Upload model files
Browse files- config.json +40 -0
- optimizer.pt +3 -0
- pytorch_model.bin +3 -0
- rng_state.pth +3 -0
- scheduler.pt +3 -0
- special_tokens_map.json +7 -0
- tokenizer_config.json +16 -0
- trainer_state.json +1716 -0
- training_args.bin +3 -0
- vocab.txt +0 -0
config.json
ADDED
|
@@ -0,0 +1,40 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "distilbert-base-multilingual-cased",
|
| 3 |
+
"activation": "gelu",
|
| 4 |
+
"architectures": [
|
| 5 |
+
"DistilBertForSequenceClassification"
|
| 6 |
+
],
|
| 7 |
+
"attention_dropout": 0.1,
|
| 8 |
+
"dim": 768,
|
| 9 |
+
"dropout": 0.1,
|
| 10 |
+
"hidden_dim": 3072,
|
| 11 |
+
"id2label": {
|
| 12 |
+
"0": "LABEL_0",
|
| 13 |
+
"1": "LABEL_1",
|
| 14 |
+
"2": "LABEL_2",
|
| 15 |
+
"3": "LABEL_3",
|
| 16 |
+
"4": "LABEL_4"
|
| 17 |
+
},
|
| 18 |
+
"initializer_range": 0.02,
|
| 19 |
+
"label2id": {
|
| 20 |
+
"LABEL_0": 0,
|
| 21 |
+
"LABEL_1": 1,
|
| 22 |
+
"LABEL_2": 2,
|
| 23 |
+
"LABEL_3": 3,
|
| 24 |
+
"LABEL_4": 4
|
| 25 |
+
},
|
| 26 |
+
"max_position_embeddings": 512,
|
| 27 |
+
"model_type": "distilbert",
|
| 28 |
+
"n_heads": 12,
|
| 29 |
+
"n_layers": 6,
|
| 30 |
+
"output_past": true,
|
| 31 |
+
"pad_token_id": 0,
|
| 32 |
+
"problem_type": "multi_label_classification",
|
| 33 |
+
"qa_dropout": 0.1,
|
| 34 |
+
"seq_classif_dropout": 0.2,
|
| 35 |
+
"sinusoidal_pos_embds": false,
|
| 36 |
+
"tie_weights_": true,
|
| 37 |
+
"torch_dtype": "float32",
|
| 38 |
+
"transformers_version": "4.24.0",
|
| 39 |
+
"vocab_size": 119547
|
| 40 |
+
}
|
optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a45e52ce05a6b657f75f0299c91bab859bc971f8a7c99e1f64d9ef5d15e42802
|
| 3 |
+
size 1082689093
|
pytorch_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:236930bc12eb148be077da5f6ff89e372954e5ae12617ce8ab3fdffdab6de29d
|
| 3 |
+
size 541349549
|
rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0f6353bc2fe2d7fef44b3357b6dbe6ee592d69d4ca2b101d397fbdb26b8d8a2f
|
| 3 |
+
size 14575
|
scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e134097bca81c3928fe283a7a9968ba629990becdb2bd3c94f6ae3538f682f2f
|
| 3 |
+
size 627
|
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 |
+
}
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"cls_token": "[CLS]",
|
| 3 |
+
"do_basic_tokenize": true,
|
| 4 |
+
"do_lower_case": false,
|
| 5 |
+
"mask_token": "[MASK]",
|
| 6 |
+
"model_max_length": 512,
|
| 7 |
+
"name_or_path": "distilbert-base-multilingual-cased",
|
| 8 |
+
"never_split": null,
|
| 9 |
+
"pad_token": "[PAD]",
|
| 10 |
+
"sep_token": "[SEP]",
|
| 11 |
+
"special_tokens_map_file": null,
|
| 12 |
+
"strip_accents": null,
|
| 13 |
+
"tokenize_chinese_chars": true,
|
| 14 |
+
"tokenizer_class": "DistilBertTokenizer",
|
| 15 |
+
"unk_token": "[UNK]"
|
| 16 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1716 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.7120954003407156,
|
| 3 |
+
"best_model_checkpoint": "Distil4\\checkpoint-40348",
|
| 4 |
+
"epoch": 100.0,
|
| 5 |
+
"global_step": 52400,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 1.0,
|
| 12 |
+
"learning_rate": 4.950095419847329e-06,
|
| 13 |
+
"loss": 0.4749,
|
| 14 |
+
"step": 523
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 1.0,
|
| 18 |
+
"eval_accuracy": 0.4026503567787971,
|
| 19 |
+
"eval_f1": 0.5392528424472117,
|
| 20 |
+
"eval_loss": 0.4015094041824341,
|
| 21 |
+
"eval_roc_auc": 0.6891349750981157,
|
| 22 |
+
"eval_runtime": 9.4449,
|
| 23 |
+
"eval_samples_per_second": 103.866,
|
| 24 |
+
"eval_steps_per_second": 6.988,
|
| 25 |
+
"step": 524
|
| 26 |
+
},
|
| 27 |
+
{
|
| 28 |
+
"epoch": 2.0,
|
| 29 |
+
"learning_rate": 4.900190839694656e-06,
|
| 30 |
+
"loss": 0.3845,
|
| 31 |
+
"step": 1046
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 2.0,
|
| 35 |
+
"eval_accuracy": 0.5168195718654435,
|
| 36 |
+
"eval_f1": 0.6353754940711462,
|
| 37 |
+
"eval_loss": 0.36452510952949524,
|
| 38 |
+
"eval_roc_auc": 0.747088659888458,
|
| 39 |
+
"eval_runtime": 9.9987,
|
| 40 |
+
"eval_samples_per_second": 98.113,
|
| 41 |
+
"eval_steps_per_second": 6.601,
|
| 42 |
+
"step": 1048
|
| 43 |
+
},
|
| 44 |
+
{
|
| 45 |
+
"epoch": 2.99,
|
| 46 |
+
"learning_rate": 4.850286259541985e-06,
|
| 47 |
+
"loss": 0.3475,
|
| 48 |
+
"step": 1569
|
| 49 |
+
},
|
| 50 |
+
{
|
| 51 |
+
"epoch": 3.0,
|
| 52 |
+
"eval_accuracy": 0.5310907237512742,
|
| 53 |
+
"eval_f1": 0.6470871449205585,
|
| 54 |
+
"eval_loss": 0.3549170196056366,
|
| 55 |
+
"eval_roc_auc": 0.7563264556700557,
|
| 56 |
+
"eval_runtime": 10.07,
|
| 57 |
+
"eval_samples_per_second": 97.418,
|
| 58 |
+
"eval_steps_per_second": 6.554,
|
| 59 |
+
"step": 1572
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 3.99,
|
| 63 |
+
"learning_rate": 4.800381679389313e-06,
|
| 64 |
+
"loss": 0.326,
|
| 65 |
+
"step": 2092
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 4.0,
|
| 69 |
+
"eval_accuracy": 0.5545361875637105,
|
| 70 |
+
"eval_f1": 0.674766355140187,
|
| 71 |
+
"eval_loss": 0.3448670506477356,
|
| 72 |
+
"eval_roc_auc": 0.7760477152234284,
|
| 73 |
+
"eval_runtime": 10.1448,
|
| 74 |
+
"eval_samples_per_second": 96.7,
|
| 75 |
+
"eval_steps_per_second": 6.506,
|
| 76 |
+
"step": 2096
|
| 77 |
+
},
|
| 78 |
+
{
|
| 79 |
+
"epoch": 4.99,
|
| 80 |
+
"learning_rate": 4.750477099236642e-06,
|
| 81 |
+
"loss": 0.3083,
|
| 82 |
+
"step": 2615
|
| 83 |
+
},
|
| 84 |
+
{
|
| 85 |
+
"epoch": 5.0,
|
| 86 |
+
"eval_accuracy": 0.5565749235474006,
|
| 87 |
+
"eval_f1": 0.6812674743709226,
|
| 88 |
+
"eval_loss": 0.34419235587120056,
|
| 89 |
+
"eval_roc_auc": 0.7803114456863509,
|
| 90 |
+
"eval_runtime": 10.1716,
|
| 91 |
+
"eval_samples_per_second": 96.445,
|
| 92 |
+
"eval_steps_per_second": 6.489,
|
| 93 |
+
"step": 2620
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 5.99,
|
| 97 |
+
"learning_rate": 4.700572519083969e-06,
|
| 98 |
+
"loss": 0.2952,
|
| 99 |
+
"step": 3138
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 6.0,
|
| 103 |
+
"eval_accuracy": 0.5718654434250765,
|
| 104 |
+
"eval_f1": 0.6965452847805789,
|
| 105 |
+
"eval_loss": 0.3325030505657196,
|
| 106 |
+
"eval_roc_auc": 0.789289320878566,
|
| 107 |
+
"eval_runtime": 10.2795,
|
| 108 |
+
"eval_samples_per_second": 95.433,
|
| 109 |
+
"eval_steps_per_second": 6.421,
|
| 110 |
+
"step": 3144
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 6.99,
|
| 114 |
+
"learning_rate": 4.650667938931298e-06,
|
| 115 |
+
"loss": 0.2799,
|
| 116 |
+
"step": 3661
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 7.0,
|
| 120 |
+
"eval_accuracy": 0.5728848114169215,
|
| 121 |
+
"eval_f1": 0.6865116279069767,
|
| 122 |
+
"eval_loss": 0.34394344687461853,
|
| 123 |
