": 8.2334,
|
| 314 |
+
"eval_gen_len": 33.6667,
|
| 315 |
+
"eval_loss": 3.177027702331543,
|
| 316 |
+
"eval_runtime": 6.4986,
|
| 317 |
+
"eval_samples_per_second": 0.462,
|
| 318 |
+
"eval_steps_per_second": 0.154,
|
| 319 |
+
"step": 31
|
| 320 |
+
},
|
| 321 |
+
{
|
| 322 |
+
"epoch": 32.0,
|
| 323 |
+
"eval_bleu": 8.2334,
|
| 324 |
+
"eval_gen_len": 33.6667,
|
| 325 |
+
"eval_loss": 3.181061029434204,
|
| 326 |
+
"eval_runtime": 6.2381,
|
| 327 |
+
"eval_samples_per_second": 0.481,
|
| 328 |
+
"eval_steps_per_second": 0.16,
|
| 329 |
+
"step": 32
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 33.0,
|
| 333 |
+
"eval_bleu": 8.2334,
|
| 334 |
+
"eval_gen_len": 33.6667,
|
| 335 |
+
"eval_loss": 3.1853837966918945,
|
| 336 |
+
"eval_runtime": 6.844,
|
| 337 |
+
"eval_samples_per_second": 0.438,
|
| 338 |
+
"eval_steps_per_second": 0.146,
|
| 339 |
+
"step": 33
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 34.0,
|
| 343 |
+
"eval_bleu": 8.2334,
|
| 344 |
+
"eval_gen_len": 33.6667,
|
| 345 |
+
"eval_loss": 3.189124345779419,
|
| 346 |
+
"eval_runtime": 7.3578,
|
| 347 |
+
"eval_samples_per_second": 0.408,
|
| 348 |
+
"eval_steps_per_second": 0.136,
|
| 349 |
+
"step": 34
|
| 350 |
+
},
|
| 351 |
+
{
|
| 352 |
+
"epoch": 35.0,
|
| 353 |
+
"eval_bleu": 8.2334,
|
| 354 |
+
"eval_gen_len": 33.6667,
|
| 355 |
+
"eval_loss": 3.192593574523926,
|
| 356 |
+
"eval_runtime": 6.8816,
|
| 357 |
+
"eval_samples_per_second": 0.436,
|
| 358 |
+
"eval_steps_per_second": 0.145,
|
| 359 |
+
"step": 35
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 36.0,
|
| 363 |
+
"eval_bleu": 8.2334,
|
| 364 |
+
"eval_gen_len": 33.6667,
|
| 365 |
+
"eval_loss": 3.195676565170288,
|
| 366 |
+
"eval_runtime": 6.1754,
|
| 367 |
+
"eval_samples_per_second": 0.486,
|
| 368 |
+
"eval_steps_per_second": 0.162,
|
| 369 |
+
"step": 36
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 37.0,
|
| 373 |
+
"eval_bleu": 9.4483,
|
| 374 |
+
"eval_gen_len": 32.6667,
|
| 375 |
+
"eval_loss": 3.197235345840454,
|
| 376 |
+
"eval_runtime": 5.8155,
|
| 377 |
+
"eval_samples_per_second": 0.516,
|
| 378 |
+
"eval_steps_per_second": 0.172,
|
| 379 |
+
"step": 37
|
| 380 |
+
},
|
| 381 |
+
{
|
| 382 |
+
"epoch": 38.0,
|
| 383 |
+
"eval_bleu": 9.4483,
|
| 384 |
+
"eval_gen_len": 32.6667,
|
| 385 |
+
"eval_loss": 3.198820114135742,
|
| 386 |
+
"eval_runtime": 6.985,
|
| 387 |
+
"eval_samples_per_second": 0.429,
|
| 388 |
+
"eval_steps_per_second": 0.143,
|
| 389 |
+
"step": 38
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 39.0,
|
| 393 |
+
"eval_bleu": 9.4588,
|
| 394 |
+
"eval_gen_len": 32.3333,
|
| 395 |
+
"eval_loss": 3.2010693550109863,
|
| 396 |
+
"eval_runtime": 6.6383,
|
| 397 |
+
"eval_samples_per_second": 0.452,
|
| 398 |
+
"eval_steps_per_second": 0.151,
|
| 399 |
+
"step": 39
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 40.0,
|
| 403 |
+
"eval_bleu": 9.4588,
|
| 404 |
+
"eval_gen_len": 32.3333,
|
| 405 |
+
"eval_loss": 3.2039597034454346,
|
| 406 |
+
"eval_runtime": 5.8356,
|
| 407 |
+
"eval_samples_per_second": 0.514,
|
| 408 |
+
"eval_steps_per_second": 0.171,
|
| 409 |
+
"step": 40
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 41.0,
|
| 413 |
+
"eval_bleu": 9.4483,
|
| 414 |
+
"eval_gen_len": 32.6667,
|
| 415 |
+
"eval_loss": 3.2071328163146973,
|
| 416 |
+
"eval_runtime": 5.936,
|
| 417 |
+
"eval_samples_per_second": 0.505,
|
| 418 |
+
"eval_steps_per_second": 0.168,
|
| 419 |
+
"step": 41
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 42.0,
|
| 423 |
+
"eval_bleu": 9.4483,
|
| 424 |
+
"eval_gen_len": 32.6667,
|
| 425 |
+
"eval_loss": 3.210134506225586,
|
| 426 |
+
"eval_runtime": 7.0844,
|
| 427 |
+
"eval_samples_per_second": 0.423,
|
| 428 |
+
"eval_steps_per_second": 0.141,
|
| 429 |
+
"step": 42
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 43.0,
|
| 433 |
+
"eval_bleu": 7.8678,
|
| 434 |
+
"eval_gen_len": 32.3333,
|
| 435 |
+
"eval_loss": 3.213345527648926,
|
| 436 |
+
"eval_runtime": 6.4907,
|
| 437 |
+
"eval_samples_per_second": 0.462,
|
| 438 |
+
"eval_steps_per_second": 0.154,
|
| 439 |
+
"step": 43
|
| 440 |
+
},
|
| 441 |
+
{
|
| 442 |
+
"epoch": 44.0,
|
| 443 |
+
"eval_bleu": 7.878,
|
| 444 |
+
"eval_gen_len": 32.0,
|
| 445 |
+
"eval_loss": 3.2164242267608643,
|
| 446 |
+
"eval_runtime": 6.8419,
|
| 447 |
+
"eval_samples_per_second": 0.438,
|
| 448 |
+
"eval_steps_per_second": 0.146,
|
| 449 |
+
"step": 44
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 45.0,
|
| 453 |
+
"eval_bleu": 7.878,
|
| 454 |
+
"eval_gen_len": 32.0,
|
| 455 |
+
"eval_loss": 3.219308853149414,
|
| 456 |
+
"eval_runtime": 5.6863,
|
| 457 |
+
"eval_samples_per_second": 0.528,
|
| 458 |
+
"eval_steps_per_second": 0.176,
|
| 459 |
+
"step": 45
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 46.0,
|
| 463 |
+
"eval_bleu": 7.878,
|
| 464 |
+
"eval_gen_len": 32.0,
|
| 465 |
+
"eval_loss": 3.221681833267212,
|
| 466 |
+
"eval_runtime": 6.9093,
|
| 467 |
+
"eval_samples_per_second": 0.434,
|
| 468 |
+
"eval_steps_per_second": 0.145,
|
| 469 |
+
"step": 46
|
| 470 |
+
},
|
| 471 |
+
{
|
| 472 |
+
"epoch": 47.0,
|
| 473 |
+
"eval_bleu": 7.878,
|
| 474 |
+
"eval_gen_len": 32.0,
|
| 475 |
+
"eval_loss": 3.223684310913086,
|
| 476 |
+
"eval_runtime": 6.7844,
|
| 477 |
+
"eval_samples_per_second": 0.442,
|
| 478 |
+
"eval_steps_per_second": 0.147,
|
| 479 |
+
"step": 47
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 48.0,
|
| 483 |
+
"eval_bleu": 7.878,
|
| 484 |
+
"eval_gen_len": 32.0,
|
| 485 |
+
"eval_loss": 3.225531578063965,
|
| 486 |
+
"eval_runtime": 5.7562,
|
| 487 |
+
"eval_samples_per_second": 0.521,
|
| 488 |
+
"eval_steps_per_second": 0.174,
|
| 489 |
+
"step": 48
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 49.0,
|
| 493 |
+
"eval_bleu": 7.878,
|
| 494 |
+
"eval_gen_len": 32.0,
|
| 495 |
+
"eval_loss": 3.2273638248443604,
|
| 496 |
+
"eval_runtime": 5.7642,
|
| 497 |
+
"eval_samples_per_second": 0.52,
|
| 498 |
+
"eval_steps_per_second": 0.173,
|
| 499 |
+
"step": 49
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 50.0,
|
| 503 |
+
"eval_bleu": 7.878,
|
| 504 |
+
"eval_gen_len": 32.0,
|
| 505 |
+
"eval_loss": 3.2290658950805664,
|
| 506 |
+
"eval_runtime": 6.9295,
|
| 507 |
+
"eval_samples_per_second": 0.433,
|
| 508 |
+
"eval_steps_per_second": 0.144,
|
| 509 |
+
"step": 50
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 51.0,
|
| 513 |
+
"eval_bleu": 7.878,
|
| 514 |
+
"eval_gen_len": 32.0,
|
| 515 |
+
"eval_loss": 3.2310614585876465,
|
| 516 |
+
"eval_runtime": 6.4631,
|
| 517 |
+
"eval_samples_per_second": 0.464,
|
| 518 |
+
"eval_steps_per_second": 0.155,
|
| 519 |
+
"step": 51
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 52.0,
|
| 523 |
+
"eval_bleu": 7.878,
|
| 524 |
+
"eval_gen_len": 32.0,
|
| 525 |
+
"eval_loss": 3.2329607009887695,
|
| 526 |
+
"eval_runtime": 5.7782,
|
| 527 |
+
"eval_samples_per_second": 0.519,
|
| 528 |
+
"eval_steps_per_second": 0.173,
|
| 529 |
+
"step": 52
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 53.0,
|
| 533 |
+
"eval_bleu": 7.878,
|
| 534 |
+
"eval_gen_len": 32.0,
|
| 535 |
+
"eval_loss": 3.2342495918273926,
|
| 536 |
+
"eval_runtime": 5.8601,
|
| 537 |
+
"eval_samples_per_second": 0.512,
|
| 538 |
+
"eval_steps_per_second": 0.171,
|
| 539 |
+
"step": 53
|
| 540 |
+
},
|
| 541 |
+
{
|
| 542 |
+
"epoch": 54.0,
|
| 543 |
+
"eval_bleu": 8.1571,
|
| 544 |
+
"eval_gen_len": 32.0,
|
| 545 |
+
"eval_loss": 3.2355172634124756,
|
| 546 |
+
"eval_runtime": 6.8845,
|
| 547 |
+
"eval_samples_per_second": 0.436,
|
| 548 |
+
"eval_steps_per_second": 0.145,
|
| 549 |
+
"step": 54
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 55.0,
