Upload 10 files
Browse files- checkpoint-95000/config.json +24 -0
- checkpoint-95000/model.safetensors +3 -0
- checkpoint-95000/optimizer.pt +3 -0
- checkpoint-95000/rng_state.pth +3 -0
- checkpoint-95000/scheduler.pt +3 -0
- checkpoint-95000/special_tokens_map.json +7 -0
- checkpoint-95000/tokenizer_config.json +58 -0
- checkpoint-95000/trainer_state.json +1981 -0
- checkpoint-95000/training_args.bin +3 -0
- checkpoint-95000/vocab.txt +0 -0
checkpoint-95000/config.json
ADDED
|
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"BertForMaskedLM"
|
| 4 |
+
],
|
| 5 |
+
"attention_probs_dropout_prob": 0.1,
|
| 6 |
+
"classifier_dropout": null,
|
| 7 |
+
"hidden_act": "gelu",
|
| 8 |
+
"hidden_dropout_prob": 0.1,
|
| 9 |
+
"hidden_size": 768,
|
| 10 |
+
"initializer_range": 0.02,
|
| 11 |
+
"intermediate_size": 3072,
|
| 12 |
+
"layer_norm_eps": 1e-12,
|
| 13 |
+
"max_position_embeddings": 512,
|
| 14 |
+
"model_type": "bert",
|
| 15 |
+
"num_attention_heads": 12,
|
| 16 |
+
"num_hidden_layers": 6,
|
| 17 |
+
"pad_token_id": 0,
|
| 18 |
+
"position_embedding_type": "absolute",
|
| 19 |
+
"torch_dtype": "float32",
|
| 20 |
+
"transformers_version": "4.50.2",
|
| 21 |
+
"type_vocab_size": 2,
|
| 22 |
+
"use_cache": true,
|
| 23 |
+
"vocab_size": 50000
|
| 24 |
+
}
|
checkpoint-95000/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:73d7834144eed44c3ecf9ee56ba68f853722a4d14143d7f63394758d3cd5f5d6
|
| 3 |
+
size 327874976
|
checkpoint-95000/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ed5f3f231fa1b2c5862cba2e4ba7026214477f6d45273c27f9aef99b777c9f3d
|
| 3 |
+
size 655814074
|
checkpoint-95000/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:091fb45f767684139f2052a8c796f176f40c3d308738e4736bf4ce24c3c7b2be
|
| 3 |
+
size 14244
|
checkpoint-95000/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:73501fb8c0426d838d543bb934170b118d4d60ad3e2531ccbcfec30df3856208
|
| 3 |
+
size 1064
|
checkpoint-95000/special_tokens_map.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"cls_token": "[CLS]",
|
| 3 |
+
"mask_token": "[MASK]",
|
| 4 |
+
"pad_token": "[PAD]",
|
| 5 |
+
"sep_token": "[SEP]",
|
| 6 |
+
"unk_token": "[UNK]"
|
| 7 |
+
}
|
checkpoint-95000/tokenizer_config.json
ADDED
|
@@ -0,0 +1,58 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 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": true,
|
| 45 |
+
"cls_token": "[CLS]",
|
| 46 |
+
"do_basic_tokenize": true,
|
| 47 |
+
"do_lower_case": true,
|
| 48 |
+
"extra_special_tokens": {},
|
| 49 |
+
"mask_token": "[MASK]",
|
| 50 |
+
"model_max_length": 512,
|
| 51 |
+
"never_split": null,
|
| 52 |
+
"pad_token": "[PAD]",
|
| 53 |
+
"sep_token": "[SEP]",
|
| 54 |
+
"strip_accents": null,
|
| 55 |
+
"tokenize_chinese_chars": true,
|
| 56 |
+
"tokenizer_class": "BertTokenizer",
|
| 57 |
+
"unk_token": "[UNK]"
|
| 58 |
+
}
|
checkpoint-95000/trainer_state.json
ADDED
|
@@ -0,0 +1,1981 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": 95000,
|
| 3 |
+
"best_metric": 2.405331611633301,
|
| 4 |
+
"best_model_checkpoint": "output/checkpoint-95000",
|
| 5 |
+
"epoch": 76.0,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 95000,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.4,
|
| 14 |
+
"grad_norm": 1.3970232009887695,
|
| 15 |
+
"learning_rate": 4.9800000000000004e-05,
|
| 16 |
+
"loss": 6.725,
|
| 17 |
+
"step": 500
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.8,
|
| 21 |
+
"grad_norm": 1.4344754219055176,
|
| 22 |
+
"learning_rate": 4.96e-05,
|
| 23 |
+
"loss": 6.1958,
|
| 24 |
+
"step": 1000
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 1.0,
|
| 28 |
+
"eval_loss": 6.076746940612793,
|
| 29 |
+
"eval_runtime": 61.2746,
|
| 30 |
+
"eval_samples_per_second": 163.2,
|
| 31 |
+
"eval_steps_per_second": 5.108,
|
| 32 |
+
"step": 1250
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 1.2,
|
| 36 |
+
"grad_norm": 1.259356141090393,
|
| 37 |
+
"learning_rate": 4.94e-05,
|
| 38 |
+
"loss": 6.118,
|
| 39 |
+
"step": 1500
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 1.6,
|
| 43 |
+
"grad_norm": 1.1414670944213867,
|
| 44 |
+
"learning_rate": 4.92e-05,
|
| 45 |
+
"loss": 6.0705,
|
| 46 |
+
"step": 2000
|
| 47 |
+
},
|
| 48 |
+
{
|
| 49 |
+
"epoch": 2.0,
|
| 50 |
+
"grad_norm": 1.2967591285705566,
|
| 51 |
+
"learning_rate": 4.9e-05,
|
| 52 |
+
"loss": 6.0444,
|
| 53 |
+
"step": 2500
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"epoch": 2.0,
|
| 57 |
+
"eval_loss": 6.011667251586914,
|
| 58 |
+
"eval_runtime": 61.2402,
|
| 59 |
+
"eval_samples_per_second": 163.291,
|
| 60 |
+
"eval_steps_per_second": 5.111,
|
| 61 |
+
"step": 2500
|
| 62 |
+
},
|
| 63 |
+
{
|
| 64 |
+
"epoch": 2.4,
|
| 65 |
+
"grad_norm": 1.572110652923584,
|
| 66 |
+
"learning_rate": 4.88e-05,
|
| 67 |
+
"loss": 6.0122,
|
| 68 |
+
"step": 3000
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 2.8,
|
| 72 |
+
"grad_norm": 1.6025742292404175,
|
| 73 |
+
"learning_rate": 4.86e-05,
|
| 74 |
+
"loss": 5.991,
|
| 75 |
+
"step": 3500
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"epoch": 3.0,
|
| 79 |
+
"eval_loss": 5.93717622756958,
|
| 80 |
+
"eval_runtime": 61.2634,
|
| 81 |
+
"eval_samples_per_second": 163.23,
|
| 82 |
+
"eval_steps_per_second": 5.109,
|
| 83 |
+
"step": 3750
|
| 84 |
+
},
|
| 85 |
+
{
|
| 86 |
+
"epoch": 3.2,
|
| 87 |
+
"grad_norm": 1.4724808931350708,
|
| 88 |
+
"learning_rate": 4.8400000000000004e-05,
|
| 89 |
+
"loss": 5.9594,
|
| 90 |
+
"step": 4000
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 3.6,
|
| 94 |
+
"grad_norm": 1.7168885469436646,
|
| 95 |
+
"learning_rate": 4.82e-05,
|
| 96 |
+
"loss": 5.8787,
|
| 97 |
+
"step": 4500
|
| 98 |
+
},
|
| 99 |
+
{
|
| 100 |
+
"epoch": 4.0,
|
| 101 |
+
"grad_norm": 1.6915048360824585,
|
| 102 |
+
"learning_rate": 4.8e-05,
|
| 103 |
+
"loss": 5.7669,
|
| 104 |
+
"step": 5000
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 4.0,
|
| 108 |
+
"eval_loss": 5.624746322631836,
|
| 109 |
+
"eval_runtime": 61.2596,
|
| 110 |
+
"eval_samples_per_second": 163.24,
|
| 111 |
+
"eval_steps_per_second": 5.109,
|
| 112 |
+
"step": 5000
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 4.4,
|
| 116 |
+
"grad_norm": 2.1659460067749023,
|
| 117 |
+
"learning_rate": 4.78e-05,
|
| 118 |
+
"loss": 5.593,
|
| 119 |
+
"step": 5500
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 4.8,
|
| 123 |
+
"grad_norm": 2.476262331008911,
|
| 124 |
+
"learning_rate": 4.76e-05,
|
| 125 |
+
"loss": 5.422,
|
| 126 |
+
"step": 6000
|
| 127 |
+
},
|
| 128 |
+
{
|
| 129 |
+
"epoch": 5.0,
|
| 130 |
+
"eval_loss": 5.117976665496826,
|
| 131 |
+
"eval_runtime": 61.2737,
|
| 132 |
+
"eval_samples_per_second": 163.202,
|
| 133 |
+
"eval_steps_per_second": 5.108,
|
| 134 |
+
"step": 6250
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 5.2,
|
| 138 |
+
"grad_norm": 2.4147756099700928,
|
| 139 |
+
"learning_rate": 4.74e-05,
|
| 140 |
+
"loss": 5.2305,
|
| 141 |
+
"step": 6500
|
| 142 |
+
},
|
| 143 |
+
{
|
| 144 |
+
"epoch": 5.6,
|
| 145 |
+
"grad_norm": 2.6457254886627197,
|
| 146 |
+
"learning_rate": 4.72e-05,
|
| 147 |
+
"loss": 4.9511,
|
| 148 |
+
"step": 7000
|
| 149 |
+
},
|
| 150 |
+
{
|
| 151 |
+
"epoch": 6.0,
|
| 152 |
+
"grad_norm": 2.7153351306915283,
|
| 153 |
+
"learning_rate": 4.7e-05,
|
| 154 |
+
"loss": 4.7339,
|
| 155 |
+
"step": 7500
|
| 156 |
+
},
|
| 157 |
+
{
|
| 158 |
+
"epoch": 6.0,
|
| 159 |
+
"eval_loss": 4.504222869873047,
|
| 160 |
+
"eval_runtime": 61.2685,
|
| 161 |
+
"eval_samples_per_second": 163.216,
|
| 162 |
+
"eval_steps_per_second": 5.109,
|
| 163 |
+
"step": 7500
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 6.4,
|
| 167 |
+
"grad_norm": 2.6116857528686523,
|
| 168 |
+
"learning_rate": 4.6800000000000006e-05,
|
| 169 |
+
"loss": 4.5647,
|
| 170 |
+
"step": 8000
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 6.8,
|
| 174 |