+
"eval_roc_auc": 0.7837168070505612,
|
| 124 |
+
"eval_runtime": 10.3609,
|
| 125 |
+
"eval_samples_per_second": 94.683,
|
| 126 |
+
"eval_steps_per_second": 6.37,
|
| 127 |
+
"step": 3668
|
| 128 |
+
},
|
| 129 |
+
{
|
| 130 |
+
"epoch": 7.98,
|
| 131 |
+
"learning_rate": 4.600763358778627e-06,
|
| 132 |
+
"loss": 0.2626,
|
| 133 |
+
"step": 4184
|
| 134 |
+
},
|
| 135 |
+
{
|
| 136 |
+
"epoch": 8.0,
|
| 137 |
+
"eval_accuracy": 0.5800203873598369,
|
| 138 |
+
"eval_f1": 0.692876965772433,
|
| 139 |
+
"eval_loss": 0.33631590008735657,
|
| 140 |
+
"eval_roc_auc": 0.7883041472539074,
|
| 141 |
+
"eval_runtime": 9.1916,
|
| 142 |
+
"eval_samples_per_second": 106.728,
|
| 143 |
+
"eval_steps_per_second": 7.18,
|
| 144 |
+
"step": 4192
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 8.98,
|
| 148 |
+
"learning_rate": 4.550858778625955e-06,
|
| 149 |
+
"loss": 0.2529,
|
| 150 |
+
"step": 4707
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 9.0,
|
| 154 |
+
"eval_accuracy": 0.5749235474006116,
|
| 155 |
+
"eval_f1": 0.6857670979667283,
|
| 156 |
+
"eval_loss": 0.34780624508857727,
|
| 157 |
+
"eval_roc_auc": 0.7840966468522641,
|
| 158 |
+
"eval_runtime": 8.7207,
|
| 159 |
+
"eval_samples_per_second": 112.491,
|
| 160 |
+
"eval_steps_per_second": 7.568,
|
| 161 |
+
"step": 4716
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"epoch": 9.98,
|
| 165 |
+
"learning_rate": 4.500954198473283e-06,
|
| 166 |
+
"loss": 0.237,
|
| 167 |
+
"step": 5230
|
| 168 |
+
},
|
| 169 |
+
{
|
| 170 |
+
"epoch": 10.0,
|
| 171 |
+
"eval_accuracy": 0.5667686034658511,
|
| 172 |
+
"eval_f1": 0.68721251149954,
|
| 173 |
+
"eval_loss": 0.34721097350120544,
|
| 174 |
+
"eval_roc_auc": 0.7855741204011842,
|
| 175 |
+
"eval_runtime": 8.7533,
|
| 176 |
+
"eval_samples_per_second": 112.072,
|
| 177 |
+
"eval_steps_per_second": 7.54,
|
| 178 |
+
"step": 5240
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 10.98,
|
| 182 |
+
"learning_rate": 4.451049618320611e-06,
|
| 183 |
+
"loss": 0.2223,
|
| 184 |
+
"step": 5753
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 11.0,
|
| 188 |
+
"eval_accuracy": 0.5657492354740061,
|
| 189 |
+
"eval_f1": 0.6845698680018207,
|
| 190 |
+
"eval_loss": 0.3609465956687927,
|
| 191 |
+
"eval_roc_auc": 0.785313626035666,
|
| 192 |
+
"eval_runtime": 8.7365,
|
| 193 |
+
"eval_samples_per_second": 112.288,
|
| 194 |
+
"eval_steps_per_second": 7.555,
|
| 195 |
+
"step": 5764
|
| 196 |
+
},
|
| 197 |
+
{
|
| 198 |
+
"epoch": 11.98,
|
| 199 |
+
"learning_rate": 4.40114503816794e-06,
|
| 200 |
+
"loss": 0.2117,
|
| 201 |
+
"step": 6276
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 12.0,
|
| 205 |
+
"eval_accuracy": 0.581039755351682,
|
| 206 |
+
"eval_f1": 0.6981818181818182,
|
| 207 |
+
"eval_loss": 0.36286601424217224,
|
| 208 |
+
"eval_roc_auc": 0.7939185467398039,
|
| 209 |
+
"eval_runtime": 8.7731,
|
| 210 |
+
"eval_samples_per_second": 111.819,
|
| 211 |
+
"eval_steps_per_second": 7.523,
|
| 212 |
+
"step": 6288
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 12.98,
|
| 216 |
+
"learning_rate": 4.351240458015267e-06,
|
| 217 |
+
"loss": 0.2,
|
| 218 |
+
"step": 6799
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 13.0,
|
| 222 |
+
"eval_accuracy": 0.5851172273190621,
|
| 223 |
+
"eval_f1": 0.7062999112688554,
|
| 224 |
+
"eval_loss": 0.3647877275943756,
|
| 225 |
+
"eval_roc_auc": 0.8024597782929012,
|
| 226 |
+
"eval_runtime": 8.7204,
|
| 227 |
+
"eval_samples_per_second": 112.495,
|
| 228 |
+
"eval_steps_per_second": 7.568,
|
| 229 |
+
"step": 6812
|
| 230 |
+
},
|
| 231 |
+
{
|
| 232 |
+
"epoch": 13.97,
|
| 233 |
+
"learning_rate": 4.301335877862596e-06,
|
| 234 |
+
"loss": 0.1909,
|
| 235 |
+
"step": 7322
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 14.0,
|
| 239 |
+
"eval_accuracy": 0.5698267074413863,
|
| 240 |
+
"eval_f1": 0.7012058954890575,
|
| 241 |
+
"eval_loss": 0.3725411891937256,
|
| 242 |
+
"eval_roc_auc": 0.7982735076082716,
|
| 243 |
+
"eval_runtime": 8.798,
|
| 244 |
+
"eval_samples_per_second": 111.503,
|
| 245 |
+
"eval_steps_per_second": 7.502,
|
| 246 |
+
"step": 7336
|
| 247 |
+
},
|
| 248 |
+
{
|
| 249 |
+
"epoch": 14.97,
|
| 250 |
+
"learning_rate": 4.2514312977099246e-06,
|
| 251 |
+
"loss": 0.1782,
|
| 252 |
+
"step": 7845
|
| 253 |
+
},
|
| 254 |
+
{
|
| 255 |
+
"epoch": 15.0,
|
| 256 |
+
"eval_accuracy": 0.5800203873598369,
|
| 257 |
+
"eval_f1": 0.7042128603104212,
|
| 258 |
+
"eval_loss": 0.37900835275650024,
|
| 259 |
+
"eval_roc_auc": 0.8012003396754722,
|
| 260 |
+
"eval_runtime": 8.7328,
|
| 261 |
+
"eval_samples_per_second": 112.335,
|
| 262 |
+
"eval_steps_per_second": 7.558,
|
| 263 |
+
"step": 7860
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 15.97,
|
| 267 |
+
"learning_rate": 4.2015267175572526e-06,
|
| 268 |
+
"loss": 0.1653,
|
| 269 |
+
"step": 8368
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 16.0,
|
| 273 |
+
"eval_accuracy": 0.5739041794087666,
|
| 274 |
+
"eval_f1": 0.7040141155712396,
|
| 275 |
+
"eval_loss": 0.3948748707771301,
|
| 276 |
+
"eval_roc_auc": 0.8018475591563196,
|
| 277 |
+
"eval_runtime": 8.6978,
|
| 278 |
+
"eval_samples_per_second": 112.788,
|
| 279 |
+
"eval_steps_per_second": 7.588,
|
| 280 |
+
"step": 8384
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 16.97,
|
| 284 |
+
"learning_rate": 4.1516221374045806e-06,
|
| 285 |
+
"loss": 0.156,
|
| 286 |
+
"step": 8891
|
| 287 |
+
},
|
| 288 |
+
{
|
| 289 |
+
"epoch": 17.0,
|
| 290 |
+
"eval_accuracy": 0.5769622833843018,
|
| 291 |
+
"eval_f1": 0.7074468085106383,
|
| 292 |
+
"eval_loss": 0.39961278438568115,
|
| 293 |
+
"eval_roc_auc": 0.8033181473916137,
|
| 294 |
+
"eval_runtime": 8.6958,
|
| 295 |
+
"eval_samples_per_second": 112.813,
|
| 296 |
+
"eval_steps_per_second": 7.59,
|
| 297 |
+
"step": 8908
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 17.97,
|
| 301 |
+
"learning_rate": 4.1017175572519085e-06,
|
| 302 |
+
"loss": 0.1478,
|
| 303 |
+
"step": 9414
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 18.0,
|
| 307 |
+
"eval_accuracy": 0.5575942915392457,
|
| 308 |
+
"eval_f1": 0.6918402777777779,
|
| 309 |
+
"eval_loss": 0.41351279616355896,
|
| 310 |
+
"eval_roc_auc": 0.79633816070322,
|
| 311 |
+
"eval_runtime": 8.7585,
|
| 312 |
+
"eval_samples_per_second": 112.006,
|
| 313 |
+
"eval_steps_per_second": 7.536,
|
| 314 |
+
"step": 9432
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 18.96,
|
| 318 |
+
"learning_rate": 4.051812977099237e-06,
|
| 319 |
+
"loss": 0.1362,
|
| 320 |
+
"step": 9937
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 19.0,
|
| 324 |
+
"eval_accuracy": 0.5616717635066258,
|
| 325 |
+
"eval_f1": 0.7035263387026557,
