Upload folder using huggingface_hub
Browse files- checkpoint-49000/config.json +25 -0
- checkpoint-49000/model.safetensors +3 -0
- checkpoint-49000/optimizer.pt +3 -0
- checkpoint-49000/rng_state.pth +3 -0
- checkpoint-49000/scheduler.pt +3 -0
- checkpoint-49000/special_tokens_map.json +37 -0
- checkpoint-49000/tokenizer.json +0 -0
- checkpoint-49000/tokenizer_config.json +56 -0
- checkpoint-49000/trainer_state.json +3464 -0
- checkpoint-49000/training_args.bin +3 -0
- checkpoint-49000/vocab.txt +0 -0
- config.json +25 -0
- model.safetensors +3 -0
- special_tokens_map.json +37 -0
- tokenizer.json +0 -0
- tokenizer_config.json +56 -0
- training_args.bin +3 -0
- vocab.txt +0 -0
checkpoint-49000/config.json
ADDED
|
@@ -0,0 +1,25 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"BertForMaskedLM"
|
| 4 |
+
],
|
| 5 |
+
"attention_probs_dropout_prob": 0.1,
|
| 6 |
+
"classifier_dropout": null,
|
| 7 |
+
"gradient_checkpointing": false,
|
| 8 |
+
"hidden_act": "gelu",
|
| 9 |
+
"hidden_dropout_prob": 0.1,
|
| 10 |
+
"hidden_size": 768,
|
| 11 |
+
"initializer_range": 0.02,
|
| 12 |
+
"intermediate_size": 3072,
|
| 13 |
+
"layer_norm_eps": 1e-12,
|
| 14 |
+
"max_position_embeddings": 512,
|
| 15 |
+
"model_type": "bert",
|
| 16 |
+
"num_attention_heads": 12,
|
| 17 |
+
"num_hidden_layers": 12,
|
| 18 |
+
"pad_token_id": 0,
|
| 19 |
+
"position_embedding_type": "absolute",
|
| 20 |
+
"torch_dtype": "float32",
|
| 21 |
+
"transformers_version": "4.52.4",
|
| 22 |
+
"type_vocab_size": 2,
|
| 23 |
+
"use_cache": true,
|
| 24 |
+
"vocab_size": 30522
|
| 25 |
+
}
|
checkpoint-49000/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:affcb1cfca592faa1c347e97043cf8ca146ba254dbbea326c4ffa126ab4daf3b
|
| 3 |
+
size 438080896
|
checkpoint-49000/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d012975978aeddfe69c2aae83038de0a5965725523e2d9c612f7819a6960b5b9
|
| 3 |
+
size 876283723
|
checkpoint-49000/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:61cfddeff87f5c88c4f271c5dc20a9b7a66ccc5cd1ca0442a2a24f9c5ce8acbf
|
| 3 |
+
size 14645
|
checkpoint-49000/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:76abedc2034b0a0c92107b5883b596c37b126a43b875db74099ede689665c760
|
| 3 |
+
size 1465
|
checkpoint-49000/special_tokens_map.json
ADDED
|
@@ -0,0 +1,37 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"cls_token": {
|
| 3 |
+
"content": "[CLS]",
|
| 4 |
+
"lstrip": false,
|
| 5 |
+
"normalized": false,
|
| 6 |
+
"rstrip": false,
|
| 7 |
+
"single_word": false
|
| 8 |
+
},
|
| 9 |
+
"mask_token": {
|
| 10 |
+
"content": "[MASK]",
|
| 11 |
+
"lstrip": false,
|
| 12 |
+
"normalized": false,
|
| 13 |
+
"rstrip": false,
|
| 14 |
+
"single_word": false
|
| 15 |
+
},
|
| 16 |
+
"pad_token": {
|
| 17 |
+
"content": "[PAD]",
|
| 18 |
+
"lstrip": false,
|
| 19 |
+
"normalized": false,
|
| 20 |
+
"rstrip": false,
|
| 21 |
+
"single_word": false
|
| 22 |
+
},
|
| 23 |
+
"sep_token": {
|
| 24 |
+
"content": "[SEP]",
|
| 25 |
+
"lstrip": false,
|
| 26 |
+
"normalized": false,
|
| 27 |
+
"rstrip": false,
|
| 28 |
+
"single_word": false
|
| 29 |
+
},
|
| 30 |
+
"unk_token": {
|
| 31 |
+
"content": "[UNK]",
|
| 32 |
+
"lstrip": false,
|
| 33 |
+
"normalized": false,
|
| 34 |
+
"rstrip": false,
|
| 35 |
+
"single_word": false
|
| 36 |
+
}
|
| 37 |
+
}
|
checkpoint-49000/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
checkpoint-49000/tokenizer_config.json
ADDED
|
@@ -0,0 +1,56 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"added_tokens_decoder": {
|
| 3 |
+
"0": {
|
| 4 |
+
"content": "[PAD]",
|
| 5 |
+
"lstrip": false,
|
| 6 |
+
"normalized": false,
|
| 7 |
+
"rstrip": false,
|
| 8 |
+
"single_word": false,
|
| 9 |
+
"special": true
|
| 10 |
+
},
|
| 11 |
+
"100": {
|
| 12 |
+
"content": "[UNK]",
|
| 13 |
+
"lstrip": false,
|
| 14 |
+
"normalized": false,
|
| 15 |
+
"rstrip": false,
|
| 16 |
+
"single_word": false,
|
| 17 |
+
"special": true
|
| 18 |
+
},
|
| 19 |
+
"101": {
|
| 20 |
+
"content": "[CLS]",
|
| 21 |
+
"lstrip": false,
|
| 22 |
+
"normalized": false,
|
| 23 |
+
"rstrip": false,
|
| 24 |
+
"single_word": false,
|
| 25 |
+
"special": true
|
| 26 |
+
},
|
| 27 |
+
"102": {
|
| 28 |
+
"content": "[SEP]",
|
| 29 |
+
"lstrip": false,
|
| 30 |
+
"normalized": false,
|
| 31 |
+
"rstrip": false,
|
| 32 |
+
"single_word": false,
|
| 33 |
+
"special": true
|
| 34 |
+
},
|
| 35 |
+
"103": {
|
| 36 |
+
"content": "[MASK]",
|
| 37 |
+
"lstrip": false,
|
| 38 |
+
"normalized": false,
|
| 39 |
+
"rstrip": false,
|
| 40 |
+
"single_word": false,
|
| 41 |
+
"special": true
|
| 42 |
+
}
|
| 43 |
+
},
|
| 44 |
+
"clean_up_tokenization_spaces": false,
|
| 45 |
+
"cls_token": "[CLS]",
|
| 46 |
+
"do_lower_case": true,
|
| 47 |
+
"extra_special_tokens": {},
|
| 48 |
+
"mask_token": "[MASK]",
|
| 49 |
+
"model_max_length": 512,
|
| 50 |
+
"pad_token": "[PAD]",
|
| 51 |
+
"sep_token": "[SEP]",
|
| 52 |
+
"strip_accents": null,
|
| 53 |
+
"tokenize_chinese_chars": true,
|
| 54 |
+
"tokenizer_class": "BertTokenizer",
|
| 55 |
+
"unk_token": "[UNK]"
|
| 56 |
+
}
|
checkpoint-49000/trainer_state.json
ADDED
|
@@ -0,0 +1,3464 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 49.69574036511156,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 49000,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.10141987829614604,
|
| 14 |
+
"grad_norm": 5.131224155426025,
|
| 15 |
+
"learning_rate": 4.989959432048682e-05,
|
| 16 |
+
"loss": 2.5384,
|
| 17 |
+
"step": 100
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.2028397565922921,
|
| 21 |
+
"grad_norm": 5.526261806488037,
|
| 22 |
+
"learning_rate": 4.979817444219067e-05,
|
| 23 |
+
"loss": 2.319,
|
| 24 |
+
"step": 200
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.30425963488843816,
|
| 28 |
+
"grad_norm": 4.514644622802734,
|
| 29 |
+
"learning_rate": 4.969675456389453e-05,
|
| 30 |
+
"loss": 2.2627,
|
| 31 |
+
"step": 300
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.4056795131845842,
|
| 35 |
+
"grad_norm": 4.743046283721924,
|
| 36 |
+
"learning_rate": 4.959533468559838e-05,
|
| 37 |
+
"loss": 2.1608,
|
| 38 |
+
"step": 400
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.5070993914807302,
|
| 42 |
+
"grad_norm": 5.047675609588623,
|
| 43 |
+
"learning_rate": 4.949391480730223e-05,
|
| 44 |
+
"loss": 2.1143,
|
| 45 |
+
"step": 500
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.6085192697768763,
|
| 49 |
+
"grad_norm": 5.081020355224609,
|
| 50 |
+
"learning_rate": 4.939249492900609e-05,
|
| 51 |
+
"loss": 2.0675,
|
| 52 |
+
"step": 600
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.7099391480730223,
|
| 56 |
+
"grad_norm": 4.293202877044678,
|
| 57 |
+
"learning_rate": 4.9291075050709936e-05,
|
| 58 |
+
"loss": 2.0743,
|
| 59 |
+
"step": 700
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.8113590263691683,
|
| 63 |
+
"grad_norm": 4.622530937194824,
|
| 64 |
+
"learning_rate": 4.91896551724138e-05,
|
| 65 |
+
"loss": 2.0565,
|
| 66 |
+
"step": 800
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.9127789046653144,
|
| 70 |
+
"grad_norm": 4.895691394805908,
|
| 71 |
+
"learning_rate": 4.9088235294117654e-05,
|
| 72 |
+
"loss": 1.9973,
|
| 73 |
+
"step": 900
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 1.0141987829614605,
|
| 77 |
+
"grad_norm": 4.207626819610596,
|
| 78 |
+
"learning_rate": 4.89868154158215e-05,
|
| 79 |
+
"loss": 1.9922,
|
| 80 |
+
"step": 1000
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 1.1156186612576064,
|
| 84 |
+
"grad_norm": 4.911829948425293,
|
| 85 |
+
"learning_rate": 4.888539553752536e-05,
|
| 86 |
+
"loss": 1.9538,
|
| 87 |
+
"step": 1100
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 1.2170385395537526,
|
| 91 |
+
"grad_norm": 4.566792011260986,
|
| 92 |
+
"learning_rate": 4.8783975659229206e-05,
|
| 93 |
+
"loss": 1.9223,
|
| 94 |
+
"step": 1200
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 1.3184584178498986,
|
| 98 |
+
"grad_norm": 4.203269004821777,
|
| 99 |
+
"learning_rate": 4.868255578093306e-05,
|
| 100 |
+
"loss": 1.9037,
|
| 101 |
+
"step": 1300
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 1.4198782961460445,
|
| 105 |
+
"grad_norm": 3.9944674968719482,
|
| 106 |
+
"learning_rate": 4.8581135902636924e-05,
|
| 107 |
+
"loss": 1.9126,
|
| 108 |
+
"step": 1400
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 1.5212981744421907,
|
| 112 |
+
"grad_norm": 3.877246856689453,
|
| 113 |
+
"learning_rate": 4.847971602434077e-05,
|
| 114 |
+
"loss": 1.8966,
|
| 115 |
+
"step": 1500
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 1.6227180527383367,
|
| 119 |
+
"grad_norm": 4.538036823272705,
|
| 120 |
+
"learning_rate": 4.837829614604463e-05,
|
| 121 |
+
"loss": 1.8839,
|
| 122 |
+
"step": 1600
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 1.7241379310344827,
|
| 126 |
+
"grad_norm": 4.315947532653809,
|
| 127 |
+
"learning_rate": 4.827687626774848e-05,
|
| 128 |
+
"loss": 1.8959,
|
| 129 |
+
"step": 1700
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 1.8255578093306288,
|
| 133 |
+
"grad_norm": 4.574951648712158,
|
| 134 |
+
"learning_rate": 4.817545638945233e-05,
|
| 135 |
+
"loss": 1.8806,
|
| 136 |
+
"step": 1800
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 1.9269776876267748,
|
| 140 |
+
"grad_norm": 4.086500644683838,
|
| 141 |
+
"learning_rate": 4.807403651115619e-05,
|
| 142 |
+
"loss": 1.8433,
|
| 143 |
+
"step": 1900
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 2.028397565922921,
|
| 147 |
+
"grad_norm": 3.9760191440582275,
|
| 148 |
+
"learning_rate": 4.797261663286004e-05,
|
| 149 |
+
"loss": 1.8676,
|
| 150 |
+
"step": 2000
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 2.129817444219067,
|
| 154 |
+
"grad_norm": 4.592931747436523,
|
| 155 |
+
"learning_rate": 4.78711967545639e-05,
|
| 156 |
+
"loss": 1.8049,
|
| 157 |
+
"step": 2100
|
| 158 |
+
},
|
| 159 |
+
{
|
| 160 |
+
"epoch": 2.231237322515213,
|
| 161 |
+
"grad_norm": 4.3724284172058105,
|
| 162 |
+
"learning_rate": 4.776977687626775e-05,
|
| 163 |
+
"loss": 1.8064,
|
| 164 |
+
"step": 2200
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 2.332657200811359,
|
| 168 |
+
"grad_norm": 3.901576042175293,
|
| 169 |
+
"learning_rate": 4.76683569979716e-05,
|
| 170 |
+
"loss": 1.8102,
|
| 171 |
+
"step": 2300
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 2.4340770791075053,
|
| 175 |
+
"grad_norm": 4.177852630615234,
|
| 176 |
+
"learning_rate": 4.756693711967546e-05,
|
| 177 |
+
"loss": 1.8145,
|
| 178 |
+
"step": 2400
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 2.535496957403651,
|
| 182 |
+
"grad_norm": 4.461359024047852,
|
| 183 |
+
"learning_rate": 4.746551724137931e-05,
|
| 184 |
+
"loss": 1.8319,
|
| 185 |
+
"step": 2500
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 2.636916835699797,
|
| 189 |
+
"grad_norm": 3.966510772705078,
|
| 190 |
+
"learning_rate": 4.736409736308317e-05,
|
| 191 |
+
"loss": 1.8097,
|
| 192 |
+
"step": 2600
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 2.738336713995943,
|
| 196 |
+
"grad_norm": 3.501115322113037,
|
| 197 |
+
"learning_rate": 4.726267748478702e-05,
|
| 198 |
+
"loss": 1.7988,
|
| 199 |
+
"step": 2700
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 2.839756592292089,
|
| 203 |
+
"grad_norm": 3.9317972660064697,
|
| 204 |
+
"learning_rate": 4.716125760649087e-05,
|
| 205 |
+
"loss": 1.7536,
|
| 206 |
+
"step": 2800
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 2.9411764705882355,
|
| 210 |
+
"grad_norm": 4.940836429595947,
|
| 211 |
+
"learning_rate": 4.705983772819473e-05,
|
| 212 |
+
"loss": 1.7633,
|
| 213 |
+
"step": 2900
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 3.0425963488843815,
|
| 217 |
+
"grad_norm": 4.480752944946289,
|
| 218 |
+
"learning_rate": 4.695841784989858e-05,
|
| 219 |
+
"loss": 1.7516,
|
| 220 |
+
"step": 3000
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 3.1440162271805274,
|
| 224 |
+
"grad_norm": 4.784350395202637,
|
| 225 |
+
"learning_rate": 4.685699797160243e-05,
|
| 226 |
+
"loss": 1.7328,
|
| 227 |
+
"step": 3100
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 3.2454361054766734,
|
| 231 |
+
"grad_norm": 4.241121292114258,
|
| 232 |
+
"learning_rate": 4.675557809330629e-05,
|
| 233 |
+
"loss": 1.7196,
|
| 234 |
+
"step": 3200
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 3.3468559837728193,
|
| 238 |
+
"grad_norm": 4.4414286613464355,
|
| 239 |
+
"learning_rate": 4.665415821501015e-05,
|
| 240 |
+
"loss": 1.7298,
|
| 241 |
+
"step": 3300
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 3.4482758620689653,
|
| 245 |
+
"grad_norm": 4.018706321716309,
|
| 246 |
+
"learning_rate": 4.6552738336714e-05,
|
| 247 |
+
"loss": 1.7579,
|
| 248 |
+
"step": 3400
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 3.5496957403651117,
|
| 252 |
+
"grad_norm": 3.7324869632720947,
|
| 253 |
+
"learning_rate": 4.645131845841785e-05,
|
| 254 |
+
"loss": 1.7483,
|
| 255 |
+
"step": 3500
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 3.6511156186612577,
|
| 259 |
+
"grad_norm": 4.362798690795898,
|
| 260 |
+
"learning_rate": 4.63498985801217e-05,
|
| 261 |
+
"loss": 1.7294,
|
| 262 |
+
"step": 3600
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 3.7525354969574036,
|
| 266 |
+
"grad_norm": 3.8625762462615967,
|
| 267 |
+
"learning_rate": 4.6248478701825556e-05,
|
| 268 |
+
"loss": 1.7201,
|
| 269 |
+
"step": 3700
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 3.8539553752535496,
|
| 273 |
+
"grad_norm": 4.202365875244141,
|
| 274 |
+
"learning_rate": 4.614705882352942e-05,
|
| 275 |
+
"loss": 1.712,
|
| 276 |
+
"step": 3800
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 3.955375253549696,
|
| 280 |
+
"grad_norm": 4.281167030334473,
|
| 281 |
+
"learning_rate": 4.604563894523327e-05,
|
| 282 |
+
"loss": 1.7253,
|
| 283 |
+
"step": 3900
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 4.056795131845842,
|
| 287 |
+
"grad_norm": 4.099128246307373,
|
| 288 |
+
"learning_rate": 4.594421906693712e-05,
|
| 289 |
+
"loss": 1.68,
|
| 290 |
+
"step": 4000
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 4.158215010141988,
|
| 294 |
+
"grad_norm": 3.8885576725006104,
|
| 295 |
+
"learning_rate": 4.584279918864098e-05,
|
| 296 |
+
"loss": 1.6609,
|
| 297 |
+
"step": 4100
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 4.259634888438134,
|
| 301 |
+
"grad_norm": 4.637487411499023,
|
| 302 |
+
"learning_rate": 4.5741379310344826e-05,
|
| 303 |
+
"loss": 1.6677,
|
| 304 |
+
"step": 4200
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 4.36105476673428,
|
| 308 |
+
"grad_norm": 4.093424320220947,
|
| 309 |
+
"learning_rate": 4.563995943204868e-05,
|
| 310 |
+
"loss": 1.6912,
|
| 311 |
+
"step": 4300
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 4.462474645030426,
|
| 315 |
+
"grad_norm": 3.5472183227539062,
|
| 316 |
+
"learning_rate": 4.553853955375254e-05,
|
| 317 |
+
"loss": 1.6828,
|
| 318 |
+
"step": 4400
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 4.563894523326572,
|
| 322 |
+
"grad_norm": 4.927292823791504,
|
| 323 |
+