+
"grad_norm": 2.3627805709838867,
|
| 175 |
+
"learning_rate": 4.660000000000001e-05,
|
| 176 |
+
"loss": 4.432,
|
| 177 |
+
"step": 8500
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 7.0,
|
| 181 |
+
"eval_loss": 4.213283061981201,
|
| 182 |
+
"eval_runtime": 61.2562,
|
| 183 |
+
"eval_samples_per_second": 163.249,
|
| 184 |
+
"eval_steps_per_second": 5.11,
|
| 185 |
+
"step": 8750
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 7.2,
|
| 189 |
+
"grad_norm": 2.548539161682129,
|
| 190 |
+
"learning_rate": 4.64e-05,
|
| 191 |
+
"loss": 4.3363,
|
| 192 |
+
"step": 9000
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 7.6,
|
| 196 |
+
"grad_norm": 2.6018478870391846,
|
| 197 |
+
"learning_rate": 4.6200000000000005e-05,
|
| 198 |
+
"loss": 4.2403,
|
| 199 |
+
"step": 9500
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 8.0,
|
| 203 |
+
"grad_norm": 2.2613413333892822,
|
| 204 |
+
"learning_rate": 4.600000000000001e-05,
|
| 205 |
+
"loss": 4.1724,
|
| 206 |
+
"step": 10000
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 8.0,
|
| 210 |
+
"eval_loss": 4.003890514373779,
|
| 211 |
+
"eval_runtime": 61.4625,
|
| 212 |
+
"eval_samples_per_second": 162.701,
|
| 213 |
+
"eval_steps_per_second": 5.093,
|
| 214 |
+
"step": 10000
|
| 215 |
+
},
|
| 216 |
+
{
|
| 217 |
+
"epoch": 8.4,
|
| 218 |
+
"grad_norm": 2.538825273513794,
|
| 219 |
+
"learning_rate": 4.58e-05,
|
| 220 |
+
"loss": 4.0805,
|
| 221 |
+
"step": 10500
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 8.8,
|
| 225 |
+
"grad_norm": 2.3357536792755127,
|
| 226 |
+
"learning_rate": 4.5600000000000004e-05,
|
| 227 |
+
"loss": 4.0316,
|
| 228 |
+
"step": 11000
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 9.0,
|
| 232 |
+
"eval_loss": 3.875485897064209,
|
| 233 |
+
"eval_runtime": 61.2731,
|
| 234 |
+
"eval_samples_per_second": 163.204,
|
| 235 |
+
"eval_steps_per_second": 5.108,
|
| 236 |
+
"step": 11250
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 9.2,
|
| 240 |
+
"grad_norm": 2.7909834384918213,
|
| 241 |
+
"learning_rate": 4.5400000000000006e-05,
|
| 242 |
+
"loss": 3.9733,
|
| 243 |
+
"step": 11500
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 9.6,
|
| 247 |
+
"grad_norm": 2.770125150680542,
|
| 248 |
+
"learning_rate": 4.52e-05,
|
| 249 |
+
"loss": 3.9146,
|
| 250 |
+
"step": 12000
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"epoch": 10.0,
|
| 254 |
+
"grad_norm": 2.6937015056610107,
|
| 255 |
+
"learning_rate": 4.5e-05,
|
| 256 |
+
"loss": 3.8765,
|
| 257 |
+
"step": 12500
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 10.0,
|
| 261 |
+
"eval_loss": 3.7373440265655518,
|
| 262 |
+
"eval_runtime": 61.2543,
|
| 263 |
+
"eval_samples_per_second": 163.254,
|
| 264 |
+
"eval_steps_per_second": 5.11,
|
| 265 |
+
"step": 12500
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 10.4,
|
| 269 |
+
"grad_norm": 2.762148380279541,
|
| 270 |
+
"learning_rate": 4.4800000000000005e-05,
|
| 271 |
+
"loss": 3.8169,
|
| 272 |
+
"step": 13000
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 10.8,
|
| 276 |
+
"grad_norm": 2.428678035736084,
|
| 277 |
+
"learning_rate": 4.46e-05,
|
| 278 |
+
"loss": 3.7702,
|
| 279 |
+
"step": 13500
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 11.0,
|
| 283 |
+
"eval_loss": 3.6505491733551025,
|
| 284 |
+
"eval_runtime": 61.3079,
|
| 285 |
+
"eval_samples_per_second": 163.111,
|
| 286 |
+
"eval_steps_per_second": 5.105,
|
| 287 |
+
"step": 13750
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 11.2,
|
| 291 |
+
"grad_norm": 2.57837176322937,
|
| 292 |
+
"learning_rate": 4.44e-05,
|
| 293 |
+
"loss": 3.7268,
|
| 294 |
+
"step": 14000
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 11.6,
|
| 298 |
+
"grad_norm": 2.6295292377471924,
|
| 299 |
+
"learning_rate": 4.4200000000000004e-05,
|
| 300 |
+
"loss": 3.6882,
|
| 301 |
+
"step": 14500
|
| 302 |
+
},
|
| 303 |
+
{
|
| 304 |
+
"epoch": 12.0,
|
| 305 |
+
"grad_norm": 2.4153482913970947,
|
| 306 |
+
"learning_rate": 4.4000000000000006e-05,
|
| 307 |
+
"loss": 3.6646,
|
| 308 |
+
"step": 15000
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 12.0,
|
| 312 |
+
"eval_loss": 3.5465199947357178,
|
| 313 |
+
"eval_runtime": 61.2796,
|
| 314 |
+
"eval_samples_per_second": 163.187,
|
| 315 |
+
"eval_steps_per_second": 5.108,
|
| 316 |
+
"step": 15000
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 12.4,
|
| 320 |
+
"grad_norm": 2.5010106563568115,
|
| 321 |
+
"learning_rate": 4.38e-05,
|
| 322 |
+
"loss": 3.617,
|
| 323 |
+
"step": 15500
|
| 324 |
+
},
|
| 325 |
+
{
|
| 326 |
+
"epoch": 12.8,
|
| 327 |
+
"grad_norm": 2.515644073486328,
|
| 328 |
+
"learning_rate": 4.36e-05,
|
| 329 |
+
"loss": 3.5834,
|
| 330 |
+
"step": 16000
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"epoch": 13.0,
|
| 334 |
+
"eval_loss": 3.478050708770752,
|
| 335 |
+
"eval_runtime": 61.4335,
|
| 336 |
+
"eval_samples_per_second": 162.778,
|
| 337 |
+
"eval_steps_per_second": 5.095,
|
| 338 |
+
"step": 16250
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 13.2,
|
| 342 |
+
"grad_norm": 3.00858211517334,
|
| 343 |
+
"learning_rate": 4.3400000000000005e-05,
|
| 344 |
+
"loss": 3.5563,
|
| 345 |
+
"step": 16500
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 13.6,
|
| 349 |
+
"grad_norm": 2.528775453567505,
|
| 350 |
+
"learning_rate": 4.32e-05,
|
| 351 |
+
"loss": 3.5116,
|
| 352 |
+
"step": 17000
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 14.0,
|
| 356 |
+
"grad_norm": 2.637566089630127,
|
| 357 |
+
"learning_rate": 4.3e-05,
|
| 358 |
+
"loss": 3.4912,
|
| 359 |
+
"step": 17500
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 14.0,
|
| 363 |
+
"eval_loss": 3.401700019836426,
|
| 364 |
+
"eval_runtime": 61.2641,
|
| 365 |
+
"eval_samples_per_second": 163.228,
|
| 366 |
+
"eval_steps_per_second": 5.109,
|
| 367 |
+
"step": 17500
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 14.4,
|
| 371 |
+
"grad_norm": 2.4223690032958984,
|
| 372 |
+
"learning_rate": 4.2800000000000004e-05,
|
| 373 |
+
"loss": 3.4544,
|
| 374 |
+
"step": 18000
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 14.8,
|
| 378 |
+
"grad_norm": 2.7772583961486816,
|
| 379 |
+
"learning_rate": 4.26e-05,
|
| 380 |
+
"loss": 3.437,
|
| 381 |
+
"step": 18500
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 15.0,
|
| 385 |
+
"eval_loss": 3.337768077850342,
|
| 386 |
+
"eval_runtime": 61.2955,
|
| 387 |
+
"eval_samples_per_second": 163.144,
|
| 388 |
+
"eval_steps_per_second": 5.106,
|
| 389 |
+
"step": 18750
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 15.2,
|
| 393 |
+
"grad_norm": 2.417879104614258,
|
| 394 |
+
"learning_rate": 4.24e-05,
|
| 395 |
+
"loss": 3.4121,
|
| 396 |
+
"step": 19000
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 15.6,
|
| 400 |
+
"grad_norm": 2.525871753692627,
|
| 401 |
+
"learning_rate": 4.22e-05,
|
| 402 |
+
"loss": 3.3835,
|
| 403 |
+
"step": 19500
|
| 404 |
+
},
|
| 405 |
+
{
|
| 406 |
+
"epoch": 16.0,
|
| 407 |
+
"grad_norm": 2.49729061126709,
|
| 408 |
+
"learning_rate": 4.2e-05,
|
| 409 |
+
"loss": 3.3679,
|
| 410 |
+
"step": 20000
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 16.0,
|
| 414 |
+
"eval_loss": 3.2797868251800537,
|
| 415 |
+
"eval_runtime": 61.2076,
|
| 416 |
+
"eval_samples_per_second": 163.378,
|
| 417 |
+
"eval_steps_per_second": 5.114,
|
| 418 |
+
"step": 20000
|
| 419 |
+
},
|
| 420 |
+
{
|
| 421 |
+
"epoch": 16.4,
|
| 422 |
+
"grad_norm": 2.6784422397613525,
|
| 423 |
+
"learning_rate": 4.18e-05,
|
| 424 |
+
"loss": 3.3326,
|
| 425 |
+
"step": 20500
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"epoch": 16.8,
|
| 429 |
+
"grad_norm": 2.5526607036590576,
|
| 430 |
+
"learning_rate": 4.16e-05,
|
| 431 |
+
"loss": 3.3217,
|
| 432 |
+
"step": 21000
|
| 433 |
+
},
|
| 434 |
+
{
|
| 435 |
+
"epoch": 17.0,
|
| 436 |