|
| 326 |
+
"eval_loss": 0.4201168119907379,
|
| 327 |
+
"eval_roc_auc": 0.8034656078584379,
|
| 328 |
+
"eval_runtime": 8.7893,
|
| 329 |
+
"eval_samples_per_second": 111.613,
|
| 330 |
+
"eval_steps_per_second": 7.509,
|
| 331 |
+
"step": 9956
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 19.96,
|
| 335 |
+
"learning_rate": 4.001908396946565e-06,
|
| 336 |
+
"loss": 0.1316,
|
| 337 |
+
"step": 10460
|
| 338 |
+
},
|
| 339 |
+
{
|
| 340 |
+
"epoch": 20.0,
|
| 341 |
+
"eval_accuracy": 0.54638124362895,
|
| 342 |
+
"eval_f1": 0.6954270923209663,
|
| 343 |
+
"eval_loss": 0.4302760362625122,
|
| 344 |
+
"eval_roc_auc": 0.7995323724495651,
|
| 345 |
+
"eval_runtime": 8.7776,
|
| 346 |
+
"eval_samples_per_second": 111.762,
|
| 347 |
+
"eval_steps_per_second": 7.519,
|
| 348 |
+
"step": 10480
|
| 349 |
+
},
|
| 350 |
+
{
|
| 351 |
+
"epoch": 20.96,
|
| 352 |
+
"learning_rate": 3.952003816793893e-06,
|
| 353 |
+
"loss": 0.1189,
|
| 354 |
+
"step": 10983
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 21.0,
|
| 358 |
+
"eval_accuracy": 0.5524974515800204,
|
| 359 |
+
"eval_f1": 0.6955767562879446,
|
| 360 |
+
"eval_loss": 0.4442458152770996,
|
| 361 |
+
"eval_roc_auc": 0.7988851529687178,
|
| 362 |
+
"eval_runtime": 9.0585,
|
| 363 |
+
"eval_samples_per_second": 108.296,
|
| 364 |
+
"eval_steps_per_second": 7.286,
|
| 365 |
+
"step": 11004
|
| 366 |
+
},
|
| 367 |
+
{
|
| 368 |
+
"epoch": 21.96,
|
| 369 |
+
"learning_rate": 3.902099236641222e-06,
|
| 370 |
+
"loss": 0.1153,
|
| 371 |
+
"step": 11506
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 22.0,
|
| 375 |
+
"eval_accuracy": 0.5800203873598369,
|
| 376 |
+
"eval_f1": 0.7025121198765976,
|
| 377 |
+
"eval_loss": 0.4538831412792206,
|
| 378 |
+
"eval_roc_auc": 0.8010173050882469,
|
| 379 |
+
"eval_runtime": 8.6472,
|
| 380 |
+
"eval_samples_per_second": 113.447,
|
| 381 |
+
"eval_steps_per_second": 7.633,
|
| 382 |
+
"step": 11528
|
| 383 |
+
},
|
| 384 |
+
{
|
| 385 |
+
"epoch": 22.96,
|
| 386 |
+
"learning_rate": 3.85219465648855e-06,
|
| 387 |
+
"loss": 0.1037,
|
| 388 |
+
"step": 12029
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 23.0,
|
| 392 |
+
"eval_accuracy": 0.5565749235474006,
|
| 393 |
+
"eval_f1": 0.7012430347192455,
|
| 394 |
+
"eval_loss": 0.4749497175216675,
|
| 395 |
+
"eval_roc_auc": 0.804281517523123,
|
| 396 |
+
"eval_runtime": 8.6607,
|
| 397 |
+
"eval_samples_per_second": 113.27,
|
| 398 |
+
"eval_steps_per_second": 7.621,
|
| 399 |
+
"step": 12052
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 23.95,
|
| 403 |
+
"learning_rate": 3.802290076335878e-06,
|
| 404 |
+
"loss": 0.0968,
|
| 405 |
+
"step": 12552
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 24.0,
|
| 409 |
+
"eval_accuracy": 0.5606523955147809,
|
| 410 |
+
"eval_f1": 0.7047124945957631,
|
| 411 |
+
"eval_loss": 0.479137659072876,
|
| 412 |
+
"eval_roc_auc": 0.8052666911477818,
|
| 413 |
+
"eval_runtime": 8.7017,
|
| 414 |
+
"eval_samples_per_second": 112.737,
|
| 415 |
+
"eval_steps_per_second": 7.585,
|
| 416 |
+
"step": 12576
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 24.95,
|
| 420 |
+
"learning_rate": 3.752385496183206e-06,
|
| 421 |
+
"loss": 0.0923,
|
| 422 |
+
"step": 13075
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 25.0,
|
| 426 |
+
"eval_accuracy": 0.5555555555555556,
|
| 427 |
+
"eval_f1": 0.6873362445414847,
|
| 428 |
+
"eval_loss": 0.50052809715271,
|
| 429 |
+
"eval_roc_auc": 0.7925810745679466,
|
| 430 |
+
"eval_runtime": 8.625,
|
| 431 |
+
"eval_samples_per_second": 113.739,
|
| 432 |
+
"eval_steps_per_second": 7.652,
|
| 433 |
+
"step": 13100
|
| 434 |
+
},
|
| 435 |
+
{
|
| 436 |
+
"epoch": 25.95,
|
| 437 |
+
"learning_rate": 3.7024809160305346e-06,
|
| 438 |
+
"loss": 0.0846,
|
| 439 |
+
"step": 13598
|
| 440 |
+
},
|
| 441 |
+
{
|
| 442 |
+
"epoch": 26.0,
|
| 443 |
+
"eval_accuracy": 0.5524974515800204,
|
| 444 |
+
"eval_f1": 0.6971770744225834,
|
| 445 |
+
"eval_loss": 0.5201558470726013,
|
| 446 |
+
"eval_roc_auc": 0.801924445158477,
|
| 447 |
+
"eval_runtime": 8.6427,
|
| 448 |
+
"eval_samples_per_second": 113.506,
|
| 449 |
+
"eval_steps_per_second": 7.636,
|
| 450 |
+
"step": 13624
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 26.95,
|
| 454 |
+
"learning_rate": 3.6525763358778626e-06,
|
| 455 |
+
"loss": 0.0807,
|
| 456 |
+
"step": 14121
|
| 457 |
+
},
|
| 458 |
+
{
|
| 459 |
+
"epoch": 27.0,
|
| 460 |
+
"eval_accuracy": 0.5688073394495413,
|
| 461 |
+
"eval_f1": 0.7090358841331603,
|
| 462 |
+
"eval_loss": 0.52068030834198,
|
| 463 |
+
"eval_roc_auc": 0.8080810630924239,
|
| 464 |
+
"eval_runtime": 8.6305,
|
| 465 |
+
"eval_samples_per_second": 113.667,
|
| 466 |
+
"eval_steps_per_second": 7.647,
|
| 467 |
+
"step": 14148
|
| 468 |
+
},
|
| 469 |
+
{
|
| 470 |
+
"epoch": 27.95,
|
| 471 |
+
"learning_rate": 3.602671755725191e-06,
|
| 472 |
+
"loss": 0.0748,
|
| 473 |
+
"step": 14644
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 28.0,
|
| 477 |
+
"eval_accuracy": 0.5728848114169215,
|
| 478 |
+
"eval_f1": 0.7102321582115221,
|
| 479 |
+
"eval_loss": 0.5269947052001953,
|
| 480 |
+
"eval_roc_auc": 0.8097203415115557,
|
| 481 |
+
"eval_runtime": 8.6597,
|
| 482 |
+
"eval_samples_per_second": 113.284,
|
| 483 |
+
"eval_steps_per_second": 7.622,
|
| 484 |
+
"step": 14672
|
| 485 |
+
},
|
| 486 |
+
{
|
| 487 |
+
"epoch": 28.94,
|
| 488 |
+
"learning_rate": 3.5527671755725195e-06,
|
| 489 |
+
"loss": 0.0727,
|
| 490 |
+
"step": 15167
|
| 491 |
+
},
|
| 492 |
+
{
|
| 493 |
+
"epoch": 29.0,
|
| 494 |
+
"eval_accuracy": 0.528032619775739,
|
| 495 |
+
"eval_f1": 0.6827004219409282,
|
| 496 |
+
"eval_loss": 0.5663571953773499,
|
| 497 |
+
"eval_roc_auc": 0.7942691239585964,
|
| 498 |
+
"eval_runtime": 8.6961,
|
| 499 |
+
"eval_samples_per_second": 112.809,
|
| 500 |
+
"eval_steps_per_second": 7.59,
|
| 501 |
+
"step": 15196
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 29.94,
|
| 505 |
+
"learning_rate": 3.5028625954198474e-06,
|
| 506 |
+
"loss": 0.0685,
|
| 507 |
+
"step": 15690
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 30.0,
|
| 511 |
+
"eval_accuracy": 0.5586136595310908,
|
| 512 |
+
"eval_f1": 0.696027633851468,
|
| 513 |
+
"eval_loss": 0.5751686096191406,
|
| 514 |
+
"eval_roc_auc": 0.7997997521287095,
|
| 515 |
+
"eval_runtime": 8.6393,
|
| 516 |
+
"eval_samples_per_second": 113.551,
|
| 517 |
+
"eval_steps_per_second": 7.639,
|
| 518 |
+
"step": 15720
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 30.94,
|
| 522 |
+
"learning_rate": 3.452958015267176e-06,
|
| 523 |
+
"loss": 0.0629,
|
| 524 |
+
"step": 16213
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 31.0,
|