"learning_rate": 4.543711967545639e-05,
|
| 324 |
+
"loss": 1.667,
|
| 325 |
+
"step": 4500
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 4.665314401622718,
|
| 329 |
+
"grad_norm": 3.901214838027954,
|
| 330 |
+
"learning_rate": 4.533569979716025e-05,
|
| 331 |
+
"loss": 1.677,
|
| 332 |
+
"step": 4600
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 4.766734279918865,
|
| 336 |
+
"grad_norm": 4.752573490142822,
|
| 337 |
+
"learning_rate": 4.5234279918864097e-05,
|
| 338 |
+
"loss": 1.6716,
|
| 339 |
+
"step": 4700
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 4.8681541582150105,
|
| 343 |
+
"grad_norm": 4.095433712005615,
|
| 344 |
+
"learning_rate": 4.513286004056795e-05,
|
| 345 |
+
"loss": 1.6351,
|
| 346 |
+
"step": 4800
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 4.9695740365111565,
|
| 350 |
+
"grad_norm": 4.0542755126953125,
|
| 351 |
+
"learning_rate": 4.503144016227181e-05,
|
| 352 |
+
"loss": 1.6574,
|
| 353 |
+
"step": 4900
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 5.070993914807302,
|
| 357 |
+
"grad_norm": 4.262263774871826,
|
| 358 |
+
"learning_rate": 4.493002028397566e-05,
|
| 359 |
+
"loss": 1.6481,
|
| 360 |
+
"step": 5000
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 5.172413793103448,
|
| 364 |
+
"grad_norm": 4.327647686004639,
|
| 365 |
+
"learning_rate": 4.482860040567952e-05,
|
| 366 |
+
"loss": 1.6278,
|
| 367 |
+
"step": 5100
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 5.273833671399594,
|
| 371 |
+
"grad_norm": 4.188023090362549,
|
| 372 |
+
"learning_rate": 4.4727180527383367e-05,
|
| 373 |
+
"loss": 1.636,
|
| 374 |
+
"step": 5200
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 5.37525354969574,
|
| 378 |
+
"grad_norm": 4.178957939147949,
|
| 379 |
+
"learning_rate": 4.462576064908722e-05,
|
| 380 |
+
"loss": 1.6177,
|
| 381 |
+
"step": 5300
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 5.476673427991886,
|
| 385 |
+
"grad_norm": 4.099201679229736,
|
| 386 |
+
"learning_rate": 4.452434077079108e-05,
|
| 387 |
+
"loss": 1.6239,
|
| 388 |
+
"step": 5400
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 5.578093306288032,
|
| 392 |
+
"grad_norm": 4.149251937866211,
|
| 393 |
+
"learning_rate": 4.4422920892494926e-05,
|
| 394 |
+
"loss": 1.6149,
|
| 395 |
+
"step": 5500
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 5.679513184584178,
|
| 399 |
+
"grad_norm": 4.668671607971191,
|
| 400 |
+
"learning_rate": 4.432150101419879e-05,
|
| 401 |
+
"loss": 1.636,
|
| 402 |
+
"step": 5600
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 5.780933062880324,
|
| 406 |
+
"grad_norm": 3.784585475921631,
|
| 407 |
+
"learning_rate": 4.4220081135902643e-05,
|
| 408 |
+
"loss": 1.611,
|
| 409 |
+
"step": 5700
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 5.882352941176471,
|
| 413 |
+
"grad_norm": 3.5564262866973877,
|
| 414 |
+
"learning_rate": 4.411866125760649e-05,
|
| 415 |
+
"loss": 1.6258,
|
| 416 |
+
"step": 5800
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 5.983772819472617,
|
| 420 |
+
"grad_norm": 4.228405475616455,
|
| 421 |
+
"learning_rate": 4.401724137931035e-05,
|
| 422 |
+
"loss": 1.6261,
|
| 423 |
+
"step": 5900
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 6.085192697768763,
|
| 427 |
+
"grad_norm": 4.541458606719971,
|
| 428 |
+
"learning_rate": 4.3915821501014196e-05,
|
| 429 |
+
"loss": 1.607,
|
| 430 |
+
"step": 6000
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 6.186612576064909,
|
| 434 |
+
"grad_norm": 3.7971739768981934,
|
| 435 |
+
"learning_rate": 4.381440162271805e-05,
|
| 436 |
+
"loss": 1.6025,
|
| 437 |
+
"step": 6100
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 6.288032454361055,
|
| 441 |
+
"grad_norm": 3.4146156311035156,
|
| 442 |
+
"learning_rate": 4.3712981744421913e-05,
|
| 443 |
+
"loss": 1.5761,
|
| 444 |
+
"step": 6200
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 6.389452332657201,
|
| 448 |
+
"grad_norm": 4.267679214477539,
|
| 449 |
+
"learning_rate": 4.361156186612576e-05,
|
| 450 |
+
"loss": 1.5996,
|
| 451 |
+
"step": 6300
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 6.490872210953347,
|
| 455 |
+
"grad_norm": 3.6959927082061768,
|
| 456 |
+
"learning_rate": 4.351014198782962e-05,
|
| 457 |
+
"loss": 1.5818,
|
| 458 |
+
"step": 6400
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 6.592292089249493,
|
| 462 |
+
"grad_norm": 4.326287269592285,
|
| 463 |
+
"learning_rate": 4.340872210953347e-05,
|
| 464 |
+
"loss": 1.5845,
|
| 465 |
+
"step": 6500
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 6.693711967545639,
|
| 469 |
+
"grad_norm": 5.447903156280518,
|
| 470 |
+
"learning_rate": 4.330730223123732e-05,
|
| 471 |
+
"loss": 1.5959,
|
| 472 |
+
"step": 6600
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 6.795131845841785,
|
| 476 |
+
"grad_norm": 3.6755692958831787,
|
| 477 |
+
"learning_rate": 4.320588235294118e-05,
|
| 478 |
+
"loss": 1.5898,
|
| 479 |
+
"step": 6700
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 6.896551724137931,
|
| 483 |
+
"grad_norm": 4.288609504699707,
|
| 484 |
+
"learning_rate": 4.310446247464503e-05,
|
| 485 |
+
"loss": 1.5715,
|
| 486 |
+
"step": 6800
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 6.9979716024340775,
|
| 490 |
+
"grad_norm": 4.3117852210998535,
|
| 491 |
+
"learning_rate": 4.300304259634889e-05,
|
| 492 |
+
"loss": 1.589,
|
| 493 |
+
"step": 6900
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 7.099391480730223,
|
| 497 |
+
"grad_norm": 3.9464619159698486,
|
| 498 |
+
"learning_rate": 4.290162271805274e-05,
|
| 499 |
+
"loss": 1.529,
|
| 500 |
+
"step": 7000
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 7.200811359026369,
|
| 504 |
+
"grad_norm": 4.832653522491455,
|
| 505 |
+
"learning_rate": 4.280020283975659e-05,
|
| 506 |
+
"loss": 1.5827,
|
| 507 |
+
"step": 7100
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 7.302231237322515,
|
| 511 |
+
"grad_norm": 4.050671100616455,
|
| 512 |
+
"learning_rate": 4.269878296146045e-05,
|
| 513 |
+
"loss": 1.5519,
|
| 514 |
+
"step": 7200
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 7.403651115618661,
|
| 518 |
+
"grad_norm": 4.140113353729248,
|
| 519 |
+
"learning_rate": 4.25973630831643e-05,
|
| 520 |
+
"loss": 1.5312,
|
| 521 |
+
"step": 7300
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 7.505070993914807,
|
| 525 |
+
"grad_norm": 4.073803901672363,
|
| 526 |
+
"learning_rate": 4.249594320486816e-05,
|
| 527 |
+
"loss": 1.5462,
|
| 528 |
+
"step": 7400
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 7.606490872210953,
|
| 532 |
+
"grad_norm": 4.087874412536621,
|
| 533 |
+
"learning_rate": 4.239452332657201e-05,
|
| 534 |
+
"loss": 1.5836,
|
| 535 |
+
"step": 7500
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 7.707910750507099,
|
| 539 |
+
"grad_norm": 4.335333824157715,
|
| 540 |
+
"learning_rate": 4.229310344827586e-05,
|
| 541 |
+
"loss": 1.553,
|
| 542 |
+
"step": 7600
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 7.809330628803245,
|
| 546 |
+
"grad_norm": 3.957524538040161,
|
| 547 |
+
"learning_rate": 4.219168356997972e-05,
|
| 548 |
+
"loss": 1.542,
|
| 549 |
+
"step": 7700
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 7.910750507099391,
|
| 553 |
+
"grad_norm": 3.965404510498047,
|
| 554 |
+
"learning_rate": 4.209026369168357e-05,
|
| 555 |
+
"loss": 1.5517,
|
| 556 |
+
"step": 7800
|
| 557 |
+
},
|
| 558 |
+
{
|
| 559 |
+
"epoch": 8.012170385395537,
|
| 560 |
+
"grad_norm": 3.691362142562866,
|
| 561 |
+
"learning_rate": 4.198884381338742e-05,
|
| 562 |
+
"loss": 1.5588,
|
| 563 |
+
"step": 7900
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 8.113590263691684,
|
| 567 |
+
"grad_norm": 3.628953456878662,
|
| 568 |
+
"learning_rate": 4.188742393509128e-05,
|
| 569 |
+
"loss": 1.5269,
|
| 570 |
+
"step": 8000
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 8.215010141987829,
|
| 574 |
+
"grad_norm": 3.479663133621216,
|
| 575 |
+
"learning_rate": 4.178600405679514e-05,
|
| 576 |
+
"loss": 1.5182,
|
| 577 |
+
"step": 8100
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 8.316430020283976,
|
| 581 |
+
"grad_norm": 3.9403600692749023,
|
| 582 |
+
"learning_rate": 4.168458417849899e-05,
|
| 583 |
+
"loss": 1.5148,
|
| 584 |
+
"step": 8200
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 8.417849898580121,
|
| 588 |
+
"grad_norm": 3.9685311317443848,
|
| 589 |
+
"learning_rate": 4.158316430020284e-05,
|
| 590 |
+
"loss": 1.5324,
|
| 591 |
+
"step": 8300
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 8.519269776876268,
|
| 595 |
+
"grad_norm": 3.9184675216674805,
|
| 596 |
+
"learning_rate": 4.148174442190669e-05,
|
| 597 |
+
"loss": 1.5326,
|
| 598 |
+
"step": 8400
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 8.620689655172415,
|
| 602 |
+
"grad_norm": 3.9011454582214355,
|
| 603 |
+
"learning_rate": 4.1380324543610546e-05,
|
| 604 |
+
"loss": 1.522,
|
| 605 |
+
"step": 8500
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 8.72210953346856,
|
| 609 |
+
"grad_norm": 4.29376745223999,
|
| 610 |
+
"learning_rate": 4.127890466531441e-05,
|
| 611 |
+
"loss": 1.5153,
|
| 612 |
+
"step": 8600
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 8.823529411764707,
|
| 616 |
+
"grad_norm": 3.6822264194488525,
|
| 617 |
+
"learning_rate": 4.117748478701826e-05,
|
| 618 |
+
"loss": 1.5131,
|
| 619 |
+
"step": 8700
|
| 620 |
+
},
|
| 621 |
+
{
|
| 622 |
+
"epoch": 8.924949290060852,
|
| 623 |
+
"grad_norm": 3.940157175064087,
|
| 624 |
+
"learning_rate": 4.107606490872211e-05,
|
| 625 |
+
"loss": 1.5201,
|
| 626 |
+
"step": 8800
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 9.026369168356998,
|
| 630 |
+
"grad_norm": 4.2661871910095215,
|
| 631 |
+
"learning_rate": 4.097464503042597e-05,
|
| 632 |
+
"loss": 1.5103,
|
| 633 |
+
"step": 8900
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 9.127789046653144,
|
| 637 |
+
"grad_norm": 4.383224010467529,
|
| 638 |
+
"learning_rate": 4.0873225152129816e-05,
|
| 639 |
+
"loss": 1.4779,
|
| 640 |
+
"step": 9000
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 9.22920892494929,
|
| 644 |
+
"grad_norm": 4.30776834487915,
|
| 645 |
+
"learning_rate": 4.077180527383367e-05,
|
| 646 |
+
"loss": 1.4829,
|
| 647 |
+
"step": 9100
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 9.330628803245435,
|
| 651 |
+
"grad_norm": 3.849860191345215,
|
| 652 |
+
"learning_rate": 4.067038539553753e-05,
|
| 653 |
+
"loss": 1.535,
|
| 654 |
+
"step": 9200
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 9.432048681541582,
|
| 658 |
+
"grad_norm": 4.159960746765137,
|
| 659 |
+
"learning_rate": 4.056896551724138e-05,
|
| 660 |
+
"loss": 1.4938,
|
| 661 |
+
"step": 9300
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 9.53346855983773,
|
| 665 |
+
"grad_norm": 4.451343059539795,
|
| 666 |
+
"learning_rate": 4.046754563894524e-05,
|
| 667 |
+
"loss": 1.51,
|
| 668 |
+
"step": 9400
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 9.634888438133874,
|
| 672 |
+
"grad_norm": 3.6356382369995117,
|
| 673 |
+
"learning_rate": 4.0366125760649086e-05,
|
| 674 |
+
"loss": 1.5104,
|
| 675 |
+
"step": 9500
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 9.736308316430021,
|
| 679 |
+
"grad_norm": 3.729419231414795,
|
| 680 |
+
"learning_rate": 4.026470588235294e-05,
|
| 681 |
+
"loss": 1.5049,
|
| 682 |
+
"step": 9600
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 9.837728194726166,
|
| 686 |
+
"grad_norm": 4.195159435272217,
|
| 687 |
+
"learning_rate": 4.01632860040568e-05,
|
| 688 |
+
"loss": 1.4732,
|
| 689 |
+
"step": 9700
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 9.939148073022313,
|
| 693 |
+
"grad_norm": 4.059690952301025,
|
| 694 |
+
"learning_rate": 4.006186612576065e-05,
|
| 695 |
+
"loss": 1.4956,
|
| 696 |
+
"step": 9800
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 10.040567951318458,
|
| 700 |
+
"grad_norm": 3.964912176132202,
|
| 701 |
+
"learning_rate": 3.996044624746451e-05,
|
| 702 |
+
"loss": 1.4773,
|
| 703 |
+
"step": 9900
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 10.141987829614605,
|
| 707 |
+
"grad_norm": 4.321567535400391,
|
| 708 |
+
"learning_rate": 3.9859026369168356e-05,
|
| 709 |
+
"loss": 1.484,
|
| 710 |
+
"step": 10000
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 10.24340770791075,
|
| 714 |
+
"grad_norm": 3.8022358417510986,
|
| 715 |
+
"learning_rate": 3.975760649087221e-05,
|
| 716 |
+
"loss": 1.4686,
|
| 717 |
+
"step": 10100
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 10.344827586206897,
|
| 721 |
+
"grad_norm": 4.274851322174072,
|
| 722 |
+
"learning_rate": 3.965618661257607e-05,
|
| 723 |
+
"loss": 1.4822,
|
| 724 |
+
"step": 10200
|
| 725 |
+
},
|
| 726 |
+
{
|
| 727 |
+
"epoch": 10.446247464503042,
|
| 728 |
+
"grad_norm": 3.9784164428710938,
|
| 729 |
+
"learning_rate": 3.9554766734279916e-05,
|
| 730 |
+
"loss": 1.4513,
|
| 731 |
+
"step": 10300
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 10.547667342799189,
|
| 735 |
+
"grad_norm": 4.056332111358643,
|
| 736 |
+
"learning_rate": 3.945334685598378e-05,
|
| 737 |
+
"loss": 1.4613,
|
| 738 |
+
"step": 10400
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"epoch": 10.649087221095336,
|
| 742 |
+
"grad_norm": 4.337247371673584,
|
| 743 |
+
"learning_rate": 3.935192697768763e-05,
|
| 744 |
+
"loss": 1.4797,
|
| 745 |
+
"step": 10500
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 10.75050709939148,
|
| 749 |
+
"grad_norm": 3.6584205627441406,
|
| 750 |
+
"learning_rate": 3.925050709939148e-05,
|
| 751 |
+
"loss": 1.4728,
|
| 752 |
+
"step": 10600
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 10.851926977687627,
|
| 756 |
+
"grad_norm": 4.399139881134033,
|
| 757 |
+
"learning_rate": 3.914908722109534e-05,
|
| 758 |
+
"loss": 1.48,
|
| 759 |
+
"step": 10700
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 10.953346855983773,
|
| 763 |
+
"grad_norm": 3.5788066387176514,
|
| 764 |
+
"learning_rate": 3.9047667342799186e-05,
|
| 765 |
+
"loss": 1.4697,
|
| 766 |
+
"step": 10800
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 11.05476673427992,
|
| 770 |
+
"grad_norm": 4.16567850112915,
|
| 771 |
+
"learning_rate": 3.894624746450304e-05,
|
| 772 |
+
"loss": 1.4553,
|
| 773 |
+
"step": 10900
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 11.156186612576064,
|
| 777 |
+
"grad_norm": 3.93194580078125,
|
| 778 |
+
"learning_rate": 3.88448275862069e-05,
|
| 779 |
+
"loss": 1.4359,
|
| 780 |
+
"step": 11000
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 11.257606490872211,
|
| 784 |
+
"grad_norm": 3.7836058139801025,
|
| 785 |
+
"learning_rate": 3.874340770791075e-05,
|
| 786 |
+
"loss": 1.4502,
|
| 787 |
+
"step": 11100
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 11.359026369168356,
|
| 791 |
+
"grad_norm": 4.094986915588379,
|
| 792 |
+
"learning_rate": 3.864198782961461e-05,
|
| 793 |
+
"loss": 1.4191,
|
| 794 |
+
"step": 11200
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 11.460446247464503,
|
| 798 |
+
"grad_norm": 4.346254825592041,