|
| 553 |
+
"eval_bleu": 8.1571,
|
| 554 |
+
"eval_gen_len": 32.0,
|
| 555 |
+
"eval_loss": 3.23667049407959,
|
| 556 |
+
"eval_runtime": 6.3856,
|
| 557 |
+
"eval_samples_per_second": 0.47,
|
| 558 |
+
"eval_steps_per_second": 0.157,
|
| 559 |
+
"step": 55
|
| 560 |
+
},
|
| 561 |
+
{
|
| 562 |
+
"epoch": 56.0,
|
| 563 |
+
"eval_bleu": 8.1571,
|
| 564 |
+
"eval_gen_len": 32.0,
|
| 565 |
+
"eval_loss": 3.239001750946045,
|
| 566 |
+
"eval_runtime": 5.7517,
|
| 567 |
+
"eval_samples_per_second": 0.522,
|
| 568 |
+
"eval_steps_per_second": 0.174,
|
| 569 |
+
"step": 56
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 57.0,
|
| 573 |
+
"eval_bleu": 8.1571,
|
| 574 |
+
"eval_gen_len": 32.0,
|
| 575 |
+
"eval_loss": 3.2409348487854004,
|
| 576 |
+
"eval_runtime": 6.43,
|
| 577 |
+
"eval_samples_per_second": 0.467,
|
| 578 |
+
"eval_steps_per_second": 0.156,
|
| 579 |
+
"step": 57
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 58.0,
|
| 583 |
+
"eval_bleu": 8.1571,
|
| 584 |
+
"eval_gen_len": 32.0,
|
| 585 |
+
"eval_loss": 3.242704391479492,
|
| 586 |
+
"eval_runtime": 6.7607,
|
| 587 |
+
"eval_samples_per_second": 0.444,
|
| 588 |
+
"eval_steps_per_second": 0.148,
|
| 589 |
+
"step": 58
|
| 590 |
+
},
|
| 591 |
+
{
|
| 592 |
+
"epoch": 59.0,
|
| 593 |
+
"eval_bleu": 8.1571,
|
| 594 |
+
"eval_gen_len": 32.0,
|
| 595 |
+
"eval_loss": 3.2443935871124268,
|
| 596 |
+
"eval_runtime": 5.9681,
|
| 597 |
+
"eval_samples_per_second": 0.503,
|
| 598 |
+
"eval_steps_per_second": 0.168,
|
| 599 |
+
"step": 59
|
| 600 |
+
},
|
| 601 |
+
{
|
| 602 |
+
"epoch": 60.0,
|
| 603 |
+
"eval_bleu": 8.1571,
|
| 604 |
+
"eval_gen_len": 32.0,
|
| 605 |
+
"eval_loss": 3.245845079421997,
|
| 606 |
+
"eval_runtime": 5.7219,
|
| 607 |
+
"eval_samples_per_second": 0.524,
|
| 608 |
+
"eval_steps_per_second": 0.175,
|
| 609 |
+
"step": 60
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 61.0,
|
| 613 |
+
"eval_bleu": 8.1571,
|
| 614 |
+
"eval_gen_len": 32.0,
|
| 615 |
+
"eval_loss": 3.2472944259643555,
|
| 616 |
+
"eval_runtime": 6.6813,
|
| 617 |
+
"eval_samples_per_second": 0.449,
|
| 618 |
+
"eval_steps_per_second": 0.15,
|
| 619 |
+
"step": 61
|
| 620 |
+
},
|
| 621 |
+
{
|
| 622 |
+
"epoch": 62.0,
|
| 623 |
+
"eval_bleu": 8.1571,
|
| 624 |
+
"eval_gen_len": 32.0,
|
| 625 |
+
"eval_loss": 3.248718500137329,
|
| 626 |
+
"eval_runtime": 6.8908,
|
| 627 |
+
"eval_samples_per_second": 0.435,
|
| 628 |
+
"eval_steps_per_second": 0.145,
|
| 629 |
+
"step": 62
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 63.0,
|
| 633 |
+
"eval_bleu": 8.1571,
|
| 634 |
+
"eval_gen_len": 32.0,
|
| 635 |
+
"eval_loss": 3.2500014305114746,
|
| 636 |
+
"eval_runtime": 5.7556,
|
| 637 |
+
"eval_samples_per_second": 0.521,
|
| 638 |
+
"eval_steps_per_second": 0.174,
|
| 639 |
+
"step": 63
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 64.0,
|
| 643 |
+
"eval_bleu": 9.4588,
|
| 644 |
+
"eval_gen_len": 32.3333,
|
| 645 |
+
"eval_loss": 3.2512311935424805,
|
| 646 |
+
"eval_runtime": 5.7165,
|
| 647 |
+
"eval_samples_per_second": 0.525,
|
| 648 |
+
"eval_steps_per_second": 0.175,
|
| 649 |
+
"step": 64
|
| 650 |
+
},
|
| 651 |
+
{
|
| 652 |
+
"epoch": 65.0,
|
| 653 |
+
"eval_bleu": 9.4588,
|
| 654 |
+
"eval_gen_len": 32.3333,
|
| 655 |
+
"eval_loss": 3.252441167831421,
|
| 656 |
+
"eval_runtime": 6.9093,
|
| 657 |
+
"eval_samples_per_second": 0.434,
|
| 658 |
+
"eval_steps_per_second": 0.145,
|
| 659 |
+
"step": 65
|
| 660 |
+
},
|
| 661 |
+
{
|
| 662 |
+
"epoch": 66.0,
|
| 663 |
+
"eval_bleu": 9.4588,
|
| 664 |
+
"eval_gen_len": 32.3333,
|
| 665 |
+
"eval_loss": 3.2536087036132812,
|
| 666 |
+
"eval_runtime": 6.5942,
|
| 667 |
+
"eval_samples_per_second": 0.455,
|
| 668 |
+
"eval_steps_per_second": 0.152,
|
| 669 |
+
"step": 66
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 67.0,
|
| 673 |
+
"eval_bleu": 9.4588,
|
| 674 |
+
"eval_gen_len": 32.3333,
|
| 675 |
+
"eval_loss": 3.254615545272827,
|
| 676 |
+
"eval_runtime": 5.7148,
|
| 677 |
+
"eval_samples_per_second": 0.525,
|
| 678 |
+
"eval_steps_per_second": 0.175,
|
| 679 |
+
"step": 67
|
| 680 |
+
},
|
| 681 |
+
{
|
| 682 |
+
"epoch": 68.0,
|
| 683 |
+
"eval_bleu": 9.4588,
|
| 684 |
+
"eval_gen_len": 32.3333,
|
| 685 |
+
"eval_loss": 3.2554357051849365,
|
| 686 |
+
"eval_runtime": 6.9043,
|
| 687 |
+
"eval_samples_per_second": 0.435,
|
| 688 |
+
"eval_steps_per_second": 0.145,
|
| 689 |
+
"step": 68
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 69.0,
|
| 693 |
+
"eval_bleu": 9.4588,
|
| 694 |
+
"eval_gen_len": 32.3333,
|
| 695 |
+
"eval_loss": 3.256138563156128,
|
| 696 |
+
"eval_runtime": 6.7504,
|
| 697 |
+
"eval_samples_per_second": 0.444,
|
| 698 |
+
"eval_steps_per_second": 0.148,
|
| 699 |
+
"step": 69
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 70.0,
|
| 703 |
+
"eval_bleu": 9.4588,
|
| 704 |
+
"eval_gen_len": 32.3333,
|
| 705 |
+
"eval_loss": 3.256850481033325,
|
| 706 |
+
"eval_runtime": 6.8705,
|
| 707 |
+
"eval_samples_per_second": 0.437,
|
| 708 |
+
"eval_steps_per_second": 0.146,
|
| 709 |
+
"step": 70
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"epoch": 71.0,
|
| 713 |
+
"eval_bleu": 9.4588,
|
| 714 |
+
"eval_gen_len": 32.3333,
|
| 715 |
+
"eval_loss": 3.257761001586914,
|
| 716 |
+
"eval_runtime": 5.7321,
|
| 717 |
+
"eval_samples_per_second": 0.523,
|
| 718 |
+
"eval_steps_per_second": 0.174,
|
| 719 |
+
"step": 71
|
| 720 |
+
},
|
| 721 |
+
{
|
| 722 |
+
"epoch": 72.0,
|
| 723 |
+
"eval_bleu": 9.4588,
|
| 724 |
+
"eval_gen_len": 32.3333,
|
| 725 |
+
"eval_loss": 3.2562146186828613,
|
| 726 |
+
"eval_runtime": 5.7598,
|
| 727 |
+
"eval_samples_per_second": 0.521,
|
| 728 |
+
"eval_steps_per_second": 0.174,
|
| 729 |
+
"step": 72
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 73.0,
|
| 733 |
+
"eval_bleu": 8.2242,
|
| 734 |
+
"eval_gen_len": 32.0,
|
| 735 |
+
"eval_loss": 3.2549028396606445,
|
| 736 |
+
"eval_runtime": 6.7394,
|
| 737 |
+
"eval_samples_per_second": 0.445,
|
| 738 |
+
"eval_steps_per_second": 0.148,
|
| 739 |
+
"step": 73
|
| 740 |
+
},
|
| 741 |
+
{
|
| 742 |
+
"epoch": 74.0,
|
| 743 |
+
"eval_bleu": 8.2242,
|
| 744 |
+
"eval_gen_len": 32.0,
|
| 745 |
+
"eval_loss": 3.2539963722229004,
|
| 746 |
+
"eval_runtime": 6.8504,
|
| 747 |
+
"eval_samples_per_second": 0.438,
|
| 748 |
+
"eval_steps_per_second": 0.146,
|
| 749 |
+
"step": 74
|
| 750 |
+
},
|
| 751 |
+
{
|
| 752 |
+
"epoch": 75.0,
|
| 753 |
+
"eval_bleu": 9.4588,
|
| 754 |
+
"eval_gen_len": 32.3333,
|
| 755 |
+
"eval_loss": 3.2532434463500977,
|
| 756 |
+
"eval_runtime": 5.7412,
|
| 757 |
+
"eval_samples_per_second": 0.523,
|
| 758 |
+
"eval_steps_per_second": 0.174,
|
| 759 |
+
"step": 75
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 76.0,
|
| 763 |
+
"eval_bleu": 9.4588,
|
| 764 |
+
"eval_gen_len": 32.3333,
|
| 765 |
+
"eval_loss": 3.2526049613952637,
|
| 766 |
+
"eval_runtime": 5.7155,
|
| 767 |
+
"eval_samples_per_second": 0.525,
|
| 768 |
+
"eval_steps_per_second": 0.175,
|
| 769 |
+
"step": 76
|
| 770 |
+
},
|
| 771 |
+
{
|
| 772 |
+
"epoch": 77.0,
|
| 773 |
+
"eval_bleu": 9.4588,
|
| 774 |
+
"eval_gen_len": 32.3333,
|
| 775 |
+
"eval_loss": 3.2521774768829346,
|
| 776 |
+
"eval_runtime": 6.901,
|
| 777 |
+
"eval_samples_per_second": 0.435,
|
| 778 |
+
"eval_steps_per_second": 0.145,
|
| 779 |
+
"step": 77
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 78.0,
|
| 783 |
+
"eval_bleu": 9.4588,
|
| 784 |
+
"eval_gen_len": 32.3333,
|
| 785 |
+
"eval_loss": 3.251833438873291,
|
| 786 |
+
"eval_runtime": 6.8378,
|
| 787 |
+
"eval_samples_per_second": 0.439,
|
| 788 |
+
"eval_steps_per_second": 0.146,
|
| 789 |
+
"step": 78