| 528 |
+
"eval_accuracy": 0.5596330275229358,
|
| 529 |
+
"eval_f1": 0.7002606429192008,
|
| 530 |
+
"eval_loss": 0.5864209532737732,
|
| 531 |
+
"eval_roc_auc": 0.8016714098827202,
|
| 532 |
+
"eval_runtime": 8.6544,
|
| 533 |
+
"eval_samples_per_second": 113.352,
|
| 534 |
+
"eval_steps_per_second": 7.626,
|
| 535 |
+
"step": 16244
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 31.94,
|
| 539 |
+
"learning_rate": 3.403053435114504e-06,
|
| 540 |
+
"loss": 0.0586,
|
| 541 |
+
"step": 16736
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 32.0,
|
| 545 |
+
"eval_accuracy": 0.5524974515800204,
|
| 546 |
+
"eval_f1": 0.7002992731936725,
|
| 547 |
+
"eval_loss": 0.5988386273384094,
|
| 548 |
+
"eval_roc_auc": 0.8040422528746184,
|
| 549 |
+
"eval_runtime": 8.6401,
|
| 550 |
+
"eval_samples_per_second": 113.54,
|
| 551 |
+
"eval_steps_per_second": 7.639,
|
| 552 |
+
"step": 16768
|
| 553 |
+
},
|
| 554 |
+
{
|
| 555 |
+
"epoch": 32.94,
|
| 556 |
+
"learning_rate": 3.3531488549618323e-06,
|
| 557 |
+
"loss": 0.0573,
|
| 558 |
+
"step": 17259
|
| 559 |
+
},
|
| 560 |
+
{
|
| 561 |
+
"epoch": 33.0,
|
| 562 |
+
"eval_accuracy": 0.5596330275229358,
|
| 563 |
+
"eval_f1": 0.7038148306900984,
|
| 564 |
+
"eval_loss": 0.5976923108100891,
|
| 565 |
+
"eval_roc_auc": 0.8059701406899085,
|
| 566 |
+
"eval_runtime": 10.028,
|
| 567 |
+
"eval_samples_per_second": 97.827,
|
| 568 |
+
"eval_steps_per_second": 6.582,
|
| 569 |
+
"step": 17292
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 33.94,
|
| 573 |
+
"learning_rate": 3.3032442748091603e-06,
|
| 574 |
+
"loss": 0.0549,
|
| 575 |
+
"step": 17782
|
| 576 |
+
},
|
| 577 |
+
{
|
| 578 |
+
"epoch": 34.0,
|
| 579 |
+
"eval_accuracy": 0.5759429153924567,
|
| 580 |
+
"eval_f1": 0.7030091583078936,
|
| 581 |
+
"eval_loss": 0.6293498873710632,
|
| 582 |
+
"eval_roc_auc": 0.8028746184388699,
|
| 583 |
+
"eval_runtime": 9.2788,
|
| 584 |
+
"eval_samples_per_second": 105.725,
|
| 585 |
+
"eval_steps_per_second": 7.113,
|
| 586 |
+
"step": 17816
|
| 587 |
+
},
|
| 588 |
+
{
|
| 589 |
+
"epoch": 34.93,
|
| 590 |
+
"learning_rate": 3.2533396946564887e-06,
|
| 591 |
+
"loss": 0.0503,
|
| 592 |
+
"step": 18305
|
| 593 |
+
},
|
| 594 |
+
{
|
| 595 |
+
"epoch": 35.0,
|
| 596 |
+
"eval_accuracy": 0.5667686034658511,
|
| 597 |
+
"eval_f1": 0.7081545064377682,
|
| 598 |
+
"eval_loss": 0.630769670009613,
|
| 599 |
+
"eval_roc_auc": 0.8086227077643385,
|
| 600 |
+
"eval_runtime": 9.6294,
|
| 601 |
+
"eval_samples_per_second": 101.876,
|
| 602 |
+
"eval_steps_per_second": 6.854,
|
| 603 |
+
"step": 18340
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 35.93,
|
| 607 |
+
"learning_rate": 3.203435114503817e-06,
|
| 608 |
+
"loss": 0.0477,
|
| 609 |
+
"step": 18828
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 36.0,
|
| 613 |
+
"eval_accuracy": 0.527013251783894,
|
| 614 |
+
"eval_f1": 0.6998341625207297,
|
| 615 |
+
"eval_loss": 0.6515944004058838,
|
| 616 |
+
"eval_roc_auc": 0.8083547543090588,
|
| 617 |
+
"eval_runtime": 10.0399,
|
| 618 |
+
"eval_samples_per_second": 97.71,
|
| 619 |
+
"eval_steps_per_second": 6.574,
|
| 620 |
+
"step": 18864
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 36.93,
|
| 624 |
+
"learning_rate": 3.153530534351145e-06,
|
| 625 |
+
"loss": 0.0464,
|
| 626 |
+
"step": 19351
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 37.0,
|
| 630 |
+
"eval_accuracy": 0.5484199796126402,
|
| 631 |
+
"eval_f1": 0.7017984107068171,
|
| 632 |
+
"eval_loss": 0.6635262370109558,
|
| 633 |
+
"eval_roc_auc": 0.8083478689954327,
|
| 634 |
+
"eval_runtime": 9.7697,
|
| 635 |
+
"eval_samples_per_second": 100.412,
|
| 636 |
+
"eval_steps_per_second": 6.756,
|
| 637 |
+
"step": 19388
|
| 638 |
+
},
|
| 639 |
+
{
|
| 640 |
+
"epoch": 37.93,
|
| 641 |
+
"learning_rate": 3.1036259541984735e-06,
|
| 642 |
+
"loss": 0.0436,
|
| 643 |
+
"step": 19874
|
| 644 |
+
},
|
| 645 |
+
{
|
| 646 |
+
"epoch": 38.0,
|
| 647 |
+
"eval_accuracy": 0.563710499490316,
|
| 648 |
+
"eval_f1": 0.7022834984920293,
|
| 649 |
+
"eval_loss": 0.6706992387771606,
|
| 650 |
+
"eval_roc_auc": 0.8041971724312043,
|
| 651 |
+
"eval_runtime": 9.7315,
|
| 652 |
+
"eval_samples_per_second": 100.807,
|
| 653 |
+
"eval_steps_per_second": 6.782,
|
| 654 |
+
"step": 19912
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 38.93,
|
| 658 |
+
"learning_rate": 3.0537213740458015e-06,
|
| 659 |
+
"loss": 0.0394,
|
| 660 |
+
"step": 20397
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 39.0,
|
| 664 |
+
"eval_accuracy": 0.527013251783894,
|
| 665 |
+
"eval_f1": 0.6885798567214496,
|
| 666 |
+
"eval_loss": 0.7015945911407471,
|
| 667 |
+
"eval_roc_auc": 0.7983710495513071,
|
| 668 |
+
"eval_runtime": 9.8867,
|
| 669 |
+
"eval_samples_per_second": 99.224,
|
| 670 |
+
"eval_steps_per_second": 6.676,
|
| 671 |
+
"step": 20436
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 39.92,
|
| 675 |
+
"learning_rate": 3.00381679389313e-06,
|
| 676 |
+
"loss": 0.038,
|
| 677 |
+
"step": 20920
|
| 678 |
+
},
|
| 679 |
+
{
|
| 680 |
+
"epoch": 40.0,
|
| 681 |
+
"eval_accuracy": 0.5372069317023446,
|
| 682 |
+
"eval_f1": 0.6997894736842106,
|
| 683 |
+
"eval_loss": 0.7073464393615723,
|
| 684 |
+
"eval_roc_auc": 0.8059839113171605,
|
| 685 |
+
"eval_runtime": 9.9063,
|
| 686 |
+
"eval_samples_per_second": 99.028,
|
| 687 |
+
"eval_steps_per_second": 6.662,
|
| 688 |
+
"step": 20960
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 40.92,
|
| 692 |
+
"learning_rate": 2.953912213740458e-06,
|
| 693 |
+
"loss": 0.036,
|
| 694 |
+
"step": 21443
|
| 695 |
+
},
|
| 696 |
+
{
|
| 697 |
+
"epoch": 41.0,
|
| 698 |
+
"eval_accuracy": 0.5474006116207951,
|
| 699 |
+
"eval_f1": 0.6956521739130435,
|
| 700 |
+
"eval_loss": 0.7225540280342102,
|
| 701 |
+
"eval_roc_auc": 0.7999896720295611,
|
| 702 |
+
"eval_runtime": 9.951,
|
| 703 |
+
"eval_samples_per_second": 98.583,
|
| 704 |
+
"eval_steps_per_second": 6.632,
|
| 705 |
+
"step": 21484
|
| 706 |
+
},
|
| 707 |
+
{
|
| 708 |
+
"epoch": 41.92,
|
| 709 |
+
"learning_rate": 2.9040076335877863e-06,
|
| 710 |
+
"loss": 0.0341,
|
| 711 |
+
"step": 21966
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 42.0,
|
| 715 |
+
"eval_accuracy": 0.564729867482161,
|
| 716 |
+
"eval_f1": 0.7089262613195343,
|
| 717 |
+
"eval_loss": 0.7202900052070618,
|
| 718 |
+
"eval_roc_auc": 0.8084046728328477,
|
| 719 |
+
"eval_runtime": 9.6425,
|
| 720 |
+
"eval_samples_per_second": 101.737,
|
| 721 |
+
"eval_steps_per_second": 6.845,
|
| 722 |
+
"step": 22008
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 42.92,
|
| 726 |
+
"learning_rate": 2.8541030534351148e-06,
|
| 727 |
+
"loss": 0.0323,
|
| 728 |