|
| 799 |
+
"learning_rate": 3.854056795131846e-05,
|
| 800 |
+
"loss": 1.456,
|
| 801 |
+
"step": 11300
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 11.561866125760648,
|
| 805 |
+
"grad_norm": 4.003499507904053,
|
| 806 |
+
"learning_rate": 3.843914807302231e-05,
|
| 807 |
+
"loss": 1.4432,
|
| 808 |
+
"step": 11400
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 11.663286004056795,
|
| 812 |
+
"grad_norm": 4.008283615112305,
|
| 813 |
+
"learning_rate": 3.8337728194726167e-05,
|
| 814 |
+
"loss": 1.4556,
|
| 815 |
+
"step": 11500
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 11.764705882352942,
|
| 819 |
+
"grad_norm": 3.6525089740753174,
|
| 820 |
+
"learning_rate": 3.823630831643002e-05,
|
| 821 |
+
"loss": 1.4604,
|
| 822 |
+
"step": 11600
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"epoch": 11.866125760649087,
|
| 826 |
+
"grad_norm": 4.341876983642578,
|
| 827 |
+
"learning_rate": 3.813488843813388e-05,
|
| 828 |
+
"loss": 1.4371,
|
| 829 |
+
"step": 11700
|
| 830 |
+
},
|
| 831 |
+
{
|
| 832 |
+
"epoch": 11.967545638945234,
|
| 833 |
+
"grad_norm": 3.9196975231170654,
|
| 834 |
+
"learning_rate": 3.803346855983773e-05,
|
| 835 |
+
"loss": 1.4428,
|
| 836 |
+
"step": 11800
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 12.068965517241379,
|
| 840 |
+
"grad_norm": 4.580789089202881,
|
| 841 |
+
"learning_rate": 3.793204868154158e-05,
|
| 842 |
+
"loss": 1.3932,
|
| 843 |
+
"step": 11900
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 12.170385395537526,
|
| 847 |
+
"grad_norm": 3.845825433731079,
|
| 848 |
+
"learning_rate": 3.783062880324544e-05,
|
| 849 |
+
"loss": 1.4258,
|
| 850 |
+
"step": 12000
|
| 851 |
+
},
|
| 852 |
+
{
|
| 853 |
+
"epoch": 12.271805273833671,
|
| 854 |
+
"grad_norm": 4.6135149002075195,
|
| 855 |
+
"learning_rate": 3.772920892494929e-05,
|
| 856 |
+
"loss": 1.4389,
|
| 857 |
+
"step": 12100
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 12.373225152129818,
|
| 861 |
+
"grad_norm": 4.0474772453308105,
|
| 862 |
+
"learning_rate": 3.762778904665315e-05,
|
| 863 |
+
"loss": 1.4344,
|
| 864 |
+
"step": 12200
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 12.474645030425963,
|
| 868 |
+
"grad_norm": 4.010397911071777,
|
| 869 |
+
"learning_rate": 3.7526369168357e-05,
|
| 870 |
+
"loss": 1.4258,
|
| 871 |
+
"step": 12300
|
| 872 |
+
},
|
| 873 |
+
{
|
| 874 |
+
"epoch": 12.57606490872211,
|
| 875 |
+
"grad_norm": 3.953735828399658,
|
| 876 |
+
"learning_rate": 3.742494929006085e-05,
|
| 877 |
+
"loss": 1.4415,
|
| 878 |
+
"step": 12400
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 12.677484787018255,
|
| 882 |
+
"grad_norm": 3.69645094871521,
|
| 883 |
+
"learning_rate": 3.732352941176471e-05,
|
| 884 |
+
"loss": 1.4134,
|
| 885 |
+
"step": 12500
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 12.778904665314402,
|
| 889 |
+
"grad_norm": 4.53965950012207,
|
| 890 |
+
"learning_rate": 3.722210953346856e-05,
|
| 891 |
+
"loss": 1.4244,
|
| 892 |
+
"step": 12600
|
| 893 |
+
},
|
| 894 |
+
{
|
| 895 |
+
"epoch": 12.880324543610548,
|
| 896 |
+
"grad_norm": 4.225129127502441,
|
| 897 |
+
"learning_rate": 3.712068965517241e-05,
|
| 898 |
+
"loss": 1.4226,
|
| 899 |
+
"step": 12700
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 12.981744421906694,
|
| 903 |
+
"grad_norm": 4.163741111755371,
|
| 904 |
+
"learning_rate": 3.701926977687627e-05,
|
| 905 |
+
"loss": 1.4363,
|
| 906 |
+
"step": 12800
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 13.08316430020284,
|
| 910 |
+
"grad_norm": 4.016907691955566,
|
| 911 |
+
"learning_rate": 3.691784989858013e-05,
|
| 912 |
+
"loss": 1.3955,
|
| 913 |
+
"step": 12900
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"epoch": 13.184584178498985,
|
| 917 |
+
"grad_norm": 3.960197925567627,
|
| 918 |
+
"learning_rate": 3.681643002028398e-05,
|
| 919 |
+
"loss": 1.4244,
|
| 920 |
+
"step": 13000
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 13.286004056795132,
|
| 924 |
+
"grad_norm": 3.9812252521514893,
|
| 925 |
+
"learning_rate": 3.671501014198783e-05,
|
| 926 |
+
"loss": 1.4129,
|
| 927 |
+
"step": 13100
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 13.387423935091277,
|
| 931 |
+
"grad_norm": 4.356760501861572,
|
| 932 |
+
"learning_rate": 3.661359026369168e-05,
|
| 933 |
+
"loss": 1.4386,
|
| 934 |
+
"step": 13200
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 13.488843813387424,
|
| 938 |
+
"grad_norm": 3.87821626663208,
|
| 939 |
+
"learning_rate": 3.6512170385395536e-05,
|
| 940 |
+
"loss": 1.3897,
|
| 941 |
+
"step": 13300
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 13.59026369168357,
|
| 945 |
+
"grad_norm": 4.03455924987793,
|
| 946 |
+
"learning_rate": 3.64107505070994e-05,
|
| 947 |
+
"loss": 1.4138,
|
| 948 |
+
"step": 13400
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 13.691683569979716,
|
| 952 |
+
"grad_norm": 3.5549731254577637,
|
| 953 |
+
"learning_rate": 3.630933062880325e-05,
|
| 954 |
+
"loss": 1.4083,
|
| 955 |
+
"step": 13500
|
| 956 |
+
},
|
| 957 |
+
{
|
| 958 |
+
"epoch": 13.793103448275861,
|
| 959 |
+
"grad_norm": 4.076521396636963,
|
| 960 |
+
"learning_rate": 3.62079107505071e-05,
|
| 961 |
+
"loss": 1.4192,
|
| 962 |
+
"step": 13600
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 13.894523326572008,
|
| 966 |
+
"grad_norm": 4.135570526123047,
|
| 967 |
+
"learning_rate": 3.610649087221096e-05,
|
| 968 |
+
"loss": 1.4029,
|
| 969 |
+
"step": 13700
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 13.995943204868155,
|
| 973 |
+
"grad_norm": 4.511417865753174,
|
| 974 |
+
"learning_rate": 3.6005070993914806e-05,
|
| 975 |
+
"loss": 1.3964,
|
| 976 |
+
"step": 13800
|
| 977 |
+
},
|
| 978 |
+
{
|
| 979 |
+
"epoch": 14.0973630831643,
|
| 980 |
+
"grad_norm": 3.7181835174560547,
|
| 981 |
+
"learning_rate": 3.590365111561866e-05,
|
| 982 |
+
"loss": 1.3828,
|
| 983 |
+
"step": 13900
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 14.198782961460447,
|
| 987 |
+
"grad_norm": 3.7762348651885986,
|
| 988 |
+
"learning_rate": 3.580223123732252e-05,
|
| 989 |
+
"loss": 1.3842,
|
| 990 |
+
"step": 14000
|
| 991 |
+
},
|
| 992 |
+
{
|
| 993 |
+
"epoch": 14.300202839756592,
|
| 994 |
+
"grad_norm": 4.129021644592285,
|
| 995 |
+
"learning_rate": 3.570081135902637e-05,
|
| 996 |
+
"loss": 1.3775,
|
| 997 |
+
"step": 14100
|
| 998 |
+
},
|
| 999 |
+
{
|
| 1000 |
+
"epoch": 14.401622718052739,
|
| 1001 |
+
"grad_norm": 4.174191474914551,
|
| 1002 |
+
"learning_rate": 3.559939148073023e-05,
|
| 1003 |
+
"loss": 1.3758,
|
| 1004 |
+
"step": 14200
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 14.503042596348884,
|
| 1008 |
+
"grad_norm": 5.045529842376709,
|
| 1009 |
+
"learning_rate": 3.5497971602434076e-05,
|
| 1010 |
+
"loss": 1.3878,
|
| 1011 |
+
"step": 14300
|
| 1012 |
+
},
|
| 1013 |
+
{
|
| 1014 |
+
"epoch": 14.60446247464503,
|
| 1015 |
+
"grad_norm": 3.996896743774414,
|
| 1016 |
+
"learning_rate": 3.539655172413793e-05,
|
| 1017 |
+
"loss": 1.3822,
|
| 1018 |
+
"step": 14400
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"epoch": 14.705882352941176,
|
| 1022 |
+
"grad_norm": 4.2597832679748535,
|
| 1023 |
+
"learning_rate": 3.529513184584179e-05,
|
| 1024 |
+
"loss": 1.3862,
|
| 1025 |
+
"step": 14500
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 14.807302231237323,
|
| 1029 |
+
"grad_norm": 4.087389945983887,
|
| 1030 |
+
"learning_rate": 3.519371196754564e-05,
|
| 1031 |
+
"loss": 1.3625,
|
| 1032 |
+
"step": 14600
|
| 1033 |
+
},
|
| 1034 |
+
{
|
| 1035 |
+
"epoch": 14.908722109533468,
|
| 1036 |
+
"grad_norm": 4.255886554718018,
|
| 1037 |
+
"learning_rate": 3.50922920892495e-05,
|
| 1038 |
+
"loss": 1.3987,
|
| 1039 |
+
"step": 14700
|
| 1040 |
+
},
|
| 1041 |
+
{
|
| 1042 |
+
"epoch": 15.010141987829615,
|
| 1043 |
+
"grad_norm": 4.261554718017578,
|
| 1044 |
+
"learning_rate": 3.4990872210953346e-05,
|
| 1045 |
+
"loss": 1.4091,
|
| 1046 |
+
"step": 14800
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 15.111561866125761,
|
| 1050 |
+
"grad_norm": 3.7017126083374023,
|
| 1051 |
+
"learning_rate": 3.48894523326572e-05,
|
| 1052 |
+
"loss": 1.3452,
|
| 1053 |
+
"step": 14900
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 15.212981744421906,
|
| 1057 |
+
"grad_norm": 4.391205310821533,
|
| 1058 |
+
"learning_rate": 3.478803245436106e-05,
|
| 1059 |
+
"loss": 1.3668,
|
| 1060 |
+
"step": 15000
|
| 1061 |
+
},
|
| 1062 |
+
{
|
| 1063 |
+
"epoch": 15.314401622718053,
|
| 1064 |
+
"grad_norm": 3.7424230575561523,
|
| 1065 |
+
"learning_rate": 3.4686612576064906e-05,
|
| 1066 |
+
"loss": 1.3573,
|
| 1067 |
+
"step": 15100
|
| 1068 |
+
},
|
| 1069 |
+
{
|
| 1070 |
+
"epoch": 15.415821501014198,
|
| 1071 |
+
"grad_norm": 3.8218424320220947,
|
| 1072 |
+
"learning_rate": 3.458519269776877e-05,
|
| 1073 |
+
"loss": 1.3595,
|
| 1074 |
+
"step": 15200
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 15.517241379310345,
|
| 1078 |
+
"grad_norm": 3.3766188621520996,
|
| 1079 |
+
"learning_rate": 3.448377281947262e-05,
|
| 1080 |
+
"loss": 1.3595,
|
| 1081 |
+
"step": 15300
|
| 1082 |
+
},
|
| 1083 |
+
{
|
| 1084 |
+
"epoch": 15.61866125760649,
|
| 1085 |
+
"grad_norm": 4.296000003814697,
|
| 1086 |
+
"learning_rate": 3.438235294117647e-05,
|
| 1087 |
+
"loss": 1.3769,
|
| 1088 |
+
"step": 15400
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 15.720081135902637,
|
| 1092 |
+
"grad_norm": 4.256558895111084,
|
| 1093 |
+
"learning_rate": 3.428093306288033e-05,
|
| 1094 |
+
"loss": 1.3774,
|
| 1095 |
+
"step": 15500
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 15.821501014198782,
|
| 1099 |
+
"grad_norm": 4.054759979248047,
|
| 1100 |
+
"learning_rate": 3.4179513184584176e-05,
|
| 1101 |
+
"loss": 1.3721,
|
| 1102 |
+
"step": 15600
|
| 1103 |
+
},
|
| 1104 |
+
{
|
| 1105 |
+
"epoch": 15.922920892494929,
|
| 1106 |
+
"grad_norm": 3.775669813156128,
|
| 1107 |
+
"learning_rate": 3.407809330628803e-05,
|
| 1108 |
+
"loss": 1.3702,
|
| 1109 |
+
"step": 15700
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 16.024340770791074,
|
| 1113 |
+
"grad_norm": 4.379668235778809,
|
| 1114 |
+
"learning_rate": 3.3976673427991886e-05,
|
| 1115 |
+
"loss": 1.3764,
|
| 1116 |
+
"step": 15800
|
| 1117 |
+
},
|
| 1118 |
+
{
|
| 1119 |
+
"epoch": 16.125760649087223,
|
| 1120 |
+
"grad_norm": 3.9183759689331055,
|
| 1121 |
+
"learning_rate": 3.387525354969574e-05,
|
| 1122 |
+
"loss": 1.3432,
|
| 1123 |
+
"step": 15900
|
| 1124 |
+
},
|
| 1125 |
+
{
|
| 1126 |
+
"epoch": 16.227180527383368,
|
| 1127 |
+
"grad_norm": 3.996525287628174,
|
| 1128 |
+
"learning_rate": 3.37738336713996e-05,
|
| 1129 |
+
"loss": 1.3529,
|
| 1130 |
+
"step": 16000
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 16.328600405679513,
|
| 1134 |
+
"grad_norm": 3.84562349319458,
|
| 1135 |
+
"learning_rate": 3.367241379310345e-05,
|
| 1136 |
+
"loss": 1.3376,
|
| 1137 |
+
"step": 16100
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 16.430020283975658,
|
| 1141 |
+
"grad_norm": 4.113460063934326,
|
| 1142 |
+
"learning_rate": 3.35709939148073e-05,
|
| 1143 |
+
"loss": 1.3346,
|
| 1144 |
+
"step": 16200
|
| 1145 |
+
},
|
| 1146 |
+
{
|
| 1147 |
+
"epoch": 16.531440162271807,
|
| 1148 |
+
"grad_norm": 5.2181396484375,
|
| 1149 |
+
"learning_rate": 3.3469574036511156e-05,
|
| 1150 |
+
"loss": 1.3603,
|
| 1151 |
+
"step": 16300
|
| 1152 |
+
},
|
| 1153 |
+
{
|
| 1154 |
+
"epoch": 16.63286004056795,
|
| 1155 |
+
"grad_norm": 4.758508205413818,
|
| 1156 |
+
"learning_rate": 3.336815415821501e-05,
|
| 1157 |
+
"loss": 1.3567,
|
| 1158 |
+
"step": 16400
|
| 1159 |
+
},
|
| 1160 |
+
{
|
| 1161 |
+
"epoch": 16.734279918864097,
|
| 1162 |
+
"grad_norm": 4.126884460449219,
|
| 1163 |
+
"learning_rate": 3.326673427991887e-05,
|
| 1164 |
+
"loss": 1.3624,
|
| 1165 |
+
"step": 16500
|
| 1166 |
+
},
|
| 1167 |
+
{
|
| 1168 |
+
"epoch": 16.835699797160242,
|
| 1169 |
+
"grad_norm": 3.5290539264678955,
|
| 1170 |
+
"learning_rate": 3.316531440162272e-05,
|
| 1171 |
+
"loss": 1.3457,
|
| 1172 |
+
"step": 16600
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 16.93711967545639,
|
| 1176 |
+
"grad_norm": 4.807733058929443,
|
| 1177 |
+
"learning_rate": 3.306389452332657e-05,
|
| 1178 |
+
"loss": 1.3525,
|
| 1179 |
+
"step": 16700
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 17.038539553752535,
|
| 1183 |
+
"grad_norm": 4.039372444152832,
|
| 1184 |
+
"learning_rate": 3.2962474645030426e-05,
|
| 1185 |
+
"loss": 1.3379,
|
| 1186 |
+
"step": 16800
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 17.13995943204868,
|
| 1190 |
+
"grad_norm": 4.763552665710449,
|
| 1191 |
+
"learning_rate": 3.286105476673428e-05,
|
| 1192 |
+
"loss": 1.3304,
|
| 1193 |
+
"step": 16900
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 17.24137931034483,
|
| 1197 |
+
"grad_norm": 4.548777103424072,
|
| 1198 |
+
"learning_rate": 3.275963488843814e-05,
|
| 1199 |
+
"loss": 1.3432,
|
| 1200 |
+
"step": 17000
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"epoch": 17.342799188640974,
|
| 1204 |
+
"grad_norm": 3.607231855392456,
|
| 1205 |
+
"learning_rate": 3.265821501014199e-05,
|
| 1206 |
+
"loss": 1.3304,
|
| 1207 |
+
"step": 17100
|
| 1208 |
+
},
|
| 1209 |
+
{
|
| 1210 |
+
"epoch": 17.44421906693712,
|
| 1211 |
+
"grad_norm": 3.489654541015625,
|
| 1212 |
+
"learning_rate": 3.255679513184584e-05,
|
| 1213 |
+
"loss": 1.3296,
|
| 1214 |
+
"step": 17200
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 17.545638945233264,
|
| 1218 |
+
"grad_norm": 4.143092155456543,
|
| 1219 |
+
"learning_rate": 3.2455375253549696e-05,
|
| 1220 |
+
"loss": 1.3296,
|
| 1221 |
+
"step": 17300
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 17.647058823529413,
|
| 1225 |
+
"grad_norm": 4.415568828582764,
|
| 1226 |
+
"learning_rate": 3.235395537525355e-05,
|
| 1227 |
+
"loss": 1.3304,
|
| 1228 |
+
"step": 17400
|
| 1229 |
+
},
|
| 1230 |
+
{
|
| 1231 |
+
"epoch": 17.748478701825558,
|
| 1232 |
+
"grad_norm": 4.408501625061035,
|
| 1233 |
+
"learning_rate": 3.22525354969574e-05,
|
| 1234 |
+
"loss": 1.3419,
|
| 1235 |
+
"step": 17500
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 17.849898580121703,
|
| 1239 |
+
"grad_norm": 3.837433099746704,
|
| 1240 |
+
"learning_rate": 3.215111561866126e-05,
|
| 1241 |
+
"loss": 1.3287,
|
| 1242 |
+
"step": 17600
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"epoch": 17.95131845841785,
|
| 1246 |
+
"grad_norm": 4.065082550048828,
|
| 1247 |
+
"learning_rate": 3.204969574036512e-05,
|
| 1248 |
+
"loss": 1.3334,
|
| 1249 |
+
"step": 17700
|
| 1250 |
+
},
|
| 1251 |
+
{
|
| 1252 |
+
"epoch": 18.052738336713997,
|
| 1253 |
+
"grad_norm": 4.563329696655273,
|
| 1254 |
+
"learning_rate": 3.1948275862068967e-05,
|
| 1255 |
+
"loss": 1.3332,
|
| 1256 |
+
"step": 17800
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 18.154158215010142,
|
| 1260 |
+
"grad_norm": 4.578526973724365,
|
| 1261 |
+
"learning_rate": 3.184685598377282e-05,
|
| 1262 |
+
"loss": 1.3357,
|
| 1263 |
+
"step": 17900
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 18.255578093306287,