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 79.0,
|
| 793 |
+
"eval_bleu": 9.4811,
|
| 794 |
+
"eval_gen_len": 33.3333,
|
| 795 |
+
"eval_loss": 3.2515926361083984,
|
| 796 |
+
"eval_runtime": 5.7488,
|
| 797 |
+
"eval_samples_per_second": 0.522,
|
| 798 |
+
"eval_steps_per_second": 0.174,
|
| 799 |
+
"step": 79
|
| 800 |
+
},
|
| 801 |
+
{
|
| 802 |
+
"epoch": 80.0,
|
| 803 |
+
"eval_bleu": 9.4811,
|
| 804 |
+
"eval_gen_len": 33.3333,
|
| 805 |
+
"eval_loss": 3.251364231109619,
|
| 806 |
+
"eval_runtime": 5.6787,
|
| 807 |
+
"eval_samples_per_second": 0.528,
|
| 808 |
+
"eval_steps_per_second": 0.176,
|
| 809 |
+
"step": 80
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 81.0,
|
| 813 |
+
"eval_bleu": 9.4811,
|
| 814 |
+
"eval_gen_len": 33.3333,
|
| 815 |
+
"eval_loss": 3.2512130737304688,
|
| 816 |
+
"eval_runtime": 6.8657,
|
| 817 |
+
"eval_samples_per_second": 0.437,
|
| 818 |
+
"eval_steps_per_second": 0.146,
|
| 819 |
+
"step": 81
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 82.0,
|
| 823 |
+
"eval_bleu": 9.4811,
|
| 824 |
+
"eval_gen_len": 33.3333,
|
| 825 |
+
"eval_loss": 3.251028060913086,
|
| 826 |
+
"eval_runtime": 6.7997,
|
| 827 |
+
"eval_samples_per_second": 0.441,
|
| 828 |
+
"eval_steps_per_second": 0.147,
|
| 829 |
+
"step": 82
|
| 830 |
+
},
|
| 831 |
+
{
|
| 832 |
+
"epoch": 83.0,
|
| 833 |
+
"eval_bleu": 9.4811,
|
| 834 |
+
"eval_gen_len": 33.3333,
|
| 835 |
+
"eval_loss": 3.2511019706726074,
|
| 836 |
+
"eval_runtime": 5.72,
|
| 837 |
+
"eval_samples_per_second": 0.524,
|
| 838 |
+
"eval_steps_per_second": 0.175,
|
| 839 |
+
"step": 83
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 84.0,
|
| 843 |
+
"eval_bleu": 9.4811,
|
| 844 |
+
"eval_gen_len": 33.3333,
|
| 845 |
+
"eval_loss": 3.2512319087982178,
|
| 846 |
+
"eval_runtime": 5.8699,
|
| 847 |
+
"eval_samples_per_second": 0.511,
|
| 848 |
+
"eval_steps_per_second": 0.17,
|
| 849 |
+
"step": 84
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 85.0,
|
| 853 |
+
"eval_bleu": 9.4811,
|
| 854 |
+
"eval_gen_len": 33.3333,
|
| 855 |
+
"eval_loss": 3.2515382766723633,
|
| 856 |
+
"eval_runtime": 6.931,
|
| 857 |
+
"eval_samples_per_second": 0.433,
|
| 858 |
+
"eval_steps_per_second": 0.144,
|
| 859 |
+
"step": 85
|
| 860 |
+
},
|
| 861 |
+
{
|
| 862 |
+
"epoch": 86.0,
|
| 863 |
+
"eval_bleu": 9.4811,
|
| 864 |
+
"eval_gen_len": 33.3333,
|
| 865 |
+
"eval_loss": 3.2519519329071045,
|
| 866 |
+
"eval_runtime": 6.3374,
|
| 867 |
+
"eval_samples_per_second": 0.473,
|
| 868 |
+
"eval_steps_per_second": 0.158,
|
| 869 |
+
"step": 86
|
| 870 |
+
},
|
| 871 |
+
{
|
| 872 |
+
"epoch": 87.0,
|
| 873 |
+
"eval_bleu": 9.4811,
|
| 874 |
+
"eval_gen_len": 33.3333,
|
| 875 |
+
"eval_loss": 3.2524967193603516,
|
| 876 |
+
"eval_runtime": 5.7106,
|
| 877 |
+
"eval_samples_per_second": 0.525,
|
| 878 |
+
"eval_steps_per_second": 0.175,
|
| 879 |
+
"step": 87
|
| 880 |
+
},
|
| 881 |
+
{
|
| 882 |
+
"epoch": 88.0,
|
| 883 |
+
"eval_bleu": 9.4811,
|
| 884 |
+
"eval_gen_len": 33.3333,
|
| 885 |
+
"eval_loss": 3.252875566482544,
|
| 886 |
+
"eval_runtime": 6.2722,
|
| 887 |
+
"eval_samples_per_second": 0.478,
|
| 888 |
+
"eval_steps_per_second": 0.159,
|
| 889 |
+
"step": 88
|
| 890 |
+
},
|
| 891 |
+
{
|
| 892 |
+
"epoch": 89.0,
|
| 893 |
+
"eval_bleu": 8.1803,
|
| 894 |
+
"eval_gen_len": 33.0,
|
| 895 |
+
"eval_loss": 3.2532503604888916,
|
| 896 |
+
"eval_runtime": 6.7099,
|
| 897 |
+
"eval_samples_per_second": 0.447,
|
| 898 |
+
"eval_steps_per_second": 0.149,
|
| 899 |
+
"step": 89
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 90.0,
|
| 903 |
+
"eval_bleu": 8.1803,
|
| 904 |
+
"eval_gen_len": 33.0,
|
| 905 |
+
"eval_loss": 3.253549575805664,
|
| 906 |
+
"eval_runtime": 6.2379,
|
| 907 |
+
"eval_samples_per_second": 0.481,
|
| 908 |
+
"eval_steps_per_second": 0.16,
|
| 909 |
+
"step": 90
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 91.0,
|
| 913 |
+
"eval_bleu": 8.1803,
|
| 914 |
+
"eval_gen_len": 33.0,
|
| 915 |
+
"eval_loss": 3.2552127838134766,
|
| 916 |
+
"eval_runtime": 5.8079,
|
| 917 |
+
"eval_samples_per_second": 0.517,
|
| 918 |
+
"eval_steps_per_second": 0.172,
|
| 919 |
+
"step": 91
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 92.0,
|
| 923 |
+
"eval_bleu": 8.1803,
|
| 924 |
+
"eval_gen_len": 33.0,
|
| 925 |
+
"eval_loss": 3.2567169666290283,
|
| 926 |
+
"eval_runtime": 6.3512,
|
| 927 |
+
"eval_samples_per_second": 0.472,
|
| 928 |
+
"eval_steps_per_second": 0.157,
|
| 929 |
+
"step": 92
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 93.0,
|
| 933 |
+
"eval_bleu": 8.1803,
|
| 934 |
+
"eval_gen_len": 33.0,
|
| 935 |
+
"eval_loss": 3.2580533027648926,
|
| 936 |
+
"eval_runtime": 6.7043,
|
| 937 |
+
"eval_samples_per_second": 0.447,
|
| 938 |
+
"eval_steps_per_second": 0.149,
|
| 939 |
+
"step": 93
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 94.0,
|
| 943 |
+
"eval_bleu": 8.1803,
|
| 944 |
+
"eval_gen_len": 33.0,
|
| 945 |
+
"eval_loss": 3.259291648864746,
|
| 946 |
+
"eval_runtime": 6.4174,
|
| 947 |
+
"eval_samples_per_second": 0.467,
|
| 948 |
+
"eval_steps_per_second": 0.156,
|
| 949 |
+
"step": 94
|
| 950 |
+
},
|
| 951 |
+
{
|
| 952 |
+
"epoch": 95.0,
|
| 953 |
+
"eval_bleu": 8.1803,
|
| 954 |
+
"eval_gen_len": 33.0,
|
| 955 |
+
"eval_loss": 3.26053524017334,
|
| 956 |
+
"eval_runtime": 5.7947,
|
| 957 |
+
"eval_samples_per_second": 0.518,
|
| 958 |
+
"eval_steps_per_second": 0.173,
|
| 959 |
+
"step": 95
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 96.0,
|
| 963 |
+
"eval_bleu": 8.1803,
|
| 964 |
+
"eval_gen_len": 33.0,
|
| 965 |
+
"eval_loss": 3.2616453170776367,
|
| 966 |
+
"eval_runtime": 6.147,
|
| 967 |
+
"eval_samples_per_second": 0.488,
|
| 968 |
+
"eval_steps_per_second": 0.163,
|
| 969 |
+
"step": 96
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 97.0,
|
| 973 |
+
"eval_bleu": 8.1803,
|
| 974 |
+
"eval_gen_len": 33.0,
|
| 975 |
+
"eval_loss": 3.262683629989624,
|
| 976 |
+
"eval_runtime": 6.9667,
|
| 977 |
+
"eval_samples_per_second": 0.431,
|
| 978 |
+
"eval_steps_per_second": 0.144,
|
| 979 |
+
"step": 97
|
| 980 |
+
},
|
| 981 |
+
{
|
| 982 |
+
"epoch": 98.0,
|
| 983 |
+
"eval_bleu": 8.1803,
|
| 984 |
+
"eval_gen_len": 33.0,
|
| 985 |
+
"eval_loss": 3.2636585235595703,
|
| 986 |
+
"eval_runtime": 6.3994,
|
| 987 |
+
"eval_samples_per_second": 0.469,
|
| 988 |
+
"eval_steps_per_second": 0.156,
|
| 989 |
+
"step": 98
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 99.0,
|
| 993 |
+
"eval_bleu": 8.1571,
|
| 994 |
+
"eval_gen_len": 32.0,
|
| 995 |
+
"eval_loss": 3.2648403644561768,
|
| 996 |
+
"eval_runtime": 5.7941,
|
| 997 |
+
"eval_samples_per_second": 0.518,
|
| 998 |
+
"eval_steps_per_second": 0.173,
|
| 999 |
+
"step": 99
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 100.0,
|
| 1003 |
+
"eval_bleu": 8.1571,
|
| 1004 |
+
"eval_gen_len": 32.0,
|
| 1005 |
+
"eval_loss": 3.266058921813965,
|
| 1006 |
+
"eval_runtime": 7.0456,
|
| 1007 |
+
"eval_samples_per_second": 0.426,
|
| 1008 |
+
"eval_steps_per_second": 0.142,
|
| 1009 |
+
"step": 100
|
| 1010 |
+
},
|
| 1011 |
+
{
|
| 1012 |
+
"epoch": 101.0,
|
| 1013 |
+
"eval_bleu": 8.1571,
|
| 1014 |
+
"eval_gen_len": 32.0,
|
| 1015 |
+
"eval_loss": 3.267348051071167,
|
| 1016 |
+
"eval_runtime": 6.9442,
|
| 1017 |
+
"eval_samples_per_second": 0.432,
|
| 1018 |
+
"eval_steps_per_second": 0.144,
|
| 1019 |
+
"step": 101
|
| 1020 |
+
},
|
| 1021 |
+
{
|
| 1022 |
+
"epoch": 102.0,
|
| 1023 |
+
"eval_bleu": 8.1571,
|
| 1024 |
+
"eval_gen_len": 32.0,
|
| 1025 |
+
"eval_loss": 3.268547296524048,
|
| 1026 |
+
"eval_runtime": 6.8288,
|
| 1027 |
+
"eval_samples_per_second": 0.439,
|
| 1028 |
+
"eval_steps_per_second": 0.146,
|
| 1029 |