+
"step": 22489
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 43.0,
|
| 732 |
+
"eval_accuracy": 0.5606523955147809,
|
| 733 |
+
"eval_f1": 0.7040552200172563,
|
| 734 |
+
"eval_loss": 0.7239031791687012,
|
| 735 |
+
"eval_roc_auc": 0.8051611163388492,
|
| 736 |
+
"eval_runtime": 9.9111,
|
| 737 |
+
"eval_samples_per_second": 98.98,
|
| 738 |
+
"eval_steps_per_second": 6.659,
|
| 739 |
+
"step": 22532
|
| 740 |
+
},
|
| 741 |
+
{
|
| 742 |
+
"epoch": 43.92,
|
| 743 |
+
"learning_rate": 2.8041984732824428e-06,
|
| 744 |
+
"loss": 0.0302,
|
| 745 |
+
"step": 23012
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 44.0,
|
| 749 |
+
"eval_accuracy": 0.5535168195718655,
|
| 750 |
+
"eval_f1": 0.6988466467321658,
|
| 751 |
+
"eval_loss": 0.7510971426963806,
|
| 752 |
+
"eval_roc_auc": 0.8032119988065456,
|
| 753 |
+
"eval_runtime": 9.9666,
|
| 754 |
+
"eval_samples_per_second": 98.429,
|
| 755 |
+
"eval_steps_per_second": 6.622,
|
| 756 |
+
"step": 23056
|
| 757 |
+
},
|
| 758 |
+
{
|
| 759 |
+
"epoch": 44.91,
|
| 760 |
+
"learning_rate": 2.754293893129771e-06,
|
| 761 |
+
"loss": 0.0286,
|
| 762 |
+
"step": 23535
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"epoch": 45.0,
|
| 766 |
+
"eval_accuracy": 0.5524974515800204,
|
| 767 |
+
"eval_f1": 0.7002132196162046,
|
| 768 |
+
"eval_loss": 0.7605226635932922,
|
| 769 |
+
"eval_roc_auc": 0.8043658626150422,
|
| 770 |
+
"eval_runtime": 9.9921,
|
| 771 |
+
"eval_samples_per_second": 98.178,
|
| 772 |
+
"eval_steps_per_second": 6.605,
|
| 773 |
+
"step": 23580
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 45.91,
|
| 777 |
+
"learning_rate": 2.704389312977099e-06,
|
| 778 |
+
"loss": 0.0275,
|
| 779 |
+
"step": 24058
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 46.0,
|
| 783 |
+
"eval_accuracy": 0.5484199796126402,
|
| 784 |
+
"eval_f1": 0.6999573196756296,
|
| 785 |
+
"eval_loss": 0.774695634841919,
|
| 786 |
+
"eval_roc_auc": 0.8040703679052581,
|
| 787 |
+
"eval_runtime": 9.9221,
|
| 788 |
+
"eval_samples_per_second": 98.87,
|
| 789 |
+
"eval_steps_per_second": 6.652,
|
| 790 |
+
"step": 24104
|
| 791 |
+
},
|
| 792 |
+
{
|
| 793 |
+
"epoch": 46.91,
|
| 794 |
+
"learning_rate": 2.6544847328244276e-06,
|
| 795 |
+
"loss": 0.026,
|
| 796 |
+
"step": 24581
|
| 797 |
+
},
|
| 798 |
+
{
|
| 799 |
+
"epoch": 47.0,
|
| 800 |
+
"eval_accuracy": 0.5372069317023446,
|
| 801 |
+
"eval_f1": 0.6884827879303017,
|
| 802 |
+
"eval_loss": 0.7950236797332764,
|
| 803 |
+
"eval_roc_auc": 0.7971047256202519,
|
| 804 |
+
"eval_runtime": 10.0992,
|
| 805 |
+
"eval_samples_per_second": 97.137,
|
| 806 |
+
"eval_steps_per_second": 6.535,
|
| 807 |
+
"step": 24628
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 47.91,
|
| 811 |
+
"learning_rate": 2.6045801526717556e-06,
|
| 812 |
+
"loss": 0.0247,
|
| 813 |
+
"step": 25104
|
| 814 |
+
},
|
| 815 |
+
{
|
| 816 |
+
"epoch": 48.0,
|
| 817 |
+
"eval_accuracy": 0.5474006116207951,
|
| 818 |
+
"eval_f1": 0.6989293361884368,
|
| 819 |
+
"eval_loss": 0.8052034974098206,
|
| 820 |
+
"eval_roc_auc": 0.802888389066122,
|
| 821 |
+
"eval_runtime": 10.0437,
|
| 822 |
+
"eval_samples_per_second": 97.673,
|
| 823 |
+
"eval_steps_per_second": 6.571,
|
| 824 |
+
"step": 25152
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 48.91,
|
| 828 |
+
"learning_rate": 2.554675572519084e-06,
|
| 829 |
+
"loss": 0.0223,
|
| 830 |
+
"step": 25627
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 49.0,
|
| 834 |
+
"eval_accuracy": 0.5524974515800204,
|
| 835 |
+
"eval_f1": 0.7021276595744681,
|
| 836 |
+
"eval_loss": 0.8275096416473389,
|
| 837 |
+
"eval_roc_auc": 0.8059489109728948,
|
| 838 |
+
"eval_runtime": 9.9609,
|
| 839 |
+
"eval_samples_per_second": 98.485,
|
| 840 |
+
"eval_steps_per_second": 6.626,
|
| 841 |
+
"step": 25676
|
| 842 |
+
},
|
| 843 |
+
{
|
| 844 |
+
"epoch": 49.9,
|
| 845 |
+
"learning_rate": 2.5047709923664124e-06,
|
| 846 |
+
"loss": 0.0239,
|
| 847 |
+
"step": 26150
|
| 848 |
+
},
|
| 849 |
+
{
|
| 850 |
+
"epoch": 50.0,
|
| 851 |
+
"eval_accuracy": 0.5382262996941896,
|
| 852 |
+
"eval_f1": 0.6999168744804655,
|
| 853 |
+
"eval_loss": 0.8239336609840393,
|
| 854 |
+
"eval_roc_auc": 0.8080311445686351,
|
| 855 |
+
"eval_runtime": 9.9615,
|
| 856 |
+
"eval_samples_per_second": 98.479,
|
| 857 |
+
"eval_steps_per_second": 6.626,
|
| 858 |
+
"step": 26200
|
| 859 |
+
},
|
| 860 |
+
{
|
| 861 |
+
"epoch": 50.9,
|
| 862 |
+
"learning_rate": 2.4548664122137404e-06,
|
| 863 |
+
"loss": 0.023,
|
| 864 |
+
"step": 26673
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 51.0,
|
| 868 |
+
"eval_accuracy": 0.5484199796126402,
|
| 869 |
+
"eval_f1": 0.7054823629409264,
|
| 870 |
+
"eval_loss": 0.8209096193313599,
|
| 871 |
+
"eval_roc_auc": 0.8083622133988203,
|
| 872 |
+
"eval_runtime": 10.0023,
|
| 873 |
+
"eval_samples_per_second": 98.078,
|
| 874 |
+
"eval_steps_per_second": 6.598,
|
| 875 |
+
"step": 26724
|
| 876 |
+
},
|
| 877 |
+
{
|
| 878 |
+
"epoch": 51.9,
|
| 879 |
+
"learning_rate": 2.404961832061069e-06,
|
| 880 |
+
"loss": 0.0199,
|
| 881 |
+
"step": 27196
|
| 882 |
+
},
|
| 883 |
+
{
|
| 884 |
+
"epoch": 52.0,
|
| 885 |
+
"eval_accuracy": 0.5484199796126402,
|
| 886 |
+
"eval_f1": 0.7072438919845693,
|
| 887 |
+
"eval_loss": 0.828449010848999,
|
| 888 |
+
"eval_roc_auc": 0.8082216382456221,
|
| 889 |
+
"eval_runtime": 10.055,
|
| 890 |
+
"eval_samples_per_second": 97.563,
|
| 891 |
+
"eval_steps_per_second": 6.564,
|
| 892 |
+
"step": 27248
|
| 893 |
+
},
|
| 894 |
+
{
|
| 895 |
+
"epoch": 52.9,
|
| 896 |
+
"learning_rate": 2.3550572519083973e-06,
|
| 897 |
+
"loss": 0.0197,
|
| 898 |
+
"step": 27719
|
| 899 |
+
},
|
| 900 |
+
{
|
| 901 |
+
"epoch": 53.0,
|
| 902 |
+
"eval_accuracy": 0.563710499490316,
|
| 903 |
+
"eval_f1": 0.7059333044608056,
|
| 904 |
+
"eval_loss": 0.8516786098480225,
|
| 905 |
+
"eval_roc_auc": 0.8058014505060705,
|
| 906 |
+
"eval_runtime": 10.0044,
|
| 907 |
+
"eval_samples_per_second": 98.057,
|
| 908 |
+
"eval_steps_per_second": 6.597,
|
| 909 |
+
"step": 27772
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 53.9,
|
| 913 |
+
"learning_rate": 2.3051526717557252e-06,
|
| 914 |
+
"loss": 0.0168,
|
| 915 |
+
"step": 28242
|
| 916 |
+
},
|
| 917 |
+
{
|
| 918 |
+
"epoch": 54.0,
|
| 919 |
+
"eval_accuracy": 0.5351681957186545,
|
| 920 |
+
"eval_f1": 0.699581589958159,
|
| 921 |
+
"eval_loss": 0.8833754658699036,
|
| 922 |
+
"eval_roc_auc": 0.8067929356682197,
|
| 923 |
+
"eval_runtime": 10.0433,
|
| 924 |
+
"eval_samples_per_second": 97.678,
|
| 925 |
+
"eval_steps_per_second": 6.572,
|
| 926 |
+
"step": 28296
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 54.9,
|
| 930 |
+