|
| 1267 |
+
"grad_norm": 3.851433753967285,
|
| 1268 |
+
"learning_rate": 3.174543610547667e-05,
|
| 1269 |
+
"loss": 1.3383,
|
| 1270 |
+
"step": 18000
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 18.356997971602436,
|
| 1274 |
+
"grad_norm": 3.335575580596924,
|
| 1275 |
+
"learning_rate": 3.1644016227180526e-05,
|
| 1276 |
+
"loss": 1.3065,
|
| 1277 |
+
"step": 18100
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 18.45841784989858,
|
| 1281 |
+
"grad_norm": 4.088294982910156,
|
| 1282 |
+
"learning_rate": 3.154259634888438e-05,
|
| 1283 |
+
"loss": 1.3242,
|
| 1284 |
+
"step": 18200
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 18.559837728194726,
|
| 1288 |
+
"grad_norm": 4.3247480392456055,
|
| 1289 |
+
"learning_rate": 3.1441176470588237e-05,
|
| 1290 |
+
"loss": 1.3226,
|
| 1291 |
+
"step": 18300
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 18.66125760649087,
|
| 1295 |
+
"grad_norm": 3.778689384460449,
|
| 1296 |
+
"learning_rate": 3.133975659229209e-05,
|
| 1297 |
+
"loss": 1.3003,
|
| 1298 |
+
"step": 18400
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 18.76267748478702,
|
| 1302 |
+
"grad_norm": 5.026618480682373,
|
| 1303 |
+
"learning_rate": 3.123833671399595e-05,
|
| 1304 |
+
"loss": 1.3047,
|
| 1305 |
+
"step": 18500
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 18.864097363083165,
|
| 1309 |
+
"grad_norm": 4.039363384246826,
|
| 1310 |
+
"learning_rate": 3.1136916835699796e-05,
|
| 1311 |
+
"loss": 1.3367,
|
| 1312 |
+
"step": 18600
|
| 1313 |
+
},
|
| 1314 |
+
{
|
| 1315 |
+
"epoch": 18.96551724137931,
|
| 1316 |
+
"grad_norm": 4.170531749725342,
|
| 1317 |
+
"learning_rate": 3.103549695740365e-05,
|
| 1318 |
+
"loss": 1.3078,
|
| 1319 |
+
"step": 18700
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 19.066937119675455,
|
| 1323 |
+
"grad_norm": 4.006753921508789,
|
| 1324 |
+
"learning_rate": 3.093407707910751e-05,
|
| 1325 |
+
"loss": 1.3025,
|
| 1326 |
+
"step": 18800
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"epoch": 19.168356997971603,
|
| 1330 |
+
"grad_norm": 4.071269512176514,
|
| 1331 |
+
"learning_rate": 3.083265720081136e-05,
|
| 1332 |
+
"loss": 1.2966,
|
| 1333 |
+
"step": 18900
|
| 1334 |
+
},
|
| 1335 |
+
{
|
| 1336 |
+
"epoch": 19.26977687626775,
|
| 1337 |
+
"grad_norm": 3.7370877265930176,
|
| 1338 |
+
"learning_rate": 3.073123732251522e-05,
|
| 1339 |
+
"loss": 1.3045,
|
| 1340 |
+
"step": 19000
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 19.371196754563893,
|
| 1344 |
+
"grad_norm": 3.4735629558563232,
|
| 1345 |
+
"learning_rate": 3.0629817444219066e-05,
|
| 1346 |
+
"loss": 1.3055,
|
| 1347 |
+
"step": 19100
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 19.472616632860042,
|
| 1351 |
+
"grad_norm": 4.281036376953125,
|
| 1352 |
+
"learning_rate": 3.052839756592292e-05,
|
| 1353 |
+
"loss": 1.3269,
|
| 1354 |
+
"step": 19200
|
| 1355 |
+
},
|
| 1356 |
+
{
|
| 1357 |
+
"epoch": 19.574036511156187,
|
| 1358 |
+
"grad_norm": 3.887519598007202,
|
| 1359 |
+
"learning_rate": 3.042697768762678e-05,
|
| 1360 |
+
"loss": 1.296,
|
| 1361 |
+
"step": 19300
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 19.675456389452332,
|
| 1365 |
+
"grad_norm": 3.7042200565338135,
|
| 1366 |
+
"learning_rate": 3.032555780933063e-05,
|
| 1367 |
+
"loss": 1.3191,
|
| 1368 |
+
"step": 19400
|
| 1369 |
+
},
|
| 1370 |
+
{
|
| 1371 |
+
"epoch": 19.776876267748477,
|
| 1372 |
+
"grad_norm": 4.007315635681152,
|
| 1373 |
+
"learning_rate": 3.0224137931034484e-05,
|
| 1374 |
+
"loss": 1.3166,
|
| 1375 |
+
"step": 19500
|
| 1376 |
+
},
|
| 1377 |
+
{
|
| 1378 |
+
"epoch": 19.878296146044626,
|
| 1379 |
+
"grad_norm": 4.224331378936768,
|
| 1380 |
+
"learning_rate": 3.0122718052738336e-05,
|
| 1381 |
+
"loss": 1.2987,
|
| 1382 |
+
"step": 19600
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 19.97971602434077,
|
| 1386 |
+
"grad_norm": 4.051445484161377,
|
| 1387 |
+
"learning_rate": 3.002129817444219e-05,
|
| 1388 |
+
"loss": 1.3106,
|
| 1389 |
+
"step": 19700
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 20.081135902636916,
|
| 1393 |
+
"grad_norm": 3.8676743507385254,
|
| 1394 |
+
"learning_rate": 2.9919878296146047e-05,
|
| 1395 |
+
"loss": 1.2985,
|
| 1396 |
+
"step": 19800
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"epoch": 20.18255578093306,
|
| 1400 |
+
"grad_norm": 3.5571846961975098,
|
| 1401 |
+
"learning_rate": 2.98184584178499e-05,
|
| 1402 |
+
"loss": 1.2675,
|
| 1403 |
+
"step": 19900
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 20.28397565922921,
|
| 1407 |
+
"grad_norm": 4.294304370880127,
|
| 1408 |
+
"learning_rate": 2.9717038539553754e-05,
|
| 1409 |
+
"loss": 1.2753,
|
| 1410 |
+
"step": 20000
|
| 1411 |
+
},
|
| 1412 |
+
{
|
| 1413 |
+
"epoch": 20.385395537525355,
|
| 1414 |
+
"grad_norm": 4.6040120124816895,
|
| 1415 |
+
"learning_rate": 2.961561866125761e-05,
|
| 1416 |
+
"loss": 1.2857,
|
| 1417 |
+
"step": 20100
|
| 1418 |
+
},
|
| 1419 |
+
{
|
| 1420 |
+
"epoch": 20.4868154158215,
|
| 1421 |
+
"grad_norm": 4.908646106719971,
|
| 1422 |
+
"learning_rate": 2.951419878296146e-05,
|
| 1423 |
+
"loss": 1.2852,
|
| 1424 |
+
"step": 20200
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 20.58823529411765,
|
| 1428 |
+
"grad_norm": 5.03817081451416,
|
| 1429 |
+
"learning_rate": 2.9412778904665317e-05,
|
| 1430 |
+
"loss": 1.3006,
|
| 1431 |
+
"step": 20300
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 20.689655172413794,
|
| 1435 |
+
"grad_norm": 4.8173909187316895,
|
| 1436 |
+
"learning_rate": 2.931135902636917e-05,
|
| 1437 |
+
"loss": 1.2853,
|
| 1438 |
+
"step": 20400
|
| 1439 |
+
},
|
| 1440 |
+
{
|
| 1441 |
+
"epoch": 20.79107505070994,
|
| 1442 |
+
"grad_norm": 4.681837558746338,
|
| 1443 |
+
"learning_rate": 2.9209939148073024e-05,
|
| 1444 |
+
"loss": 1.2876,
|
| 1445 |
+
"step": 20500
|
| 1446 |
+
},
|
| 1447 |
+
{
|
| 1448 |
+
"epoch": 20.892494929006084,
|
| 1449 |
+
"grad_norm": 4.08661413192749,
|
| 1450 |
+
"learning_rate": 2.910851926977688e-05,
|
| 1451 |
+
"loss": 1.2944,
|
| 1452 |
+
"step": 20600
|
| 1453 |
+
},
|
| 1454 |
+
{
|
| 1455 |
+
"epoch": 20.993914807302232,
|
| 1456 |
+
"grad_norm": 4.02225399017334,
|
| 1457 |
+
"learning_rate": 2.900709939148073e-05,
|
| 1458 |
+
"loss": 1.3013,
|
| 1459 |
+
"step": 20700
|
| 1460 |
+
},
|
| 1461 |
+
{
|
| 1462 |
+
"epoch": 21.095334685598377,
|
| 1463 |
+
"grad_norm": 4.421962261199951,
|
| 1464 |
+
"learning_rate": 2.8905679513184587e-05,
|
| 1465 |
+
"loss": 1.279,
|
| 1466 |
+
"step": 20800
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"epoch": 21.196754563894523,
|
| 1470 |
+
"grad_norm": 4.088225841522217,
|
| 1471 |
+
"learning_rate": 2.8804259634888442e-05,
|
| 1472 |
+
"loss": 1.2638,
|
| 1473 |
+
"step": 20900
|
| 1474 |
+
},
|
| 1475 |
+
{
|
| 1476 |
+
"epoch": 21.29817444219067,
|
| 1477 |
+
"grad_norm": 4.2725701332092285,
|
| 1478 |
+
"learning_rate": 2.870283975659229e-05,
|
| 1479 |
+
"loss": 1.2612,
|
| 1480 |
+
"step": 21000
|
| 1481 |
+
},
|
| 1482 |
+
{
|
| 1483 |
+
"epoch": 21.399594320486816,
|
| 1484 |
+
"grad_norm": 4.10544490814209,
|
| 1485 |
+
"learning_rate": 2.860141987829615e-05,
|
| 1486 |
+
"loss": 1.3006,
|
| 1487 |
+
"step": 21100
|
| 1488 |
+
},
|
| 1489 |
+
{
|
| 1490 |
+
"epoch": 21.50101419878296,
|
| 1491 |
+
"grad_norm": 4.1848626136779785,
|
| 1492 |
+
"learning_rate": 2.8499999999999998e-05,
|
| 1493 |
+
"loss": 1.2772,
|
| 1494 |
+
"step": 21200
|
| 1495 |
+
},
|
| 1496 |
+
{
|
| 1497 |
+
"epoch": 21.602434077079106,
|
| 1498 |
+
"grad_norm": 3.46309494972229,
|
| 1499 |
+
"learning_rate": 2.8398580121703854e-05,
|
| 1500 |
+
"loss": 1.2798,
|
| 1501 |
+
"step": 21300
|
| 1502 |
+
},
|
| 1503 |
+
{
|
| 1504 |
+
"epoch": 21.703853955375255,
|
| 1505 |
+
"grad_norm": 4.1652727127075195,
|
| 1506 |
+
"learning_rate": 2.8297160243407712e-05,
|
| 1507 |
+
"loss": 1.2728,
|
| 1508 |
+
"step": 21400
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 21.8052738336714,
|
| 1512 |
+
"grad_norm": 4.085648059844971,
|
| 1513 |
+
"learning_rate": 2.819574036511156e-05,
|
| 1514 |
+
"loss": 1.2774,
|
| 1515 |
+
"step": 21500
|
| 1516 |
+
},
|
| 1517 |
+
{
|
| 1518 |
+
"epoch": 21.906693711967545,
|
| 1519 |
+
"grad_norm": 4.772132873535156,
|
| 1520 |
+
"learning_rate": 2.8094320486815416e-05,
|
| 1521 |
+
"loss": 1.286,
|
| 1522 |
+
"step": 21600
|
| 1523 |
+
},
|
| 1524 |
+
{
|
| 1525 |
+
"epoch": 22.00811359026369,
|
| 1526 |
+
"grad_norm": 4.132139205932617,
|
| 1527 |
+
"learning_rate": 2.7992900608519275e-05,
|
| 1528 |
+
"loss": 1.2715,
|
| 1529 |
+
"step": 21700
|
| 1530 |
+
},
|
| 1531 |
+
{
|
| 1532 |
+
"epoch": 22.10953346855984,
|
| 1533 |
+
"grad_norm": 3.8104045391082764,
|
| 1534 |
+
"learning_rate": 2.7891480730223124e-05,
|
| 1535 |
+
"loss": 1.2377,
|
| 1536 |
+
"step": 21800
|
| 1537 |
+
},
|
| 1538 |
+
{
|
| 1539 |
+
"epoch": 22.210953346855984,
|
| 1540 |
+
"grad_norm": 4.074591636657715,
|
| 1541 |
+
"learning_rate": 2.779006085192698e-05,
|
| 1542 |
+
"loss": 1.2486,
|
| 1543 |
+
"step": 21900
|
| 1544 |
+
},
|
| 1545 |
+
{
|
| 1546 |
+
"epoch": 22.31237322515213,
|
| 1547 |
+
"grad_norm": 4.18817138671875,
|
| 1548 |
+
"learning_rate": 2.768864097363083e-05,
|
| 1549 |
+
"loss": 1.2921,
|
| 1550 |
+
"step": 22000
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 22.413793103448278,
|
| 1554 |
+
"grad_norm": 4.426843643188477,
|
| 1555 |
+
"learning_rate": 2.7587221095334686e-05,
|
| 1556 |
+
"loss": 1.2686,
|
| 1557 |
+
"step": 22100
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 22.515212981744423,
|
| 1561 |
+
"grad_norm": 3.9790797233581543,
|
| 1562 |
+
"learning_rate": 2.748580121703854e-05,
|
| 1563 |
+
"loss": 1.2648,
|
| 1564 |
+
"step": 22200
|
| 1565 |
+
},
|
| 1566 |
+
{
|
| 1567 |
+
"epoch": 22.616632860040568,
|
| 1568 |
+
"grad_norm": 4.080827236175537,
|
| 1569 |
+
"learning_rate": 2.7384381338742394e-05,
|
| 1570 |
+
"loss": 1.2681,
|
| 1571 |
+
"step": 22300
|
| 1572 |
+
},
|
| 1573 |
+
{
|
| 1574 |
+
"epoch": 22.718052738336713,
|
| 1575 |
+
"grad_norm": 4.053229808807373,
|
| 1576 |
+
"learning_rate": 2.728296146044625e-05,
|
| 1577 |
+
"loss": 1.2844,
|
| 1578 |
+
"step": 22400
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 22.81947261663286,
|
| 1582 |
+
"grad_norm": 4.299985408782959,
|
| 1583 |
+
"learning_rate": 2.7181541582150104e-05,
|
| 1584 |
+
"loss": 1.2867,
|
| 1585 |
+
"step": 22500
|
| 1586 |
+
},
|
| 1587 |
+
{
|
| 1588 |
+
"epoch": 22.920892494929006,
|
| 1589 |
+
"grad_norm": 4.402357578277588,
|
| 1590 |
+
"learning_rate": 2.7080121703853956e-05,
|
| 1591 |
+
"loss": 1.2533,
|
| 1592 |
+
"step": 22600
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 23.02231237322515,
|
| 1596 |
+
"grad_norm": 4.21071195602417,
|
| 1597 |
+
"learning_rate": 2.6978701825557812e-05,
|
| 1598 |
+
"loss": 1.2553,
|
| 1599 |
+
"step": 22700
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 23.123732251521297,
|
| 1603 |
+
"grad_norm": 4.171008110046387,
|
| 1604 |
+
"learning_rate": 2.6877281947261664e-05,
|
| 1605 |
+
"loss": 1.2512,
|
| 1606 |
+
"step": 22800
|
| 1607 |
+
},
|
| 1608 |
+
{
|
| 1609 |
+
"epoch": 23.225152129817445,
|
| 1610 |
+
"grad_norm": 4.028729438781738,
|
| 1611 |
+
"learning_rate": 2.677586206896552e-05,
|
| 1612 |
+
"loss": 1.236,
|
| 1613 |
+
"step": 22900
|
| 1614 |
+
},
|
| 1615 |
+
{
|
| 1616 |
+
"epoch": 23.32657200811359,
|
| 1617 |
+
"grad_norm": 4.201495170593262,
|
| 1618 |
+
"learning_rate": 2.6674442190669374e-05,
|
| 1619 |
+
"loss": 1.2578,
|
| 1620 |
+
"step": 23000
|
| 1621 |
+
},
|
| 1622 |
+
{
|
| 1623 |
+
"epoch": 23.427991886409735,
|
| 1624 |
+
"grad_norm": 3.6161012649536133,
|
| 1625 |
+
"learning_rate": 2.6573022312373223e-05,
|
| 1626 |
+
"loss": 1.2377,
|
| 1627 |
+
"step": 23100
|
| 1628 |
+
},
|
| 1629 |
+
{
|
| 1630 |
+
"epoch": 23.529411764705884,
|
| 1631 |
+
"grad_norm": 3.7652299404144287,
|
| 1632 |
+
"learning_rate": 2.6471602434077082e-05,
|
| 1633 |
+
"loss": 1.2407,
|
| 1634 |
+
"step": 23200
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 23.63083164300203,
|
| 1638 |
+
"grad_norm": 3.6019327640533447,
|
| 1639 |
+
"learning_rate": 2.6370182555780937e-05,
|
| 1640 |
+
"loss": 1.2437,
|
| 1641 |
+
"step": 23300
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 23.732251521298174,
|
| 1645 |
+
"grad_norm": 4.130826950073242,
|
| 1646 |
+
"learning_rate": 2.6268762677484786e-05,
|
| 1647 |
+
"loss": 1.2598,
|
| 1648 |
+
"step": 23400
|
| 1649 |
+
},
|
| 1650 |
+
{
|
| 1651 |
+
"epoch": 23.83367139959432,
|
| 1652 |
+
"grad_norm": 4.149597644805908,
|
| 1653 |
+
"learning_rate": 2.6167342799188644e-05,
|
| 1654 |
+
"loss": 1.2476,
|
| 1655 |
+
"step": 23500
|
| 1656 |
+
},
|
| 1657 |
+
{
|
| 1658 |
+
"epoch": 23.935091277890468,
|
| 1659 |
+
"grad_norm": 4.324525833129883,
|
| 1660 |
+
"learning_rate": 2.6065922920892493e-05,
|
| 1661 |
+
"loss": 1.2775,
|
| 1662 |
+
"step": 23600
|
| 1663 |
+
},
|
| 1664 |
+
{
|
| 1665 |
+
"epoch": 24.036511156186613,
|
| 1666 |
+
"grad_norm": 3.9695489406585693,
|
| 1667 |
+
"learning_rate": 2.596450304259635e-05,
|
| 1668 |
+
"loss": 1.2451,
|
| 1669 |
+
"step": 23700
|
| 1670 |
+
},
|
| 1671 |
+
{
|
| 1672 |
+
"epoch": 24.137931034482758,
|
| 1673 |
+
"grad_norm": 4.289801597595215,
|
| 1674 |
+
"learning_rate": 2.5863083164300207e-05,
|
| 1675 |
+
"loss": 1.2404,
|
| 1676 |
+
"step": 23800
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 24.239350912778903,
|
| 1680 |
+
"grad_norm": 3.993157148361206,
|
| 1681 |
+
"learning_rate": 2.5761663286004056e-05,
|
| 1682 |
+
"loss": 1.2461,
|
| 1683 |
+
"step": 23900
|
| 1684 |
+
},
|
| 1685 |
+
{
|
| 1686 |
+
"epoch": 24.34077079107505,
|
| 1687 |
+
"grad_norm": 3.966215133666992,
|
| 1688 |
+
"learning_rate": 2.566024340770791e-05,
|
| 1689 |
+
"loss": 1.227,
|
| 1690 |
+
"step": 24000
|
| 1691 |
+
},
|
| 1692 |
+
{
|
| 1693 |
+
"epoch": 24.442190669371197,
|
| 1694 |
+
"grad_norm": 4.6348676681518555,
|
| 1695 |
+
"learning_rate": 2.555882352941177e-05,
|
| 1696 |
+
"loss": 1.2079,
|
| 1697 |
+
"step": 24100
|
| 1698 |
+
},
|
| 1699 |
+
{
|
| 1700 |
+
"epoch": 24.543610547667342,
|
| 1701 |
+
"grad_norm": 3.782102108001709,
|
| 1702 |
+
"learning_rate": 2.545740365111562e-05,
|
| 1703 |
+
"loss": 1.2395,
|
| 1704 |
+
"step": 24200
|
| 1705 |
+
},
|
| 1706 |
+
{
|
| 1707 |
+
"epoch": 24.64503042596349,
|
| 1708 |
+
"grad_norm": 4.354336261749268,
|
| 1709 |
+
"learning_rate": 2.5355983772819474e-05,
|
| 1710 |
+
"loss": 1.2447,
|
| 1711 |
+
"step": 24300
|
| 1712 |
+
},
|
| 1713 |
+
{
|
| 1714 |
+
"epoch": 24.746450304259636,
|
| 1715 |
+
"grad_norm": 4.102418422698975,
|
| 1716 |
+
"learning_rate": 2.5254563894523326e-05,
|
| 1717 |
+
"loss": 1.2574,
|
| 1718 |
+
"step": 24400
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"epoch": 24.84787018255578,
|
| 1722 |
+
"grad_norm": 4.408313751220703,
|
| 1723 |
+
"learning_rate": 2.515314401622718e-05,
|
| 1724 |
+
"loss": 1.2685,
|
| 1725 |
+
"step": 24500
|
| 1726 |
+
},
|
| 1727 |
+
{
|
| 1728 |
+
"epoch": 24.949290060851926,
|
| 1729 |
+