+
"eval_loss": 3.233443021774292,
|
| 437 |
+
"eval_runtime": 61.2163,
|
| 438 |
+
"eval_samples_per_second": 163.355,
|
| 439 |
+
"eval_steps_per_second": 5.113,
|
| 440 |
+
"step": 21250
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 17.2,
|
| 444 |
+
"grad_norm": 2.685425281524658,
|
| 445 |
+
"learning_rate": 4.14e-05,
|
| 446 |
+
"loss": 3.2826,
|
| 447 |
+
"step": 21500
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 17.6,
|
| 451 |
+
"grad_norm": 2.7774910926818848,
|
| 452 |
+
"learning_rate": 4.12e-05,
|
| 453 |
+
"loss": 3.2716,
|
| 454 |
+
"step": 22000
|
| 455 |
+
},
|
| 456 |
+
{
|
| 457 |
+
"epoch": 18.0,
|
| 458 |
+
"grad_norm": 2.404536008834839,
|
| 459 |
+
"learning_rate": 4.1e-05,
|
| 460 |
+
"loss": 3.2583,
|
| 461 |
+
"step": 22500
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 18.0,
|
| 465 |
+
"eval_loss": 3.1886167526245117,
|
| 466 |
+
"eval_runtime": 61.4623,
|
| 467 |
+
"eval_samples_per_second": 162.701,
|
| 468 |
+
"eval_steps_per_second": 5.093,
|
| 469 |
+
"step": 22500
|
| 470 |
+
},
|
| 471 |
+
{
|
| 472 |
+
"epoch": 18.4,
|
| 473 |
+
"grad_norm": 2.5723462104797363,
|
| 474 |
+
"learning_rate": 4.08e-05,
|
| 475 |
+
"loss": 3.2232,
|
| 476 |
+
"step": 23000
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 18.8,
|
| 480 |
+
"grad_norm": 2.5137925148010254,
|
| 481 |
+
"learning_rate": 4.0600000000000004e-05,
|
| 482 |
+
"loss": 3.2212,
|
| 483 |
+
"step": 23500
|
| 484 |
+
},
|
| 485 |
+
{
|
| 486 |
+
"epoch": 19.0,
|
| 487 |
+
"eval_loss": 3.1394808292388916,
|
| 488 |
+
"eval_runtime": 61.2181,
|
| 489 |
+
"eval_samples_per_second": 163.35,
|
| 490 |
+
"eval_steps_per_second": 5.113,
|
| 491 |
+
"step": 23750
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 19.2,
|
| 495 |
+
"grad_norm": 2.7127292156219482,
|
| 496 |
+
"learning_rate": 4.0400000000000006e-05,
|
| 497 |
+
"loss": 3.1803,
|
| 498 |
+
"step": 24000
|
| 499 |
+
},
|
| 500 |
+
{
|
| 501 |
+
"epoch": 19.6,
|
| 502 |
+
"grad_norm": 2.5438954830169678,
|
| 503 |
+
"learning_rate": 4.02e-05,
|
| 504 |
+
"loss": 3.1692,
|
| 505 |
+
"step": 24500
|
| 506 |
+
},
|
| 507 |
+
{
|
| 508 |
+
"epoch": 20.0,
|
| 509 |
+
"grad_norm": 2.503960132598877,
|
| 510 |
+
"learning_rate": 4e-05,
|
| 511 |
+
"loss": 3.1697,
|
| 512 |
+
"step": 25000
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 20.0,
|
| 516 |
+
"eval_loss": 3.112931251525879,
|
| 517 |
+
"eval_runtime": 61.2263,
|
| 518 |
+
"eval_samples_per_second": 163.329,
|
| 519 |
+
"eval_steps_per_second": 5.112,
|
| 520 |
+
"step": 25000
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 20.4,
|
| 524 |
+
"grad_norm": 2.501166820526123,
|
| 525 |
+
"learning_rate": 3.9800000000000005e-05,
|
| 526 |
+
"loss": 3.1396,
|
| 527 |
+
"step": 25500
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 20.8,
|
| 531 |
+
"grad_norm": 2.7387282848358154,
|
| 532 |
+
"learning_rate": 3.960000000000001e-05,
|
| 533 |
+
"loss": 3.1209,
|
| 534 |
+
"step": 26000
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 21.0,
|
| 538 |
+
"eval_loss": 3.0616869926452637,
|
| 539 |
+
"eval_runtime": 61.2523,
|
| 540 |
+
"eval_samples_per_second": 163.259,
|
| 541 |
+
"eval_steps_per_second": 5.11,
|
| 542 |
+
"step": 26250
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 21.2,
|
| 546 |
+
"grad_norm": 2.641589641571045,
|
| 547 |
+
"learning_rate": 3.94e-05,
|
| 548 |
+
"loss": 3.1079,
|
| 549 |
+
"step": 26500
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 21.6,
|
| 553 |
+
"grad_norm": 2.8861656188964844,
|
| 554 |
+
"learning_rate": 3.9200000000000004e-05,
|
| 555 |
+
"loss": 3.0907,
|
| 556 |
+
"step": 27000
|
| 557 |
+
},
|
| 558 |
+
{
|
| 559 |
+
"epoch": 22.0,
|
| 560 |
+
"grad_norm": 2.9420666694641113,
|
| 561 |
+
"learning_rate": 3.9000000000000006e-05,
|
| 562 |
+
"loss": 3.0772,
|
| 563 |
+
"step": 27500
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 22.0,
|
| 567 |
+
"eval_loss": 3.029193639755249,
|
| 568 |
+
"eval_runtime": 61.3008,
|
| 569 |
+
"eval_samples_per_second": 163.13,
|
| 570 |
+
"eval_steps_per_second": 5.106,
|
| 571 |
+
"step": 27500
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"epoch": 22.4,
|
| 575 |
+
"grad_norm": 2.6903202533721924,
|
| 576 |
+
"learning_rate": 3.88e-05,
|
| 577 |
+
"loss": 3.0618,
|
| 578 |
+
"step": 28000
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 22.8,
|
| 582 |
+
"grad_norm": 2.4258873462677,
|
| 583 |
+
"learning_rate": 3.86e-05,
|
| 584 |
+
"loss": 3.0482,
|
| 585 |
+
"step": 28500
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 23.0,
|
| 589 |
+
"eval_loss": 2.998987913131714,
|
| 590 |
+
"eval_runtime": 61.2446,
|
| 591 |
+
"eval_samples_per_second": 163.28,
|
| 592 |
+
"eval_steps_per_second": 5.111,
|
| 593 |
+
"step": 28750
|
| 594 |
+
},
|
| 595 |
+
{
|
| 596 |
+
"epoch": 23.2,
|
| 597 |
+
"grad_norm": 2.8380045890808105,
|
| 598 |
+
"learning_rate": 3.8400000000000005e-05,
|
| 599 |
+
"loss": 3.0271,
|
| 600 |
+
"step": 29000
|
| 601 |
+
},
|
| 602 |
+
{
|
| 603 |
+
"epoch": 23.6,
|
| 604 |
+
"grad_norm": 2.936922788619995,
|
| 605 |
+
"learning_rate": 3.82e-05,
|
| 606 |
+
"loss": 3.016,
|
| 607 |
+
"step": 29500
|
| 608 |
+
},
|
| 609 |
+
{
|
| 610 |
+
"epoch": 24.0,
|
| 611 |
+
"grad_norm": 2.829270362854004,
|
| 612 |
+
"learning_rate": 3.8e-05,
|
| 613 |
+
"loss": 3.0102,
|
| 614 |
+
"step": 30000
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 24.0,
|
| 618 |
+
"eval_loss": 2.972419023513794,
|
| 619 |
+
"eval_runtime": 61.3686,
|
| 620 |
+
"eval_samples_per_second": 162.95,
|
| 621 |
+
"eval_steps_per_second": 5.1,
|
| 622 |
+
"step": 30000
|
| 623 |
+
},
|
| 624 |
+
{
|
| 625 |
+
"epoch": 24.4,
|
| 626 |
+
"grad_norm": 2.6595258712768555,
|
| 627 |
+
"learning_rate": 3.7800000000000004e-05,
|
| 628 |
+
"loss": 2.9857,
|
| 629 |
+
"step": 30500
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 24.8,
|
| 633 |
+
"grad_norm": 2.9528753757476807,
|
| 634 |
+
"learning_rate": 3.76e-05,
|
| 635 |
+
"loss": 2.9833,
|
| 636 |
+
"step": 31000
|
| 637 |
+
},
|
| 638 |
+
{
|
| 639 |
+
"epoch": 25.0,
|
| 640 |
+
"eval_loss": 2.943739414215088,
|
| 641 |
+
"eval_runtime": 61.233,
|
| 642 |
+
"eval_samples_per_second": 163.311,
|
| 643 |
+
"eval_steps_per_second": 5.112,
|
| 644 |
+
"step": 31250
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 25.2,
|
| 648 |
+
"grad_norm": 2.8180456161499023,
|
| 649 |
+
"learning_rate": 3.74e-05,
|
| 650 |
+
"loss": 2.9631,
|
| 651 |
+
"step": 31500
|
| 652 |
+
},
|
| 653 |
+
{
|
| 654 |
+
"epoch": 25.6,
|
| 655 |
+
"grad_norm": 3.0738282203674316,
|
| 656 |
+
"learning_rate": 3.72e-05,
|
| 657 |
+
"loss": 2.9521,
|
| 658 |
+
"step": 32000
|
| 659 |
+
},
|
| 660 |
+
{
|
| 661 |
+
"epoch": 26.0,
|
| 662 |
+
"grad_norm": 3.0022482872009277,
|
| 663 |
+
"learning_rate": 3.7e-05,
|
| 664 |
+
"loss": 2.9429,
|
| 665 |
+
"step": 32500
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 26.0,
|
| 669 |
+
"eval_loss": 2.918297290802002,
|
| 670 |
+
"eval_runtime": 61.2635,
|
| 671 |
+
"eval_samples_per_second": 163.229,
|
| 672 |
+
"eval_steps_per_second": 5.109,
|
| 673 |
+
"step": 32500
|
| 674 |
+
},
|
| 675 |
+
{
|
| 676 |
+
"epoch": 26.4,
|
| 677 |
+
"grad_norm": 2.5933921337127686,
|
| 678 |
+
"learning_rate": 3.68e-05,
|
| 679 |
+
"loss": 2.9205,
|
| 680 |
+
"step": 33000
|
| 681 |
+
},
|
| 682 |
+
{
|
| 683 |
+
"epoch": 26.8,
|
| 684 |
+
"grad_norm": 2.725785493850708,
|
| 685 |
+
"learning_rate": 3.66e-05,
|
| 686 |
+
"loss": 2.9138,
|
| 687 |
+
"step": 33500
|
| 688 |
+
},
|
| 689 |
+
{
|
| 690 |
+
"epoch": 27.0,
|
| 691 |
+
"eval_loss": 2.8853907585144043,
|
| 692 |
+
"eval_runtime": 61.2544,
|
| 693 |
+
"eval_samples_per_second": 163.254,
|
| 694 |
+
"eval_steps_per_second": 5.11,
|
| 695 |
+
"step": 33750
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 27.2,
|
| 699 |
+