+
"step": 102
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 103.0,
|
| 1033 |
+
"eval_bleu": 8.1803,
|
| 1034 |
+
"eval_gen_len": 33.0,
|
| 1035 |
+
"eval_loss": 3.2697231769561768,
|
| 1036 |
+
"eval_runtime": 5.8621,
|
| 1037 |
+
"eval_samples_per_second": 0.512,
|
| 1038 |
+
"eval_steps_per_second": 0.171,
|
| 1039 |
+
"step": 103
|
| 1040 |
+
},
|
| 1041 |
+
{
|
| 1042 |
+
"epoch": 104.0,
|
| 1043 |
+
"eval_bleu": 8.1803,
|
| 1044 |
+
"eval_gen_len": 33.0,
|
| 1045 |
+
"eval_loss": 3.2708990573883057,
|
| 1046 |
+
"eval_runtime": 5.8145,
|
| 1047 |
+
"eval_samples_per_second": 0.516,
|
| 1048 |
+
"eval_steps_per_second": 0.172,
|
| 1049 |
+
"step": 104
|
| 1050 |
+
},
|
| 1051 |
+
{
|
| 1052 |
+
"epoch": 105.0,
|
| 1053 |
+
"eval_bleu": 8.1803,
|
| 1054 |
+
"eval_gen_len": 33.0,
|
| 1055 |
+
"eval_loss": 3.2720344066619873,
|
| 1056 |
+
"eval_runtime": 6.9784,
|
| 1057 |
+
"eval_samples_per_second": 0.43,
|
| 1058 |
+
"eval_steps_per_second": 0.143,
|
| 1059 |
+
"step": 105
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 106.0,
|
| 1063 |
+
"eval_bleu": 8.1803,
|
| 1064 |
+
"eval_gen_len": 33.0,
|
| 1065 |
+
"eval_loss": 3.273036003112793,
|
| 1066 |
+
"eval_runtime": 6.773,
|
| 1067 |
+
"eval_samples_per_second": 0.443,
|
| 1068 |
+
"eval_steps_per_second": 0.148,
|
| 1069 |
+
"step": 106
|
| 1070 |
+
},
|
| 1071 |
+
{
|
| 1072 |
+
"epoch": 107.0,
|
| 1073 |
+
"eval_bleu": 9.4811,
|
| 1074 |
+
"eval_gen_len": 33.3333,
|
| 1075 |
+
"eval_loss": 3.2738993167877197,
|
| 1076 |
+
"eval_runtime": 5.8039,
|
| 1077 |
+
"eval_samples_per_second": 0.517,
|
| 1078 |
+
"eval_steps_per_second": 0.172,
|
| 1079 |
+
"step": 107
|
| 1080 |
+
},
|
| 1081 |
+
{
|
| 1082 |
+
"epoch": 108.0,
|
| 1083 |
+
"eval_bleu": 9.4811,
|
| 1084 |
+
"eval_gen_len": 33.3333,
|
| 1085 |
+
"eval_loss": 3.2747609615325928,
|
| 1086 |
+
"eval_runtime": 5.8853,
|
| 1087 |
+
"eval_samples_per_second": 0.51,
|
| 1088 |
+
"eval_steps_per_second": 0.17,
|
| 1089 |
+
"step": 108
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 109.0,
|
| 1093 |
+
"eval_bleu": 9.4811,
|
| 1094 |
+
"eval_gen_len": 33.3333,
|
| 1095 |
+
"eval_loss": 3.2754225730895996,
|
| 1096 |
+
"eval_runtime": 7.051,
|
| 1097 |
+
"eval_samples_per_second": 0.425,
|
| 1098 |
+
"eval_steps_per_second": 0.142,
|
| 1099 |
+
"step": 109
|
| 1100 |
+
},
|
| 1101 |
+
{
|
| 1102 |
+
"epoch": 110.0,
|
| 1103 |
+
"eval_bleu": 9.4811,
|
| 1104 |
+
"eval_gen_len": 33.3333,
|
| 1105 |
+
"eval_loss": 3.276052951812744,
|
| 1106 |
+
"eval_runtime": 6.6331,
|
| 1107 |
+
"eval_samples_per_second": 0.452,
|
| 1108 |
+
"eval_steps_per_second": 0.151,
|
| 1109 |
+
"step": 110
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 111.0,
|
| 1113 |
+
"eval_bleu": 9.4811,
|
| 1114 |
+
"eval_gen_len": 33.3333,
|
| 1115 |
+
"eval_loss": 3.2766623497009277,
|
| 1116 |
+
"eval_runtime": 6.2513,
|
| 1117 |
+
"eval_samples_per_second": 0.48,
|
| 1118 |
+
"eval_steps_per_second": 0.16,
|
| 1119 |
+
"step": 111
|
| 1120 |
+
},
|
| 1121 |
+
{
|
| 1122 |
+
"epoch": 112.0,
|
| 1123 |
+
"eval_bleu": 9.4811,
|
| 1124 |
+
"eval_gen_len": 33.3333,
|
| 1125 |
+
"eval_loss": 3.2771129608154297,
|
| 1126 |
+
"eval_runtime": 5.7987,
|
| 1127 |
+
"eval_samples_per_second": 0.517,
|
| 1128 |
+
"eval_steps_per_second": 0.172,
|
| 1129 |
+
"step": 112
|
| 1130 |
+
},
|
| 1131 |
+
{
|
| 1132 |
+
"epoch": 113.0,
|
| 1133 |
+
"eval_bleu": 9.4811,
|
| 1134 |
+
"eval_gen_len": 33.3333,
|
| 1135 |
+
"eval_loss": 3.2771925926208496,
|
| 1136 |
+
"eval_runtime": 6.9685,
|
| 1137 |
+
"eval_samples_per_second": 0.431,
|
| 1138 |
+
"eval_steps_per_second": 0.144,
|
| 1139 |
+
"step": 113
|
| 1140 |
+
},
|
| 1141 |
+
{
|
| 1142 |
+
"epoch": 114.0,
|
| 1143 |
+
"eval_bleu": 9.4811,
|
| 1144 |
+
"eval_gen_len": 33.3333,
|
| 1145 |
+
"eval_loss": 3.277334690093994,
|
| 1146 |
+
"eval_runtime": 6.9521,
|
| 1147 |
+
"eval_samples_per_second": 0.432,
|
| 1148 |
+
"eval_steps_per_second": 0.144,
|
| 1149 |
+
"step": 114
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 115.0,
|
| 1153 |
+
"eval_bleu": 8.2476,
|
| 1154 |
+
"eval_gen_len": 33.0,
|
| 1155 |
+
"eval_loss": 3.277435541152954,
|
| 1156 |
+
"eval_runtime": 5.7965,
|
| 1157 |
+
"eval_samples_per_second": 0.518,
|
| 1158 |
+
"eval_steps_per_second": 0.173,
|
| 1159 |
+
"step": 115
|
| 1160 |
+
},
|
| 1161 |
+
{
|
| 1162 |
+
"epoch": 116.0,
|
| 1163 |
+
"eval_bleu": 8.2476,
|
| 1164 |
+
"eval_gen_len": 33.0,
|
| 1165 |
+
"eval_loss": 3.2774922847747803,
|
| 1166 |
+
"eval_runtime": 5.7606,
|
| 1167 |
+
"eval_samples_per_second": 0.521,
|
| 1168 |
+
"eval_steps_per_second": 0.174,
|
| 1169 |
+
"step": 116
|
| 1170 |
+
},
|
| 1171 |
+
{
|
| 1172 |
+
"epoch": 117.0,
|
| 1173 |
+
"eval_bleu": 8.2476,
|
| 1174 |
+
"eval_gen_len": 33.0,
|
| 1175 |
+
"eval_loss": 3.277543306350708,
|
| 1176 |
+
"eval_runtime": 6.9958,
|
| 1177 |
+
"eval_samples_per_second": 0.429,
|
| 1178 |
+
"eval_steps_per_second": 0.143,
|
| 1179 |
+
"step": 117
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 118.0,
|
| 1183 |
+
"eval_bleu": 8.2476,
|
| 1184 |
+
"eval_gen_len": 33.0,
|
| 1185 |
+
"eval_loss": 3.2776224613189697,
|
| 1186 |
+
"eval_runtime": 6.6883,
|
| 1187 |
+
"eval_samples_per_second": 0.449,
|
| 1188 |
+
"eval_steps_per_second": 0.15,
|
| 1189 |
+
"step": 118
|
| 1190 |
+
},
|
| 1191 |
+
{
|
| 1192 |
+
"epoch": 119.0,
|
| 1193 |
+
"eval_bleu": 8.2476,
|
| 1194 |
+
"eval_gen_len": 33.0,
|
| 1195 |
+
"eval_loss": 3.2776474952697754,
|
| 1196 |
+
"eval_runtime": 5.7558,
|
| 1197 |
+
"eval_samples_per_second": 0.521,
|
| 1198 |
+
"eval_steps_per_second": 0.174,
|
| 1199 |
+
"step": 119
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 120.0,
|
| 1203 |
+
"eval_bleu": 8.2476,
|
| 1204 |
+
"eval_gen_len": 33.0,
|
| 1205 |
+
"eval_loss": 3.2776882648468018,
|
| 1206 |
+
"eval_runtime": 5.9124,
|
| 1207 |
+
"eval_samples_per_second": 0.507,
|
| 1208 |
+
"eval_steps_per_second": 0.169,
|
| 1209 |
+
"step": 120
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"epoch": 121.0,
|
| 1213 |
+
"eval_bleu": 8.2476,
|
| 1214 |
+
"eval_gen_len": 33.0,
|
| 1215 |
+
"eval_loss": 3.2777838706970215,
|
| 1216 |
+
"eval_runtime": 7.011,
|
| 1217 |
+
"eval_samples_per_second": 0.428,
|
| 1218 |
+
"eval_steps_per_second": 0.143,
|
| 1219 |
+
"step": 121
|
| 1220 |
+
},
|
| 1221 |
+
{
|
| 1222 |
+
"epoch": 122.0,
|
| 1223 |
+
"eval_bleu": 8.2476,
|
| 1224 |
+
"eval_gen_len": 33.0,
|
| 1225 |
+
"eval_loss": 3.2778778076171875,
|
| 1226 |
+
"eval_runtime": 6.4551,
|
| 1227 |
+
"eval_samples_per_second": 0.465,
|
| 1228 |
+
"eval_steps_per_second": 0.155,
|
| 1229 |
+
"step": 122
|
| 1230 |
+
},
|
| 1231 |
+
{
|
| 1232 |
+
"epoch": 123.0,
|
| 1233 |
+
"eval_bleu": 8.2476,
|
| 1234 |
+
"eval_gen_len": 33.0,
|
| 1235 |
+
"eval_loss": 3.277864933013916,
|
| 1236 |
+
"eval_runtime": 5.8337,
|
| 1237 |
+
"eval_samples_per_second": 0.514,
|
| 1238 |
+
"eval_steps_per_second": 0.171,
|
| 1239 |
+
"step": 123
|
| 1240 |
+
},
|
| 1241 |
+
{
|
| 1242 |
+
"epoch": 124.0,
|
| 1243 |
+
"eval_bleu": 8.2476,
|
| 1244 |
+
"eval_gen_len": 33.0,
|
| 1245 |
+
"eval_loss": 3.277876853942871,
|
| 1246 |
+
"eval_runtime": 6.2362,
|
| 1247 |
+
"eval_samples_per_second": 0.481,
|
| 1248 |
+
"eval_steps_per_second": 0.16,
|
| 1249 |
+
"step": 124
|
| 1250 |
+
},
|
| 1251 |
+
{
|
| 1252 |
+
"epoch": 125.0,
|
| 1253 |
+
"eval_bleu": 8.2476,
|
| 1254 |
+
"eval_gen_len": 33.0,
|
| 1255 |
+
"eval_loss": 3.277902603149414,
|
| 1256 |
+
"eval_runtime": 6.928,
|
| 1257 |
+
"eval_samples_per_second": 0.433,
|
| 1258 |
+
"eval_steps_per_second": 0.144,
|
| 1259 |
+
"step": 125
|
| 1260 |
+
},
|
| 1261 |
+
{
|
| 1262 |
+
"epoch": 126.0,