"grad_norm": 4.105988025665283,
|
| 1730 |
+
"learning_rate": 2.5051724137931037e-05,
|
| 1731 |
+
"loss": 1.2458,
|
| 1732 |
+
"step": 24600
|
| 1733 |
+
},
|
| 1734 |
+
{
|
| 1735 |
+
"epoch": 25.050709939148074,
|
| 1736 |
+
"grad_norm": 3.945258855819702,
|
| 1737 |
+
"learning_rate": 2.4950304259634892e-05,
|
| 1738 |
+
"loss": 1.2296,
|
| 1739 |
+
"step": 24700
|
| 1740 |
+
},
|
| 1741 |
+
{
|
| 1742 |
+
"epoch": 25.15212981744422,
|
| 1743 |
+
"grad_norm": 4.094118118286133,
|
| 1744 |
+
"learning_rate": 2.4848884381338744e-05,
|
| 1745 |
+
"loss": 1.2418,
|
| 1746 |
+
"step": 24800
|
| 1747 |
+
},
|
| 1748 |
+
{
|
| 1749 |
+
"epoch": 25.253549695740364,
|
| 1750 |
+
"grad_norm": 3.9714670181274414,
|
| 1751 |
+
"learning_rate": 2.4747464503042596e-05,
|
| 1752 |
+
"loss": 1.193,
|
| 1753 |
+
"step": 24900
|
| 1754 |
+
},
|
| 1755 |
+
{
|
| 1756 |
+
"epoch": 25.35496957403651,
|
| 1757 |
+
"grad_norm": 4.001987457275391,
|
| 1758 |
+
"learning_rate": 2.464604462474645e-05,
|
| 1759 |
+
"loss": 1.2221,
|
| 1760 |
+
"step": 25000
|
| 1761 |
+
},
|
| 1762 |
+
{
|
| 1763 |
+
"epoch": 25.456389452332658,
|
| 1764 |
+
"grad_norm": 3.931515693664551,
|
| 1765 |
+
"learning_rate": 2.4544624746450307e-05,
|
| 1766 |
+
"loss": 1.2198,
|
| 1767 |
+
"step": 25100
|
| 1768 |
+
},
|
| 1769 |
+
{
|
| 1770 |
+
"epoch": 25.557809330628803,
|
| 1771 |
+
"grad_norm": 4.188907623291016,
|
| 1772 |
+
"learning_rate": 2.444320486815416e-05,
|
| 1773 |
+
"loss": 1.2354,
|
| 1774 |
+
"step": 25200
|
| 1775 |
+
},
|
| 1776 |
+
{
|
| 1777 |
+
"epoch": 25.65922920892495,
|
| 1778 |
+
"grad_norm": 4.085264205932617,
|
| 1779 |
+
"learning_rate": 2.4341784989858014e-05,
|
| 1780 |
+
"loss": 1.2217,
|
| 1781 |
+
"step": 25300
|
| 1782 |
+
},
|
| 1783 |
+
{
|
| 1784 |
+
"epoch": 25.760649087221097,
|
| 1785 |
+
"grad_norm": 4.202343463897705,
|
| 1786 |
+
"learning_rate": 2.4240365111561866e-05,
|
| 1787 |
+
"loss": 1.2179,
|
| 1788 |
+
"step": 25400
|
| 1789 |
+
},
|
| 1790 |
+
{
|
| 1791 |
+
"epoch": 25.862068965517242,
|
| 1792 |
+
"grad_norm": 4.085913181304932,
|
| 1793 |
+
"learning_rate": 2.413894523326572e-05,
|
| 1794 |
+
"loss": 1.2296,
|
| 1795 |
+
"step": 25500
|
| 1796 |
+
},
|
| 1797 |
+
{
|
| 1798 |
+
"epoch": 25.963488843813387,
|
| 1799 |
+
"grad_norm": 4.521745204925537,
|
| 1800 |
+
"learning_rate": 2.4037525354969577e-05,
|
| 1801 |
+
"loss": 1.2358,
|
| 1802 |
+
"step": 25600
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 26.064908722109532,
|
| 1806 |
+
"grad_norm": 4.399627685546875,
|
| 1807 |
+
"learning_rate": 2.393610547667343e-05,
|
| 1808 |
+
"loss": 1.2181,
|
| 1809 |
+
"step": 25700
|
| 1810 |
+
},
|
| 1811 |
+
{
|
| 1812 |
+
"epoch": 26.16632860040568,
|
| 1813 |
+
"grad_norm": 3.6188008785247803,
|
| 1814 |
+
"learning_rate": 2.383468559837728e-05,
|
| 1815 |
+
"loss": 1.213,
|
| 1816 |
+
"step": 25800
|
| 1817 |
+
},
|
| 1818 |
+
{
|
| 1819 |
+
"epoch": 26.267748478701826,
|
| 1820 |
+
"grad_norm": 4.37255334854126,
|
| 1821 |
+
"learning_rate": 2.373326572008114e-05,
|
| 1822 |
+
"loss": 1.2282,
|
| 1823 |
+
"step": 25900
|
| 1824 |
+
},
|
| 1825 |
+
{
|
| 1826 |
+
"epoch": 26.36916835699797,
|
| 1827 |
+
"grad_norm": 4.154409408569336,
|
| 1828 |
+
"learning_rate": 2.363184584178499e-05,
|
| 1829 |
+
"loss": 1.2046,
|
| 1830 |
+
"step": 26000
|
| 1831 |
+
},
|
| 1832 |
+
{
|
| 1833 |
+
"epoch": 26.470588235294116,
|
| 1834 |
+
"grad_norm": 4.17800235748291,
|
| 1835 |
+
"learning_rate": 2.3530425963488843e-05,
|
| 1836 |
+
"loss": 1.2119,
|
| 1837 |
+
"step": 26100
|
| 1838 |
+
},
|
| 1839 |
+
{
|
| 1840 |
+
"epoch": 26.572008113590265,
|
| 1841 |
+
"grad_norm": 3.617309808731079,
|
| 1842 |
+
"learning_rate": 2.34290060851927e-05,
|
| 1843 |
+
"loss": 1.2079,
|
| 1844 |
+
"step": 26200
|
| 1845 |
+
},
|
| 1846 |
+
{
|
| 1847 |
+
"epoch": 26.67342799188641,
|
| 1848 |
+
"grad_norm": 3.834677219390869,
|
| 1849 |
+
"learning_rate": 2.3327586206896554e-05,
|
| 1850 |
+
"loss": 1.2204,
|
| 1851 |
+
"step": 26300
|
| 1852 |
+
},
|
| 1853 |
+
{
|
| 1854 |
+
"epoch": 26.774847870182555,
|
| 1855 |
+
"grad_norm": 4.5550336837768555,
|
| 1856 |
+
"learning_rate": 2.3226166328600406e-05,
|
| 1857 |
+
"loss": 1.2035,
|
| 1858 |
+
"step": 26400
|
| 1859 |
+
},
|
| 1860 |
+
{
|
| 1861 |
+
"epoch": 26.876267748478703,
|
| 1862 |
+
"grad_norm": 3.776331901550293,
|
| 1863 |
+
"learning_rate": 2.312474645030426e-05,
|
| 1864 |
+
"loss": 1.2337,
|
| 1865 |
+
"step": 26500
|
| 1866 |
+
},
|
| 1867 |
+
{
|
| 1868 |
+
"epoch": 26.97768762677485,
|
| 1869 |
+
"grad_norm": 3.869279623031616,
|
| 1870 |
+
"learning_rate": 2.3023326572008113e-05,
|
| 1871 |
+
"loss": 1.2285,
|
| 1872 |
+
"step": 26600
|
| 1873 |
+
},
|
| 1874 |
+
{
|
| 1875 |
+
"epoch": 27.079107505070994,
|
| 1876 |
+
"grad_norm": 3.822451591491699,
|
| 1877 |
+
"learning_rate": 2.292190669371197e-05,
|
| 1878 |
+
"loss": 1.2021,
|
| 1879 |
+
"step": 26700
|
| 1880 |
+
},
|
| 1881 |
+
{
|
| 1882 |
+
"epoch": 27.18052738336714,
|
| 1883 |
+
"grad_norm": 4.070253372192383,
|
| 1884 |
+
"learning_rate": 2.2820486815415824e-05,
|
| 1885 |
+
"loss": 1.1907,
|
| 1886 |
+
"step": 26800
|
| 1887 |
+
},
|
| 1888 |
+
{
|
| 1889 |
+
"epoch": 27.281947261663287,
|
| 1890 |
+
"grad_norm": 4.30029821395874,
|
| 1891 |
+
"learning_rate": 2.2719066937119676e-05,
|
| 1892 |
+
"loss": 1.2131,
|
| 1893 |
+
"step": 26900
|
| 1894 |
+
},
|
| 1895 |
+
{
|
| 1896 |
+
"epoch": 27.383367139959432,
|
| 1897 |
+
"grad_norm": 3.917759418487549,
|
| 1898 |
+
"learning_rate": 2.2617647058823528e-05,
|
| 1899 |
+
"loss": 1.2138,
|
| 1900 |
+
"step": 27000
|
| 1901 |
+
},
|
| 1902 |
+
{
|
| 1903 |
+
"epoch": 27.484787018255577,
|
| 1904 |
+
"grad_norm": 3.384033679962158,
|
| 1905 |
+
"learning_rate": 2.2516227180527387e-05,
|
| 1906 |
+
"loss": 1.202,
|
| 1907 |
+
"step": 27100
|
| 1908 |
+
},
|
| 1909 |
+
{
|
| 1910 |
+
"epoch": 27.586206896551722,
|
| 1911 |
+
"grad_norm": 3.878225564956665,
|
| 1912 |
+
"learning_rate": 2.241480730223124e-05,
|
| 1913 |
+
"loss": 1.1805,
|
| 1914 |
+
"step": 27200
|
| 1915 |
+
},
|
| 1916 |
+
{
|
| 1917 |
+
"epoch": 27.68762677484787,
|
| 1918 |
+
"grad_norm": 4.346697807312012,
|
| 1919 |
+
"learning_rate": 2.231338742393509e-05,
|
| 1920 |
+
"loss": 1.2233,
|
| 1921 |
+
"step": 27300
|
| 1922 |
+
},
|
| 1923 |
+
{
|
| 1924 |
+
"epoch": 27.789046653144016,
|
| 1925 |
+
"grad_norm": 4.826686382293701,
|
| 1926 |
+
"learning_rate": 2.2211967545638946e-05,
|
| 1927 |
+
"loss": 1.1878,
|
| 1928 |
+
"step": 27400
|
| 1929 |
+
},
|
| 1930 |
+
{
|
| 1931 |
+
"epoch": 27.89046653144016,
|
| 1932 |
+
"grad_norm": 4.52904748916626,
|
| 1933 |
+
"learning_rate": 2.21105476673428e-05,
|
| 1934 |
+
"loss": 1.2054,
|
| 1935 |
+
"step": 27500
|
| 1936 |
+
},
|
| 1937 |
+
{
|
| 1938 |
+
"epoch": 27.99188640973631,
|
| 1939 |
+
"grad_norm": 3.991633892059326,
|
| 1940 |
+
"learning_rate": 2.2009127789046654e-05,
|
| 1941 |
+
"loss": 1.2113,
|
| 1942 |
+
"step": 27600
|
| 1943 |
+
},
|
| 1944 |
+
{
|
| 1945 |
+
"epoch": 28.093306288032455,
|
| 1946 |
+
"grad_norm": 4.163396835327148,
|
| 1947 |
+
"learning_rate": 2.190770791075051e-05,
|
| 1948 |
+
"loss": 1.1967,
|
| 1949 |
+
"step": 27700
|
| 1950 |
+
},
|
| 1951 |
+
{
|
| 1952 |
+
"epoch": 28.1947261663286,
|
| 1953 |
+
"grad_norm": 4.316657543182373,
|
| 1954 |
+
"learning_rate": 2.180628803245436e-05,
|
| 1955 |
+
"loss": 1.1923,
|
| 1956 |
+
"step": 27800
|
| 1957 |
+
},
|
| 1958 |
+
{
|
| 1959 |
+
"epoch": 28.296146044624745,
|
| 1960 |
+
"grad_norm": 3.8177928924560547,
|
| 1961 |
+
"learning_rate": 2.1704868154158216e-05,
|
| 1962 |
+
"loss": 1.176,
|
| 1963 |
+
"step": 27900
|
| 1964 |
+
},
|
| 1965 |
+
{
|
| 1966 |
+
"epoch": 28.397565922920894,
|
| 1967 |
+
"grad_norm": 4.520811080932617,
|
| 1968 |
+
"learning_rate": 2.160344827586207e-05,
|
| 1969 |
+
"loss": 1.1999,
|
| 1970 |
+
"step": 28000
|
| 1971 |
+
},
|
| 1972 |
+
{
|
| 1973 |
+
"epoch": 28.49898580121704,
|
| 1974 |
+
"grad_norm": 3.9443113803863525,
|
| 1975 |
+
"learning_rate": 2.1502028397565924e-05,
|
| 1976 |
+
"loss": 1.1806,
|
| 1977 |
+
"step": 28100
|
| 1978 |
+
},
|
| 1979 |
+
{
|
| 1980 |
+
"epoch": 28.600405679513184,
|
| 1981 |
+
"grad_norm": 4.400629043579102,
|
| 1982 |
+
"learning_rate": 2.1400608519269776e-05,
|
| 1983 |
+
"loss": 1.2019,
|
| 1984 |
+
"step": 28200
|
| 1985 |
+
},
|
| 1986 |
+
{
|
| 1987 |
+
"epoch": 28.70182555780933,
|
| 1988 |
+
"grad_norm": 4.546008110046387,
|
| 1989 |
+
"learning_rate": 2.1299188640973634e-05,
|
| 1990 |
+
"loss": 1.2033,
|
| 1991 |
+
"step": 28300
|
| 1992 |
+
},
|
| 1993 |
+
{
|
| 1994 |
+
"epoch": 28.803245436105477,
|
| 1995 |
+
"grad_norm": 3.394660711288452,
|
| 1996 |
+
"learning_rate": 2.1197768762677486e-05,
|
| 1997 |
+
"loss": 1.1699,
|
| 1998 |
+
"step": 28400
|
| 1999 |
+
},
|
| 2000 |
+
{
|
| 2001 |
+
"epoch": 28.904665314401623,
|
| 2002 |
+
"grad_norm": 4.7550530433654785,
|
| 2003 |
+
"learning_rate": 2.1096348884381338e-05,
|
| 2004 |
+
"loss": 1.2116,
|
| 2005 |
+
"step": 28500
|
| 2006 |
+
},
|
| 2007 |
+
{
|
| 2008 |
+
"epoch": 29.006085192697768,
|
| 2009 |
+
"grad_norm": 4.705698490142822,
|
| 2010 |
+
"learning_rate": 2.0994929006085194e-05,
|
| 2011 |
+
"loss": 1.1926,
|
| 2012 |
+
"step": 28600
|
| 2013 |
+
},
|
| 2014 |
+
{
|
| 2015 |
+
"epoch": 29.107505070993916,
|
| 2016 |
+
"grad_norm": 3.963656187057495,
|
| 2017 |
+
"learning_rate": 2.089350912778905e-05,
|
| 2018 |
+
"loss": 1.1836,
|
| 2019 |
+
"step": 28700
|
| 2020 |
+
},
|
| 2021 |
+
{
|
| 2022 |
+
"epoch": 29.20892494929006,
|
| 2023 |
+
"grad_norm": 4.29594087600708,
|
| 2024 |
+
"learning_rate": 2.07920892494929e-05,
|
| 2025 |
+
"loss": 1.1751,
|
| 2026 |
+
"step": 28800
|
| 2027 |
+
},
|
| 2028 |
+
{
|
| 2029 |
+
"epoch": 29.310344827586206,
|
| 2030 |
+
"grad_norm": 4.209803581237793,
|
| 2031 |
+
"learning_rate": 2.0690669371196756e-05,
|
| 2032 |
+
"loss": 1.1957,
|
| 2033 |
+
"step": 28900
|
| 2034 |
+
},
|
| 2035 |
+
{
|
| 2036 |
+
"epoch": 29.41176470588235,
|
| 2037 |
+
"grad_norm": 3.5338592529296875,
|
| 2038 |
+
"learning_rate": 2.058924949290061e-05,
|
| 2039 |
+
"loss": 1.1754,
|
| 2040 |
+
"step": 29000
|
| 2041 |
+
},
|
| 2042 |
+
{
|
| 2043 |
+
"epoch": 29.5131845841785,
|
| 2044 |
+
"grad_norm": 4.474703788757324,
|
| 2045 |
+
"learning_rate": 2.0487829614604464e-05,
|
| 2046 |
+
"loss": 1.1816,
|
| 2047 |
+
"step": 29100
|
| 2048 |
+
},
|
| 2049 |
+
{
|
| 2050 |
+
"epoch": 29.614604462474645,
|
| 2051 |
+
"grad_norm": 4.149474620819092,
|
| 2052 |
+
"learning_rate": 2.038640973630832e-05,
|
| 2053 |
+
"loss": 1.1857,
|
| 2054 |
+
"step": 29200
|
| 2055 |
+
},
|
| 2056 |
+
{
|
| 2057 |
+
"epoch": 29.71602434077079,
|
| 2058 |
+
"grad_norm": 4.727872371673584,
|
| 2059 |
+
"learning_rate": 2.028498985801217e-05,
|
| 2060 |
+
"loss": 1.1688,
|
| 2061 |
+
"step": 29300
|
| 2062 |
+
},
|
| 2063 |
+
{
|
| 2064 |
+
"epoch": 29.817444219066935,
|
| 2065 |
+
"grad_norm": 4.274538040161133,
|
| 2066 |
+
"learning_rate": 2.0183569979716023e-05,
|
| 2067 |
+
"loss": 1.1858,
|
| 2068 |
+
"step": 29400
|
| 2069 |
+
},
|
| 2070 |
+
{
|
| 2071 |
+
"epoch": 29.918864097363084,
|
| 2072 |
+
"grad_norm": 3.7311525344848633,
|
| 2073 |
+
"learning_rate": 2.0082150101419882e-05,
|
| 2074 |
+
"loss": 1.1812,
|
| 2075 |
+
"step": 29500
|
| 2076 |
+
},
|
| 2077 |
+
{
|
| 2078 |
+
"epoch": 30.02028397565923,
|
| 2079 |
+
"grad_norm": 3.8032689094543457,
|
| 2080 |
+
"learning_rate": 1.9980730223123734e-05,
|
| 2081 |
+
"loss": 1.1811,
|
| 2082 |
+
"step": 29600
|
| 2083 |
+
},
|
| 2084 |
+
{
|
| 2085 |
+
"epoch": 30.121703853955374,
|
| 2086 |
+
"grad_norm": 4.139228343963623,
|
| 2087 |
+
"learning_rate": 1.9879310344827586e-05,
|
| 2088 |
+
"loss": 1.1719,
|
| 2089 |
+
"step": 29700
|
| 2090 |
+
},
|
| 2091 |
+
{
|
| 2092 |
+
"epoch": 30.223123732251523,
|
| 2093 |
+
"grad_norm": 4.439788341522217,
|
| 2094 |
+
"learning_rate": 1.977789046653144e-05,
|
| 2095 |
+
"loss": 1.1729,
|
| 2096 |
+
"step": 29800
|
| 2097 |
+
},
|
| 2098 |
+
{
|
| 2099 |
+
"epoch": 30.324543610547668,
|
| 2100 |
+
"grad_norm": 4.718489646911621,
|
| 2101 |
+
"learning_rate": 1.9676470588235296e-05,
|
| 2102 |
+
"loss": 1.1695,
|
| 2103 |
+
"step": 29900
|
| 2104 |
+
},
|
| 2105 |
+
{
|
| 2106 |
+
"epoch": 30.425963488843813,
|
| 2107 |
+
"grad_norm": 4.7541422843933105,
|
| 2108 |
+
"learning_rate": 1.957505070993915e-05,
|
| 2109 |
+
"loss": 1.1844,
|
| 2110 |
+
"step": 30000
|
| 2111 |
+
},
|
| 2112 |
+
{
|
| 2113 |
+
"epoch": 30.527383367139958,
|
| 2114 |
+
"grad_norm": 4.507453441619873,
|
| 2115 |
+
"learning_rate": 1.9473630831643004e-05,
|
| 2116 |
+
"loss": 1.166,
|
| 2117 |
+
"step": 30100
|
| 2118 |
+
},
|
| 2119 |
+
{
|
| 2120 |
+
"epoch": 30.628803245436107,
|
| 2121 |
+
"grad_norm": 4.363028526306152,
|
| 2122 |
+
"learning_rate": 1.9372210953346856e-05,
|
| 2123 |
+
"loss": 1.1816,
|
| 2124 |
+
"step": 30200
|
| 2125 |
+
},
|
| 2126 |
+
{
|
| 2127 |
+
"epoch": 30.73022312373225,
|
| 2128 |
+
"grad_norm": 4.432469844818115,
|
| 2129 |
+
"learning_rate": 1.927079107505071e-05,
|
| 2130 |
+
"loss": 1.1724,
|
| 2131 |
+
"step": 30300
|
| 2132 |
+
},
|
| 2133 |
+
{
|
| 2134 |
+
"epoch": 30.831643002028397,
|
| 2135 |
+
"grad_norm": 4.204954624176025,
|
| 2136 |
+
"learning_rate": 1.9169371196754567e-05,
|
| 2137 |
+
"loss": 1.1793,
|
| 2138 |
+
"step": 30400
|
| 2139 |
+
},
|
| 2140 |
+
{
|
| 2141 |
+
"epoch": 30.933062880324545,
|
| 2142 |
+
"grad_norm": 3.8087048530578613,
|
| 2143 |
+
"learning_rate": 1.906795131845842e-05,
|
| 2144 |
+
"loss": 1.1573,
|
| 2145 |
+
"step": 30500
|
| 2146 |
+
},
|
| 2147 |
+
{
|
| 2148 |
+
"epoch": 31.03448275862069,
|
| 2149 |
+
"grad_norm": 4.493447303771973,
|
| 2150 |
+
"learning_rate": 1.896653144016227e-05,
|
| 2151 |
+
"loss": 1.1674,
|
| 2152 |
+
"step": 30600
|
| 2153 |
+
},
|
| 2154 |
+
{
|
| 2155 |
+
"epoch": 31.135902636916835,
|
| 2156 |
+
"grad_norm": 3.788645029067993,
|
| 2157 |
+
"learning_rate": 1.886511156186613e-05,
|
| 2158 |
+
"loss": 1.166,
|
| 2159 |
+
"step": 30700
|
| 2160 |
+
},
|
| 2161 |
+
{
|
| 2162 |
+
"epoch": 31.23732251521298,
|
| 2163 |
+
"grad_norm": 4.687623977661133,
|
| 2164 |
+
"learning_rate": 1.876369168356998e-05,
|
| 2165 |
+
"loss": 1.1738,
|
| 2166 |
+
"step": 30800
|
| 2167 |
+
},
|
| 2168 |
+
{
|
| 2169 |
+
"epoch": 31.33874239350913,
|
| 2170 |
+
"grad_norm": 3.6606080532073975,
|
| 2171 |
+
"learning_rate": 1.8662271805273833e-05,
|
| 2172 |
+
"loss": 1.1562,
|
| 2173 |
+
"step": 30900
|
| 2174 |
+
},
|
| 2175 |
+
{
|
| 2176 |
+
"epoch": 31.440162271805274,
|
| 2177 |
+
"grad_norm": 4.326778888702393,
|
| 2178 |
+
"learning_rate": 1.856085192697769e-05,
|
| 2179 |
+
"loss": 1.1662,
|
| 2180 |
+
"step": 31000
|
| 2181 |
+
},
|
| 2182 |
+
{
|
| 2183 |
+
"epoch": 31.54158215010142,
|
| 2184 |
+
"grad_norm": 4.338911533355713,
|
| 2185 |
+
"learning_rate": 1.8459432048681544e-05,
|
| 2186 |
+
"loss": 1.1653,
|
| 2187 |
+
"step": 31100
|
| 2188 |
+
},
|
| 2189 |
+
{
|
| 2190 |
+
"epoch": 31.643002028397564,
|
| 2191 |
+
"grad_norm": 4.431127548217773,
|
| 2192 |
+
"learning_rate": 1.8358012170385396e-05,
|
| 2193 |
+
"loss": 1.1755,