"grad_norm": 2.8924100399017334,
|
| 700 |
+
"learning_rate": 3.6400000000000004e-05,
|
| 701 |
+
"loss": 2.9061,
|
| 702 |
+
"step": 34000
|
| 703 |
+
},
|
| 704 |
+
{
|
| 705 |
+
"epoch": 27.6,
|
| 706 |
+
"grad_norm": 2.9126296043395996,
|
| 707 |
+
"learning_rate": 3.62e-05,
|
| 708 |
+
"loss": 2.8858,
|
| 709 |
+
"step": 34500
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"epoch": 28.0,
|
| 713 |
+
"grad_norm": 2.768714189529419,
|
| 714 |
+
"learning_rate": 3.6e-05,
|
| 715 |
+
"loss": 2.8815,
|
| 716 |
+
"step": 35000
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 28.0,
|
| 720 |
+
"eval_loss": 2.8625736236572266,
|
| 721 |
+
"eval_runtime": 61.266,
|
| 722 |
+
"eval_samples_per_second": 163.223,
|
| 723 |
+
"eval_steps_per_second": 5.109,
|
| 724 |
+
"step": 35000
|
| 725 |
+
},
|
| 726 |
+
{
|
| 727 |
+
"epoch": 28.4,
|
| 728 |
+
"grad_norm": 2.8135673999786377,
|
| 729 |
+
"learning_rate": 3.58e-05,
|
| 730 |
+
"loss": 2.8635,
|
| 731 |
+
"step": 35500
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 28.8,
|
| 735 |
+
"grad_norm": 2.8309783935546875,
|
| 736 |
+
"learning_rate": 3.56e-05,
|
| 737 |
+
"loss": 2.848,
|
| 738 |
+
"step": 36000
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"epoch": 29.0,
|
| 742 |
+
"eval_loss": 2.8396055698394775,
|
| 743 |
+
"eval_runtime": 61.5143,
|
| 744 |
+
"eval_samples_per_second": 162.564,
|
| 745 |
+
"eval_steps_per_second": 5.088,
|
| 746 |
+
"step": 36250
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 29.2,
|
| 750 |
+
"grad_norm": 2.9002137184143066,
|
| 751 |
+
"learning_rate": 3.54e-05,
|
| 752 |
+
"loss": 2.8476,
|
| 753 |
+
"step": 36500
|
| 754 |
+
},
|
| 755 |
+
{
|
| 756 |
+
"epoch": 29.6,
|
| 757 |
+
"grad_norm": 2.79744815826416,
|
| 758 |
+
"learning_rate": 3.52e-05,
|
| 759 |
+
"loss": 2.8341,
|
| 760 |
+
"step": 37000
|
| 761 |
+
},
|
| 762 |
+
{
|
| 763 |
+
"epoch": 30.0,
|
| 764 |
+
"grad_norm": 3.041928768157959,
|
| 765 |
+
"learning_rate": 3.5e-05,
|
| 766 |
+
"loss": 2.8276,
|
| 767 |
+
"step": 37500
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 30.0,
|
| 771 |
+
"eval_loss": 2.805696487426758,
|
| 772 |
+
"eval_runtime": 61.2638,
|
| 773 |
+
"eval_samples_per_second": 163.229,
|
| 774 |
+
"eval_steps_per_second": 5.109,
|
| 775 |
+
"step": 37500
|
| 776 |
+
},
|
| 777 |
+
{
|
| 778 |
+
"epoch": 30.4,
|
| 779 |
+
"grad_norm": 2.7281506061553955,
|
| 780 |
+
"learning_rate": 3.48e-05,
|
| 781 |
+
"loss": 2.8067,
|
| 782 |
+
"step": 38000
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 30.8,
|
| 786 |
+
"grad_norm": 2.6974635124206543,
|
| 787 |
+
"learning_rate": 3.46e-05,
|
| 788 |
+
"loss": 2.8114,
|
| 789 |
+
"step": 38500
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 31.0,
|
| 793 |
+
"eval_loss": 2.7848477363586426,
|
| 794 |
+
"eval_runtime": 61.2633,
|
| 795 |
+
"eval_samples_per_second": 163.23,
|
| 796 |
+
"eval_steps_per_second": 5.109,
|
| 797 |
+
"step": 38750
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 31.2,
|
| 801 |
+
"grad_norm": 3.201719045639038,
|
| 802 |
+
"learning_rate": 3.4399999999999996e-05,
|
| 803 |
+
"loss": 2.7928,
|
| 804 |
+
"step": 39000
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 31.6,
|
| 808 |
+
"grad_norm": 2.7298779487609863,
|
| 809 |
+
"learning_rate": 3.4200000000000005e-05,
|
| 810 |
+
"loss": 2.7814,
|
| 811 |
+
"step": 39500
|
| 812 |
+
},
|
| 813 |
+
{
|
| 814 |
+
"epoch": 32.0,
|
| 815 |
+
"grad_norm": 2.873957872390747,
|
| 816 |
+
"learning_rate": 3.4000000000000007e-05,
|
| 817 |
+
"loss": 2.7806,
|
| 818 |
+
"step": 40000
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 32.0,
|
| 822 |
+
"eval_loss": 2.773723840713501,
|
| 823 |
+
"eval_runtime": 61.1994,
|
| 824 |
+
"eval_samples_per_second": 163.4,
|
| 825 |
+
"eval_steps_per_second": 5.114,
|
| 826 |
+
"step": 40000
|
| 827 |
+
},
|
| 828 |
+
{
|
| 829 |
+
"epoch": 32.4,
|
| 830 |
+
"grad_norm": 2.733128547668457,
|
| 831 |
+
"learning_rate": 3.38e-05,
|
| 832 |
+
"loss": 2.7619,
|
| 833 |
+
"step": 40500
|
| 834 |
+
},
|
| 835 |
+
{
|
| 836 |
+
"epoch": 32.8,
|
| 837 |
+
"grad_norm": 2.825956106185913,
|
| 838 |
+
"learning_rate": 3.3600000000000004e-05,
|
| 839 |
+
"loss": 2.7608,
|
| 840 |
+
"step": 41000
|
| 841 |
+
},
|
| 842 |
+
{
|
| 843 |
+
"epoch": 33.0,
|
| 844 |
+
"eval_loss": 2.7562143802642822,
|
| 845 |
+
"eval_runtime": 61.2126,
|
| 846 |
+
"eval_samples_per_second": 163.365,
|
| 847 |
+
"eval_steps_per_second": 5.113,
|
| 848 |
+
"step": 41250
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 33.2,
|
| 852 |
+
"grad_norm": 2.9455862045288086,
|
| 853 |
+
"learning_rate": 3.3400000000000005e-05,
|
| 854 |
+
"loss": 2.742,
|
| 855 |
+
"step": 41500
|
| 856 |
+
},
|
| 857 |
+
{
|
| 858 |
+
"epoch": 33.6,
|
| 859 |
+
"grad_norm": 2.9023234844207764,
|
| 860 |
+
"learning_rate": 3.32e-05,
|
| 861 |
+
"loss": 2.7347,
|
| 862 |
+
"step": 42000
|
| 863 |
+
},
|
| 864 |
+
{
|
| 865 |
+
"epoch": 34.0,
|
| 866 |
+
"grad_norm": 2.7190394401550293,
|
| 867 |
+
"learning_rate": 3.3e-05,
|
| 868 |
+
"loss": 2.7366,
|
| 869 |
+
"step": 42500
|
| 870 |
+
},
|
| 871 |
+
{
|
| 872 |
+
"epoch": 34.0,
|
| 873 |
+
"eval_loss": 2.7386631965637207,
|
| 874 |
+
"eval_runtime": 61.2856,
|
| 875 |
+
"eval_samples_per_second": 163.17,
|
| 876 |
+
"eval_steps_per_second": 5.107,
|
| 877 |
+
"step": 42500
|
| 878 |
+
},
|
| 879 |
+
{
|
| 880 |
+
"epoch": 34.4,
|
| 881 |
+
"grad_norm": 2.9011335372924805,
|
| 882 |
+
"learning_rate": 3.2800000000000004e-05,
|
| 883 |
+
"loss": 2.7113,
|
| 884 |
+
"step": 43000
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 34.8,
|
| 888 |
+
"grad_norm": 2.6606903076171875,
|
| 889 |
+
"learning_rate": 3.26e-05,
|
| 890 |
+
"loss": 2.7157,
|
| 891 |
+
"step": 43500
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 35.0,
|
| 895 |
+
"eval_loss": 2.713857889175415,
|
| 896 |
+
"eval_runtime": 61.228,
|
| 897 |
+
"eval_samples_per_second": 163.324,
|
| 898 |
+
"eval_steps_per_second": 5.112,
|
| 899 |
+
"step": 43750
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 35.2,
|
| 903 |
+
"grad_norm": 2.7517592906951904,
|
| 904 |
+
"learning_rate": 3.24e-05,
|
| 905 |
+
"loss": 2.7046,
|
| 906 |
+
"step": 44000
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 35.6,
|
| 910 |
+
"grad_norm": 3.1010994911193848,
|
| 911 |
+
"learning_rate": 3.2200000000000003e-05,
|
| 912 |
+
"loss": 2.6955,
|
| 913 |
+
"step": 44500
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"epoch": 36.0,
|
| 917 |
+
"grad_norm": 2.7236251831054688,
|
| 918 |
+
"learning_rate": 3.2000000000000005e-05,
|
| 919 |
+
"loss": 2.6894,
|
| 920 |
+
"step": 45000
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 36.0,
|
| 924 |
+
"eval_loss": 2.707624673843384,
|
| 925 |
+
"eval_runtime": 61.2324,
|
| 926 |
+
"eval_samples_per_second": 163.312,
|
| 927 |
+
"eval_steps_per_second": 5.112,
|
| 928 |
+
"step": 45000
|
| 929 |
+
},
|
| 930 |
+
{
|
| 931 |
+
"epoch": 36.4,
|
| 932 |
+
"grad_norm": 2.961221933364868,
|
| 933 |
+
"learning_rate": 3.18e-05,
|
| 934 |
+
"loss": 2.6825,
|
| 935 |
+
"step": 45500
|
| 936 |
+
},
|
| 937 |
+
{
|
| 938 |
+
"epoch": 36.8,
|
| 939 |
+
"grad_norm": 2.9229893684387207,
|
| 940 |
+
"learning_rate": 3.16e-05,
|
| 941 |
+
"loss": 2.6708,
|
| 942 |
+
"step": 46000
|
| 943 |
+
},
|
| 944 |
+
{
|
| 945 |
+
"epoch": 37.0,
|
| 946 |
+
"eval_loss": 2.686065196990967,
|
| 947 |
+
"eval_runtime": 61.2242,
|
| 948 |
+
"eval_samples_per_second": 163.334,
|
| 949 |
+
"eval_steps_per_second": 5.112,
|
| 950 |
+
"step": 46250
|
| 951 |
+
},
|
| 952 |
+
{
|
| 953 |
+
"epoch": 37.2,
|
| 954 |
+
"grad_norm": 3.1210429668426514,
|
| 955 |
+
"learning_rate": 3.1400000000000004e-05,
|
| 956 |
+
"loss": 2.6717,
|
| 957 |
+
"step": 46500
|
| 958 |
+
},
|
| 959 |
+
{
|
| 960 |
+
"epoch": 37.6,
|
| 961 |
+
"grad_norm": 2.785784959793091,
|