|
| 1263 |
+
"eval_bleu": 8.2476,
|
| 1264 |
+
"eval_gen_len": 33.0,
|
| 1265 |
+
"eval_loss": 3.277930974960327,
|
| 1266 |
+
"eval_runtime": 6.3826,
|
| 1267 |
+
"eval_samples_per_second": 0.47,
|
| 1268 |
+
"eval_steps_per_second": 0.157,
|
| 1269 |
+
"step": 126
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 127.0,
|
| 1273 |
+
"eval_bleu": 8.2476,
|
| 1274 |
+
"eval_gen_len": 33.0,
|
| 1275 |
+
"eval_loss": 3.2780230045318604,
|
| 1276 |
+
"eval_runtime": 5.8018,
|
| 1277 |
+
"eval_samples_per_second": 0.517,
|
| 1278 |
+
"eval_steps_per_second": 0.172,
|
| 1279 |
+
"step": 127
|
| 1280 |
+
},
|
| 1281 |
+
{
|
| 1282 |
+
"epoch": 128.0,
|
| 1283 |
+
"eval_bleu": 8.2476,
|
| 1284 |
+
"eval_gen_len": 33.0,
|
| 1285 |
+
"eval_loss": 3.2781319618225098,
|
| 1286 |
+
"eval_runtime": 6.3692,
|
| 1287 |
+
"eval_samples_per_second": 0.471,
|
| 1288 |
+
"eval_steps_per_second": 0.157,
|
| 1289 |
+
"step": 128
|
| 1290 |
+
},
|
| 1291 |
+
{
|
| 1292 |
+
"epoch": 129.0,
|
| 1293 |
+
"eval_bleu": 8.2476,
|
| 1294 |
+
"eval_gen_len": 33.0,
|
| 1295 |
+
"eval_loss": 3.278233528137207,
|
| 1296 |
+
"eval_runtime": 6.9854,
|
| 1297 |
+
"eval_samples_per_second": 0.429,
|
| 1298 |
+
"eval_steps_per_second": 0.143,
|
| 1299 |
+
"step": 129
|
| 1300 |
+
},
|
| 1301 |
+
{
|
| 1302 |
+
"epoch": 130.0,
|
| 1303 |
+
"eval_bleu": 8.2476,
|
| 1304 |
+
"eval_gen_len": 33.0,
|
| 1305 |
+
"eval_loss": 3.278286933898926,
|
| 1306 |
+
"eval_runtime": 5.9741,
|
| 1307 |
+
"eval_samples_per_second": 0.502,
|
| 1308 |
+
"eval_steps_per_second": 0.167,
|
| 1309 |
+
"step": 130
|
| 1310 |
+
},
|
| 1311 |
+
{
|
| 1312 |
+
"epoch": 131.0,
|
| 1313 |
+
"eval_bleu": 8.2476,
|
| 1314 |
+
"eval_gen_len": 33.0,
|
| 1315 |
+
"eval_loss": 3.2783098220825195,
|
| 1316 |
+
"eval_runtime": 5.7425,
|
| 1317 |
+
"eval_samples_per_second": 0.522,
|
| 1318 |
+
"eval_steps_per_second": 0.174,
|
| 1319 |
+
"step": 131
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 132.0,
|
| 1323 |
+
"eval_bleu": 8.1803,
|
| 1324 |
+
"eval_gen_len": 33.0,
|
| 1325 |
+
"eval_loss": 3.278249740600586,
|
| 1326 |
+
"eval_runtime": 6.8038,
|
| 1327 |
+
"eval_samples_per_second": 0.441,
|
| 1328 |
+
"eval_steps_per_second": 0.147,
|
| 1329 |
+
"step": 132
|
| 1330 |
+
},
|
| 1331 |
+
{
|
| 1332 |
+
"epoch": 133.0,
|
| 1333 |
+
"eval_bleu": 8.1803,
|
| 1334 |
+
"eval_gen_len": 33.0,
|
| 1335 |
+
"eval_loss": 3.2781460285186768,
|
| 1336 |
+
"eval_runtime": 7.0042,
|
| 1337 |
+
"eval_samples_per_second": 0.428,
|
| 1338 |
+
"eval_steps_per_second": 0.143,
|
| 1339 |
+
"step": 133
|
| 1340 |
+
},
|
| 1341 |
+
{
|
| 1342 |
+
"epoch": 134.0,
|
| 1343 |
+
"eval_bleu": 8.1803,
|
| 1344 |
+
"eval_gen_len": 33.0,
|
| 1345 |
+
"eval_loss": 3.2780065536499023,
|
| 1346 |
+
"eval_runtime": 6.4337,
|
| 1347 |
+
"eval_samples_per_second": 0.466,
|
| 1348 |
+
"eval_steps_per_second": 0.155,
|
| 1349 |
+
"step": 134
|
| 1350 |
+
},
|
| 1351 |
+
{
|
| 1352 |
+
"epoch": 135.0,
|
| 1353 |
+
"eval_bleu": 8.1803,
|
| 1354 |
+
"eval_gen_len": 33.0,
|
| 1355 |
+
"eval_loss": 3.277865409851074,
|
| 1356 |
+
"eval_runtime": 5.7507,
|
| 1357 |
+
"eval_samples_per_second": 0.522,
|
| 1358 |
+
"eval_steps_per_second": 0.174,
|
| 1359 |
+
"step": 135
|
| 1360 |
+
},
|
| 1361 |
+
{
|
| 1362 |
+
"epoch": 136.0,
|
| 1363 |
+
"eval_bleu": 8.1803,
|
| 1364 |
+
"eval_gen_len": 33.0,
|
| 1365 |
+
"eval_loss": 3.277766227722168,
|
| 1366 |
+
"eval_runtime": 6.314,
|
| 1367 |
+
"eval_samples_per_second": 0.475,
|
| 1368 |
+
"eval_steps_per_second": 0.158,
|
| 1369 |
+
"step": 136
|
| 1370 |
+
},
|
| 1371 |
+
{
|
| 1372 |
+
"epoch": 137.0,
|
| 1373 |
+
"eval_bleu": 8.1803,
|
| 1374 |
+
"eval_gen_len": 33.0,
|
| 1375 |
+
"eval_loss": 3.277714252471924,
|
| 1376 |
+
"eval_runtime": 7.0339,
|
| 1377 |
+
"eval_samples_per_second": 0.427,
|
| 1378 |
+
"eval_steps_per_second": 0.142,
|
| 1379 |
+
"step": 137
|
| 1380 |
+
},
|
| 1381 |
+
{
|
| 1382 |
+
"epoch": 138.0,
|
| 1383 |
+
"eval_bleu": 8.1803,
|
| 1384 |
+
"eval_gen_len": 33.0,
|
| 1385 |
+
"eval_loss": 3.2776453495025635,
|
| 1386 |
+
"eval_runtime": 6.2724,
|
| 1387 |
+
"eval_samples_per_second": 0.478,
|
| 1388 |
+
"eval_steps_per_second": 0.159,
|
| 1389 |
+
"step": 138
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 139.0,
|
| 1393 |
+
"eval_bleu": 8.1803,
|
| 1394 |
+
"eval_gen_len": 33.0,
|
| 1395 |
+
"eval_loss": 3.277595281600952,
|
| 1396 |
+
"eval_runtime": 5.808,
|
| 1397 |
+
"eval_samples_per_second": 0.517,
|
| 1398 |
+
"eval_steps_per_second": 0.172,
|
| 1399 |
+
"step": 139
|
| 1400 |
+
},
|
| 1401 |
+
{
|
| 1402 |
+
"epoch": 140.0,
|
| 1403 |
+
"eval_bleu": 8.1803,
|
| 1404 |
+
"eval_gen_len": 33.0,
|
| 1405 |
+
"eval_loss": 3.2775936126708984,
|
| 1406 |
+
"eval_runtime": 6.5459,
|
| 1407 |
+
"eval_samples_per_second": 0.458,
|
| 1408 |
+
"eval_steps_per_second": 0.153,
|
| 1409 |
+
"step": 140
|
| 1410 |
+
},
|
| 1411 |
+
{
|
| 1412 |
+
"epoch": 141.0,
|
| 1413 |
+
"eval_bleu": 8.1803,
|
| 1414 |
+
"eval_gen_len": 33.0,
|
| 1415 |
+
"eval_loss": 3.2775943279266357,
|
| 1416 |
+
"eval_runtime": 6.9577,
|
| 1417 |
+
"eval_samples_per_second": 0.431,
|
| 1418 |
+
"eval_steps_per_second": 0.144,
|
| 1419 |
+
"step": 141
|
| 1420 |
+
},
|
| 1421 |
+
{
|
| 1422 |
+
"epoch": 142.0,
|
| 1423 |
+
"eval_bleu": 8.1803,
|
| 1424 |
+
"eval_gen_len": 33.0,
|
| 1425 |
+
"eval_loss": 3.277588129043579,
|
| 1426 |
+
"eval_runtime": 5.8955,
|
| 1427 |
+
"eval_samples_per_second": 0.509,
|
| 1428 |
+
"eval_steps_per_second": 0.17,
|
| 1429 |
+
"step": 142
|
| 1430 |
+
},
|
| 1431 |
+
{
|
| 1432 |
+
"epoch": 143.0,
|
| 1433 |
+
"eval_bleu": 8.1803,
|
| 1434 |
+
"eval_gen_len": 33.0,
|
| 1435 |
+
"eval_loss": 3.2775893211364746,
|
| 1436 |
+
"eval_runtime": 5.8029,
|
| 1437 |
+
"eval_samples_per_second": 0.517,
|
| 1438 |
+
"eval_steps_per_second": 0.172,
|
| 1439 |
+
"step": 143
|
| 1440 |
+
},
|
| 1441 |
+
{
|
| 1442 |
+
"epoch": 144.0,
|
| 1443 |
+
"eval_bleu": 8.1803,
|
| 1444 |
+
"eval_gen_len": 33.0,
|
| 1445 |
+
"eval_loss": 3.2775719165802,
|
| 1446 |
+
"eval_runtime": 7.9979,
|
| 1447 |
+
"eval_samples_per_second": 0.375,
|
| 1448 |
+
"eval_steps_per_second": 0.125,
|
| 1449 |
+
"step": 144
|
| 1450 |
+
},
|
| 1451 |
+
{
|
| 1452 |
+
"epoch": 145.0,
|
| 1453 |
+
"eval_bleu": 8.1803,
|
| 1454 |
+
"eval_gen_len": 33.0,
|
| 1455 |
+
"eval_loss": 3.277585983276367,
|
| 1456 |
+
"eval_runtime": 6.5177,
|
| 1457 |
+
"eval_samples_per_second": 0.46,
|
| 1458 |
+
"eval_steps_per_second": 0.153,
|
| 1459 |
+
"step": 145
|
| 1460 |
+
},
|
| 1461 |
+
{
|
| 1462 |
+
"epoch": 146.0,
|
| 1463 |
+
"eval_bleu": 8.1803,
|
| 1464 |
+
"eval_gen_len": 33.0,
|
| 1465 |
+
"eval_loss": 3.2776174545288086,
|
| 1466 |
+
"eval_runtime": 5.7505,
|
| 1467 |
+
"eval_samples_per_second": 0.522,
|
| 1468 |
+
"eval_steps_per_second": 0.174,
|
| 1469 |
+
"step": 146
|
| 1470 |
+
},
|
| 1471 |
+
{
|
| 1472 |
+
"epoch": 147.0,
|
| 1473 |
+
"eval_bleu": 8.1803,
|
| 1474 |
+
"eval_gen_len": 33.0,
|
| 1475 |
+
"eval_loss": 3.277653694152832,
|
| 1476 |
+
"eval_runtime": 6.0717,
|
| 1477 |
+
"eval_samples_per_second": 0.494,
|
| 1478 |
+
"eval_steps_per_second": 0.165,
|
| 1479 |
+
"step": 147
|
| 1480 |
+
},
|
| 1481 |
+
{
|
| 1482 |
+
"epoch": 148.0,
|
| 1483 |
+
"eval_bleu": 8.1803,
|
| 1484 |
+
"eval_gen_len": 33.0,
|
| 1485 |
+
"eval_loss": 3.2776741981506348,
|
| 1486 |
+
"eval_runtime": 6.9074,
|
| 1487 |
+
"eval_samples_per_second": 0.434,
|
| 1488 |
+
"eval_steps_per_second": 0.145,
|
| 1489 |
+
"step": 148
|
| 1490 |
+
},
|
| 1491 |
+
{
|
| 1492 |
+
"epoch": 149.0,
|
| 1493 |
+
"eval_bleu": 8.1803,
|
| 1494 |
+
"eval_gen_len": 33.0,
|
| 1495 |
+
"eval_loss": 3.2777059078216553,