|
| 2194 |
+
"step": 31200
|
| 2195 |
+
},
|
| 2196 |
+
{
|
| 2197 |
+
"epoch": 31.744421906693713,
|
| 2198 |
+
"grad_norm": 4.019861221313477,
|
| 2199 |
+
"learning_rate": 1.825659229208925e-05,
|
| 2200 |
+
"loss": 1.1637,
|
| 2201 |
+
"step": 31300
|
| 2202 |
+
},
|
| 2203 |
+
{
|
| 2204 |
+
"epoch": 31.845841784989858,
|
| 2205 |
+
"grad_norm": 4.0842156410217285,
|
| 2206 |
+
"learning_rate": 1.8155172413793103e-05,
|
| 2207 |
+
"loss": 1.155,
|
| 2208 |
+
"step": 31400
|
| 2209 |
+
},
|
| 2210 |
+
{
|
| 2211 |
+
"epoch": 31.947261663286003,
|
| 2212 |
+
"grad_norm": 3.6525912284851074,
|
| 2213 |
+
"learning_rate": 1.805375253549696e-05,
|
| 2214 |
+
"loss": 1.1771,
|
| 2215 |
+
"step": 31500
|
| 2216 |
+
},
|
| 2217 |
+
{
|
| 2218 |
+
"epoch": 32.04868154158215,
|
| 2219 |
+
"grad_norm": 4.539126873016357,
|
| 2220 |
+
"learning_rate": 1.7952332657200814e-05,
|
| 2221 |
+
"loss": 1.1716,
|
| 2222 |
+
"step": 31600
|
| 2223 |
+
},
|
| 2224 |
+
{
|
| 2225 |
+
"epoch": 32.1501014198783,
|
| 2226 |
+
"grad_norm": 3.795107364654541,
|
| 2227 |
+
"learning_rate": 1.7850912778904666e-05,
|
| 2228 |
+
"loss": 1.1386,
|
| 2229 |
+
"step": 31700
|
| 2230 |
+
},
|
| 2231 |
+
{
|
| 2232 |
+
"epoch": 32.251521298174445,
|
| 2233 |
+
"grad_norm": 3.7753000259399414,
|
| 2234 |
+
"learning_rate": 1.7749492900608518e-05,
|
| 2235 |
+
"loss": 1.1484,
|
| 2236 |
+
"step": 31800
|
| 2237 |
+
},
|
| 2238 |
+
{
|
| 2239 |
+
"epoch": 32.35294117647059,
|
| 2240 |
+
"grad_norm": 3.8549020290374756,
|
| 2241 |
+
"learning_rate": 1.7648073022312377e-05,
|
| 2242 |
+
"loss": 1.1395,
|
| 2243 |
+
"step": 31900
|
| 2244 |
+
},
|
| 2245 |
+
{
|
| 2246 |
+
"epoch": 32.454361054766736,
|
| 2247 |
+
"grad_norm": 4.203573226928711,
|
| 2248 |
+
"learning_rate": 1.754665314401623e-05,
|
| 2249 |
+
"loss": 1.1709,
|
| 2250 |
+
"step": 32000
|
| 2251 |
+
},
|
| 2252 |
+
{
|
| 2253 |
+
"epoch": 32.55578093306288,
|
| 2254 |
+
"grad_norm": 4.275285720825195,
|
| 2255 |
+
"learning_rate": 1.744523326572008e-05,
|
| 2256 |
+
"loss": 1.1622,
|
| 2257 |
+
"step": 32100
|
| 2258 |
+
},
|
| 2259 |
+
{
|
| 2260 |
+
"epoch": 32.657200811359026,
|
| 2261 |
+
"grad_norm": 3.709843873977661,
|
| 2262 |
+
"learning_rate": 1.7343813387423936e-05,
|
| 2263 |
+
"loss": 1.1742,
|
| 2264 |
+
"step": 32200
|
| 2265 |
+
},
|
| 2266 |
+
{
|
| 2267 |
+
"epoch": 32.758620689655174,
|
| 2268 |
+
"grad_norm": 4.47550106048584,
|
| 2269 |
+
"learning_rate": 1.724239350912779e-05,
|
| 2270 |
+
"loss": 1.153,
|
| 2271 |
+
"step": 32300
|
| 2272 |
+
},
|
| 2273 |
+
{
|
| 2274 |
+
"epoch": 32.860040567951316,
|
| 2275 |
+
"grad_norm": 4.055787563323975,
|
| 2276 |
+
"learning_rate": 1.7140973630831643e-05,
|
| 2277 |
+
"loss": 1.1765,
|
| 2278 |
+
"step": 32400
|
| 2279 |
+
},
|
| 2280 |
+
{
|
| 2281 |
+
"epoch": 32.961460446247465,
|
| 2282 |
+
"grad_norm": 3.901118278503418,
|
| 2283 |
+
"learning_rate": 1.70395537525355e-05,
|
| 2284 |
+
"loss": 1.1538,
|
| 2285 |
+
"step": 32500
|
| 2286 |
+
},
|
| 2287 |
+
{
|
| 2288 |
+
"epoch": 33.06288032454361,
|
| 2289 |
+
"grad_norm": 3.3739051818847656,
|
| 2290 |
+
"learning_rate": 1.693813387423935e-05,
|
| 2291 |
+
"loss": 1.1574,
|
| 2292 |
+
"step": 32600
|
| 2293 |
+
},
|
| 2294 |
+
{
|
| 2295 |
+
"epoch": 33.164300202839755,
|
| 2296 |
+
"grad_norm": 5.091039180755615,
|
| 2297 |
+
"learning_rate": 1.6836713995943206e-05,
|
| 2298 |
+
"loss": 1.147,
|
| 2299 |
+
"step": 32700
|
| 2300 |
+
},
|
| 2301 |
+
{
|
| 2302 |
+
"epoch": 33.2657200811359,
|
| 2303 |
+
"grad_norm": 3.774888277053833,
|
| 2304 |
+
"learning_rate": 1.673529411764706e-05,
|
| 2305 |
+
"loss": 1.1539,
|
| 2306 |
+
"step": 32800
|
| 2307 |
+
},
|
| 2308 |
+
{
|
| 2309 |
+
"epoch": 33.36713995943205,
|
| 2310 |
+
"grad_norm": 3.8296284675598145,
|
| 2311 |
+
"learning_rate": 1.6633874239350913e-05,
|
| 2312 |
+
"loss": 1.1381,
|
| 2313 |
+
"step": 32900
|
| 2314 |
+
},
|
| 2315 |
+
{
|
| 2316 |
+
"epoch": 33.46855983772819,
|
| 2317 |
+
"grad_norm": 4.447118282318115,
|
| 2318 |
+
"learning_rate": 1.6532454361054765e-05,
|
| 2319 |
+
"loss": 1.1588,
|
| 2320 |
+
"step": 33000
|
| 2321 |
+
},
|
| 2322 |
+
{
|
| 2323 |
+
"epoch": 33.56997971602434,
|
| 2324 |
+
"grad_norm": 3.8956475257873535,
|
| 2325 |
+
"learning_rate": 1.6431034482758624e-05,
|
| 2326 |
+
"loss": 1.1368,
|
| 2327 |
+
"step": 33100
|
| 2328 |
+
},
|
| 2329 |
+
{
|
| 2330 |
+
"epoch": 33.671399594320484,
|
| 2331 |
+
"grad_norm": 4.509363174438477,
|
| 2332 |
+
"learning_rate": 1.6329614604462476e-05,
|
| 2333 |
+
"loss": 1.1397,
|
| 2334 |
+
"step": 33200
|
| 2335 |
+
},
|
| 2336 |
+
{
|
| 2337 |
+
"epoch": 33.77281947261663,
|
| 2338 |
+
"grad_norm": 4.429649353027344,
|
| 2339 |
+
"learning_rate": 1.6228194726166328e-05,
|
| 2340 |
+
"loss": 1.1567,
|
| 2341 |
+
"step": 33300
|
| 2342 |
+
},
|
| 2343 |
+
{
|
| 2344 |
+
"epoch": 33.87423935091278,
|
| 2345 |
+
"grad_norm": 4.257778167724609,
|
| 2346 |
+
"learning_rate": 1.6126774847870183e-05,
|
| 2347 |
+
"loss": 1.1719,
|
| 2348 |
+
"step": 33400
|
| 2349 |
+
},
|
| 2350 |
+
{
|
| 2351 |
+
"epoch": 33.97565922920892,
|
| 2352 |
+
"grad_norm": 3.7069602012634277,
|
| 2353 |
+
"learning_rate": 1.602535496957404e-05,
|
| 2354 |
+
"loss": 1.1788,
|
| 2355 |
+
"step": 33500
|
| 2356 |
+
},
|
| 2357 |
+
{
|
| 2358 |
+
"epoch": 34.07707910750507,
|
| 2359 |
+
"grad_norm": 3.900291919708252,
|
| 2360 |
+
"learning_rate": 1.592393509127789e-05,
|
| 2361 |
+
"loss": 1.1463,
|
| 2362 |
+
"step": 33600
|
| 2363 |
+
},
|
| 2364 |
+
{
|
| 2365 |
+
"epoch": 34.17849898580122,
|
| 2366 |
+
"grad_norm": 4.337148666381836,
|
| 2367 |
+
"learning_rate": 1.5822515212981746e-05,
|
| 2368 |
+
"loss": 1.136,
|
| 2369 |
+
"step": 33700
|
| 2370 |
+
},
|
| 2371 |
+
{
|
| 2372 |
+
"epoch": 34.27991886409736,
|
| 2373 |
+
"grad_norm": 4.398409843444824,
|
| 2374 |
+
"learning_rate": 1.5721095334685598e-05,
|
| 2375 |
+
"loss": 1.1318,
|
| 2376 |
+
"step": 33800
|
| 2377 |
+
},
|
| 2378 |
+
{
|
| 2379 |
+
"epoch": 34.38133874239351,
|
| 2380 |
+
"grad_norm": 3.7657828330993652,
|
| 2381 |
+
"learning_rate": 1.5619675456389454e-05,
|
| 2382 |
+
"loss": 1.1257,
|
| 2383 |
+
"step": 33900
|
| 2384 |
+
},
|
| 2385 |
+
{
|
| 2386 |
+
"epoch": 34.48275862068966,
|
| 2387 |
+
"grad_norm": 3.638970136642456,
|
| 2388 |
+
"learning_rate": 1.551825557809331e-05,
|
| 2389 |
+
"loss": 1.1322,
|
| 2390 |
+
"step": 34000
|
| 2391 |
+
},
|
| 2392 |
+
{
|
| 2393 |
+
"epoch": 34.5841784989858,
|
| 2394 |
+
"grad_norm": 4.614083290100098,
|
| 2395 |
+
"learning_rate": 1.541683569979716e-05,
|
| 2396 |
+
"loss": 1.1476,
|
| 2397 |
+
"step": 34100
|
| 2398 |
+
},
|
| 2399 |
+
{
|
| 2400 |
+
"epoch": 34.68559837728195,
|
| 2401 |
+
"grad_norm": 4.460692405700684,
|
| 2402 |
+
"learning_rate": 1.5315415821501013e-05,
|
| 2403 |
+
"loss": 1.1371,
|
| 2404 |
+
"step": 34200
|
| 2405 |
+
},
|
| 2406 |
+
{
|
| 2407 |
+
"epoch": 34.78701825557809,
|
| 2408 |
+
"grad_norm": 4.388944625854492,
|
| 2409 |
+
"learning_rate": 1.521399594320487e-05,
|
| 2410 |
+
"loss": 1.1717,
|
| 2411 |
+
"step": 34300
|
| 2412 |
+
},
|
| 2413 |
+
{
|
| 2414 |
+
"epoch": 34.88843813387424,
|
| 2415 |
+
"grad_norm": 4.264405250549316,
|
| 2416 |
+
"learning_rate": 1.5112576064908724e-05,
|
| 2417 |
+
"loss": 1.1289,
|
| 2418 |
+
"step": 34400
|
| 2419 |
+
},
|
| 2420 |
+
{
|
| 2421 |
+
"epoch": 34.98985801217039,
|
| 2422 |
+
"grad_norm": 4.518435955047607,
|
| 2423 |
+
"learning_rate": 1.5011156186612577e-05,
|
| 2424 |
+
"loss": 1.1403,
|
| 2425 |
+
"step": 34500
|
| 2426 |
+
},
|
| 2427 |
+
{
|
| 2428 |
+
"epoch": 35.09127789046653,
|
| 2429 |
+
"grad_norm": 4.1524977684021,
|
| 2430 |
+
"learning_rate": 1.490973630831643e-05,
|
| 2431 |
+
"loss": 1.1242,
|
| 2432 |
+
"step": 34600
|
| 2433 |
+
},
|
| 2434 |
+
{
|
| 2435 |
+
"epoch": 35.19269776876268,
|
| 2436 |
+
"grad_norm": 4.083790302276611,
|
| 2437 |
+
"learning_rate": 1.4808316430020286e-05,
|
| 2438 |
+
"loss": 1.1244,
|
| 2439 |
+
"step": 34700
|
| 2440 |
+
},
|
| 2441 |
+
{
|
| 2442 |
+
"epoch": 35.294117647058826,
|
| 2443 |
+
"grad_norm": 4.083959102630615,
|
| 2444 |
+
"learning_rate": 1.470689655172414e-05,
|
| 2445 |
+
"loss": 1.1382,
|
| 2446 |
+
"step": 34800
|
| 2447 |
+
},
|
| 2448 |
+
{
|
| 2449 |
+
"epoch": 35.39553752535497,
|
| 2450 |
+
"grad_norm": 4.660854816436768,
|
| 2451 |
+
"learning_rate": 1.4605476673427992e-05,
|
| 2452 |
+
"loss": 1.1182,
|
| 2453 |
+
"step": 34900
|
| 2454 |
+
},
|
| 2455 |
+
{
|
| 2456 |
+
"epoch": 35.496957403651116,
|
| 2457 |
+
"grad_norm": 4.251002788543701,
|
| 2458 |
+
"learning_rate": 1.4504056795131846e-05,
|
| 2459 |
+
"loss": 1.1291,
|
| 2460 |
+
"step": 35000
|
| 2461 |
+
},
|
| 2462 |
+
{
|
| 2463 |
+
"epoch": 35.598377281947265,
|
| 2464 |
+
"grad_norm": 4.217166900634766,
|
| 2465 |
+
"learning_rate": 1.4402636916835701e-05,
|
| 2466 |
+
"loss": 1.1382,
|
| 2467 |
+
"step": 35100
|
| 2468 |
+
},
|
| 2469 |
+
{
|
| 2470 |
+
"epoch": 35.699797160243406,
|
| 2471 |
+
"grad_norm": 4.404730796813965,
|
| 2472 |
+
"learning_rate": 1.4301217038539555e-05,
|
| 2473 |
+
"loss": 1.1463,
|
| 2474 |
+
"step": 35200
|
| 2475 |
+
},
|
| 2476 |
+
{
|
| 2477 |
+
"epoch": 35.801217038539555,
|
| 2478 |
+
"grad_norm": 4.128324508666992,
|
| 2479 |
+
"learning_rate": 1.4199797160243408e-05,
|
| 2480 |
+
"loss": 1.1307,
|
| 2481 |
+
"step": 35300
|
| 2482 |
+
},
|
| 2483 |
+
{
|
| 2484 |
+
"epoch": 35.9026369168357,
|
| 2485 |
+
"grad_norm": 4.593958377838135,
|
| 2486 |
+
"learning_rate": 1.4098377281947262e-05,
|
| 2487 |
+
"loss": 1.1583,
|
| 2488 |
+
"step": 35400
|
| 2489 |
+
},
|
| 2490 |
+
{
|
| 2491 |
+
"epoch": 36.004056795131845,
|
| 2492 |
+
"grad_norm": 3.714672803878784,
|
| 2493 |
+
"learning_rate": 1.3996957403651117e-05,
|
| 2494 |
+
"loss": 1.1383,
|
| 2495 |
+
"step": 35500
|
| 2496 |
+
},
|
| 2497 |
+
{
|
| 2498 |
+
"epoch": 36.105476673427994,
|
| 2499 |
+
"grad_norm": 4.306432247161865,
|
| 2500 |
+
"learning_rate": 1.3895537525354971e-05,
|
| 2501 |
+
"loss": 1.1218,
|
| 2502 |
+
"step": 35600
|
| 2503 |
+
},
|
| 2504 |
+
{
|
| 2505 |
+
"epoch": 36.206896551724135,
|
| 2506 |
+
"grad_norm": 3.848845958709717,
|
| 2507 |
+
"learning_rate": 1.3794117647058825e-05,
|
| 2508 |
+
"loss": 1.1095,
|
| 2509 |
+
"step": 35700
|
| 2510 |
+
},
|
| 2511 |
+
{
|
| 2512 |
+
"epoch": 36.308316430020284,
|
| 2513 |
+
"grad_norm": 4.119749069213867,
|
| 2514 |
+
"learning_rate": 1.3692697768762677e-05,
|
| 2515 |
+
"loss": 1.1235,
|
| 2516 |
+
"step": 35800
|
| 2517 |
+
},
|
| 2518 |
+
{
|
| 2519 |
+
"epoch": 36.40973630831643,
|
| 2520 |
+
"grad_norm": 4.287354946136475,
|
| 2521 |
+
"learning_rate": 1.3591277890466534e-05,
|
| 2522 |
+
"loss": 1.1498,
|
| 2523 |
+
"step": 35900
|
| 2524 |
+
},
|
| 2525 |
+
{
|
| 2526 |
+
"epoch": 36.511156186612574,
|
| 2527 |
+
"grad_norm": 4.493227958679199,
|
| 2528 |
+
"learning_rate": 1.3489858012170387e-05,
|
| 2529 |
+
"loss": 1.1127,
|
| 2530 |
+
"step": 36000
|
| 2531 |
+
},
|
| 2532 |
+
{
|
| 2533 |
+
"epoch": 36.61257606490872,
|
| 2534 |
+
"grad_norm": 4.363001346588135,
|
| 2535 |
+
"learning_rate": 1.338843813387424e-05,
|
| 2536 |
+
"loss": 1.13,
|
| 2537 |
+
"step": 36100
|
| 2538 |
+
},
|
| 2539 |
+
{
|
| 2540 |
+
"epoch": 36.71399594320487,
|
| 2541 |
+
"grad_norm": 3.56654691696167,
|
| 2542 |
+
"learning_rate": 1.3287018255578093e-05,
|
| 2543 |
+
"loss": 1.1461,
|
| 2544 |
+
"step": 36200
|
| 2545 |
+
},
|
| 2546 |
+
{
|
| 2547 |
+
"epoch": 36.81541582150101,
|
| 2548 |
+
"grad_norm": 3.8014466762542725,
|
| 2549 |
+
"learning_rate": 1.3185598377281948e-05,
|
| 2550 |
+
"loss": 1.1334,
|
| 2551 |
+
"step": 36300
|
| 2552 |
+
},
|
| 2553 |
+
{
|
| 2554 |
+
"epoch": 36.91683569979716,
|
| 2555 |
+
"grad_norm": 5.109997749328613,
|
| 2556 |
+
"learning_rate": 1.3084178498985802e-05,
|
| 2557 |
+
"loss": 1.1266,
|
| 2558 |
+
"step": 36400
|
| 2559 |
+
},
|
| 2560 |
+
{
|
| 2561 |
+
"epoch": 37.0182555780933,
|
| 2562 |
+
"grad_norm": 4.2766594886779785,
|
| 2563 |
+
"learning_rate": 1.2982758620689656e-05,
|
| 2564 |
+
"loss": 1.116,
|
| 2565 |
+
"step": 36500
|
| 2566 |
+
},
|
| 2567 |
+
{
|
| 2568 |
+
"epoch": 37.11967545638945,
|
| 2569 |
+
"grad_norm": 4.092167377471924,
|
| 2570 |
+
"learning_rate": 1.288133874239351e-05,
|
| 2571 |
+
"loss": 1.1126,
|
| 2572 |
+
"step": 36600
|
| 2573 |
+
},
|
| 2574 |
+
{
|
| 2575 |
+
"epoch": 37.2210953346856,
|
| 2576 |
+
"grad_norm": 4.259754180908203,
|
| 2577 |
+
"learning_rate": 1.2779918864097365e-05,
|
| 2578 |
+
"loss": 1.1125,
|
| 2579 |
+
"step": 36700
|
| 2580 |
+
},
|
| 2581 |
+
{
|
| 2582 |
+
"epoch": 37.32251521298174,
|
| 2583 |
+
"grad_norm": 4.275333404541016,
|
| 2584 |
+
"learning_rate": 1.2678498985801218e-05,
|
| 2585 |
+
"loss": 1.1198,
|
| 2586 |
+
"step": 36800
|
| 2587 |
+
},
|
| 2588 |
+
{
|
| 2589 |
+
"epoch": 37.42393509127789,
|
| 2590 |
+
"grad_norm": 4.391400337219238,
|
| 2591 |
+
"learning_rate": 1.2577079107505072e-05,
|
| 2592 |
+
"loss": 1.1236,
|
| 2593 |
+
"step": 36900
|
| 2594 |
+
},
|
| 2595 |
+
{
|
| 2596 |
+
"epoch": 37.52535496957404,
|
| 2597 |
+
"grad_norm": 3.9459197521209717,
|
| 2598 |
+
"learning_rate": 1.2475659229208926e-05,
|
| 2599 |
+
"loss": 1.125,
|
| 2600 |
+
"step": 37000
|
| 2601 |
+
},
|
| 2602 |
+
{
|
| 2603 |
+
"epoch": 37.62677484787018,
|
| 2604 |
+
"grad_norm": 3.889401435852051,
|
| 2605 |
+
"learning_rate": 1.237423935091278e-05,
|
| 2606 |
+
"loss": 1.126,
|
| 2607 |
+
"step": 37100
|
| 2608 |
+
},
|
| 2609 |
+
{
|
| 2610 |
+
"epoch": 37.72819472616633,
|
| 2611 |
+
"grad_norm": 3.894930839538574,
|
| 2612 |
+
"learning_rate": 1.2272819472616633e-05,
|
| 2613 |
+
"loss": 1.1064,
|
| 2614 |
+
"step": 37200
|
| 2615 |
+
},
|
| 2616 |
+
{
|
| 2617 |
+
"epoch": 37.82961460446248,
|
| 2618 |
+
"grad_norm": 3.794931650161743,
|
| 2619 |
+
"learning_rate": 1.2171399594320487e-05,
|
| 2620 |
+
"loss": 1.1164,
|
| 2621 |
+
"step": 37300
|
| 2622 |
+
},
|
| 2623 |
+
{
|
| 2624 |
+
"epoch": 37.93103448275862,
|
| 2625 |
+
"grad_norm": 4.259538650512695,
|
| 2626 |
+
"learning_rate": 1.2069979716024342e-05,
|
| 2627 |
+
"loss": 1.1299,
|
| 2628 |
+
"step": 37400
|
| 2629 |
+
},
|
| 2630 |
+
{
|
| 2631 |
+
"epoch": 38.03245436105477,
|
| 2632 |
+
"grad_norm": 3.636101722717285,
|
| 2633 |
+
"learning_rate": 1.1968559837728196e-05,
|
| 2634 |
+
"loss": 1.0973,
|
| 2635 |
+
"step": 37500
|
| 2636 |
+
},
|
| 2637 |
+
{
|
| 2638 |
+
"epoch": 38.13387423935091,
|
| 2639 |
+
"grad_norm": 3.7862744331359863,
|
| 2640 |
+
"learning_rate": 1.186713995943205e-05,
|
| 2641 |
+
"loss": 1.108,
|
| 2642 |
+
"step": 37600
|
| 2643 |
+
},
|
| 2644 |
+
{
|
| 2645 |
+
"epoch": 38.23529411764706,
|
| 2646 |
+
"grad_norm": 3.8924472332000732,
|
| 2647 |
+
"learning_rate": 1.1765720081135903e-05,
|
| 2648 |
+
"loss": 1.094,
|
| 2649 |
+
"step": 37700
|
| 2650 |
+
},
|
| 2651 |
+
{
|
| 2652 |
+
"epoch": 38.33671399594321,
|
| 2653 |
+
"grad_norm": 4.239724636077881,
|
| 2654 |
+
"learning_rate": 1.1664300202839757e-05,
|
| 2655 |
+
"loss": 1.1103,
|
| 2656 |
+
"step": 37800
|
| 2657 |
+
},
|
| 2658 |
+
{
|
| 2659 |
+
"epoch": 38.43813387423935,
|
| 2660 |