| 962 |
+
"learning_rate": 3.12e-05,
|
| 963 |
+
"loss": 2.6594,
|
| 964 |
+
"step": 47000
|
| 965 |
+
},
|
| 966 |
+
{
|
| 967 |
+
"epoch": 38.0,
|
| 968 |
+
"grad_norm": 2.8745310306549072,
|
| 969 |
+
"learning_rate": 3.1e-05,
|
| 970 |
+
"loss": 2.6577,
|
| 971 |
+
"step": 47500
|
| 972 |
+
},
|
| 973 |
+
{
|
| 974 |
+
"epoch": 38.0,
|
| 975 |
+
"eval_loss": 2.6833102703094482,
|
| 976 |
+
"eval_runtime": 61.2268,
|
| 977 |
+
"eval_samples_per_second": 163.327,
|
| 978 |
+
"eval_steps_per_second": 5.112,
|
| 979 |
+
"step": 47500
|
| 980 |
+
},
|
| 981 |
+
{
|
| 982 |
+
"epoch": 38.4,
|
| 983 |
+
"grad_norm": 2.848191022872925,
|
| 984 |
+
"learning_rate": 3.08e-05,
|
| 985 |
+
"loss": 2.6407,
|
| 986 |
+
"step": 48000
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 38.8,
|
| 990 |
+
"grad_norm": 2.9842936992645264,
|
| 991 |
+
"learning_rate": 3.06e-05,
|
| 992 |
+
"loss": 2.6437,
|
| 993 |
+
"step": 48500
|
| 994 |
+
},
|
| 995 |
+
{
|
| 996 |
+
"epoch": 39.0,
|
| 997 |
+
"eval_loss": 2.6599860191345215,
|
| 998 |
+
"eval_runtime": 61.3483,
|
| 999 |
+
"eval_samples_per_second": 163.004,
|
| 1000 |
+
"eval_steps_per_second": 5.102,
|
| 1001 |
+
"step": 48750
|
| 1002 |
+
},
|
| 1003 |
+
{
|
| 1004 |
+
"epoch": 39.2,
|
| 1005 |
+
"grad_norm": 2.9027788639068604,
|
| 1006 |
+
"learning_rate": 3.04e-05,
|
| 1007 |
+
"loss": 2.6275,
|
| 1008 |
+
"step": 49000
|
| 1009 |
+
},
|
| 1010 |
+
{
|
| 1011 |
+
"epoch": 39.6,
|
| 1012 |
+
"grad_norm": 2.8184854984283447,
|
| 1013 |
+
"learning_rate": 3.02e-05,
|
| 1014 |
+
"loss": 2.6188,
|
| 1015 |
+
"step": 49500
|
| 1016 |
+
},
|
| 1017 |
+
{
|
| 1018 |
+
"epoch": 40.0,
|
| 1019 |
+
"grad_norm": 2.8621346950531006,
|
| 1020 |
+
"learning_rate": 3e-05,
|
| 1021 |
+
"loss": 2.6255,
|
| 1022 |
+
"step": 50000
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 40.0,
|
| 1026 |
+
"eval_loss": 2.6438729763031006,
|
| 1027 |
+
"eval_runtime": 61.2542,
|
| 1028 |
+
"eval_samples_per_second": 163.254,
|
| 1029 |
+
"eval_steps_per_second": 5.11,
|
| 1030 |
+
"step": 50000
|
| 1031 |
+
},
|
| 1032 |
+
{
|
| 1033 |
+
"epoch": 40.4,
|
| 1034 |
+
"grad_norm": 2.661895275115967,
|
| 1035 |
+
"learning_rate": 2.98e-05,
|
| 1036 |
+
"loss": 2.6124,
|
| 1037 |
+
"step": 50500
|
| 1038 |
+
},
|
| 1039 |
+
{
|
| 1040 |
+
"epoch": 40.8,
|
| 1041 |
+
"grad_norm": 2.913214683532715,
|
| 1042 |
+
"learning_rate": 2.96e-05,
|
| 1043 |
+
"loss": 2.6128,
|
| 1044 |
+
"step": 51000
|
| 1045 |
+
},
|
| 1046 |
+
{
|
| 1047 |
+
"epoch": 41.0,
|
| 1048 |
+
"eval_loss": 2.634756088256836,
|
| 1049 |
+
"eval_runtime": 61.2199,
|
| 1050 |
+
"eval_samples_per_second": 163.346,
|
| 1051 |
+
"eval_steps_per_second": 5.113,
|
| 1052 |
+
"step": 51250
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 41.2,
|
| 1056 |
+
"grad_norm": 2.8805391788482666,
|
| 1057 |
+
"learning_rate": 2.94e-05,
|
| 1058 |
+
"loss": 2.5944,
|
| 1059 |
+
"step": 51500
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 41.6,
|
| 1063 |
+
"grad_norm": 3.3086748123168945,
|
| 1064 |
+
"learning_rate": 2.9199999999999998e-05,
|
| 1065 |
+
"loss": 2.5968,
|
| 1066 |
+
"step": 52000
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 42.0,
|
| 1070 |
+
"grad_norm": 2.8824660778045654,
|
| 1071 |
+
"learning_rate": 2.9e-05,
|
| 1072 |
+
"loss": 2.5929,
|
| 1073 |
+
"step": 52500
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 42.0,
|
| 1077 |
+
"eval_loss": 2.6286351680755615,
|
| 1078 |
+
"eval_runtime": 61.2491,
|
| 1079 |
+
"eval_samples_per_second": 163.268,
|
| 1080 |
+
"eval_steps_per_second": 5.11,
|
| 1081 |
+
"step": 52500
|
| 1082 |
+
},
|
| 1083 |
+
{
|
| 1084 |
+
"epoch": 42.4,
|
| 1085 |
+
"grad_norm": 3.0503251552581787,
|
| 1086 |
+
"learning_rate": 2.88e-05,
|
| 1087 |
+
"loss": 2.5691,
|
| 1088 |
+
"step": 53000
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 42.8,
|
| 1092 |
+
"grad_norm": 2.777174711227417,
|
| 1093 |
+
"learning_rate": 2.86e-05,
|
| 1094 |
+
"loss": 2.5861,
|
| 1095 |
+
"step": 53500
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 43.0,
|
| 1099 |
+
"eval_loss": 2.605225086212158,
|
| 1100 |
+
"eval_runtime": 61.226,
|
| 1101 |
+
"eval_samples_per_second": 163.329,
|
| 1102 |
+
"eval_steps_per_second": 5.112,
|
| 1103 |
+
"step": 53750
|
| 1104 |
+
},
|
| 1105 |
+
{
|
| 1106 |
+
"epoch": 43.2,
|
| 1107 |
+
"grad_norm": 2.6447839736938477,
|
| 1108 |
+
"learning_rate": 2.84e-05,
|
| 1109 |
+
"loss": 2.5709,
|
| 1110 |
+
"step": 54000
|
| 1111 |
+
},
|
| 1112 |
+
{
|
| 1113 |
+
"epoch": 43.6,
|
| 1114 |
+
"grad_norm": 2.7956857681274414,
|
| 1115 |
+
"learning_rate": 2.8199999999999998e-05,
|
| 1116 |
+
"loss": 2.5672,
|
| 1117 |
+
"step": 54500
|
| 1118 |
+
},
|
| 1119 |
+
{
|
| 1120 |
+
"epoch": 44.0,
|
| 1121 |
+
"grad_norm": 3.081850528717041,
|
| 1122 |
+
"learning_rate": 2.8000000000000003e-05,
|
| 1123 |
+
"loss": 2.5665,
|
| 1124 |
+
"step": 55000
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 44.0,
|
| 1128 |
+
"eval_loss": 2.588907480239868,
|
| 1129 |
+
"eval_runtime": 61.2915,
|
| 1130 |
+
"eval_samples_per_second": 163.155,
|
| 1131 |
+
"eval_steps_per_second": 5.107,
|
| 1132 |
+
"step": 55000
|
| 1133 |
+
},
|
| 1134 |
+
{
|
| 1135 |
+
"epoch": 44.4,
|
| 1136 |
+
"grad_norm": 2.979550838470459,
|
| 1137 |
+
"learning_rate": 2.7800000000000005e-05,
|
| 1138 |
+
"loss": 2.5553,
|
| 1139 |
+
"step": 55500
|
| 1140 |
+
},
|
| 1141 |
+
{
|
| 1142 |
+
"epoch": 44.8,
|
| 1143 |
+
"grad_norm": 2.8044862747192383,
|
| 1144 |
+
"learning_rate": 2.7600000000000003e-05,
|
| 1145 |
+
"loss": 2.5428,
|
| 1146 |
+
"step": 56000
|
| 1147 |
+
},
|
| 1148 |
+
{
|
| 1149 |
+
"epoch": 45.0,
|
| 1150 |
+
"eval_loss": 2.5998992919921875,
|
| 1151 |
+
"eval_runtime": 61.2365,
|
| 1152 |
+
"eval_samples_per_second": 163.301,
|
| 1153 |
+
"eval_steps_per_second": 5.111,
|
| 1154 |
+
"step": 56250
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 45.2,
|
| 1158 |
+
"grad_norm": 2.7438066005706787,
|
| 1159 |
+
"learning_rate": 2.7400000000000002e-05,
|
| 1160 |
+
"loss": 2.5435,
|
| 1161 |
+
"step": 56500
|
| 1162 |
+
},
|
| 1163 |
+
{
|
| 1164 |
+
"epoch": 45.6,
|
| 1165 |
+
"grad_norm": 2.8336572647094727,
|
| 1166 |
+
"learning_rate": 2.7200000000000004e-05,
|
| 1167 |
+
"loss": 2.5333,
|
| 1168 |
+
"step": 57000
|
| 1169 |
+
},
|
| 1170 |
+
{
|
| 1171 |
+
"epoch": 46.0,
|
| 1172 |
+
"grad_norm": 2.8354380130767822,
|
| 1173 |
+
"learning_rate": 2.7000000000000002e-05,
|
| 1174 |
+
"loss": 2.5385,
|
| 1175 |
+
"step": 57500
|
| 1176 |
+
},
|
| 1177 |
+
{
|
| 1178 |
+
"epoch": 46.0,
|
| 1179 |
+
"eval_loss": 2.5822722911834717,
|
| 1180 |
+
"eval_runtime": 61.2292,
|
| 1181 |
+
"eval_samples_per_second": 163.321,
|
| 1182 |
+
"eval_steps_per_second": 5.112,
|
| 1183 |
+
"step": 57500
|
| 1184 |
+
},
|
| 1185 |
+
{
|
| 1186 |
+
"epoch": 46.4,
|
| 1187 |
+
"grad_norm": 3.035831928253174,
|
| 1188 |
+
"learning_rate": 2.6800000000000004e-05,
|
| 1189 |
+
"loss": 2.5185,
|
| 1190 |
+
"step": 58000
|
| 1191 |
+
},
|
| 1192 |
+
{
|
| 1193 |
+
"epoch": 46.8,
|
| 1194 |
+
"grad_norm": 3.1147818565368652,
|
| 1195 |
+
"learning_rate": 2.6600000000000003e-05,
|
| 1196 |
+
"loss": 2.5273,
|
| 1197 |
+
"step": 58500
|
| 1198 |
+
},
|
| 1199 |
+
{
|
| 1200 |
+
"epoch": 47.0,
|
| 1201 |
+
"eval_loss": 2.5733859539031982,
|
| 1202 |
+
"eval_runtime": 61.2138,
|
| 1203 |
+
"eval_samples_per_second": 163.362,
|
| 1204 |
+
"eval_steps_per_second": 5.113,
|
| 1205 |
+
"step": 58750
|
| 1206 |
+
},
|
| 1207 |
+
{
|
| 1208 |
+
"epoch": 47.2,
|
| 1209 |
+
"grad_norm": 2.987078905105591,
|
| 1210 |
+
"learning_rate": 2.64e-05,
|
| 1211 |
+
"loss": 2.5207,
|
| 1212 |
+
"step": 59000
|
| 1213 |
+
},
|
| 1214 |
+
{
|
| 1215 |
+
"epoch": 47.6,
|
| 1216 |
+
"grad_norm": 2.7736542224884033,
|
| 1217 |
+
"learning_rate": 2.6200000000000003e-05,
|
| 1218 |
+
"loss": 2.5038,