|
| 1496 |
+
"eval_runtime": 6.2214,
|
| 1497 |
+
"eval_samples_per_second": 0.482,
|
| 1498 |
+
"eval_steps_per_second": 0.161,
|
| 1499 |
+
"step": 149
|
| 1500 |
+
},
|
| 1501 |
+
{
|
| 1502 |
+
"epoch": 150.0,
|
| 1503 |
+
"eval_bleu": 8.1803,
|
| 1504 |
+
"eval_gen_len": 33.0,
|
| 1505 |
+
"eval_loss": 3.2777509689331055,
|
| 1506 |
+
"eval_runtime": 5.7516,
|
| 1507 |
+
"eval_samples_per_second": 0.522,
|
| 1508 |
+
"eval_steps_per_second": 0.174,
|
| 1509 |
+
"step": 150
|
| 1510 |
+
},
|
| 1511 |
+
{
|
| 1512 |
+
"epoch": 151.0,
|
| 1513 |
+
"eval_bleu": 8.1803,
|
| 1514 |
+
"eval_gen_len": 33.0,
|
| 1515 |
+
"eval_loss": 3.2778258323669434,
|
| 1516 |
+
"eval_runtime": 6.3204,
|
| 1517 |
+
"eval_samples_per_second": 0.475,
|
| 1518 |
+
"eval_steps_per_second": 0.158,
|
| 1519 |
+
"step": 151
|
| 1520 |
+
},
|
| 1521 |
+
{
|
| 1522 |
+
"epoch": 152.0,
|
| 1523 |
+
"eval_bleu": 8.1803,
|
| 1524 |
+
"eval_gen_len": 33.0,
|
| 1525 |
+
"eval_loss": 3.27789306640625,
|
| 1526 |
+
"eval_runtime": 6.9032,
|
| 1527 |
+
"eval_samples_per_second": 0.435,
|
| 1528 |
+
"eval_steps_per_second": 0.145,
|
| 1529 |
+
"step": 152
|
| 1530 |
+
},
|
| 1531 |
+
{
|
| 1532 |
+
"epoch": 153.0,
|
| 1533 |
+
"eval_bleu": 8.1803,
|
| 1534 |
+
"eval_gen_len": 33.0,
|
| 1535 |
+
"eval_loss": 3.2779464721679688,
|
| 1536 |
+
"eval_runtime": 6.0847,
|
| 1537 |
+
"eval_samples_per_second": 0.493,
|
| 1538 |
+
"eval_steps_per_second": 0.164,
|
| 1539 |
+
"step": 153
|
| 1540 |
+
},
|
| 1541 |
+
{
|
| 1542 |
+
"epoch": 154.0,
|
| 1543 |
+
"eval_bleu": 8.1803,
|
| 1544 |
+
"eval_gen_len": 33.0,
|
| 1545 |
+
"eval_loss": 3.278012752532959,
|
| 1546 |
+
"eval_runtime": 5.726,
|
| 1547 |
+
"eval_samples_per_second": 0.524,
|
| 1548 |
+
"eval_steps_per_second": 0.175,
|
| 1549 |
+
"step": 154
|
| 1550 |
+
},
|
| 1551 |
+
{
|
| 1552 |
+
"epoch": 155.0,
|
| 1553 |
+
"eval_bleu": 8.1803,
|
| 1554 |
+
"eval_gen_len": 33.0,
|
| 1555 |
+
"eval_loss": 3.278102397918701,
|
| 1556 |
+
"eval_runtime": 6.6526,
|
| 1557 |
+
"eval_samples_per_second": 0.451,
|
| 1558 |
+
"eval_steps_per_second": 0.15,
|
| 1559 |
+
"step": 155
|
| 1560 |
+
},
|
| 1561 |
+
{
|
| 1562 |
+
"epoch": 156.0,
|
| 1563 |
+
"eval_bleu": 8.1803,
|
| 1564 |
+
"eval_gen_len": 33.0,
|
| 1565 |
+
"eval_loss": 3.278223991394043,
|
| 1566 |
+
"eval_runtime": 6.9359,
|
| 1567 |
+
"eval_samples_per_second": 0.433,
|
| 1568 |
+
"eval_steps_per_second": 0.144,
|
| 1569 |
+
"step": 156
|
| 1570 |
+
},
|
| 1571 |
+
{
|
| 1572 |
+
"epoch": 157.0,
|
| 1573 |
+
"eval_bleu": 8.1803,
|
| 1574 |
+
"eval_gen_len": 33.0,
|
| 1575 |
+
"eval_loss": 3.2783420085906982,
|
| 1576 |
+
"eval_runtime": 5.7917,
|
| 1577 |
+
"eval_samples_per_second": 0.518,
|
| 1578 |
+
"eval_steps_per_second": 0.173,
|
| 1579 |
+
"step": 157
|
| 1580 |
+
},
|
| 1581 |
+
{
|
| 1582 |
+
"epoch": 158.0,
|
| 1583 |
+
"eval_bleu": 8.1803,
|
| 1584 |
+
"eval_gen_len": 33.0,
|
| 1585 |
+
"eval_loss": 3.2784695625305176,
|
| 1586 |
+
"eval_runtime": 5.7905,
|
| 1587 |
+
"eval_samples_per_second": 0.518,
|
| 1588 |
+
"eval_steps_per_second": 0.173,
|
| 1589 |
+
"step": 158
|
| 1590 |
+
},
|
| 1591 |
+
{
|
| 1592 |
+
"epoch": 159.0,
|
| 1593 |
+
"eval_bleu": 8.1803,
|
| 1594 |
+
"eval_gen_len": 33.0,
|
| 1595 |
+
"eval_loss": 3.2785534858703613,
|
| 1596 |
+
"eval_runtime": 6.6377,
|
| 1597 |
+
"eval_samples_per_second": 0.452,
|
| 1598 |
+
"eval_steps_per_second": 0.151,
|
| 1599 |
+
"step": 159
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 160.0,
|
| 1603 |
+
"eval_bleu": 8.1803,
|
| 1604 |
+
"eval_gen_len": 33.0,
|
| 1605 |
+
"eval_loss": 3.2786056995391846,
|
| 1606 |
+
"eval_runtime": 6.9389,
|
| 1607 |
+
"eval_samples_per_second": 0.432,
|
| 1608 |
+
"eval_steps_per_second": 0.144,
|
| 1609 |
+
"step": 160
|
| 1610 |
+
},
|
| 1611 |
+
{
|
| 1612 |
+
"epoch": 161.0,
|
| 1613 |
+
"eval_bleu": 8.1803,
|
| 1614 |
+
"eval_gen_len": 33.0,
|
| 1615 |
+
"eval_loss": 3.278670310974121,
|
| 1616 |
+
"eval_runtime": 5.9014,
|
| 1617 |
+
"eval_samples_per_second": 0.508,
|
| 1618 |
+
"eval_steps_per_second": 0.169,
|
| 1619 |
+
"step": 161
|
| 1620 |
+
},
|
| 1621 |
+
{
|
| 1622 |
+
"epoch": 162.0,
|
| 1623 |
+
"eval_bleu": 8.1803,
|
| 1624 |
+
"eval_gen_len": 33.0,
|
| 1625 |
+
"eval_loss": 3.278756618499756,
|
| 1626 |
+
"eval_runtime": 5.8521,
|
| 1627 |
+
"eval_samples_per_second": 0.513,
|
| 1628 |
+
"eval_steps_per_second": 0.171,
|
| 1629 |
+
"step": 162
|
| 1630 |
+
},
|
| 1631 |
+
{
|
| 1632 |
+
"epoch": 163.0,
|
| 1633 |
+
"eval_bleu": 8.1803,
|
| 1634 |
+
"eval_gen_len": 33.0,
|
| 1635 |
+
"eval_loss": 3.2788338661193848,
|
| 1636 |
+
"eval_runtime": 6.8527,
|
| 1637 |
+
"eval_samples_per_second": 0.438,
|
| 1638 |
+
"eval_steps_per_second": 0.146,
|
| 1639 |
+
"step": 163
|
| 1640 |
+
},
|
| 1641 |
+
{
|
| 1642 |
+
"epoch": 164.0,
|
| 1643 |
+
"eval_bleu": 8.1803,
|
| 1644 |
+
"eval_gen_len": 33.0,
|
| 1645 |
+
"eval_loss": 3.2789201736450195,
|
| 1646 |
+
"eval_runtime": 6.7198,
|
| 1647 |
+
"eval_samples_per_second": 0.446,
|
| 1648 |
+
"eval_steps_per_second": 0.149,
|
| 1649 |
+
"step": 164
|
| 1650 |
+
},
|
| 1651 |
+
{
|
| 1652 |
+
"epoch": 165.0,
|
| 1653 |
+
"eval_bleu": 8.1803,
|
| 1654 |
+
"eval_gen_len": 33.0,
|
| 1655 |
+
"eval_loss": 3.2789995670318604,
|
| 1656 |
+
"eval_runtime": 7.4502,
|
| 1657 |
+
"eval_samples_per_second": 0.403,
|
| 1658 |
+
"eval_steps_per_second": 0.134,
|
| 1659 |
+
"step": 165
|
| 1660 |
+
},
|
| 1661 |
+
{
|
| 1662 |
+
"epoch": 166.0,
|
| 1663 |
+
"eval_bleu": 8.1803,
|
| 1664 |
+
"eval_gen_len": 33.0,
|
| 1665 |
+
"eval_loss": 3.2790791988372803,
|
| 1666 |
+
"eval_runtime": 5.8218,
|
| 1667 |
+
"eval_samples_per_second": 0.515,
|
| 1668 |
+
"eval_steps_per_second": 0.172,
|
| 1669 |
+
"step": 166
|
| 1670 |
+
},
|
| 1671 |
+
{
|
| 1672 |
+
"epoch": 167.0,
|
| 1673 |
+
"eval_bleu": 8.1803,
|
| 1674 |
+
"eval_gen_len": 33.0,
|
| 1675 |
+
"eval_loss": 3.2791695594787598,
|
| 1676 |
+
"eval_runtime": 6.2315,
|
| 1677 |
+
"eval_samples_per_second": 0.481,
|
| 1678 |
+
"eval_steps_per_second": 0.16,
|
| 1679 |
+
"step": 167
|
| 1680 |
+
},
|
| 1681 |
+
{
|
| 1682 |
+
"epoch": 168.0,
|
| 1683 |
+
"eval_bleu": 8.1803,
|
| 1684 |
+
"eval_gen_len": 33.0,
|
| 1685 |
+
"eval_loss": 3.2792553901672363,
|
| 1686 |
+
"eval_runtime": 6.9869,
|
| 1687 |
+
"eval_samples_per_second": 0.429,
|
| 1688 |
+
"eval_steps_per_second": 0.143,
|
| 1689 |
+
"step": 168
|
| 1690 |
+
},
|
| 1691 |
+
{
|
| 1692 |
+
"epoch": 169.0,
|
| 1693 |
+
"eval_bleu": 8.1803,
|
| 1694 |
+
"eval_gen_len": 33.0,
|
| 1695 |
+
"eval_loss": 3.2793426513671875,
|
| 1696 |
+
"eval_runtime": 6.2576,
|
| 1697 |
+
"eval_samples_per_second": 0.479,
|
| 1698 |
+
"eval_steps_per_second": 0.16,
|
| 1699 |
+
"step": 169
|
| 1700 |
+
},
|
| 1701 |
+
{
|
| 1702 |
+
"epoch": 170.0,
|
| 1703 |
+
"eval_bleu": 8.1803,
|
| 1704 |
+
"eval_gen_len": 33.0,
|
| 1705 |
+
"eval_loss": 3.2794339656829834,
|
| 1706 |
+
"eval_runtime": 5.7989,
|
| 1707 |
+
"eval_samples_per_second": 0.517,
|
| 1708 |
+
"eval_steps_per_second": 0.172,
|
| 1709 |
+
"step": 170
|
| 1710 |
+
},
|
| 1711 |
+
{
|
| 1712 |
+
"epoch": 171.0,
|
| 1713 |
+
"eval_bleu": 8.1803,
|
| 1714 |
+
"eval_gen_len": 33.0,
|
| 1715 |
+
"eval_loss": 3.2795324325561523,
|
| 1716 |
+
"eval_runtime": 6.4942,
|
| 1717 |
+
"eval_samples_per_second": 0.462,
|
| 1718 |
+
"eval_steps_per_second": 0.154,
|
| 1719 |
+
"step": 171
|
| 1720 |
+
},
|
| 1721 |
+
{
|
| 1722 |
+
"epoch": 172.0,
|
| 1723 |
+
"eval_bleu": 8.1803,
|
| 1724 |
+
"eval_gen_len": 33.0,
|
| 1725 |
+
"eval_loss": 3.279637098312378,
|
| 1726 |
+
"eval_runtime": 6.9341,
|
| 1727 |
+
"eval_samples_per_second": 0.433,
|
| 1728 |
+