+
"grad_norm": 4.01432466506958,
|
| 2661 |
+
"learning_rate": 1.156288032454361e-05,
|
| 2662 |
+
"loss": 1.1315,
|
| 2663 |
+
"step": 37900
|
| 2664 |
+
},
|
| 2665 |
+
{
|
| 2666 |
+
"epoch": 38.5395537525355,
|
| 2667 |
+
"grad_norm": 3.942206859588623,
|
| 2668 |
+
"learning_rate": 1.1461460446247466e-05,
|
| 2669 |
+
"loss": 1.117,
|
| 2670 |
+
"step": 38000
|
| 2671 |
+
},
|
| 2672 |
+
{
|
| 2673 |
+
"epoch": 38.640973630831645,
|
| 2674 |
+
"grad_norm": 4.0454607009887695,
|
| 2675 |
+
"learning_rate": 1.136004056795132e-05,
|
| 2676 |
+
"loss": 1.1158,
|
| 2677 |
+
"step": 38100
|
| 2678 |
+
},
|
| 2679 |
+
{
|
| 2680 |
+
"epoch": 38.74239350912779,
|
| 2681 |
+
"grad_norm": 4.053074359893799,
|
| 2682 |
+
"learning_rate": 1.1258620689655173e-05,
|
| 2683 |
+
"loss": 1.0937,
|
| 2684 |
+
"step": 38200
|
| 2685 |
+
},
|
| 2686 |
+
{
|
| 2687 |
+
"epoch": 38.843813387423936,
|
| 2688 |
+
"grad_norm": 4.3812971115112305,
|
| 2689 |
+
"learning_rate": 1.1157200811359027e-05,
|
| 2690 |
+
"loss": 1.0867,
|
| 2691 |
+
"step": 38300
|
| 2692 |
+
},
|
| 2693 |
+
{
|
| 2694 |
+
"epoch": 38.945233265720084,
|
| 2695 |
+
"grad_norm": 3.714223861694336,
|
| 2696 |
+
"learning_rate": 1.105578093306288e-05,
|
| 2697 |
+
"loss": 1.1241,
|
| 2698 |
+
"step": 38400
|
| 2699 |
+
},
|
| 2700 |
+
{
|
| 2701 |
+
"epoch": 39.046653144016226,
|
| 2702 |
+
"grad_norm": 4.840576171875,
|
| 2703 |
+
"learning_rate": 1.0954361054766734e-05,
|
| 2704 |
+
"loss": 1.1074,
|
| 2705 |
+
"step": 38500
|
| 2706 |
+
},
|
| 2707 |
+
{
|
| 2708 |
+
"epoch": 39.148073022312374,
|
| 2709 |
+
"grad_norm": 3.7746899127960205,
|
| 2710 |
+
"learning_rate": 1.085294117647059e-05,
|
| 2711 |
+
"loss": 1.0931,
|
| 2712 |
+
"step": 38600
|
| 2713 |
+
},
|
| 2714 |
+
{
|
| 2715 |
+
"epoch": 39.249492900608516,
|
| 2716 |
+
"grad_norm": 4.381807327270508,
|
| 2717 |
+
"learning_rate": 1.0751521298174443e-05,
|
| 2718 |
+
"loss": 1.0957,
|
| 2719 |
+
"step": 38700
|
| 2720 |
+
},
|
| 2721 |
+
{
|
| 2722 |
+
"epoch": 39.350912778904664,
|
| 2723 |
+
"grad_norm": 4.536212921142578,
|
| 2724 |
+
"learning_rate": 1.0650101419878297e-05,
|
| 2725 |
+
"loss": 1.0937,
|
| 2726 |
+
"step": 38800
|
| 2727 |
+
},
|
| 2728 |
+
{
|
| 2729 |
+
"epoch": 39.45233265720081,
|
| 2730 |
+
"grad_norm": 4.437652587890625,
|
| 2731 |
+
"learning_rate": 1.054868154158215e-05,
|
| 2732 |
+
"loss": 1.089,
|
| 2733 |
+
"step": 38900
|
| 2734 |
+
},
|
| 2735 |
+
{
|
| 2736 |
+
"epoch": 39.553752535496955,
|
| 2737 |
+
"grad_norm": 4.1285858154296875,
|
| 2738 |
+
"learning_rate": 1.0447261663286004e-05,
|
| 2739 |
+
"loss": 1.0851,
|
| 2740 |
+
"step": 39000
|
| 2741 |
+
},
|
| 2742 |
+
{
|
| 2743 |
+
"epoch": 39.6551724137931,
|
| 2744 |
+
"grad_norm": 4.137048244476318,
|
| 2745 |
+
"learning_rate": 1.0345841784989858e-05,
|
| 2746 |
+
"loss": 1.1018,
|
| 2747 |
+
"step": 39100
|
| 2748 |
+
},
|
| 2749 |
+
{
|
| 2750 |
+
"epoch": 39.75659229208925,
|
| 2751 |
+
"grad_norm": 4.798439979553223,
|
| 2752 |
+
"learning_rate": 1.0244421906693713e-05,
|
| 2753 |
+
"loss": 1.1038,
|
| 2754 |
+
"step": 39200
|
| 2755 |
+
},
|
| 2756 |
+
{
|
| 2757 |
+
"epoch": 39.85801217038539,
|
| 2758 |
+
"grad_norm": 3.76686429977417,
|
| 2759 |
+
"learning_rate": 1.0143002028397567e-05,
|
| 2760 |
+
"loss": 1.1211,
|
| 2761 |
+
"step": 39300
|
| 2762 |
+
},
|
| 2763 |
+
{
|
| 2764 |
+
"epoch": 39.95943204868154,
|
| 2765 |
+
"grad_norm": 4.3406453132629395,
|
| 2766 |
+
"learning_rate": 1.004158215010142e-05,
|
| 2767 |
+
"loss": 1.1148,
|
| 2768 |
+
"step": 39400
|
| 2769 |
+
},
|
| 2770 |
+
{
|
| 2771 |
+
"epoch": 40.06085192697769,
|
| 2772 |
+
"grad_norm": 4.008460998535156,
|
| 2773 |
+
"learning_rate": 9.940162271805274e-06,
|
| 2774 |
+
"loss": 1.1072,
|
| 2775 |
+
"step": 39500
|
| 2776 |
+
},
|
| 2777 |
+
{
|
| 2778 |
+
"epoch": 40.16227180527383,
|
| 2779 |
+
"grad_norm": 3.8538854122161865,
|
| 2780 |
+
"learning_rate": 9.838742393509128e-06,
|
| 2781 |
+
"loss": 1.1041,
|
| 2782 |
+
"step": 39600
|
| 2783 |
+
},
|
| 2784 |
+
{
|
| 2785 |
+
"epoch": 40.26369168356998,
|
| 2786 |
+
"grad_norm": 4.347944259643555,
|
| 2787 |
+
"learning_rate": 9.737322515212982e-06,
|
| 2788 |
+
"loss": 1.093,
|
| 2789 |
+
"step": 39700
|
| 2790 |
+
},
|
| 2791 |
+
{
|
| 2792 |
+
"epoch": 40.36511156186612,
|
| 2793 |
+
"grad_norm": 4.429162979125977,
|
| 2794 |
+
"learning_rate": 9.635902636916837e-06,
|
| 2795 |
+
"loss": 1.0924,
|
| 2796 |
+
"step": 39800
|
| 2797 |
+
},
|
| 2798 |
+
{
|
| 2799 |
+
"epoch": 40.46653144016227,
|
| 2800 |
+
"grad_norm": 4.08125638961792,
|
| 2801 |
+
"learning_rate": 9.53448275862069e-06,
|
| 2802 |
+
"loss": 1.1159,
|
| 2803 |
+
"step": 39900
|
| 2804 |
+
},
|
| 2805 |
+
{
|
| 2806 |
+
"epoch": 40.56795131845842,
|
| 2807 |
+
"grad_norm": 4.317127704620361,
|
| 2808 |
+
"learning_rate": 9.433062880324544e-06,
|
| 2809 |
+
"loss": 1.1102,
|
| 2810 |
+
"step": 40000
|
| 2811 |
+
},
|
| 2812 |
+
{
|
| 2813 |
+
"epoch": 40.66937119675456,
|
| 2814 |
+
"grad_norm": 4.39460563659668,
|
| 2815 |
+
"learning_rate": 9.331643002028398e-06,
|
| 2816 |
+
"loss": 1.0964,
|
| 2817 |
+
"step": 40100
|
| 2818 |
+
},
|
| 2819 |
+
{
|
| 2820 |
+
"epoch": 40.77079107505071,
|
| 2821 |
+
"grad_norm": 3.9972689151763916,
|
| 2822 |
+
"learning_rate": 9.230223123732252e-06,
|
| 2823 |
+
"loss": 1.0857,
|
| 2824 |
+
"step": 40200
|
| 2825 |
+
},
|
| 2826 |
+
{
|
| 2827 |
+
"epoch": 40.87221095334686,
|
| 2828 |
+
"grad_norm": 3.5488409996032715,
|
| 2829 |
+
"learning_rate": 9.128803245436105e-06,
|
| 2830 |
+
"loss": 1.0919,
|
| 2831 |
+
"step": 40300
|
| 2832 |
+
},
|
| 2833 |
+
{
|
| 2834 |
+
"epoch": 40.973630831643,
|
| 2835 |
+
"grad_norm": 4.233036994934082,
|
| 2836 |
+
"learning_rate": 9.02738336713996e-06,
|
| 2837 |
+
"loss": 1.0592,
|
| 2838 |
+
"step": 40400
|
| 2839 |
+
},
|
| 2840 |
+
{
|
| 2841 |
+
"epoch": 41.07505070993915,
|
| 2842 |
+
"grad_norm": 4.364148139953613,
|
| 2843 |
+
"learning_rate": 8.925963488843815e-06,
|
| 2844 |
+
"loss": 1.1155,
|
| 2845 |
+
"step": 40500
|
| 2846 |
+
},
|
| 2847 |
+
{
|
| 2848 |
+
"epoch": 41.1764705882353,
|
| 2849 |
+
"grad_norm": 4.002589225769043,
|
| 2850 |
+
"learning_rate": 8.824543610547668e-06,
|
| 2851 |
+
"loss": 1.09,
|
| 2852 |
+
"step": 40600
|
| 2853 |
+
},
|
| 2854 |
+
{
|
| 2855 |
+
"epoch": 41.27789046653144,
|
| 2856 |
+
"grad_norm": 4.327565670013428,
|
| 2857 |
+
"learning_rate": 8.723123732251522e-06,
|
| 2858 |
+
"loss": 1.0841,
|
| 2859 |
+
"step": 40700
|
| 2860 |
+
},
|
| 2861 |
+
{
|
| 2862 |
+
"epoch": 41.37931034482759,
|
| 2863 |
+
"grad_norm": 4.477043628692627,
|
| 2864 |
+
"learning_rate": 8.621703853955376e-06,
|
| 2865 |
+
"loss": 1.0837,
|
| 2866 |
+
"step": 40800
|
| 2867 |
+
},
|
| 2868 |
+
{
|
| 2869 |
+
"epoch": 41.48073022312373,
|
| 2870 |
+
"grad_norm": 4.210312366485596,
|
| 2871 |
+
"learning_rate": 8.52028397565923e-06,
|
| 2872 |
+
"loss": 1.0897,
|
| 2873 |
+
"step": 40900
|
| 2874 |
+
},
|
| 2875 |
+
{
|
| 2876 |
+
"epoch": 41.58215010141988,
|
| 2877 |
+
"grad_norm": 4.245460033416748,
|
| 2878 |
+
"learning_rate": 8.418864097363085e-06,
|
| 2879 |
+
"loss": 1.0995,
|
| 2880 |
+
"step": 41000
|
| 2881 |
+
},
|
| 2882 |
+
{
|
| 2883 |
+
"epoch": 41.683569979716026,
|
| 2884 |
+
"grad_norm": 4.513962745666504,
|
| 2885 |
+
"learning_rate": 8.317444219066938e-06,
|
| 2886 |
+
"loss": 1.0833,
|
| 2887 |
+
"step": 41100
|
| 2888 |
+
},
|
| 2889 |
+
{
|
| 2890 |
+
"epoch": 41.78498985801217,
|
| 2891 |
+
"grad_norm": 4.622128963470459,
|
| 2892 |
+
"learning_rate": 8.216024340770792e-06,
|
| 2893 |
+
"loss": 1.0836,
|
| 2894 |
+
"step": 41200
|
| 2895 |
+
},
|
| 2896 |
+
{
|
| 2897 |
+
"epoch": 41.886409736308316,
|
| 2898 |
+
"grad_norm": 3.732800245285034,
|
| 2899 |
+
"learning_rate": 8.114604462474646e-06,
|
| 2900 |
+
"loss": 1.1055,
|
| 2901 |
+
"step": 41300
|
| 2902 |
+
},
|
| 2903 |
+
{
|
| 2904 |
+
"epoch": 41.987829614604465,
|
| 2905 |
+
"grad_norm": 3.7526845932006836,
|
| 2906 |
+
"learning_rate": 8.0131845841785e-06,
|
| 2907 |
+
"loss": 1.0852,
|
| 2908 |
+
"step": 41400
|
| 2909 |
+
},
|
| 2910 |
+
{
|
| 2911 |
+
"epoch": 42.089249492900606,
|
| 2912 |
+
"grad_norm": 4.781412601470947,
|
| 2913 |
+
"learning_rate": 7.911764705882353e-06,
|
| 2914 |
+
"loss": 1.0722,
|
| 2915 |
+
"step": 41500
|
| 2916 |
+
},
|
| 2917 |
+
{
|
| 2918 |
+
"epoch": 42.190669371196755,
|
| 2919 |
+
"grad_norm": 4.005661487579346,
|
| 2920 |
+
"learning_rate": 7.810344827586208e-06,
|
| 2921 |
+
"loss": 1.0778,
|
| 2922 |
+
"step": 41600
|
| 2923 |
+
},
|
| 2924 |
+
{
|
| 2925 |
+
"epoch": 42.2920892494929,
|
| 2926 |
+
"grad_norm": 4.243937015533447,
|
| 2927 |
+
"learning_rate": 7.708924949290062e-06,
|
| 2928 |
+
"loss": 1.0819,
|
| 2929 |
+
"step": 41700
|
| 2930 |
+
},
|
| 2931 |
+
{
|
| 2932 |
+
"epoch": 42.393509127789045,
|
| 2933 |
+
"grad_norm": 3.724163293838501,
|
| 2934 |
+
"learning_rate": 7.607505070993916e-06,
|
| 2935 |
+
"loss": 1.0831,
|
| 2936 |
+
"step": 41800
|
| 2937 |
+
},
|
| 2938 |
+
{
|
| 2939 |
+
"epoch": 42.494929006085194,
|
| 2940 |
+
"grad_norm": 3.9272992610931396,
|
| 2941 |
+
"learning_rate": 7.506085192697769e-06,
|
| 2942 |
+
"loss": 1.082,
|
| 2943 |
+
"step": 41900
|
| 2944 |
+
},
|
| 2945 |
+
{
|
| 2946 |
+
"epoch": 42.59634888438134,
|
| 2947 |
+
"grad_norm": 4.436212539672852,
|
| 2948 |
+
"learning_rate": 7.404665314401624e-06,
|
| 2949 |
+
"loss": 1.0976,
|
| 2950 |
+
"step": 42000
|
| 2951 |
+
},
|
| 2952 |
+
{
|
| 2953 |
+
"epoch": 42.697768762677484,
|
| 2954 |
+
"grad_norm": 5.445650577545166,
|
| 2955 |
+
"learning_rate": 7.303245436105477e-06,
|
| 2956 |
+
"loss": 1.0983,
|
| 2957 |
+
"step": 42100
|
| 2958 |
+
},
|
| 2959 |
+
{
|
| 2960 |
+
"epoch": 42.79918864097363,
|
| 2961 |
+
"grad_norm": 4.699320316314697,
|
| 2962 |
+
"learning_rate": 7.201825557809332e-06,
|
| 2963 |
+
"loss": 1.1122,
|
| 2964 |
+
"step": 42200
|
| 2965 |
+
},
|
| 2966 |
+
{
|
| 2967 |
+
"epoch": 42.900608519269774,
|
| 2968 |
+
"grad_norm": 5.361352443695068,
|
| 2969 |
+
"learning_rate": 7.100405679513185e-06,
|
| 2970 |
+
"loss": 1.0897,
|
| 2971 |
+
"step": 42300
|
| 2972 |
+
},
|
| 2973 |
+
{
|
| 2974 |
+
"epoch": 43.00202839756592,
|
| 2975 |
+
"grad_norm": 4.925732612609863,
|
| 2976 |
+
"learning_rate": 6.998985801217039e-06,
|
| 2977 |
+
"loss": 1.0879,
|
| 2978 |
+
"step": 42400
|
| 2979 |
+
},
|
| 2980 |
+
{
|
| 2981 |
+
"epoch": 43.10344827586207,
|
| 2982 |
+
"grad_norm": 4.510122776031494,
|
| 2983 |
+
"learning_rate": 6.897565922920893e-06,
|
| 2984 |
+
"loss": 1.077,
|
| 2985 |
+
"step": 42500
|
| 2986 |
+
},
|
| 2987 |
+
{
|
| 2988 |
+
"epoch": 43.20486815415821,
|
| 2989 |
+
"grad_norm": 4.398224830627441,
|
| 2990 |
+
"learning_rate": 6.7961460446247476e-06,
|
| 2991 |
+
"loss": 1.0481,
|
| 2992 |
+
"step": 42600
|
| 2993 |
+
},
|
| 2994 |
+
{
|
| 2995 |
+
"epoch": 43.30628803245436,
|
| 2996 |
+
"grad_norm": 3.701659917831421,
|
| 2997 |
+
"learning_rate": 6.6947261663286e-06,
|
| 2998 |
+
"loss": 1.0913,
|
| 2999 |
+
"step": 42700
|
| 3000 |
+
},
|
| 3001 |
+
{
|
| 3002 |
+
"epoch": 43.40770791075051,
|
| 3003 |
+
"grad_norm": 4.113113880157471,
|
| 3004 |
+
"learning_rate": 6.593306288032455e-06,
|
| 3005 |
+
"loss": 1.0645,
|
| 3006 |
+
"step": 42800
|
| 3007 |
+
},
|
| 3008 |
+
{
|
| 3009 |
+
"epoch": 43.50912778904665,
|
| 3010 |
+
"grad_norm": 4.151520729064941,
|
| 3011 |
+
"learning_rate": 6.4918864097363086e-06,
|
| 3012 |
+
"loss": 1.085,
|
| 3013 |
+
"step": 42900
|
| 3014 |
+
},
|
| 3015 |
+
{
|
| 3016 |
+
"epoch": 43.6105476673428,
|
| 3017 |
+
"grad_norm": 4.146597385406494,
|
| 3018 |
+
"learning_rate": 6.390466531440163e-06,
|
| 3019 |
+
"loss": 1.0769,
|
| 3020 |
+
"step": 43000
|
| 3021 |
+
},
|
| 3022 |
+
{
|
| 3023 |
+
"epoch": 43.71196754563895,
|
| 3024 |
+
"grad_norm": 4.135023593902588,
|
| 3025 |
+
"learning_rate": 6.289046653144017e-06,
|
| 3026 |
+
"loss": 1.1016,
|
| 3027 |
+
"step": 43100
|
| 3028 |
+
},
|
| 3029 |
+
{
|
| 3030 |
+
"epoch": 43.81338742393509,
|
| 3031 |
+
"grad_norm": 4.028172492980957,
|
| 3032 |
+
"learning_rate": 6.1876267748478704e-06,
|
| 3033 |
+
"loss": 1.0609,
|
| 3034 |
+
"step": 43200
|
| 3035 |
+
},
|
| 3036 |
+
{
|
| 3037 |
+
"epoch": 43.91480730223124,
|
| 3038 |
+
"grad_norm": 4.14988899230957,
|
| 3039 |
+
"learning_rate": 6.086206896551724e-06,
|
| 3040 |
+
"loss": 1.0915,
|
| 3041 |
+
"step": 43300
|
| 3042 |
+
},
|
| 3043 |
+
{
|
| 3044 |
+
"epoch": 44.01622718052738,
|
| 3045 |
+
"grad_norm": 4.097161769866943,
|
| 3046 |
+
"learning_rate": 5.984787018255579e-06,
|
| 3047 |
+
"loss": 1.0787,
|
| 3048 |
+
"step": 43400
|
| 3049 |
+
},
|
| 3050 |
+
{
|
| 3051 |
+
"epoch": 44.11764705882353,
|
| 3052 |
+
"grad_norm": 4.110335350036621,
|
| 3053 |
+
"learning_rate": 5.883367139959432e-06,
|
| 3054 |
+
"loss": 1.0762,
|
| 3055 |
+
"step": 43500
|
| 3056 |
+
},
|
| 3057 |
+
{
|
| 3058 |
+
"epoch": 44.21906693711968,
|
| 3059 |
+
"grad_norm": 4.515265464782715,
|
| 3060 |
+
"learning_rate": 5.781947261663286e-06,
|
| 3061 |
+
"loss": 1.0901,
|
| 3062 |
+
"step": 43600
|
| 3063 |
+
},
|
| 3064 |
+
{
|
| 3065 |
+
"epoch": 44.32048681541582,
|
| 3066 |
+
"grad_norm": 4.0620951652526855,
|
| 3067 |
+
"learning_rate": 5.6805273833671405e-06,
|
| 3068 |
+
"loss": 1.0834,
|
| 3069 |
+
"step": 43700
|
| 3070 |
+
},
|
| 3071 |
+
{
|
| 3072 |
+
"epoch": 44.42190669371197,
|
| 3073 |
+
"grad_norm": 3.9556586742401123,
|
| 3074 |
+
"learning_rate": 5.579107505070994e-06,
|
| 3075 |
+
"loss": 1.0814,
|
| 3076 |
+
"step": 43800
|
| 3077 |
+
},
|
| 3078 |
+
{
|
| 3079 |
+
"epoch": 44.523326572008116,
|
| 3080 |
+
"grad_norm": 4.868508338928223,
|
| 3081 |
+
"learning_rate": 5.477687626774848e-06,
|
| 3082 |
+
"loss": 1.0802,
|
| 3083 |
+
"step": 43900
|
| 3084 |
+
},
|
| 3085 |
+
{
|
| 3086 |
+
"epoch": 44.62474645030426,
|
| 3087 |
+
"grad_norm": 4.4028143882751465,
|
| 3088 |
+
"learning_rate": 5.376267748478702e-06,
|
| 3089 |
+
"loss": 1.0713,
|
| 3090 |
+
"step": 44000
|
| 3091 |
+
},
|
| 3092 |
+
{
|
| 3093 |
+
"epoch": 44.72616632860041,
|
| 3094 |
+
"grad_norm": 3.419487476348877,
|
| 3095 |
+
"learning_rate": 5.274847870182556e-06,
|
| 3096 |
+
"loss": 1.0874,
|
| 3097 |
+
"step": 44100
|
| 3098 |
+
},
|
| 3099 |
+
{
|
| 3100 |
+
"epoch": 44.827586206896555,
|
| 3101 |
+
"grad_norm": 4.371777057647705,
|
| 3102 |
+
"learning_rate": 5.17342799188641e-06,
|
| 3103 |
+
"loss": 1.097,
|
| 3104 |
+
"step": 44200
|
| 3105 |
+
},
|
| 3106 |
+
{
|
| 3107 |
+
"epoch": 44.9290060851927,
|
| 3108 |
+
"grad_norm": 4.513946533203125,
|
| 3109 |
+
"learning_rate": 5.072008113590264e-06,
|
| 3110 |
+
"loss": 1.0732,
|
| 3111 |
+
"step": 44300
|
| 3112 |
+
},
|
| 3113 |
+
{
|
| 3114 |
+
"epoch": 45.030425963488845,
|
| 3115 |
+
"grad_norm": 5.1020283699035645,
|
| 3116 |
+
"learning_rate": 4.970588235294118e-06,
|
| 3117 |
+
"loss": 1.0658,
|
| 3118 |
+
"step": 44400
|
| 3119 |
+
},
|
| 3120 |
+
{
|
| 3121 |
+
"epoch": 45.13184584178499,
|
| 3122 |
+
"grad_norm": 4.232724189758301,
|
| 3123 |
+
"learning_rate": 4.8691683569979716e-06,
|
| 3124 |
+
"loss": 1.0713,
|
| 3125 |
+
"step": 44500
|
| 3126 |
+
},
|