"learning_rate": 2.2552480916030537e-06,
|
| 931 |
+
"loss": 0.018,
|
| 932 |
+
"step": 28765
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"epoch": 55.0,
|
| 936 |
+
"eval_accuracy": 0.5524974515800204,
|
| 937 |
+
"eval_f1": 0.7060839760068551,
|
| 938 |
+
"eval_loss": 0.8679118156433105,
|
| 939 |
+
"eval_roc_auc": 0.8075250740171215,
|
| 940 |
+
"eval_runtime": 9.9308,
|
| 941 |
+
"eval_samples_per_second": 98.784,
|
| 942 |
+
"eval_steps_per_second": 6.646,
|
| 943 |
+
"step": 28820
|
| 944 |
+
},
|
| 945 |
+
{
|
| 946 |
+
"epoch": 55.89,
|
| 947 |
+
"learning_rate": 2.2053435114503817e-06,
|
| 948 |
+
"loss": 0.0169,
|
| 949 |
+
"step": 29288
|
| 950 |
+
},
|
| 951 |
+
{
|
| 952 |
+
"epoch": 56.0,
|
| 953 |
+
"eval_accuracy": 0.5575942915392457,
|
| 954 |
+
"eval_f1": 0.7040417209908736,
|
| 955 |
+
"eval_loss": 0.8795809149742126,
|
| 956 |
+
"eval_roc_auc": 0.8040565972780059,
|
| 957 |
+
"eval_runtime": 9.9865,
|
| 958 |
+
"eval_samples_per_second": 98.232,
|
| 959 |
+
"eval_steps_per_second": 6.609,
|
| 960 |
+
"step": 29344
|
| 961 |
+
},
|
| 962 |
+
{
|
| 963 |
+
"epoch": 56.89,
|
| 964 |
+
"learning_rate": 2.15543893129771e-06,
|
| 965 |
+
"loss": 0.0168,
|
| 966 |
+
"step": 29811
|
| 967 |
+
},
|
| 968 |
+
{
|
| 969 |
+
"epoch": 57.0,
|
| 970 |
+
"eval_accuracy": 0.5433231396534148,
|
| 971 |
+
"eval_f1": 0.6988879384088965,
|
| 972 |
+
"eval_loss": 0.9083885550498962,
|
| 973 |
+
"eval_roc_auc": 0.8030501939363338,
|
| 974 |
+
"eval_runtime": 10.0982,
|
| 975 |
+
"eval_samples_per_second": 97.146,
|
| 976 |
+
"eval_steps_per_second": 6.536,
|
| 977 |
+
"step": 29868
|
| 978 |
+
},
|
| 979 |
+
{
|
| 980 |
+
"epoch": 57.89,
|
| 981 |
+
"learning_rate": 2.105534351145038e-06,
|
| 982 |
+
"loss": 0.0156,
|
| 983 |
+
"step": 30334
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 58.0,
|
| 987 |
+
"eval_accuracy": 0.5606523955147809,
|
| 988 |
+
"eval_f1": 0.710651142733937,
|
| 989 |
+
"eval_loss": 0.9000456929206848,
|
| 990 |
+
"eval_roc_auc": 0.8095304216107044,
|
| 991 |
+
"eval_runtime": 10.0091,
|
| 992 |
+
"eval_samples_per_second": 98.011,
|
| 993 |
+
"eval_steps_per_second": 6.594,
|
| 994 |
+
"step": 30392
|
| 995 |
+
},
|
| 996 |
+
{
|
| 997 |
+
"epoch": 58.89,
|
| 998 |
+
"learning_rate": 2.055629770992367e-06,
|
| 999 |
+
"loss": 0.0138,
|
| 1000 |
+
"step": 30857
|
| 1001 |
+
},
|
| 1002 |
+
{
|
| 1003 |
+
"epoch": 59.0,
|
| 1004 |
+
"eval_accuracy": 0.5382262996941896,
|
| 1005 |
+
"eval_f1": 0.6970849176172369,
|
| 1006 |
+
"eval_loss": 0.9262450337409973,
|
| 1007 |
+
"eval_roc_auc": 0.8036761837001675,
|
| 1008 |
+
"eval_runtime": 9.8719,
|
| 1009 |
+
"eval_samples_per_second": 99.373,
|
| 1010 |
+
"eval_steps_per_second": 6.686,
|
| 1011 |
+
"step": 30916
|
| 1012 |
+
},
|
| 1013 |
+
{
|
| 1014 |
+
"epoch": 59.89,
|
| 1015 |
+
"learning_rate": 2.005725190839695e-06,
|
| 1016 |
+
"loss": 0.0139,
|
| 1017 |
+
"step": 31380
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 60.0,
|
| 1021 |
+
"eval_accuracy": 0.5596330275229358,
|
| 1022 |
+
"eval_f1": 0.7065868263473054,
|
| 1023 |
+
"eval_loss": 0.923125684261322,
|
| 1024 |
+
"eval_roc_auc": 0.8081160634366895,
|
| 1025 |
+
"eval_runtime": 10.0204,
|
| 1026 |
+
"eval_samples_per_second": 97.901,
|
| 1027 |
+
"eval_steps_per_second": 6.587,
|
| 1028 |
+
"step": 31440
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"epoch": 60.88,
|
| 1032 |
+
"learning_rate": 1.955820610687023e-06,
|
| 1033 |
+
"loss": 0.0155,
|
| 1034 |
+
"step": 31903
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 61.0,
|
| 1038 |
+
"eval_accuracy": 0.5596330275229358,
|
| 1039 |
+
"eval_f1": 0.7081545064377682,
|
| 1040 |
+
"eval_loss": 0.9300869107246399,
|
| 1041 |
+
"eval_roc_auc": 0.8086227077643385,
|
| 1042 |
+
"eval_runtime": 9.9625,
|
| 1043 |
+
"eval_samples_per_second": 98.47,
|
| 1044 |
+
"eval_steps_per_second": 6.625,
|
| 1045 |
+
"step": 31964
|
| 1046 |
+
},
|
| 1047 |
+
{
|
| 1048 |
+
"epoch": 61.88,
|
| 1049 |
+
"learning_rate": 1.9059160305343513e-06,
|
| 1050 |
+
"loss": 0.0149,
|
| 1051 |
+
"step": 32426
|
| 1052 |
+
},
|
| 1053 |
+
{
|
| 1054 |
+
"epoch": 62.0,
|
| 1055 |
+
"eval_accuracy": 0.5504587155963303,
|
| 1056 |
+
"eval_f1": 0.7104930467762326,
|
| 1057 |
+
"eval_loss": 0.9461256861686707,
|
| 1058 |
+
"eval_roc_auc": 0.8130057836634459,
|
| 1059 |
+
"eval_runtime": 10.0052,
|
| 1060 |
+
"eval_samples_per_second": 98.049,
|
| 1061 |
+
"eval_steps_per_second": 6.597,
|
| 1062 |
+
"step": 32488
|
| 1063 |
+
},
|
| 1064 |
+
{
|
| 1065 |
+
"epoch": 62.88,
|
| 1066 |
+
"learning_rate": 1.8560114503816795e-06,
|
| 1067 |
+
"loss": 0.0124,
|
| 1068 |
+
"step": 32949
|
| 1069 |
+
},
|
| 1070 |
+
{
|
| 1071 |
+
"epoch": 63.0,
|
| 1072 |
+
"eval_accuracy": 0.5565749235474006,
|
| 1073 |
+
"eval_f1": 0.7081380485726461,
|
| 1074 |
+
"eval_loss": 0.9584424495697021,
|
| 1075 |
+
"eval_roc_auc": 0.809727226825182,
|
| 1076 |
+
"eval_runtime": 9.989,
|
| 1077 |
+
"eval_samples_per_second": 98.208,
|
| 1078 |
+
"eval_steps_per_second": 6.607,
|
| 1079 |
+
"step": 33012
|
| 1080 |
+
},
|
| 1081 |
+
{
|
| 1082 |
+
"epoch": 63.88,
|
| 1083 |
+
"learning_rate": 1.8061068702290077e-06,
|
| 1084 |
+
"loss": 0.011,
|
| 1085 |
+
"step": 33472
|
| 1086 |
+
},
|
| 1087 |
+
{
|
| 1088 |
+
"epoch": 64.0,
|
| 1089 |
+
"eval_accuracy": 0.5545361875637105,
|
| 1090 |
+
"eval_f1": 0.7051226861816616,
|
| 1091 |
+
"eval_loss": 0.9570773243904114,
|
| 1092 |
+
"eval_roc_auc": 0.8061812903077735,
|
| 1093 |
+
"eval_runtime": 10.0472,
|
| 1094 |
+
"eval_samples_per_second": 97.639,
|
| 1095 |
+
"eval_steps_per_second": 6.569,
|
| 1096 |
+
"step": 33536
|
| 1097 |
+
},
|
| 1098 |
+
{
|
| 1099 |
+
"epoch": 64.88,
|
| 1100 |
+
"learning_rate": 1.756202290076336e-06,
|
| 1101 |
+
"loss": 0.0114,
|
| 1102 |
+
"step": 33995
|
| 1103 |
+
},
|
| 1104 |
+
{
|
| 1105 |
+
"epoch": 65.0,
|
| 1106 |
+
"eval_accuracy": 0.5565749235474006,
|
| 1107 |
+
"eval_f1": 0.7043701799485862,
|
| 1108 |
+
"eval_loss": 0.9560405015945435,
|
| 1109 |
+
"eval_roc_auc": 0.8063993252392647,
|
| 1110 |
+
"eval_runtime": 10.0628,
|
| 1111 |
+
"eval_samples_per_second": 97.488,
|
| 1112 |
+
"eval_steps_per_second": 6.559,
|
| 1113 |
+
"step": 34060
|
| 1114 |
+
},
|
| 1115 |
+
{
|
| 1116 |
+
"epoch": 65.87,
|
| 1117 |
+
"learning_rate": 1.7062977099236644e-06,
|
| 1118 |
+
"loss": 0.011,
|
| 1119 |
+
"step": 34518
|
| 1120 |
+
},
|
| 1121 |
+
{
|
| 1122 |
+
"epoch": 66.0,
|
| 1123 |
+
"eval_accuracy": 0.5504587155963303,
|
| 1124 |
+
"eval_f1": 0.7023809523809524,
|
| 1125 |
+
"eval_loss": 0.9797949194908142,
|
| 1126 |
+
"eval_roc_auc": 0.8062444056826787,
|
| 1127 |
+
"eval_runtime": 10.0681,
|
| 1128 |
+