|
| 1219 |
+
"step": 59500
|
| 1220 |
+
},
|
| 1221 |
+
{
|
| 1222 |
+
"epoch": 48.0,
|
| 1223 |
+
"grad_norm": 2.8549487590789795,
|
| 1224 |
+
"learning_rate": 2.6000000000000002e-05,
|
| 1225 |
+
"loss": 2.5175,
|
| 1226 |
+
"step": 60000
|
| 1227 |
+
},
|
| 1228 |
+
{
|
| 1229 |
+
"epoch": 48.0,
|
| 1230 |
+
"eval_loss": 2.561167001724243,
|
| 1231 |
+
"eval_runtime": 61.2382,
|
| 1232 |
+
"eval_samples_per_second": 163.297,
|
| 1233 |
+
"eval_steps_per_second": 5.111,
|
| 1234 |
+
"step": 60000
|
| 1235 |
+
},
|
| 1236 |
+
{
|
| 1237 |
+
"epoch": 48.4,
|
| 1238 |
+
"grad_norm": 3.058418035507202,
|
| 1239 |
+
"learning_rate": 2.58e-05,
|
| 1240 |
+
"loss": 2.4956,
|
| 1241 |
+
"step": 60500
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 48.8,
|
| 1245 |
+
"grad_norm": 2.955167293548584,
|
| 1246 |
+
"learning_rate": 2.5600000000000002e-05,
|
| 1247 |
+
"loss": 2.5064,
|
| 1248 |
+
"step": 61000
|
| 1249 |
+
},
|
| 1250 |
+
{
|
| 1251 |
+
"epoch": 49.0,
|
| 1252 |
+
"eval_loss": 2.555574417114258,
|
| 1253 |
+
"eval_runtime": 61.2242,
|
| 1254 |
+
"eval_samples_per_second": 163.334,
|
| 1255 |
+
"eval_steps_per_second": 5.112,
|
| 1256 |
+
"step": 61250
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 49.2,
|
| 1260 |
+
"grad_norm": 2.9953746795654297,
|
| 1261 |
+
"learning_rate": 2.54e-05,
|
| 1262 |
+
"loss": 2.4899,
|
| 1263 |
+
"step": 61500
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 49.6,
|
| 1267 |
+
"grad_norm": 3.176257610321045,
|
| 1268 |
+
"learning_rate": 2.5200000000000003e-05,
|
| 1269 |
+
"loss": 2.4862,
|
| 1270 |
+
"step": 62000
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 50.0,
|
| 1274 |
+
"grad_norm": 3.131290912628174,
|
| 1275 |
+
"learning_rate": 2.5e-05,
|
| 1276 |
+
"loss": 2.4931,
|
| 1277 |
+
"step": 62500
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 50.0,
|
| 1281 |
+
"eval_loss": 2.5496270656585693,
|
| 1282 |
+
"eval_runtime": 61.2301,
|
| 1283 |
+
"eval_samples_per_second": 163.319,
|
| 1284 |
+
"eval_steps_per_second": 5.112,
|
| 1285 |
+
"step": 62500
|
| 1286 |
+
},
|
| 1287 |
+
{
|
| 1288 |
+
"epoch": 50.4,
|
| 1289 |
+
"grad_norm": 2.871742010116577,
|
| 1290 |
+
"learning_rate": 2.48e-05,
|
| 1291 |
+
"loss": 2.4794,
|
| 1292 |
+
"step": 63000
|
| 1293 |
+
},
|
| 1294 |
+
{
|
| 1295 |
+
"epoch": 50.8,
|
| 1296 |
+
"grad_norm": 2.7357499599456787,
|
| 1297 |
+
"learning_rate": 2.46e-05,
|
| 1298 |
+
"loss": 2.4802,
|
| 1299 |
+
"step": 63500
|
| 1300 |
+
},
|
| 1301 |
+
{
|
| 1302 |
+
"epoch": 51.0,
|
| 1303 |
+
"eval_loss": 2.528775691986084,
|
| 1304 |
+
"eval_runtime": 61.2246,
|
| 1305 |
+
"eval_samples_per_second": 163.333,
|
| 1306 |
+
"eval_steps_per_second": 5.112,
|
| 1307 |
+
"step": 63750
|
| 1308 |
+
},
|
| 1309 |
+
{
|
| 1310 |
+
"epoch": 51.2,
|
| 1311 |
+
"grad_norm": 2.903444528579712,
|
| 1312 |
+
"learning_rate": 2.44e-05,
|
| 1313 |
+
"loss": 2.4712,
|
| 1314 |
+
"step": 64000
|
| 1315 |
+
},
|
| 1316 |
+
{
|
| 1317 |
+
"epoch": 51.6,
|
| 1318 |
+
"grad_norm": 3.0363874435424805,
|
| 1319 |
+
"learning_rate": 2.4200000000000002e-05,
|
| 1320 |
+
"loss": 2.4673,
|
| 1321 |
+
"step": 64500
|
| 1322 |
+
},
|
| 1323 |
+
{
|
| 1324 |
+
"epoch": 52.0,
|
| 1325 |
+
"grad_norm": 2.935192823410034,
|
| 1326 |
+
"learning_rate": 2.4e-05,
|
| 1327 |
+
"loss": 2.4604,
|
| 1328 |
+
"step": 65000
|
| 1329 |
+
},
|
| 1330 |
+
{
|
| 1331 |
+
"epoch": 52.0,
|
| 1332 |
+
"eval_loss": 2.5387279987335205,
|
| 1333 |
+
"eval_runtime": 61.5389,
|
| 1334 |
+
"eval_samples_per_second": 162.499,
|
| 1335 |
+
"eval_steps_per_second": 5.086,
|
| 1336 |
+
"step": 65000
|
| 1337 |
+
},
|
| 1338 |
+
{
|
| 1339 |
+
"epoch": 52.4,
|
| 1340 |
+
"grad_norm": 2.8958323001861572,
|
| 1341 |
+
"learning_rate": 2.38e-05,
|
| 1342 |
+
"loss": 2.4495,
|
| 1343 |
+
"step": 65500
|
| 1344 |
+
},
|
| 1345 |
+
{
|
| 1346 |
+
"epoch": 52.8,
|
| 1347 |
+
"grad_norm": 2.9605252742767334,
|
| 1348 |
+
"learning_rate": 2.36e-05,
|
| 1349 |
+
"loss": 2.4585,
|
| 1350 |
+
"step": 66000
|
| 1351 |
+
},
|
| 1352 |
+
{
|
| 1353 |
+
"epoch": 53.0,
|
| 1354 |
+
"eval_loss": 2.5214128494262695,
|
| 1355 |
+
"eval_runtime": 61.2198,
|
| 1356 |
+
"eval_samples_per_second": 163.346,
|
| 1357 |
+
"eval_steps_per_second": 5.113,
|
| 1358 |
+
"step": 66250
|
| 1359 |
+
},
|
| 1360 |
+
{
|
| 1361 |
+
"epoch": 53.2,
|
| 1362 |
+
"grad_norm": 2.8114893436431885,
|
| 1363 |
+
"learning_rate": 2.3400000000000003e-05,
|
| 1364 |
+
"loss": 2.4386,
|
| 1365 |
+
"step": 66500
|
| 1366 |
+
},
|
| 1367 |
+
{
|
| 1368 |
+
"epoch": 53.6,
|
| 1369 |
+
"grad_norm": 2.9136197566986084,
|
| 1370 |
+
"learning_rate": 2.32e-05,
|
| 1371 |
+
"loss": 2.442,
|
| 1372 |
+
"step": 67000
|
| 1373 |
+
},
|
| 1374 |
+
{
|
| 1375 |
+
"epoch": 54.0,
|
| 1376 |
+
"grad_norm": 2.891444683074951,
|
| 1377 |
+
"learning_rate": 2.3000000000000003e-05,
|
| 1378 |
+
"loss": 2.4517,
|
| 1379 |
+
"step": 67500
|
| 1380 |
+
},
|
| 1381 |
+
{
|
| 1382 |
+
"epoch": 54.0,
|
| 1383 |
+
"eval_loss": 2.520853042602539,
|
| 1384 |
+
"eval_runtime": 61.2251,
|
| 1385 |
+
"eval_samples_per_second": 163.332,
|
| 1386 |
+
"eval_steps_per_second": 5.112,
|
| 1387 |
+
"step": 67500
|
| 1388 |
+
},
|
| 1389 |
+
{
|
| 1390 |
+
"epoch": 54.4,
|
| 1391 |
+
"grad_norm": 2.8234751224517822,
|
| 1392 |
+
"learning_rate": 2.2800000000000002e-05,
|
| 1393 |
+
"loss": 2.4323,
|
| 1394 |
+
"step": 68000
|
| 1395 |
+
},
|
| 1396 |
+
{
|
| 1397 |
+
"epoch": 54.8,
|
| 1398 |
+
"grad_norm": 2.6390535831451416,
|
| 1399 |
+
"learning_rate": 2.26e-05,
|
| 1400 |
+
"loss": 2.4389,
|
| 1401 |
+
"step": 68500
|
| 1402 |
+
},
|
| 1403 |
+
{
|
| 1404 |
+
"epoch": 55.0,
|
| 1405 |
+
"eval_loss": 2.508610725402832,
|
| 1406 |
+
"eval_runtime": 61.2599,
|
| 1407 |
+
"eval_samples_per_second": 163.239,
|
| 1408 |
+
"eval_steps_per_second": 5.109,
|
| 1409 |
+
"step": 68750
|
| 1410 |
+
},
|
| 1411 |
+
{
|
| 1412 |
+
"epoch": 55.2,
|
| 1413 |
+
"grad_norm": 2.766709566116333,
|
| 1414 |
+
"learning_rate": 2.2400000000000002e-05,
|
| 1415 |
+
"loss": 2.4248,
|
| 1416 |
+
"step": 69000
|
| 1417 |
+
},
|
| 1418 |
+
{
|
| 1419 |
+
"epoch": 55.6,
|
| 1420 |
+
"grad_norm": 2.9248626232147217,
|
| 1421 |
+
"learning_rate": 2.22e-05,
|
| 1422 |
+
"loss": 2.4301,
|
| 1423 |
+
"step": 69500
|
| 1424 |
+
},
|
| 1425 |
+
{
|
| 1426 |
+
"epoch": 56.0,
|
| 1427 |
+
"grad_norm": 3.262801170349121,
|
| 1428 |
+
"learning_rate": 2.2000000000000003e-05,
|
| 1429 |
+
"loss": 2.4225,
|
| 1430 |
+
"step": 70000
|
| 1431 |
+
},
|
| 1432 |
+
{
|
| 1433 |
+
"epoch": 56.0,
|
| 1434 |
+
"eval_loss": 2.5070488452911377,
|
| 1435 |
+
"eval_runtime": 61.2656,
|
| 1436 |
+
"eval_samples_per_second": 163.224,
|
| 1437 |
+
"eval_steps_per_second": 5.109,
|
| 1438 |
+
"step": 70000
|
| 1439 |
+
},
|
| 1440 |
+
{
|
| 1441 |
+
"epoch": 56.4,
|
| 1442 |
+
"grad_norm": 3.05898118019104,
|
| 1443 |
+
"learning_rate": 2.18e-05,
|
| 1444 |
+
"loss": 2.414,
|
| 1445 |
+
"step": 70500
|
| 1446 |
+
},
|
| 1447 |
+
{
|
| 1448 |
+
"epoch": 56.8,
|
| 1449 |
+
"grad_norm": 2.8624179363250732,
|
| 1450 |
+
"learning_rate": 2.16e-05,
|
| 1451 |
+
"loss": 2.4145,
|
| 1452 |
+
"step": 71000
|
| 1453 |
+
},
|
| 1454 |
+
{
|
| 1455 |
+
"epoch": 57.0,
|
| 1456 |
+
"eval_loss": 2.4903688430786133,
|
| 1457 |
+
"eval_runtime": 61.3344,
|
| 1458 |
+
"eval_samples_per_second": 163.041,
|
| 1459 |
+
"eval_steps_per_second": 5.103,
|
| 1460 |
+
"step": 71250
|
| 1461 |
+
},
|
| 1462 |
+
{
|
| 1463 |
+
"epoch": 57.2,
|
| 1464 |
+
"grad_norm": 2.8718996047973633,
|
| 1465 |
+
"learning_rate": 2.1400000000000002e-05,
|
| 1466 |
+
"loss": 2.4152,
|
| 1467 |
+
"step": 71500
|
| 1468 |
+
},
|
| 1469 |
+
{
|
| 1470 |
+
"epoch": 57.6,
|
| 1471 |
+
"grad_norm": 3.02909517288208,
|
| 1472 |
+
"learning_rate": 2.12e-05,
|
| 1473 |
+
"loss": 2.4107,
|
| 1474 |
+
"step": 72000
|
| 1475 |
+
},
|
| 1476 |
+