| 3127 |
+
{
|
| 3128 |
+
"epoch": 45.233265720081135,
|
| 3129 |
+
"grad_norm": 4.947866439819336,
|
| 3130 |
+
"learning_rate": 4.767748478701826e-06,
|
| 3131 |
+
"loss": 1.0807,
|
| 3132 |
+
"step": 44600
|
| 3133 |
+
},
|
| 3134 |
+
{
|
| 3135 |
+
"epoch": 45.334685598377284,
|
| 3136 |
+
"grad_norm": 4.363738536834717,
|
| 3137 |
+
"learning_rate": 4.66632860040568e-06,
|
| 3138 |
+
"loss": 1.0748,
|
| 3139 |
+
"step": 44700
|
| 3140 |
+
},
|
| 3141 |
+
{
|
| 3142 |
+
"epoch": 45.436105476673426,
|
| 3143 |
+
"grad_norm": 4.23621129989624,
|
| 3144 |
+
"learning_rate": 4.5649087221095334e-06,
|
| 3145 |
+
"loss": 1.0598,
|
| 3146 |
+
"step": 44800
|
| 3147 |
+
},
|
| 3148 |
+
{
|
| 3149 |
+
"epoch": 45.537525354969574,
|
| 3150 |
+
"grad_norm": 3.6266748905181885,
|
| 3151 |
+
"learning_rate": 4.463488843813388e-06,
|
| 3152 |
+
"loss": 1.0529,
|
| 3153 |
+
"step": 44900
|
| 3154 |
+
},
|
| 3155 |
+
{
|
| 3156 |
+
"epoch": 45.63894523326572,
|
| 3157 |
+
"grad_norm": 4.692754745483398,
|
| 3158 |
+
"learning_rate": 4.362068965517242e-06,
|
| 3159 |
+
"loss": 1.0769,
|
| 3160 |
+
"step": 45000
|
| 3161 |
+
},
|
| 3162 |
+
{
|
| 3163 |
+
"epoch": 45.740365111561864,
|
| 3164 |
+
"grad_norm": 4.276831150054932,
|
| 3165 |
+
"learning_rate": 4.260649087221095e-06,
|
| 3166 |
+
"loss": 1.0499,
|
| 3167 |
+
"step": 45100
|
| 3168 |
+
},
|
| 3169 |
+
{
|
| 3170 |
+
"epoch": 45.84178498985801,
|
| 3171 |
+
"grad_norm": 3.9113900661468506,
|
| 3172 |
+
"learning_rate": 4.15922920892495e-06,
|
| 3173 |
+
"loss": 1.0744,
|
| 3174 |
+
"step": 45200
|
| 3175 |
+
},
|
| 3176 |
+
{
|
| 3177 |
+
"epoch": 45.94320486815416,
|
| 3178 |
+
"grad_norm": 4.487521171569824,
|
| 3179 |
+
"learning_rate": 4.0578093306288035e-06,
|
| 3180 |
+
"loss": 1.064,
|
| 3181 |
+
"step": 45300
|
| 3182 |
+
},
|
| 3183 |
+
{
|
| 3184 |
+
"epoch": 46.0446247464503,
|
| 3185 |
+
"grad_norm": 4.446450233459473,
|
| 3186 |
+
"learning_rate": 3.956389452332657e-06,
|
| 3187 |
+
"loss": 1.0726,
|
| 3188 |
+
"step": 45400
|
| 3189 |
+
},
|
| 3190 |
+
{
|
| 3191 |
+
"epoch": 46.14604462474645,
|
| 3192 |
+
"grad_norm": 4.588889122009277,
|
| 3193 |
+
"learning_rate": 3.854969574036512e-06,
|
| 3194 |
+
"loss": 1.081,
|
| 3195 |
+
"step": 45500
|
| 3196 |
+
},
|
| 3197 |
+
{
|
| 3198 |
+
"epoch": 46.24746450304259,
|
| 3199 |
+
"grad_norm": 3.8698606491088867,
|
| 3200 |
+
"learning_rate": 3.7535496957403653e-06,
|
| 3201 |
+
"loss": 1.0754,
|
| 3202 |
+
"step": 45600
|
| 3203 |
+
},
|
| 3204 |
+
{
|
| 3205 |
+
"epoch": 46.34888438133874,
|
| 3206 |
+
"grad_norm": 4.251828193664551,
|
| 3207 |
+
"learning_rate": 3.652129817444219e-06,
|
| 3208 |
+
"loss": 1.0799,
|
| 3209 |
+
"step": 45700
|
| 3210 |
+
},
|
| 3211 |
+
{
|
| 3212 |
+
"epoch": 46.45030425963489,
|
| 3213 |
+
"grad_norm": 3.970852851867676,
|
| 3214 |
+
"learning_rate": 3.550709939148073e-06,
|
| 3215 |
+
"loss": 1.0801,
|
| 3216 |
+
"step": 45800
|
| 3217 |
+
},
|
| 3218 |
+
{
|
| 3219 |
+
"epoch": 46.55172413793103,
|
| 3220 |
+
"grad_norm": 4.236978054046631,
|
| 3221 |
+
"learning_rate": 3.449290060851927e-06,
|
| 3222 |
+
"loss": 1.0828,
|
| 3223 |
+
"step": 45900
|
| 3224 |
+
},
|
| 3225 |
+
{
|
| 3226 |
+
"epoch": 46.65314401622718,
|
| 3227 |
+
"grad_norm": 3.778306722640991,
|
| 3228 |
+
"learning_rate": 3.347870182555781e-06,
|
| 3229 |
+
"loss": 1.0699,
|
| 3230 |
+
"step": 46000
|
| 3231 |
+
},
|
| 3232 |
+
{
|
| 3233 |
+
"epoch": 46.75456389452333,
|
| 3234 |
+
"grad_norm": 4.238831520080566,
|
| 3235 |
+
"learning_rate": 3.246450304259635e-06,
|
| 3236 |
+
"loss": 1.0609,
|
| 3237 |
+
"step": 46100
|
| 3238 |
+
},
|
| 3239 |
+
{
|
| 3240 |
+
"epoch": 46.85598377281947,
|
| 3241 |
+
"grad_norm": 4.496991157531738,
|
| 3242 |
+
"learning_rate": 3.145030425963489e-06,
|
| 3243 |
+
"loss": 1.0888,
|
| 3244 |
+
"step": 46200
|
| 3245 |
+
},
|
| 3246 |
+
{
|
| 3247 |
+
"epoch": 46.95740365111562,
|
| 3248 |
+
"grad_norm": 4.271062850952148,
|
| 3249 |
+
"learning_rate": 3.0436105476673427e-06,
|
| 3250 |
+
"loss": 1.0596,
|
| 3251 |
+
"step": 46300
|
| 3252 |
+
},
|
| 3253 |
+
{
|
| 3254 |
+
"epoch": 47.05882352941177,
|
| 3255 |
+
"grad_norm": 3.694577932357788,
|
| 3256 |
+
"learning_rate": 2.942190669371197e-06,
|
| 3257 |
+
"loss": 1.0647,
|
| 3258 |
+
"step": 46400
|
| 3259 |
+
},
|
| 3260 |
+
{
|
| 3261 |
+
"epoch": 47.16024340770791,
|
| 3262 |
+
"grad_norm": 3.7699005603790283,
|
| 3263 |
+
"learning_rate": 2.840770791075051e-06,
|
| 3264 |
+
"loss": 1.0652,
|
| 3265 |
+
"step": 46500
|
| 3266 |
+
},
|
| 3267 |
+
{
|
| 3268 |
+
"epoch": 47.26166328600406,
|
| 3269 |
+
"grad_norm": 4.500093460083008,
|
| 3270 |
+
"learning_rate": 2.7393509127789046e-06,
|
| 3271 |
+
"loss": 1.0635,
|
| 3272 |
+
"step": 46600
|
| 3273 |
+
},
|
| 3274 |
+
{
|
| 3275 |
+
"epoch": 47.3630831643002,
|
| 3276 |
+
"grad_norm": 4.237122535705566,
|
| 3277 |
+
"learning_rate": 2.6379310344827587e-06,
|
| 3278 |
+
"loss": 1.0608,
|
| 3279 |
+
"step": 46700
|
| 3280 |
+
},
|
| 3281 |
+
{
|
| 3282 |
+
"epoch": 47.46450304259635,
|
| 3283 |
+
"grad_norm": 4.547657012939453,
|
| 3284 |
+
"learning_rate": 2.536511156186613e-06,
|
| 3285 |
+
"loss": 1.0588,
|
| 3286 |
+
"step": 46800
|
| 3287 |
+
},
|
| 3288 |
+
{
|
| 3289 |
+
"epoch": 47.5659229208925,
|
| 3290 |
+
"grad_norm": 4.323256969451904,
|
| 3291 |
+
"learning_rate": 2.4350912778904665e-06,
|
| 3292 |
+
"loss": 1.0615,
|
| 3293 |
+
"step": 46900
|
| 3294 |
+
},
|
| 3295 |
+
{
|
| 3296 |
+
"epoch": 47.66734279918864,
|
| 3297 |
+
"grad_norm": 4.30881404876709,
|
| 3298 |
+
"learning_rate": 2.3336713995943206e-06,
|
| 3299 |
+
"loss": 1.0713,
|
| 3300 |
+
"step": 47000
|
| 3301 |
+
},
|
| 3302 |
+
{
|
| 3303 |
+
"epoch": 47.76876267748479,
|
| 3304 |
+
"grad_norm": 4.941198348999023,
|
| 3305 |
+
"learning_rate": 2.2322515212981747e-06,
|
| 3306 |
+
"loss": 1.089,
|
| 3307 |
+
"step": 47100
|
| 3308 |
+
},
|
| 3309 |
+
{
|
| 3310 |
+
"epoch": 47.870182555780936,
|
| 3311 |
+
"grad_norm": 4.09073543548584,
|
| 3312 |
+
"learning_rate": 2.1308316430020283e-06,
|
| 3313 |
+
"loss": 1.0756,
|
| 3314 |
+
"step": 47200
|
| 3315 |
+
},
|
| 3316 |
+
{
|
| 3317 |
+
"epoch": 47.97160243407708,
|
| 3318 |
+
"grad_norm": 4.901280879974365,
|
| 3319 |
+
"learning_rate": 2.0294117647058824e-06,
|
| 3320 |
+
"loss": 1.0529,
|
| 3321 |
+
"step": 47300
|
| 3322 |
+
},
|
| 3323 |
+
{
|
| 3324 |
+
"epoch": 48.073022312373226,
|
| 3325 |
+
"grad_norm": 5.106020927429199,
|
| 3326 |
+
"learning_rate": 1.9279918864097365e-06,
|
| 3327 |
+
"loss": 1.0634,
|
| 3328 |
+
"step": 47400
|
| 3329 |
+
},
|
| 3330 |
+
{
|
| 3331 |
+
"epoch": 48.174442190669374,
|
| 3332 |
+
"grad_norm": 5.030118465423584,
|
| 3333 |
+
"learning_rate": 1.8265720081135904e-06,
|
| 3334 |
+
"loss": 1.0843,
|
| 3335 |
+
"step": 47500
|
| 3336 |
+
},
|
| 3337 |
+
{
|
| 3338 |
+
"epoch": 48.275862068965516,
|
| 3339 |
+
"grad_norm": 4.184380054473877,
|
| 3340 |
+
"learning_rate": 1.7251521298174443e-06,
|
| 3341 |
+
"loss": 1.0644,
|
| 3342 |
+
"step": 47600
|
| 3343 |
+
},
|
| 3344 |
+
{
|
| 3345 |
+
"epoch": 48.377281947261665,
|
| 3346 |
+
"grad_norm": 4.010962009429932,
|
| 3347 |
+
"learning_rate": 1.6237322515212982e-06,
|
| 3348 |
+
"loss": 1.0706,
|
| 3349 |
+
"step": 47700
|
| 3350 |
+
},
|
| 3351 |
+
{
|
| 3352 |
+
"epoch": 48.478701825557806,
|
| 3353 |
+
"grad_norm": 4.617873191833496,
|
| 3354 |
+
"learning_rate": 1.5223123732251523e-06,
|
| 3355 |
+
"loss": 1.0822,
|
| 3356 |
+
"step": 47800
|
| 3357 |
+
},
|
| 3358 |
+
{
|
| 3359 |
+
"epoch": 48.580121703853955,
|
| 3360 |
+
"grad_norm": 3.919419288635254,
|
| 3361 |
+
"learning_rate": 1.4208924949290062e-06,
|
| 3362 |
+
"loss": 1.0616,
|
| 3363 |
+
"step": 47900
|
| 3364 |
+
},
|
| 3365 |
+
{
|
| 3366 |
+
"epoch": 48.6815415821501,
|
| 3367 |
+
"grad_norm": 4.350607872009277,
|
| 3368 |
+
"learning_rate": 1.31947261663286e-06,
|
| 3369 |
+
"loss": 1.0538,
|
| 3370 |
+
"step": 48000
|
| 3371 |
+
},
|
| 3372 |
+
{
|
| 3373 |
+
"epoch": 48.782961460446245,
|
| 3374 |
+
"grad_norm": 3.5256316661834717,
|
| 3375 |
+
"learning_rate": 1.2180527383367141e-06,
|
| 3376 |
+
"loss": 1.0627,
|
| 3377 |
+
"step": 48100
|
| 3378 |
+
},
|
| 3379 |
+
{
|
| 3380 |
+
"epoch": 48.88438133874239,
|
| 3381 |
+
"grad_norm": 4.878030776977539,
|
| 3382 |
+
"learning_rate": 1.116632860040568e-06,
|
| 3383 |
+
"loss": 1.0522,
|
| 3384 |
+
"step": 48200
|
| 3385 |
+
},
|
| 3386 |
+
{
|
| 3387 |
+
"epoch": 48.98580121703854,
|
| 3388 |
+
"grad_norm": 3.217900276184082,
|
| 3389 |
+
"learning_rate": 1.015212981744422e-06,
|
| 3390 |
+
"loss": 1.0564,
|
| 3391 |
+
"step": 48300
|
| 3392 |
+
},
|
| 3393 |
+
{
|
| 3394 |
+
"epoch": 49.087221095334684,
|
| 3395 |
+
"grad_norm": 4.753799915313721,
|
| 3396 |
+
"learning_rate": 9.137931034482759e-07,
|
| 3397 |
+
"loss": 1.0808,
|
| 3398 |
+
"step": 48400
|
| 3399 |
+
},
|
| 3400 |
+
{
|
| 3401 |
+
"epoch": 49.18864097363083,
|
| 3402 |
+
"grad_norm": 4.744013786315918,
|
| 3403 |
+
"learning_rate": 8.123732251521299e-07,
|
| 3404 |
+
"loss": 1.0728,
|
| 3405 |
+
"step": 48500
|
| 3406 |
+
},
|
| 3407 |
+
{
|
| 3408 |
+
"epoch": 49.29006085192698,
|
| 3409 |
+
"grad_norm": 4.291200637817383,
|
| 3410 |
+
"learning_rate": 7.109533468559839e-07,
|
| 3411 |
+
"loss": 1.0773,
|
| 3412 |
+
"step": 48600
|
| 3413 |
+
},
|
| 3414 |
+
{
|
| 3415 |
+
"epoch": 49.39148073022312,
|
| 3416 |
+
"grad_norm": 4.640216827392578,
|
| 3417 |
+
"learning_rate": 6.095334685598378e-07,
|
| 3418 |
+
"loss": 1.0684,
|
| 3419 |
+
"step": 48700
|
| 3420 |
+
},
|
| 3421 |
+
{
|
| 3422 |
+
"epoch": 49.49290060851927,
|
| 3423 |
+
"grad_norm": 4.025397300720215,
|
| 3424 |
+
"learning_rate": 5.081135902636917e-07,
|
| 3425 |
+
"loss": 1.0615,
|
| 3426 |
+
"step": 48800
|
| 3427 |
+
},
|
| 3428 |
+
{
|
| 3429 |
+
"epoch": 49.59432048681541,
|
| 3430 |
+
"grad_norm": 4.3919830322265625,
|
| 3431 |
+
"learning_rate": 4.066937119675457e-07,
|
| 3432 |
+
"loss": 1.0726,
|
| 3433 |
+
"step": 48900
|
| 3434 |
+
},
|
| 3435 |
+
{
|
| 3436 |
+
"epoch": 49.69574036511156,
|
| 3437 |
+
"grad_norm": 4.383048057556152,
|
| 3438 |
+
"learning_rate": 3.052738336713996e-07,
|
| 3439 |
+
"loss": 1.0462,
|
| 3440 |
+
"step": 49000
|
| 3441 |
+
}
|
| 3442 |
+
],
|
| 3443 |
+
"logging_steps": 100,
|
| 3444 |
+
"max_steps": 49300,
|
| 3445 |
+
"num_input_tokens_seen": 0,
|
| 3446 |
+
"num_train_epochs": 50,
|
| 3447 |
+
"save_steps": 500,
|
| 3448 |
+
"stateful_callbacks": {
|
| 3449 |
+
"TrainerControl": {
|
| 3450 |
+
"args": {
|
| 3451 |
+
"should_epoch_stop": false,
|
| 3452 |
+
"should_evaluate": false,
|
| 3453 |
+
"should_log": false,
|
| 3454 |
+
"should_save": true,
|
| 3455 |
+
"should_training_stop": false
|
| 3456 |
+
},
|
| 3457 |
+
"attributes": {}
|
| 3458 |
+
}
|
| 3459 |
+
},
|
| 3460 |
+
"total_flos": 8.249331026151936e+17,
|
| 3461 |
+
"train_batch_size": 64,
|
| 3462 |
+
"trial_name": null,
|
| 3463 |
+
"trial_params": null
|
| 3464 |
+
}
|
checkpoint-49000/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:14cced3a0f66dcdc1f8741f93d1ffe33c67abb84152654c2a418a0788b639afb
|
| 3 |
+
size 5713
|
checkpoint-49000/vocab.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
config.json
ADDED
|
@@ -0,0 +1,25 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"BertForMaskedLM"
|
| 4 |
+
],
|
| 5 |
+
"attention_probs_dropout_prob": 0.1,
|
| 6 |
+
"classifier_dropout": null,
|
| 7 |
+
"gradient_checkpointing": false,
|
| 8 |
+
"hidden_act": "gelu",
|
| 9 |
+
"hidden_dropout_prob": 0.1,
|
| 10 |
+
"hidden_size": 768,
|
| 11 |
+
"initializer_range": 0.02,
|
| 12 |
+
"intermediate_size": 3072,
|
| 13 |
+
"layer_norm_eps": 1e-12,
|
| 14 |
+
"max_position_embeddings": 512,
|
| 15 |
+
"model_type": "bert",
|
| 16 |
+
"num_attention_heads": 12,
|
| 17 |
+
"num_hidden_layers": 12,
|
| 18 |
+
"pad_token_id": 0,
|
| 19 |
+
"position_embedding_type": "absolute",
|
| 20 |
+
"torch_dtype": "float32",
|
| 21 |
+
"transformers_version": "4.52.4",
|
| 22 |
+
"type_vocab_size": 2,
|
| 23 |
+
"use_cache": true,
|
| 24 |
+
"vocab_size": 30522
|
| 25 |
+
}
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9217cbf18182c7d56a27d35d5ca6c72015ca5223ad491ce42289d0c0b148a6bd
|
| 3 |
+
size 438080896
|
special_tokens_map.json
ADDED
|
@@ -0,0 +1,37 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"cls_token": {
|
| 3 |
+
"content": "[CLS]",
|
| 4 |
+
"lstrip": false,
|
| 5 |
+
"normalized": false,
|
| 6 |
+
"rstrip": false,
|
| 7 |
+
"single_word": false
|
| 8 |
+
},
|
| 9 |
+
"mask_token": {
|
| 10 |
+
"content": "[MASK]",
|
| 11 |
+
"lstrip": false,
|
| 12 |
+
"normalized": false,
|
| 13 |
+
"rstrip": false,
|
| 14 |
+
"single_word": false
|
| 15 |
+
},
|
| 16 |
+
"pad_token": {
|
| 17 |
+
"content": "[PAD]",
|
| 18 |
+
"lstrip": false,
|
| 19 |
+
"normalized": false,
|
| 20 |
+
"rstrip": false,
|
| 21 |
+
"single_word": false
|
| 22 |
+
},
|
| 23 |
+
"sep_token": {
|
| 24 |
+
"content": "[SEP]",
|
| 25 |
+
"lstrip": false,
|
| 26 |
+
"normalized": false,
|
| 27 |
+
"rstrip": false,
|
| 28 |
+
"single_word": false
|
| 29 |
+
},
|
| 30 |
+
"unk_token": {
|
| 31 |
+
"content": "[UNK]",
|
| 32 |
+
"lstrip": false,
|
| 33 |
+
"normalized": false,
|
| 34 |
+
"rstrip": false,
|
| 35 |
+
"single_word": false
|
| 36 |
+
}
|
| 37 |
+
}
|
tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1,56 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"added_tokens_decoder": {
|
| 3 |
+
"0": {
|
| 4 |
+
"content": "[PAD]",
|
| 5 |
+
"lstrip": false,
|
| 6 |
+
"normalized": false,
|
| 7 |
+
"rstrip": false,
|
| 8 |
+
"single_word": false,
|
| 9 |
+
"special": true
|
| 10 |
+
},
|
| 11 |
+
"100": {
|
| 12 |
+
"content": "[UNK]",
|
| 13 |
+
"lstrip": false,
|
| 14 |
+
"normalized": false,
|
| 15 |
+
"rstrip": false,
|
| 16 |
+
"single_word": false,
|
| 17 |
+
"special": true
|
| 18 |
+
},
|
| 19 |
+
"101": {
|
| 20 |
+
"content": "[CLS]",
|
| 21 |
+
"lstrip": false,
|
| 22 |
+
"normalized": false,
|
| 23 |
+
"rstrip": false,
|
| 24 |
+
"single_word": false,
|
| 25 |
+
"special": true
|
| 26 |
+
},
|
| 27 |
+
"102": {
|
| 28 |
+
"content": "[SEP]",
|
| 29 |
+
"lstrip": false,
|
| 30 |
+
"normalized": false,
|
| 31 |
+
"rstrip": false,
|
| 32 |
+
"single_word": false,
|
| 33 |
+
"special": true
|
| 34 |
+
},
|
| 35 |
+
"103": {
|
| 36 |
+
"content": "[MASK]",
|
| 37 |
+
"lstrip": false,
|
| 38 |
+
"normalized": false,
|
| 39 |
+
"rstrip": false,
|
| 40 |
+
"single_word": false,
|
| 41 |
+
"special": true
|
| 42 |
+
}
|
| 43 |
+
},
|
| 44 |
+
"clean_up_tokenization_spaces": false,
|
| 45 |
+
"cls_token": "[CLS]",
|
| 46 |
+
"do_lower_case": true,
|
| 47 |
+
"extra_special_tokens": {},
|
| 48 |
+
"mask_token": "[MASK]",
|
| 49 |
+
"model_max_length": 512,
|
| 50 |
+
"pad_token": "[PAD]",
|
| 51 |
+
"sep_token": "[SEP]",
|
| 52 |
+
"strip_accents": null,
|
| 53 |
+
"tokenize_chinese_chars": true,
|
| 54 |
+
"tokenizer_class": "BertTokenizer",
|
| 55 |
+
"unk_token": "[UNK]"
|
| 56 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:14cced3a0f66dcdc1f8741f93d1ffe33c67abb84152654c2a418a0788b639afb
|
| 3 |
+
size 5713
|
vocab.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|