"eval_steps_per_second": 0.144,
|
| 1729 |
+
"step": 172
|
| 1730 |
+
},
|
| 1731 |
+
{
|
| 1732 |
+
"epoch": 173.0,
|
| 1733 |
+
"eval_bleu": 8.1803,
|
| 1734 |
+
"eval_gen_len": 33.0,
|
| 1735 |
+
"eval_loss": 3.279742479324341,
|
| 1736 |
+
"eval_runtime": 5.9588,
|
| 1737 |
+
"eval_samples_per_second": 0.503,
|
| 1738 |
+
"eval_steps_per_second": 0.168,
|
| 1739 |
+
"step": 173
|
| 1740 |
+
},
|
| 1741 |
+
{
|
| 1742 |
+
"epoch": 174.0,
|
| 1743 |
+
"eval_bleu": 8.1803,
|
| 1744 |
+
"eval_gen_len": 33.0,
|
| 1745 |
+
"eval_loss": 3.2798376083374023,
|
| 1746 |
+
"eval_runtime": 5.8286,
|
| 1747 |
+
"eval_samples_per_second": 0.515,
|
| 1748 |
+
"eval_steps_per_second": 0.172,
|
| 1749 |
+
"step": 174
|
| 1750 |
+
},
|
| 1751 |
+
{
|
| 1752 |
+
"epoch": 175.0,
|
| 1753 |
+
"eval_bleu": 8.1803,
|
| 1754 |
+
"eval_gen_len": 33.0,
|
| 1755 |
+
"eval_loss": 3.2799010276794434,
|
| 1756 |
+
"eval_runtime": 6.7052,
|
| 1757 |
+
"eval_samples_per_second": 0.447,
|
| 1758 |
+
"eval_steps_per_second": 0.149,
|
| 1759 |
+
"step": 175
|
| 1760 |
+
},
|
| 1761 |
+
{
|
| 1762 |
+
"epoch": 176.0,
|
| 1763 |
+
"eval_bleu": 8.1803,
|
| 1764 |
+
"eval_gen_len": 33.0,
|
| 1765 |
+
"eval_loss": 3.279953718185425,
|
| 1766 |
+
"eval_runtime": 6.7267,
|
| 1767 |
+
"eval_samples_per_second": 0.446,
|
| 1768 |
+
"eval_steps_per_second": 0.149,
|
| 1769 |
+
"step": 176
|
| 1770 |
+
},
|
| 1771 |
+
{
|
| 1772 |
+
"epoch": 177.0,
|
| 1773 |
+
"eval_bleu": 8.1803,
|
| 1774 |
+
"eval_gen_len": 33.0,
|
| 1775 |
+
"eval_loss": 3.280021905899048,
|
| 1776 |
+
"eval_runtime": 7.6558,
|
| 1777 |
+
"eval_samples_per_second": 0.392,
|
| 1778 |
+
"eval_steps_per_second": 0.131,
|
| 1779 |
+
"step": 177
|
| 1780 |
+
},
|
| 1781 |
+
{
|
| 1782 |
+
"epoch": 178.0,
|
| 1783 |
+
"eval_bleu": 8.1803,
|
| 1784 |
+
"eval_gen_len": 33.0,
|
| 1785 |
+
"eval_loss": 3.2801008224487305,
|
| 1786 |
+
"eval_runtime": 5.7795,
|
| 1787 |
+
"eval_samples_per_second": 0.519,
|
| 1788 |
+
"eval_steps_per_second": 0.173,
|
| 1789 |
+
"step": 178
|
| 1790 |
+
},
|
| 1791 |
+
{
|
| 1792 |
+
"epoch": 179.0,
|
| 1793 |
+
"eval_bleu": 8.1803,
|
| 1794 |
+
"eval_gen_len": 33.0,
|
| 1795 |
+
"eval_loss": 3.2801709175109863,
|
| 1796 |
+
"eval_runtime": 5.9609,
|
| 1797 |
+
"eval_samples_per_second": 0.503,
|
| 1798 |
+
"eval_steps_per_second": 0.168,
|
| 1799 |
+
"step": 179
|
| 1800 |
+
},
|
| 1801 |
+
{
|
| 1802 |
+
"epoch": 180.0,
|
| 1803 |
+
"eval_bleu": 8.1803,
|
| 1804 |
+
"eval_gen_len": 33.0,
|
| 1805 |
+
"eval_loss": 3.2802419662475586,
|
| 1806 |
+
"eval_runtime": 6.974,
|
| 1807 |
+
"eval_samples_per_second": 0.43,
|
| 1808 |
+
"eval_steps_per_second": 0.143,
|
| 1809 |
+
"step": 180
|
| 1810 |
+
},
|
| 1811 |
+
{
|
| 1812 |
+
"epoch": 181.0,
|
| 1813 |
+
"eval_bleu": 8.1803,
|
| 1814 |
+
"eval_gen_len": 33.0,
|
| 1815 |
+
"eval_loss": 3.280299425125122,
|
| 1816 |
+
"eval_runtime": 6.4189,
|
| 1817 |
+
"eval_samples_per_second": 0.467,
|
| 1818 |
+
"eval_steps_per_second": 0.156,
|
| 1819 |
+
"step": 181
|
| 1820 |
+
},
|
| 1821 |
+
{
|
| 1822 |
+
"epoch": 182.0,
|
| 1823 |
+
"eval_bleu": 8.1803,
|
| 1824 |
+
"eval_gen_len": 33.0,
|
| 1825 |
+
"eval_loss": 3.2803618907928467,
|
| 1826 |
+
"eval_runtime": 5.8011,
|
| 1827 |
+
"eval_samples_per_second": 0.517,
|
| 1828 |
+
"eval_steps_per_second": 0.172,
|
| 1829 |
+
"step": 182
|
| 1830 |
+
},
|
| 1831 |
+
{
|
| 1832 |
+
"epoch": 183.0,
|
| 1833 |
+
"eval_bleu": 8.1803,
|
| 1834 |
+
"eval_gen_len": 33.0,
|
| 1835 |
+
"eval_loss": 3.280426025390625,
|
| 1836 |
+
"eval_runtime": 6.3455,
|
| 1837 |
+
"eval_samples_per_second": 0.473,
|
| 1838 |
+
"eval_steps_per_second": 0.158,
|
| 1839 |
+
"step": 183
|
| 1840 |
+
},
|
| 1841 |
+
{
|
| 1842 |
+
"epoch": 184.0,
|
| 1843 |
+
"eval_bleu": 8.1803,
|
| 1844 |
+
"eval_gen_len": 33.0,
|
| 1845 |
+
"eval_loss": 3.280491590499878,
|
| 1846 |
+
"eval_runtime": 6.9298,
|
| 1847 |
+
"eval_samples_per_second": 0.433,
|
| 1848 |
+
"eval_steps_per_second": 0.144,
|
| 1849 |
+
"step": 184
|
| 1850 |
+
},
|
| 1851 |
+
{
|
| 1852 |
+
"epoch": 185.0,
|
| 1853 |
+
"eval_bleu": 8.1803,
|
| 1854 |
+
"eval_gen_len": 33.0,
|
| 1855 |
+
"eval_loss": 3.2805566787719727,
|
| 1856 |
+
"eval_runtime": 6.0973,
|
| 1857 |
+
"eval_samples_per_second": 0.492,
|
| 1858 |
+
"eval_steps_per_second": 0.164,
|
| 1859 |
+
"step": 185
|
| 1860 |
+
},
|
| 1861 |
+
{
|
| 1862 |
+
"epoch": 186.0,
|
| 1863 |
+
"eval_bleu": 8.1803,
|
| 1864 |
+
"eval_gen_len": 33.0,
|
| 1865 |
+
"eval_loss": 3.280604839324951,
|
| 1866 |
+
"eval_runtime": 5.7806,
|
| 1867 |
+
"eval_samples_per_second": 0.519,
|
| 1868 |
+
"eval_steps_per_second": 0.173,
|
| 1869 |
+
"step": 186
|
| 1870 |
+
},
|
| 1871 |
+
{
|
| 1872 |
+
"epoch": 187.0,
|
| 1873 |
+
"eval_bleu": 8.1803,
|
| 1874 |
+
"eval_gen_len": 33.0,
|
| 1875 |
+
"eval_loss": 3.280635356903076,
|
| 1876 |
+
"eval_runtime": 6.4001,
|
| 1877 |
+
"eval_samples_per_second": 0.469,
|
| 1878 |
+
"eval_steps_per_second": 0.156,
|
| 1879 |
+
"step": 187
|
| 1880 |
+
},
|
| 1881 |
+
{
|
| 1882 |
+
"epoch": 188.0,
|
| 1883 |
+
"eval_bleu": 8.1803,
|
| 1884 |
+
"eval_gen_len": 33.0,
|
| 1885 |
+
"eval_loss": 3.2806549072265625,
|
| 1886 |
+
"eval_runtime": 6.9369,
|
| 1887 |
+
"eval_samples_per_second": 0.432,
|
| 1888 |
+
"eval_steps_per_second": 0.144,
|
| 1889 |
+
"step": 188
|
| 1890 |
+
},
|
| 1891 |
+
{
|
| 1892 |
+
"epoch": 189.0,
|
| 1893 |
+
"eval_bleu": 8.1803,
|
| 1894 |
+
"eval_gen_len": 33.0,
|
| 1895 |
+
"eval_loss": 3.2806756496429443,
|
| 1896 |
+
"eval_runtime": 6.0988,
|
| 1897 |
+
"eval_samples_per_second": 0.492,
|
| 1898 |
+
"eval_steps_per_second": 0.164,
|
| 1899 |
+
"step": 189
|
| 1900 |
+
},
|
| 1901 |
+
{
|
| 1902 |
+
"epoch": 190.0,
|
| 1903 |
+
"eval_bleu": 8.1803,
|
| 1904 |
+
"eval_gen_len": 33.0,
|
| 1905 |
+
"eval_loss": 3.2806994915008545,
|
| 1906 |
+
"eval_runtime": 5.7929,
|
| 1907 |
+
"eval_samples_per_second": 0.518,
|
| 1908 |
+
"eval_steps_per_second": 0.173,
|
| 1909 |
+
"step": 190
|
| 1910 |
+
},
|
| 1911 |
+
{
|
| 1912 |
+
"epoch": 191.0,
|
| 1913 |
+
"eval_bleu": 8.1803,
|
| 1914 |
+
"eval_gen_len": 33.0,
|
| 1915 |
+
"eval_loss": 3.280724287033081,
|
| 1916 |
+
"eval_runtime": 6.5767,
|
| 1917 |
+
"eval_samples_per_second": 0.456,
|
| 1918 |
+
"eval_steps_per_second": 0.152,
|
| 1919 |
+
"step": 191
|
| 1920 |
+
},
|
| 1921 |
+
{
|
| 1922 |
+
"epoch": 192.0,
|
| 1923 |
+
"eval_bleu": 8.1803,
|
| 1924 |
+
"eval_gen_len": 33.0,
|
| 1925 |
+
"eval_loss": 3.280754804611206,
|
| 1926 |
+
"eval_runtime": 6.94,
|
| 1927 |
+
"eval_samples_per_second": 0.432,
|
| 1928 |
+
"eval_steps_per_second": 0.144,
|
| 1929 |
+
"step": 192
|
| 1930 |
+
},
|
| 1931 |
+
{
|
| 1932 |
+
"epoch": 193.0,
|
| 1933 |
+
"eval_bleu": 8.1803,
|
| 1934 |
+
"eval_gen_len": 33.0,
|
| 1935 |
+
"eval_loss": 3.280787706375122,
|
| 1936 |
+
"eval_runtime": 5.867,
|
| 1937 |
+
"eval_samples_per_second": 0.511,
|
| 1938 |
+
"eval_steps_per_second": 0.17,
|
| 1939 |
+
"step": 193
|
| 1940 |
+
},
|
| 1941 |
+
{
|
| 1942 |
+
"epoch": 194.0,
|
| 1943 |
+
"eval_bleu": 8.1803,
|
| 1944 |
+
"eval_gen_len": 33.0,
|
| 1945 |
+
"eval_loss": 3.280816078186035,
|
| 1946 |
+
"eval_runtime": 5.8122,
|
| 1947 |
+
"eval_samples_per_second": 0.516,
|
| 1948 |
+
"eval_steps_per_second": 0.172,
|
| 1949 |
+
"step": 194
|
| 1950 |
+
},
|
| 1951 |
+
{
|
| 1952 |
+
"epoch": 195.0,
|
| 1953 |
+
"eval_bleu": 8.1803,
|
| 1954 |
+
"eval_gen_len": 33.0,
|
| 1955 |
+
"eval_loss": 3.2808401584625244,
|
| 1956 |
+
"eval_runtime": 6.7074,
|
| 1957 |
+
"eval_samples_per_second": 0.447,
|
| 1958 |
+
"eval_steps_per_second": 0.149,
|
| 1959 |
+
"step": 195
|
| 1960 |
+
},
|
| 1961 |
+
{
|
| 1962 |
+
"epoch": 196.0,
|