"eval_samples_per_second": 97.436,
|
| 1129 |
+
"eval_steps_per_second": 6.555,
|
| 1130 |
+
"step": 34584
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 66.87,
|
| 1134 |
+
"learning_rate": 1.6563931297709926e-06,
|
| 1135 |
+
"loss": 0.0107,
|
| 1136 |
+
"step": 35041
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 67.0,
|
| 1140 |
+
"eval_accuracy": 0.545361875637105,
|
| 1141 |
+
"eval_f1": 0.6986301369863014,
|
| 1142 |
+
"eval_loss": 0.9825329780578613,
|
| 1143 |
+
"eval_roc_auc": 0.8027546992265496,
|
| 1144 |
+
"eval_runtime": 10.0516,
|
| 1145 |
+
"eval_samples_per_second": 97.596,
|
| 1146 |
+
"eval_steps_per_second": 6.566,
|
| 1147 |
+
"step": 35108
|
| 1148 |
+
},
|
| 1149 |
+
{
|
| 1150 |
+
"epoch": 67.87,
|
| 1151 |
+
"learning_rate": 1.6064885496183208e-06,
|
| 1152 |
+
"loss": 0.0091,
|
| 1153 |
+
"step": 35564
|
| 1154 |
+
},
|
| 1155 |
+
{
|
| 1156 |
+
"epoch": 68.0,
|
| 1157 |
+
"eval_accuracy": 0.5606523955147809,
|
| 1158 |
+
"eval_f1": 0.7040552200172563,
|
| 1159 |
+
"eval_loss": 0.9886102676391602,
|
| 1160 |
+
"eval_roc_auc": 0.8051611163388492,
|
| 1161 |
+
"eval_runtime": 10.0044,
|
| 1162 |
+
"eval_samples_per_second": 98.056,
|
| 1163 |
+
"eval_steps_per_second": 6.597,
|
| 1164 |
+
"step": 35632
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 68.87,
|
| 1168 |
+
"learning_rate": 1.556583969465649e-06,
|
| 1169 |
+
"loss": 0.0095,
|
| 1170 |
+
"step": 36087
|
| 1171 |
+
},
|
| 1172 |
+
{
|
| 1173 |
+
"epoch": 69.0,
|
| 1174 |
+
"eval_accuracy": 0.5392456676860347,
|
| 1175 |
+
"eval_f1": 0.7075,
|
| 1176 |
+
"eval_loss": 1.007102131843567,
|
| 1177 |
+
"eval_roc_auc": 0.8127734043285672,
|
| 1178 |
+
"eval_runtime": 10.0072,
|
| 1179 |
+
"eval_samples_per_second": 98.029,
|
| 1180 |
+
"eval_steps_per_second": 6.595,
|
| 1181 |
+
"step": 36156
|
| 1182 |
+
},
|
| 1183 |
+
{
|
| 1184 |
+
"epoch": 69.87,
|
| 1185 |
+
"learning_rate": 1.5066793893129772e-06,
|
| 1186 |
+
"loss": 0.0088,
|
| 1187 |
+
"step": 36610
|
| 1188 |
+
},
|
| 1189 |
+
{
|
| 1190 |
+
"epoch": 70.0,
|
| 1191 |
+
"eval_accuracy": 0.5443425076452599,
|
| 1192 |
+
"eval_f1": 0.7051336444633007,
|
| 1193 |
+
"eval_loss": 1.004128098487854,
|
| 1194 |
+
"eval_roc_auc": 0.8083903284294599,
|
| 1195 |
+
"eval_runtime": 10.0021,
|
| 1196 |
+
"eval_samples_per_second": 98.079,
|
| 1197 |
+
"eval_steps_per_second": 6.599,
|
| 1198 |
+
"step": 36680
|
| 1199 |
+
},
|
| 1200 |
+
{
|
| 1201 |
+
"epoch": 70.86,
|
| 1202 |
+
"learning_rate": 1.4567748091603054e-06,
|
| 1203 |
+
"loss": 0.0102,
|
| 1204 |
+
"step": 37133
|
| 1205 |
+
},
|
| 1206 |
+
{
|
| 1207 |
+
"epoch": 71.0,
|
| 1208 |
+
"eval_accuracy": 0.5474006116207951,
|
| 1209 |
+
"eval_f1": 0.7023354564755838,
|
| 1210 |
+
"eval_loss": 1.023705244064331,
|
| 1211 |
+
"eval_roc_auc": 0.8064062105528907,
|
| 1212 |
+
"eval_runtime": 10.0301,
|
| 1213 |
+
"eval_samples_per_second": 97.806,
|
| 1214 |
+
"eval_steps_per_second": 6.58,
|
| 1215 |
+
"step": 37204
|
| 1216 |
+
},
|
| 1217 |
+
{
|
| 1218 |
+
"epoch": 71.86,
|
| 1219 |
+
"learning_rate": 1.4068702290076336e-06,
|
| 1220 |
+
"loss": 0.0086,
|
| 1221 |
+
"step": 37656
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 72.0,
|
| 1225 |
+
"eval_accuracy": 0.5606523955147809,
|
| 1226 |
+
"eval_f1": 0.7097887020267357,
|
| 1227 |
+
"eval_loss": 1.0078336000442505,
|
| 1228 |
+
"eval_roc_auc": 0.808967547221776,
|
| 1229 |
+
"eval_runtime": 10.0138,
|
| 1230 |
+
"eval_samples_per_second": 97.964,
|
| 1231 |
+
"eval_steps_per_second": 6.591,
|
| 1232 |
+
"step": 37728
|
| 1233 |
+
},
|
| 1234 |
+
{
|
| 1235 |
+
"epoch": 72.86,
|
| 1236 |
+
"learning_rate": 1.356965648854962e-06,
|
| 1237 |
+
"loss": 0.0084,
|
| 1238 |
+
"step": 38179
|
| 1239 |
+
},
|
| 1240 |
+
{
|
| 1241 |
+
"epoch": 73.0,
|
| 1242 |
+
"eval_accuracy": 0.5504587155963303,
|
| 1243 |
+
"eval_f1": 0.7039249146757679,
|
| 1244 |
+
"eval_loss": 1.0251305103302002,
|
| 1245 |
+
"eval_roc_auc": 0.8067510500103279,
|
| 1246 |
+
"eval_runtime": 9.9979,
|
| 1247 |
+
"eval_samples_per_second": 98.12,
|
| 1248 |
+
"eval_steps_per_second": 6.601,
|
| 1249 |
+
"step": 38252
|
| 1250 |
+
},
|
| 1251 |
+
{
|
| 1252 |
+
"epoch": 73.86,
|
| 1253 |
+
"learning_rate": 1.3070610687022902e-06,
|
| 1254 |
+
"loss": 0.0084,
|
| 1255 |
+
"step": 38702
|
| 1256 |
+
},
|
| 1257 |
+
{
|
| 1258 |
+
"epoch": 74.0,
|
| 1259 |
+
"eval_accuracy": 0.5565749235474006,
|
| 1260 |
+
"eval_f1": 0.7060839760068551,
|
| 1261 |
+
"eval_loss": 1.0233700275421143,
|
| 1262 |
+
"eval_roc_auc": 0.8075250740171215,
|
| 1263 |
+
"eval_runtime": 10.0267,
|
| 1264 |
+
"eval_samples_per_second": 97.839,
|
| 1265 |
+
"eval_steps_per_second": 6.582,
|
| 1266 |
+
"step": 38776
|
| 1267 |
+
},
|
| 1268 |
+
{
|
| 1269 |
+
"epoch": 74.86,
|
| 1270 |
+
"learning_rate": 1.2571564885496184e-06,
|
| 1271 |
+
"loss": 0.0076,
|
| 1272 |
+
"step": 39225
|
| 1273 |
+
},
|
| 1274 |
+
{
|
| 1275 |
+
"epoch": 75.0,
|
| 1276 |
+
"eval_accuracy": 0.5433231396534148,
|
| 1277 |
+
"eval_f1": 0.7029787234042555,
|
| 1278 |
+
"eval_loss": 1.0505975484848022,
|
| 1279 |
+
"eval_roc_auc": 0.8065117853618233,
|
| 1280 |
+
"eval_runtime": 9.9951,
|
| 1281 |
+
"eval_samples_per_second": 98.148,
|
| 1282 |
+
"eval_steps_per_second": 6.603,
|
| 1283 |
+
"step": 39300
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 75.85,
|
| 1287 |
+
"learning_rate": 1.2072519083969466e-06,
|
| 1288 |
+
"loss": 0.0089,
|
| 1289 |
+
"step": 39748
|
| 1290 |
+
},
|
| 1291 |
+
{
|
| 1292 |
+
"epoch": 76.0,
|
| 1293 |
+
"eval_accuracy": 0.5575942915392457,
|
| 1294 |
+
"eval_f1": 0.7084398976982098,
|
| 1295 |
+
"eval_loss": 1.0305790901184082,
|
| 1296 |
+
"eval_roc_auc": 0.8098609166647541,
|
| 1297 |
+
"eval_runtime": 10.0015,
|
| 1298 |
+
"eval_samples_per_second": 98.086,
|
| 1299 |
+
"eval_steps_per_second": 6.599,
|
| 1300 |
+
"step": 39824
|
| 1301 |
+
},
|
| 1302 |
+
{
|
| 1303 |
+
"epoch": 76.85,
|
| 1304 |
+
"learning_rate": 1.1573473282442748e-06,
|
| 1305 |
+
"loss": 0.0074,
|
| 1306 |
+
"step": 40271
|
| 1307 |
+
},
|
| 1308 |
+
{
|
| 1309 |
+
"epoch": 77.0,
|
| 1310 |
+
"eval_accuracy": 0.5514780835881753,
|
| 1311 |
+
"eval_f1": 0.7120954003407156,
|
| 1312 |
+
"eval_loss": 1.037413477897644,
|
| 1313 |
+
"eval_roc_auc": 0.8124079089302518,
|
| 1314 |
+
"eval_runtime": 10.0466,
|
| 1315 |
+
"eval_samples_per_second": 97.645,
|
| 1316 |
+
"eval_steps_per_second": 6.569,
|
| 1317 |
+
"step": 40348
|
| 1318 |
+
},
|
| 1319 |
+
{
|
| 1320 |
+
"epoch": 77.85,
|
| 1321 |
+
"learning_rate": 1.1074427480916033e-06,
|
| 1322 |
+
"loss": 0.0064,
|
| 1323 |
+
"step": 40794
|
| 1324 |
+
},
|
| 1325 |
+
{
|
| 1326 |
+
"epoch": 78.0,
|
| 1327 |
+