{
|
| 1477 |
+
"epoch": 58.0,
|
| 1478 |
+
"grad_norm": 3.092914581298828,
|
| 1479 |
+
"learning_rate": 2.1e-05,
|
| 1480 |
+
"loss": 2.4058,
|
| 1481 |
+
"step": 72500
|
| 1482 |
+
},
|
| 1483 |
+
{
|
| 1484 |
+
"epoch": 58.0,
|
| 1485 |
+
"eval_loss": 2.4901366233825684,
|
| 1486 |
+
"eval_runtime": 61.3186,
|
| 1487 |
+
"eval_samples_per_second": 163.083,
|
| 1488 |
+
"eval_steps_per_second": 5.104,
|
| 1489 |
+
"step": 72500
|
| 1490 |
+
},
|
| 1491 |
+
{
|
| 1492 |
+
"epoch": 58.4,
|
| 1493 |
+
"grad_norm": 3.0763566493988037,
|
| 1494 |
+
"learning_rate": 2.08e-05,
|
| 1495 |
+
"loss": 2.3941,
|
| 1496 |
+
"step": 73000
|
| 1497 |
+
},
|
| 1498 |
+
{
|
| 1499 |
+
"epoch": 58.8,
|
| 1500 |
+
"grad_norm": 2.809154748916626,
|
| 1501 |
+
"learning_rate": 2.06e-05,
|
| 1502 |
+
"loss": 2.398,
|
| 1503 |
+
"step": 73500
|
| 1504 |
+
},
|
| 1505 |
+
{
|
| 1506 |
+
"epoch": 59.0,
|
| 1507 |
+
"eval_loss": 2.491445302963257,
|
| 1508 |
+
"eval_runtime": 61.2316,
|
| 1509 |
+
"eval_samples_per_second": 163.314,
|
| 1510 |
+
"eval_steps_per_second": 5.112,
|
| 1511 |
+
"step": 73750
|
| 1512 |
+
},
|
| 1513 |
+
{
|
| 1514 |
+
"epoch": 59.2,
|
| 1515 |
+
"grad_norm": 2.948514699935913,
|
| 1516 |
+
"learning_rate": 2.04e-05,
|
| 1517 |
+
"loss": 2.3977,
|
| 1518 |
+
"step": 74000
|
| 1519 |
+
},
|
| 1520 |
+
{
|
| 1521 |
+
"epoch": 59.6,
|
| 1522 |
+
"grad_norm": 2.9708919525146484,
|
| 1523 |
+
"learning_rate": 2.0200000000000003e-05,
|
| 1524 |
+
"loss": 2.3849,
|
| 1525 |
+
"step": 74500
|
| 1526 |
+
},
|
| 1527 |
+
{
|
| 1528 |
+
"epoch": 60.0,
|
| 1529 |
+
"grad_norm": 3.0955567359924316,
|
| 1530 |
+
"learning_rate": 2e-05,
|
| 1531 |
+
"loss": 2.3847,
|
| 1532 |
+
"step": 75000
|
| 1533 |
+
},
|
| 1534 |
+
{
|
| 1535 |
+
"epoch": 60.0,
|
| 1536 |
+
"eval_loss": 2.481839418411255,
|
| 1537 |
+
"eval_runtime": 61.2301,
|
| 1538 |
+
"eval_samples_per_second": 163.318,
|
| 1539 |
+
"eval_steps_per_second": 5.112,
|
| 1540 |
+
"step": 75000
|
| 1541 |
+
},
|
| 1542 |
+
{
|
| 1543 |
+
"epoch": 60.4,
|
| 1544 |
+
"grad_norm": 2.8540799617767334,
|
| 1545 |
+
"learning_rate": 1.9800000000000004e-05,
|
| 1546 |
+
"loss": 2.3831,
|
| 1547 |
+
"step": 75500
|
| 1548 |
+
},
|
| 1549 |
+
{
|
| 1550 |
+
"epoch": 60.8,
|
| 1551 |
+
"grad_norm": 2.79695987701416,
|
| 1552 |
+
"learning_rate": 1.9600000000000002e-05,
|
| 1553 |
+
"loss": 2.3833,
|
| 1554 |
+
"step": 76000
|
| 1555 |
+
},
|
| 1556 |
+
{
|
| 1557 |
+
"epoch": 61.0,
|
| 1558 |
+
"eval_loss": 2.4663026332855225,
|
| 1559 |
+
"eval_runtime": 61.2384,
|
| 1560 |
+
"eval_samples_per_second": 163.296,
|
| 1561 |
+
"eval_steps_per_second": 5.111,
|
| 1562 |
+
"step": 76250
|
| 1563 |
+
},
|
| 1564 |
+
{
|
| 1565 |
+
"epoch": 61.2,
|
| 1566 |
+
"grad_norm": 3.144937515258789,
|
| 1567 |
+
"learning_rate": 1.94e-05,
|
| 1568 |
+
"loss": 2.3802,
|
| 1569 |
+
"step": 76500
|
| 1570 |
+
},
|
| 1571 |
+
{
|
| 1572 |
+
"epoch": 61.6,
|
| 1573 |
+
"grad_norm": 3.230477809906006,
|
| 1574 |
+
"learning_rate": 1.9200000000000003e-05,
|
| 1575 |
+
"loss": 2.369,
|
| 1576 |
+
"step": 77000
|
| 1577 |
+
},
|
| 1578 |
+
{
|
| 1579 |
+
"epoch": 62.0,
|
| 1580 |
+
"grad_norm": 3.04854416847229,
|
| 1581 |
+
"learning_rate": 1.9e-05,
|
| 1582 |
+
"loss": 2.3772,
|
| 1583 |
+
"step": 77500
|
| 1584 |
+
},
|
| 1585 |
+
{
|
| 1586 |
+
"epoch": 62.0,
|
| 1587 |
+
"eval_loss": 2.475891351699829,
|
| 1588 |
+
"eval_runtime": 61.2453,
|
| 1589 |
+
"eval_samples_per_second": 163.278,
|
| 1590 |
+
"eval_steps_per_second": 5.111,
|
| 1591 |
+
"step": 77500
|
| 1592 |
+
},
|
| 1593 |
+
{
|
| 1594 |
+
"epoch": 62.4,
|
| 1595 |
+
"grad_norm": 2.8289458751678467,
|
| 1596 |
+
"learning_rate": 1.88e-05,
|
| 1597 |
+
"loss": 2.3607,
|
| 1598 |
+
"step": 78000
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 62.8,
|
| 1602 |
+
"grad_norm": 2.9443745613098145,
|
| 1603 |
+
"learning_rate": 1.86e-05,
|
| 1604 |
+
"loss": 2.3711,
|
| 1605 |
+
"step": 78500
|
| 1606 |
+
},
|
| 1607 |
+
{
|
| 1608 |
+
"epoch": 63.0,
|
| 1609 |
+
"eval_loss": 2.45942759513855,
|
| 1610 |
+
"eval_runtime": 61.2995,
|
| 1611 |
+
"eval_samples_per_second": 163.133,
|
| 1612 |
+
"eval_steps_per_second": 5.106,
|
| 1613 |
+
"step": 78750
|
| 1614 |
+
},
|
| 1615 |
+
{
|
| 1616 |
+
"epoch": 63.2,
|
| 1617 |
+
"grad_norm": 3.4686219692230225,
|
| 1618 |
+
"learning_rate": 1.84e-05,
|
| 1619 |
+
"loss": 2.3635,
|
| 1620 |
+
"step": 79000
|
| 1621 |
+
},
|
| 1622 |
+
{
|
| 1623 |
+
"epoch": 63.6,
|
| 1624 |
+
"grad_norm": 2.9754679203033447,
|
| 1625 |
+
"learning_rate": 1.8200000000000002e-05,
|
| 1626 |
+
"loss": 2.3544,
|
| 1627 |
+
"step": 79500
|
| 1628 |
+
},
|
| 1629 |
+
{
|
| 1630 |
+
"epoch": 64.0,
|
| 1631 |
+
"grad_norm": 2.959327459335327,
|
| 1632 |
+
"learning_rate": 1.8e-05,
|
| 1633 |
+
"loss": 2.3603,
|
| 1634 |
+
"step": 80000
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 64.0,
|
| 1638 |
+
"eval_loss": 2.454479455947876,
|
| 1639 |
+
"eval_runtime": 61.2094,
|
| 1640 |
+
"eval_samples_per_second": 163.374,
|
| 1641 |
+
"eval_steps_per_second": 5.114,
|
| 1642 |
+
"step": 80000
|
| 1643 |
+
},
|
| 1644 |
+
{
|
| 1645 |
+
"epoch": 64.4,
|
| 1646 |
+
"grad_norm": 3.086289882659912,
|
| 1647 |
+
"learning_rate": 1.78e-05,
|
| 1648 |
+
"loss": 2.348,
|
| 1649 |
+
"step": 80500
|
| 1650 |
+
},
|
| 1651 |
+
{
|
| 1652 |
+
"epoch": 64.8,
|
| 1653 |
+
"grad_norm": 2.840416193008423,
|
| 1654 |
+
"learning_rate": 1.76e-05,
|
| 1655 |
+
"loss": 2.3546,
|
| 1656 |
+
"step": 81000
|
| 1657 |
+
},
|
| 1658 |
+
{
|
| 1659 |
+
"epoch": 65.0,
|
| 1660 |
+
"eval_loss": 2.4469666481018066,
|
| 1661 |
+
"eval_runtime": 61.2942,
|
| 1662 |
+
"eval_samples_per_second": 163.147,
|
| 1663 |
+
"eval_steps_per_second": 5.107,
|
| 1664 |
+
"step": 81250
|
| 1665 |
+
},
|
| 1666 |
+
{
|
| 1667 |
+
"epoch": 65.2,
|
| 1668 |
+
"grad_norm": 3.062312364578247,
|
| 1669 |
+
"learning_rate": 1.74e-05,
|
| 1670 |
+
"loss": 2.3485,
|
| 1671 |
+
"step": 81500
|
| 1672 |
+
},
|
| 1673 |
+
{
|
| 1674 |
+
"epoch": 65.6,
|
| 1675 |
+
"grad_norm": 2.8818204402923584,
|
| 1676 |
+
"learning_rate": 1.7199999999999998e-05,
|
| 1677 |
+
"loss": 2.3434,
|
| 1678 |
+
"step": 82000
|
| 1679 |
+
},
|
| 1680 |
+
{
|
| 1681 |
+
"epoch": 66.0,
|
| 1682 |
+
"grad_norm": 2.827214002609253,
|
| 1683 |
+
"learning_rate": 1.7000000000000003e-05,
|
| 1684 |
+
"loss": 2.3497,
|
| 1685 |
+
"step": 82500
|
| 1686 |
+
},
|
| 1687 |
+
{
|
| 1688 |
+
"epoch": 66.0,
|
| 1689 |
+
"eval_loss": 2.4515304565429688,
|
| 1690 |
+
"eval_runtime": 61.2229,
|
| 1691 |
+
"eval_samples_per_second": 163.338,
|
| 1692 |
+
"eval_steps_per_second": 5.112,
|
| 1693 |
+
"step": 82500
|
| 1694 |
+
},
|
| 1695 |
+
{
|
| 1696 |
+
"epoch": 66.4,
|
| 1697 |
+
"grad_norm": 2.871967315673828,
|
| 1698 |
+
"learning_rate": 1.6800000000000002e-05,
|
| 1699 |
+
"loss": 2.3415,
|
| 1700 |
+
"step": 83000
|
| 1701 |
+
},
|
| 1702 |
+
{
|
| 1703 |
+
"epoch": 66.8,
|
| 1704 |
+
"grad_norm": 3.2059788703918457,
|
| 1705 |
+
"learning_rate": 1.66e-05,
|
| 1706 |
+
"loss": 2.3328,
|
| 1707 |
+
"step": 83500
|
| 1708 |
+
},
|
| 1709 |
+
{
|
| 1710 |
+
"epoch": 67.0,
|
| 1711 |
+
"eval_loss": 2.4459116458892822,
|
| 1712 |
+
"eval_runtime": 61.2313,
|
| 1713 |
+
"eval_samples_per_second": 163.315,
|
| 1714 |
+
"eval_steps_per_second": 5.112,
|
| 1715 |
+
"step": 83750
|
| 1716 |
+
},
|
| 1717 |
+
{
|
| 1718 |
+
"epoch": 67.2,
|
| 1719 |
+
"grad_norm": 2.958317518234253,
|
| 1720 |
+
"learning_rate": 1.6400000000000002e-05,
|
| 1721 |
+
"loss": 2.3355,
|
| 1722 |
+
"step": 84000
|
| 1723 |
+
},
|
| 1724 |
+
{
|
| 1725 |
+
"epoch": 67.6,
|
| 1726 |
+
"grad_norm": 3.2304327487945557,
|
| 1727 |
+
"learning_rate": 1.62e-05,
|
| 1728 |
+
"loss": 2.329,
|
| 1729 |
+
"step": 84500
|
| 1730 |
+
},
|
| 1731 |
+
{
|
| 1732 |
+
"epoch": 68.0,
|
| 1733 |
+
"grad_norm": 3.0539510250091553,
|