| 1963 |
+
"eval_bleu": 8.1803,
|
| 1964 |
+
"eval_gen_len": 33.0,
|
| 1965 |
+
"eval_loss": 3.2808632850646973,
|
| 1966 |
+
"eval_runtime": 6.9623,
|
| 1967 |
+
"eval_samples_per_second": 0.431,
|
| 1968 |
+
"eval_steps_per_second": 0.144,
|
| 1969 |
+
"step": 196
|
| 1970 |
+
},
|
| 1971 |
+
{
|
| 1972 |
+
"epoch": 197.0,
|
| 1973 |
+
"eval_bleu": 8.1803,
|
| 1974 |
+
"eval_gen_len": 33.0,
|
| 1975 |
+
"eval_loss": 3.2808849811553955,
|
| 1976 |
+
"eval_runtime": 7.5232,
|
| 1977 |
+
"eval_samples_per_second": 0.399,
|
| 1978 |
+
"eval_steps_per_second": 0.133,
|
| 1979 |
+
"step": 197
|
| 1980 |
+
},
|
| 1981 |
+
{
|
| 1982 |
+
"epoch": 198.0,
|
| 1983 |
+
"eval_bleu": 8.1803,
|
| 1984 |
+
"eval_gen_len": 33.0,
|
| 1985 |
+
"eval_loss": 3.2809011936187744,
|
| 1986 |
+
"eval_runtime": 5.7406,
|
| 1987 |
+
"eval_samples_per_second": 0.523,
|
| 1988 |
+
"eval_steps_per_second": 0.174,
|
| 1989 |
+
"step": 198
|
| 1990 |
+
},
|
| 1991 |
+
{
|
| 1992 |
+
"epoch": 199.0,
|
| 1993 |
+
"eval_bleu": 8.1803,
|
| 1994 |
+
"eval_gen_len": 33.0,
|
| 1995 |
+
"eval_loss": 3.280911922454834,
|
| 1996 |
+
"eval_runtime": 6.4384,
|
| 1997 |
+
"eval_samples_per_second": 0.466,
|
| 1998 |
+
"eval_steps_per_second": 0.155,
|
| 1999 |
+
"step": 199
|
| 2000 |
+
}
|
| 2001 |
+
],
|
| 2002 |
+
"logging_steps": 500,
|
| 2003 |
+
"max_steps": 200,
|
| 2004 |
+
"num_input_tokens_seen": 0,
|
| 2005 |
+
"num_train_epochs": 200,
|
| 2006 |
+
"save_steps": 500,
|
| 2007 |
+
"stateful_callbacks": {
|
| 2008 |
+
"TrainerControl": {
|
| 2009 |
+
"args": {
|
| 2010 |
+
"should_epoch_stop": false,
|
| 2011 |
+
"should_evaluate": false,
|
| 2012 |
+
"should_log": false,
|
| 2013 |
+
"should_save": true,
|
| 2014 |
+
"should_training_stop": true
|
| 2015 |
+
},
|
| 2016 |
+
"attributes": {}
|
| 2017 |
+
}
|
| 2018 |
+
},
|
| 2019 |
+
"total_flos": 8421625036800.0,
|
| 2020 |
+
"train_batch_size": 16,
|
| 2021 |
+
"trial_name": null,
|
| 2022 |
+
"trial_params": null
|
| 2023 |
+
}
|
checkpoint-200/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8dd641c3ecc2393712db1ca82f7499ded93d486edf2f931e6ad00e5502e7a446
|
| 3 |
+
size 5432
|
config.json
ADDED
|
@@ -0,0 +1,56 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "Helsinki-NLP/opus-mt-zh-en",
|
| 3 |
+
"activation_dropout": 0.0,
|
| 4 |
+
"activation_function": "swish",
|
| 5 |
+
"add_bias_logits": false,
|
| 6 |
+
"add_final_layer_norm": false,
|
| 7 |
+
"architectures": [
|
| 8 |
+
"MarianMTModel"
|
| 9 |
+
],
|
| 10 |
+
"attention_dropout": 0.0,
|
| 11 |
+
"bos_token_id": 0,
|
| 12 |
+
"classif_dropout": 0.0,
|
| 13 |
+
"classifier_dropout": 0.0,
|
| 14 |
+
"d_model": 512,
|
| 15 |
+
"decoder_attention_heads": 8,
|
| 16 |
+
"decoder_ffn_dim": 2048,
|
| 17 |
+
"decoder_layerdrop": 0.0,
|
| 18 |
+
"decoder_layers": 6,
|
| 19 |
+
"decoder_start_token_id": 65000,
|
| 20 |
+
"decoder_vocab_size": 65001,
|
| 21 |
+
"dropout": 0.1,
|
| 22 |
+
"encoder_attention_heads": 8,
|
| 23 |
+
"encoder_ffn_dim": 2048,
|
| 24 |
+
"encoder_layerdrop": 0.0,
|
| 25 |
+
"encoder_layers": 6,
|
| 26 |
+
"eos_token_id": 0,
|
| 27 |
+
"extra_pos_embeddings": 65001,
|
| 28 |
+
"forced_eos_token_id": 0,
|
| 29 |
+
"id2label": {
|
| 30 |
+
"0": "LABEL_0",
|
| 31 |
+
"1": "LABEL_1",
|
| 32 |
+
"2": "LABEL_2"
|
| 33 |
+
},
|
| 34 |
+
"init_std": 0.02,
|
| 35 |
+
"is_encoder_decoder": true,
|
| 36 |
+
"label2id": {
|
| 37 |
+
"LABEL_0": 0,
|
| 38 |
+
"LABEL_1": 1,
|
| 39 |
+
"LABEL_2": 2
|
| 40 |
+
},
|
| 41 |
+
"max_length": null,
|
| 42 |
+
"max_position_embeddings": 512,
|
| 43 |
+
"model_type": "marian",
|
| 44 |
+
"normalize_before": false,
|
| 45 |
+
"normalize_embedding": false,
|
| 46 |
+
"num_beams": null,
|
| 47 |
+
"num_hidden_layers": 6,
|
| 48 |
+
"pad_token_id": 65000,
|
| 49 |
+
"scale_embedding": true,
|
| 50 |
+
"share_encoder_decoder_embeddings": true,
|
| 51 |
+
"static_position_embeddings": true,
|
| 52 |
+
"torch_dtype": "float32",
|
| 53 |
+
"transformers_version": "4.46.2",
|
| 54 |
+
"use_cache": true,
|
| 55 |
+
"vocab_size": 65001
|
| 56 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bad_words_ids": [
|
| 3 |
+
[
|
| 4 |
+
65000
|
| 5 |
+
]
|
| 6 |
+
],
|
| 7 |
+
"bos_token_id": 0,
|
| 8 |
+
"decoder_start_token_id": 65000,
|
| 9 |
+
"eos_token_id": 0,
|
| 10 |
+
"forced_eos_token_id": 0,
|
| 11 |
+
"max_length": 512,
|
| 12 |
+
"num_beams": 6,
|
| 13 |
+
"pad_token_id": 65000,
|
| 14 |
+
"renormalize_logits": true,
|
| 15 |
+
"transformers_version": "4.46.2"
|
| 16 |
+
}
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:056c0879a45c67724d9458dbe29f7e80bc0512df9d78bcd528604342602a417e
|
| 3 |
+
size 309965092
|
runs/Dec03_08-07-34_e9eef5465588/events.out.tfevents.1733213606.e9eef5465588.150.0
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:36a58dee3bf115e6afc5125f5da21e8660cb4005dc840c523ea27750125f4721
|
| 3 |
+
size 11886
|
runs/Dec03_08-07-34_e9eef5465588/events.out.tfevents.1733213681.e9eef5465588.150.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b3915bf89a79e1932bcba0b319fd1e6eaef86b26da2794df52d75ccedc5a3ceb
|
| 3 |
+
size 5987
|
runs/Dec03_08-07-34_e9eef5465588/events.out.tfevents.1733213721.e9eef5465588.150.2
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1270c95fe6a19bb3f2d4763ad1156036d860c0d8b2216cfbdf8a4a4dede155ba
|
| 3 |
+
size 7802
|
runs/Dec03_08-19-26_e9eef5465588/events.out.tfevents.1733214041.e9eef5465588.150.3
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:25a512735c7a36b8ad4e32064be1172dd5f4285d215a9225754446218d422107
|
| 3 |
+
size 5987
|
runs/Dec03_08-19-26_e9eef5465588/events.out.tfevents.1733214098.e9eef5465588.150.4
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b4fb15a06dc08f003461cf2ae61c6905ed9cd93222cc3cde07dd6ade03a98b2c
|
| 3 |
+
size 13595
|
runs/Dec03_08-48-25_e9eef5465588/events.out.tfevents.1733215721.e9eef5465588.150.5
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cad6ed1068827cee96099ab05baf3dc4398906fa1d399d4cb996ba048e428ed0
|
| 3 |
+
size 7062
|
runs/Dec03_08-52-34_e9eef5465588/events.out.tfevents.1733215972.e9eef5465588.150.6
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:60be4c3f3bfa5307d8b4ef6dbc7ff258dc991065b97cfae5b07dd958cac44494
|
| 3 |
+
size 79441
|
source.spm
ADDED
|
Binary file (805 kB). View file
|
|
|
special_tokens_map.json
ADDED
|
@@ -0,0 +1,5 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"eos_token": "</s>",
|
| 3 |
+
"pad_token": "<pad>",
|
| 4 |
+
"unk_token": "<unk>"
|
| 5 |
+
}
|
target.spm
ADDED
|
Binary file (807 kB). View file
|
|
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"added_tokens_decoder": {
|
| 3 |
+
"0": {
|
| 4 |
+
"content": "</s>",
|
| 5 |
+
"lstrip": false,
|
| 6 |
+
"normalized": false,
|
| 7 |
+
"rstrip": false,
|
| 8 |
+
"single_word": false,
|
| 9 |
+
"special": true
|
| 10 |
+
},
|
| 11 |
+
"1": {
|
| 12 |
+
"content": "<unk>",
|
| 13 |
+
"lstrip": false,
|
| 14 |
+
"normalized": false,
|
| 15 |
+
"rstrip": false,
|
| 16 |
+
"single_word": false,
|
| 17 |
+
"special": true
|
| 18 |
+
},
|
| 19 |
+
"65000": {
|
| 20 |
+
"content": "<pad>",
|
| 21 |
+
"lstrip": false,
|
| 22 |
+
"normalized": false,
|
| 23 |
+
"rstrip": false,
|
| 24 |
+
"single_word": false,
|
| 25 |
+
"special": true
|
| 26 |
+
}
|
| 27 |
+
},
|
| 28 |
+
"clean_up_tokenization_spaces": false,
|
| 29 |
+
"eos_token": "</s>",
|
| 30 |
+
"model_max_length": 512,
|
| 31 |
+
"pad_token": "<pad>",
|
| 32 |
+
"separate_vocabs": false,
|
| 33 |
+
"source_lang": "zho",
|
| 34 |
+
"sp_model_kwargs": {},
|
| 35 |
+
"target_lang": "eng",
|
| 36 |
+
"tokenizer_class": "MarianTokenizer",
|
| 37 |
+
"unk_token": "<unk>"
|
| 38 |
+
}
|
vocab.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|