| 1734 |
+
"learning_rate": 1.6000000000000003e-05,
|
| 1735 |
+
"loss": 2.3281,
|
| 1736 |
+
"step": 85000
|
| 1737 |
+
},
|
| 1738 |
+
{
|
| 1739 |
+
"epoch": 68.0,
|
| 1740 |
+
"eval_loss": 2.4377968311309814,
|
| 1741 |
+
"eval_runtime": 61.2555,
|
| 1742 |
+
"eval_samples_per_second": 163.251,
|
| 1743 |
+
"eval_steps_per_second": 5.11,
|
| 1744 |
+
"step": 85000
|
| 1745 |
+
},
|
| 1746 |
+
{
|
| 1747 |
+
"epoch": 68.4,
|
| 1748 |
+
"grad_norm": 3.338268995285034,
|
| 1749 |
+
"learning_rate": 1.58e-05,
|
| 1750 |
+
"loss": 2.3221,
|
| 1751 |
+
"step": 85500
|
| 1752 |
+
},
|
| 1753 |
+
{
|
| 1754 |
+
"epoch": 68.8,
|
| 1755 |
+
"grad_norm": 3.2005858421325684,
|
| 1756 |
+
"learning_rate": 1.56e-05,
|
| 1757 |
+
"loss": 2.3272,
|
| 1758 |
+
"step": 86000
|
| 1759 |
+
},
|
| 1760 |
+
{
|
| 1761 |
+
"epoch": 69.0,
|
| 1762 |
+
"eval_loss": 2.4320311546325684,
|
| 1763 |
+
"eval_runtime": 61.3445,
|
| 1764 |
+
"eval_samples_per_second": 163.014,
|
| 1765 |
+
"eval_steps_per_second": 5.102,
|
| 1766 |
+
"step": 86250
|
| 1767 |
+
},
|
| 1768 |
+
{
|
| 1769 |
+
"epoch": 69.2,
|
| 1770 |
+
"grad_norm": 3.010361671447754,
|
| 1771 |
+
"learning_rate": 1.54e-05,
|
| 1772 |
+
"loss": 2.312,
|
| 1773 |
+
"step": 86500
|
| 1774 |
+
},
|
| 1775 |
+
{
|
| 1776 |
+
"epoch": 69.6,
|
| 1777 |
+
"grad_norm": 3.110023021697998,
|
| 1778 |
+
"learning_rate": 1.52e-05,
|
| 1779 |
+
"loss": 2.3175,
|
| 1780 |
+
"step": 87000
|
| 1781 |
+
},
|
| 1782 |
+
{
|
| 1783 |
+
"epoch": 70.0,
|
| 1784 |
+
"grad_norm": 2.972137212753296,
|
| 1785 |
+
"learning_rate": 1.5e-05,
|
| 1786 |
+
"loss": 2.3172,
|
| 1787 |
+
"step": 87500
|
| 1788 |
+
},
|
| 1789 |
+
{
|
| 1790 |
+
"epoch": 70.0,
|
| 1791 |
+
"eval_loss": 2.428523063659668,
|
| 1792 |
+
"eval_runtime": 61.2024,
|
| 1793 |
+
"eval_samples_per_second": 163.392,
|
| 1794 |
+
"eval_steps_per_second": 5.114,
|
| 1795 |
+
"step": 87500
|
| 1796 |
+
},
|
| 1797 |
+
{
|
| 1798 |
+
"epoch": 70.4,
|
| 1799 |
+
"grad_norm": 2.8720614910125732,
|
| 1800 |
+
"learning_rate": 1.48e-05,
|
| 1801 |
+
"loss": 2.3103,
|
| 1802 |
+
"step": 88000
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 70.8,
|
| 1806 |
+
"grad_norm": 2.9163401126861572,
|
| 1807 |
+
"learning_rate": 1.4599999999999999e-05,
|
| 1808 |
+
"loss": 2.3128,
|
| 1809 |
+
"step": 88500
|
| 1810 |
+
},
|
| 1811 |
+
{
|
| 1812 |
+
"epoch": 71.0,
|
| 1813 |
+
"eval_loss": 2.4292280673980713,
|
| 1814 |
+
"eval_runtime": 61.2279,
|
| 1815 |
+
"eval_samples_per_second": 163.324,
|
| 1816 |
+
"eval_steps_per_second": 5.112,
|
| 1817 |
+
"step": 88750
|
| 1818 |
+
},
|
| 1819 |
+
{
|
| 1820 |
+
"epoch": 71.2,
|
| 1821 |
+
"grad_norm": 2.890698194503784,
|
| 1822 |
+
"learning_rate": 1.44e-05,
|
| 1823 |
+
"loss": 2.304,
|
| 1824 |
+
"step": 89000
|
| 1825 |
+
},
|
| 1826 |
+
{
|
| 1827 |
+
"epoch": 71.6,
|
| 1828 |
+
"grad_norm": 3.2436020374298096,
|
| 1829 |
+
"learning_rate": 1.42e-05,
|
| 1830 |
+
"loss": 2.3067,
|
| 1831 |
+
"step": 89500
|
| 1832 |
+
},
|
| 1833 |
+
{
|
| 1834 |
+
"epoch": 72.0,
|
| 1835 |
+
"grad_norm": 2.849823236465454,
|
| 1836 |
+
"learning_rate": 1.4000000000000001e-05,
|
| 1837 |
+
"loss": 2.3082,
|
| 1838 |
+
"step": 90000
|
| 1839 |
+
},
|
| 1840 |
+
{
|
| 1841 |
+
"epoch": 72.0,
|
| 1842 |
+
"eval_loss": 2.41927170753479,
|
| 1843 |
+
"eval_runtime": 61.2261,
|
| 1844 |
+
"eval_samples_per_second": 163.329,
|
| 1845 |
+
"eval_steps_per_second": 5.112,
|
| 1846 |
+
"step": 90000
|
| 1847 |
+
},
|
| 1848 |
+
{
|
| 1849 |
+
"epoch": 72.4,
|
| 1850 |
+
"grad_norm": 3.191131353378296,
|
| 1851 |
+
"learning_rate": 1.3800000000000002e-05,
|
| 1852 |
+
"loss": 2.2925,
|
| 1853 |
+
"step": 90500
|
| 1854 |
+
},
|
| 1855 |
+
{
|
| 1856 |
+
"epoch": 72.8,
|
| 1857 |
+
"grad_norm": 3.203021764755249,
|
| 1858 |
+
"learning_rate": 1.3600000000000002e-05,
|
| 1859 |
+
"loss": 2.3004,
|
| 1860 |
+
"step": 91000
|
| 1861 |
+
},
|
| 1862 |
+
{
|
| 1863 |
+
"epoch": 73.0,
|
| 1864 |
+
"eval_loss": 2.4278526306152344,
|
| 1865 |
+
"eval_runtime": 61.2358,
|
| 1866 |
+
"eval_samples_per_second": 163.303,
|
| 1867 |
+
"eval_steps_per_second": 5.111,
|
| 1868 |
+
"step": 91250
|
| 1869 |
+
},
|
| 1870 |
+
{
|
| 1871 |
+
"epoch": 73.2,
|
| 1872 |
+
"grad_norm": 3.002692461013794,
|
| 1873 |
+
"learning_rate": 1.3400000000000002e-05,
|
| 1874 |
+
"loss": 2.2949,
|
| 1875 |
+
"step": 91500
|
| 1876 |
+
},
|
| 1877 |
+
{
|
| 1878 |
+
"epoch": 73.6,
|
| 1879 |
+
"grad_norm": 3.137084722518921,
|
| 1880 |
+
"learning_rate": 1.32e-05,
|
| 1881 |
+
"loss": 2.2898,
|
| 1882 |
+
"step": 92000
|
| 1883 |
+
},
|
| 1884 |
+
{
|
| 1885 |
+
"epoch": 74.0,
|
| 1886 |
+
"grad_norm": 3.325843095779419,
|
| 1887 |
+
"learning_rate": 1.3000000000000001e-05,
|
| 1888 |
+
"loss": 2.2955,
|
| 1889 |
+
"step": 92500
|
| 1890 |
+
},
|
| 1891 |
+
{
|
| 1892 |
+
"epoch": 74.0,
|
| 1893 |
+
"eval_loss": 2.409475326538086,
|
| 1894 |
+
"eval_runtime": 61.3551,
|
| 1895 |
+
"eval_samples_per_second": 162.986,
|
| 1896 |
+
"eval_steps_per_second": 5.101,
|
| 1897 |
+
"step": 92500
|
| 1898 |
+
},
|
| 1899 |
+
{
|
| 1900 |
+
"epoch": 74.4,
|
| 1901 |
+
"grad_norm": 3.207425594329834,
|
| 1902 |
+
"learning_rate": 1.2800000000000001e-05,
|
| 1903 |
+
"loss": 2.2895,
|
| 1904 |
+
"step": 93000
|
| 1905 |
+
},
|
| 1906 |
+
{
|
| 1907 |
+
"epoch": 74.8,
|
| 1908 |
+
"grad_norm": 3.2221837043762207,
|
| 1909 |
+
"learning_rate": 1.2600000000000001e-05,
|
| 1910 |
+
"loss": 2.2849,
|
| 1911 |
+
"step": 93500
|
| 1912 |
+
},
|
| 1913 |
+
{
|
| 1914 |
+
"epoch": 75.0,
|
| 1915 |
+
"eval_loss": 2.4111411571502686,
|
| 1916 |
+
"eval_runtime": 61.2648,
|
| 1917 |
+
"eval_samples_per_second": 163.226,
|
| 1918 |
+
"eval_steps_per_second": 5.109,
|
| 1919 |
+
"step": 93750
|
| 1920 |
+
},
|
| 1921 |
+
{
|
| 1922 |
+
"epoch": 75.2,
|
| 1923 |
+
"grad_norm": 3.0594756603240967,
|
| 1924 |
+
"learning_rate": 1.24e-05,
|
| 1925 |
+
"loss": 2.2791,
|
| 1926 |
+
"step": 94000
|
| 1927 |
+
},
|
| 1928 |
+
{
|
| 1929 |
+
"epoch": 75.6,
|
| 1930 |
+
"grad_norm": 3.013936996459961,
|
| 1931 |
+
"learning_rate": 1.22e-05,
|
| 1932 |
+
"loss": 2.2935,
|
| 1933 |
+
"step": 94500
|
| 1934 |
+
},
|
| 1935 |
+
{
|
| 1936 |
+
"epoch": 76.0,
|
| 1937 |
+
"grad_norm": 3.296487808227539,
|
| 1938 |
+
"learning_rate": 1.2e-05,
|
| 1939 |
+
"loss": 2.2816,
|
| 1940 |
+
"step": 95000
|
| 1941 |
+
},
|
| 1942 |
+
{
|
| 1943 |
+
"epoch": 76.0,
|
| 1944 |
+
"eval_loss": 2.405331611633301,
|
| 1945 |
+
"eval_runtime": 61.2866,
|
| 1946 |
+
"eval_samples_per_second": 163.168,
|
| 1947 |
+
"eval_steps_per_second": 5.107,
|
| 1948 |
+
"step": 95000
|
| 1949 |
+
}
|
| 1950 |
+
],
|
| 1951 |
+
"logging_steps": 500,
|
| 1952 |
+
"max_steps": 125000,
|
| 1953 |
+
"num_input_tokens_seen": 0,
|
| 1954 |
+
"num_train_epochs": 100,
|
| 1955 |
+
"save_steps": 500,
|
| 1956 |
+
"stateful_callbacks": {
|
| 1957 |
+
"EarlyStoppingCallback": {
|
| 1958 |
+
"args": {
|
| 1959 |
+
"early_stopping_patience": 2,
|
| 1960 |
+
"early_stopping_threshold": 0.0
|
| 1961 |
+
},
|
| 1962 |
+
"attributes": {
|
| 1963 |
+
"early_stopping_patience_counter": 0
|
| 1964 |
+
}
|
| 1965 |
+
},
|
| 1966 |
+
"TrainerControl": {
|
| 1967 |
+
"args": {
|
| 1968 |
+
"should_epoch_stop": false,
|
| 1969 |
+
"should_evaluate": false,
|
| 1970 |
+
"should_log": false,
|
| 1971 |
+
"should_save": true,
|
| 1972 |
+
"should_training_stop": false
|
| 1973 |
+
},
|
| 1974 |
+
"attributes": {}
|
| 1975 |
+
}
|
| 1976 |
+
},
|
| 1977 |
+
"total_flos": 4.0316781723648e+17,
|
| 1978 |
+
"train_batch_size": 32,
|
| 1979 |
+
"trial_name": null,
|
| 1980 |
+
"trial_params": null
|
| 1981 |
+
}
|
checkpoint-95000/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5d3d02d48e9e4591dab8aa13e7587cd64d29bd125855f9fb9651702a72f0b64b
|
| 3 |
+
size 5304
|
checkpoint-95000/vocab.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|