"eval_accuracy": 0.563710499490316,
|
| 1328 |
+
"eval_f1": 0.7064418504107222,
|
| 1329 |
+
"eval_loss": 1.0435516834259033,
|
| 1330 |
+
"eval_roc_auc": 0.8063924399256387,
|
| 1331 |
+
"eval_runtime": 9.9844,
|
| 1332 |
+
"eval_samples_per_second": 98.254,
|
| 1333 |
+
"eval_steps_per_second": 6.61,
|
| 1334 |
+
"step": 40872
|
| 1335 |
+
},
|
| 1336 |
+
{
|
| 1337 |
+
"epoch": 78.85,
|
| 1338 |
+
"learning_rate": 1.0575381679389315e-06,
|
| 1339 |
+
"loss": 0.0079,
|
| 1340 |
+
"step": 41317
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 79.0,
|
| 1344 |
+
"eval_accuracy": 0.5535168195718655,
|
| 1345 |
+
"eval_f1": 0.7080479452054794,
|
| 1346 |
+
"eval_loss": 1.0525715351104736,
|
| 1347 |
+
"eval_roc_auc": 0.8089463175047622,
|
| 1348 |
+
"eval_runtime": 10.0182,
|
| 1349 |
+
"eval_samples_per_second": 97.922,
|
| 1350 |
+
"eval_steps_per_second": 6.588,
|
| 1351 |
+
"step": 41396
|
| 1352 |
+
},
|
| 1353 |
+
{
|
| 1354 |
+
"epoch": 79.85,
|
| 1355 |
+
"learning_rate": 1.0076335877862597e-06,
|
| 1356 |
+
"loss": 0.0059,
|
| 1357 |
+
"step": 41840
|
| 1358 |
+
},
|
| 1359 |
+
{
|
| 1360 |
+
"epoch": 80.0,
|
| 1361 |
+
"eval_accuracy": 0.545361875637105,
|
| 1362 |
+
"eval_f1": 0.7050359712230215,
|
| 1363 |
+
"eval_loss": 1.0556447505950928,
|
| 1364 |
+
"eval_roc_auc": 0.8087139381698836,
|
| 1365 |
+
"eval_runtime": 9.9919,
|
| 1366 |
+
"eval_samples_per_second": 98.18,
|
| 1367 |
+
"eval_steps_per_second": 6.605,
|
| 1368 |
+
"step": 41920
|
| 1369 |
+
},
|
| 1370 |
+
{
|
| 1371 |
+
"epoch": 80.85,
|
| 1372 |
+
"learning_rate": 9.577290076335879e-07,
|
| 1373 |
+
"loss": 0.0063,
|
| 1374 |
+
"step": 42363
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 81.0,
|
| 1378 |
+
"eval_accuracy": 0.5412844036697247,
|
| 1379 |
+
"eval_f1": 0.7034834324553951,
|
| 1380 |
+
"eval_loss": 1.0627212524414062,
|
| 1381 |
+
"eval_roc_auc": 0.8071027747813913,
|
| 1382 |
+
"eval_runtime": 10.1039,
|
| 1383 |
+
"eval_samples_per_second": 97.091,
|
| 1384 |
+
"eval_steps_per_second": 6.532,
|
| 1385 |
+
"step": 42444
|
| 1386 |
+
},
|
| 1387 |
+
{
|
| 1388 |
+
"epoch": 81.84,
|
| 1389 |
+
"learning_rate": 9.078244274809162e-07,
|
| 1390 |
+
"loss": 0.0064,
|
| 1391 |
+
"step": 42886
|
| 1392 |
+
},
|
| 1393 |
+
{
|
| 1394 |
+
"epoch": 82.0,
|
| 1395 |
+
"eval_accuracy": 0.5382262996941896,
|
| 1396 |
+
"eval_f1": 0.7048903878583473,
|
| 1397 |
+
"eval_loss": 1.0669602155685425,
|
| 1398 |
+
"eval_roc_auc": 0.8091993527805191,
|
| 1399 |
+
"eval_runtime": 10.2231,
|
| 1400 |
+
"eval_samples_per_second": 95.96,
|
| 1401 |
+
"eval_steps_per_second": 6.456,
|
| 1402 |
+
"step": 42968
|
| 1403 |
+
},
|
| 1404 |
+
{
|
| 1405 |
+
"epoch": 82.84,
|
| 1406 |
+
"learning_rate": 8.579198473282444e-07,
|
| 1407 |
+
"loss": 0.0064,
|
| 1408 |
+
"step": 43409
|
| 1409 |
+
},
|
| 1410 |
+
{
|
| 1411 |
+
"epoch": 83.0,
|
| 1412 |
+
"eval_accuracy": 0.5575942915392457,
|
| 1413 |
+
"eval_f1": 0.70824434002563,
|
| 1414 |
+
"eval_loss": 1.0584640502929688,
|
| 1415 |
+
"eval_roc_auc": 0.8094036170847583,
|
| 1416 |
+
"eval_runtime": 10.0727,
|
| 1417 |
+
"eval_samples_per_second": 97.392,
|
| 1418 |
+
"eval_steps_per_second": 6.552,
|
| 1419 |
+
"step": 43492
|
| 1420 |
+
},
|
| 1421 |
+
{
|
| 1422 |
+
"epoch": 83.84,
|
| 1423 |
+
"learning_rate": 8.080152671755725e-07,
|
| 1424 |
+
"loss": 0.006,
|
| 1425 |
+
"step": 43932
|
| 1426 |
+
},
|
| 1427 |
+
{
|
| 1428 |
+
"epoch": 84.0,
|
| 1429 |
+
"eval_accuracy": 0.5524974515800204,
|
| 1430 |
+
"eval_f1": 0.7100340136054422,
|
| 1431 |
+
"eval_loss": 1.0683461427688599,
|
| 1432 |
+
"eval_roc_auc": 0.8113102751830346,
|
| 1433 |
+
"eval_runtime": 10.0298,
|
| 1434 |
+
"eval_samples_per_second": 97.809,
|
| 1435 |
+
"eval_steps_per_second": 6.58,
|
| 1436 |
+
"step": 44016
|
| 1437 |
+
},
|
| 1438 |
+
{
|
| 1439 |
+
"epoch": 84.84,
|
| 1440 |
+
"learning_rate": 7.581106870229009e-07,
|
| 1441 |
+
"loss": 0.0056,
|
| 1442 |
+
"step": 44455
|
| 1443 |
+
},
|
| 1444 |
+
{
|
| 1445 |
+
"epoch": 85.0,
|
| 1446 |
+
"eval_accuracy": 0.5484199796126402,
|
| 1447 |
+
"eval_f1": 0.7041294167730949,
|
| 1448 |
+
"eval_loss": 1.0729014873504639,
|
| 1449 |
+
"eval_roc_auc": 0.8072083495903238,
|
| 1450 |
+
"eval_runtime": 10.0817,
|
| 1451 |
+
"eval_samples_per_second": 97.305,
|
| 1452 |
+
"eval_steps_per_second": 6.547,
|
| 1453 |
+
"step": 44540
|
| 1454 |
+
},
|
| 1455 |
+
{
|
| 1456 |
+
"epoch": 85.84,
|
| 1457 |
+
"learning_rate": 7.082061068702291e-07,
|
| 1458 |
+
"loss": 0.0063,
|
| 1459 |
+
"step": 44978
|
| 1460 |
+
},
|
| 1461 |
+
{
|
| 1462 |
+
"epoch": 86.0,
|
| 1463 |
+
"eval_accuracy": 0.564729867482161,
|
| 1464 |
+
"eval_f1": 0.7094274644855791,
|
| 1465 |
+
"eval_loss": 1.0694997310638428,
|
| 1466 |
+
"eval_roc_auc": 0.8089956622524155,
|
| 1467 |
+
"eval_runtime": 10.0647,
|
| 1468 |
+
"eval_samples_per_second": 97.469,
|
| 1469 |
+
"eval_steps_per_second": 6.558,
|
| 1470 |
+
"step": 45064
|
| 1471 |
+
},
|
| 1472 |
+
{
|
| 1473 |
+
"epoch": 86.83,
|
| 1474 |
+
"learning_rate": 6.583015267175573e-07,
|
| 1475 |
+
"loss": 0.0051,
|
| 1476 |
+
"step": 45501
|
| 1477 |
+
},
|
| 1478 |
+
{
|
| 1479 |
+
"epoch": 87.0,
|
| 1480 |
+
"eval_accuracy": 0.5555555555555556,
|
| 1481 |
+
"eval_f1": 0.7070967741935484,
|
| 1482 |
+
"eval_loss": 1.0698884725570679,
|
| 1483 |
+
"eval_roc_auc": 0.8076025337954145,
|
| 1484 |
+
"eval_runtime": 10.0544,
|
| 1485 |
+
"eval_samples_per_second": 97.569,
|
| 1486 |
+
"eval_steps_per_second": 6.564,
|
| 1487 |
+
"step": 45588
|
| 1488 |
+
},
|
| 1489 |
+
{
|
| 1490 |
+
"epoch": 87.83,
|
| 1491 |
+
"learning_rate": 6.083969465648855e-07,
|
| 1492 |
+
"loss": 0.0059,
|
| 1493 |
+
"step": 46024
|
| 1494 |
+
},
|
| 1495 |
+
{
|
| 1496 |
+
"epoch": 88.0,
|
| 1497 |
+
"eval_accuracy": 0.5494393476044852,
|
| 1498 |
+
"eval_f1": 0.7032119914346895,
|
| 1499 |
+
"eval_loss": 1.073889970779419,
|
| 1500 |
+
"eval_roc_auc": 0.8057027610107641,
|
| 1501 |
+
"eval_runtime": 10.0443,
|
| 1502 |
+
"eval_samples_per_second": 97.668,
|
| 1503 |
+
"eval_steps_per_second": 6.571,
|
| 1504 |
+
"step": 46112
|
| 1505 |
+
},
|
| 1506 |
+
{
|
| 1507 |
+
"epoch": 88.83,
|
| 1508 |
+
"learning_rate": 5.584923664122137e-07,
|
| 1509 |
+
"loss": 0.0067,
|
| 1510 |
+
"step": 46547
|
| 1511 |
+
},
|
| 1512 |
+
{
|
| 1513 |
+
"epoch": 89.0,
|
| 1514 |
+
"eval_accuracy": 0.5535168195718655,
|
| 1515 |
+
"eval_f1": 0.7074422583404619,
|
| 1516 |
+
"eval_loss": 1.0765037536621094,
|
| 1517 |
+
"eval_roc_auc": 0.808678937825618,
|
| 1518 |
+
"eval_runtime": 10.0208,
|
| 1519 |
+
"eval_samples_per_second": 97.897,
|
| 1520 |
+
"eval_steps_per_second": 6.586,
|
| 1521 |
+
"step": 46636
|
| 1522 |
+
},
|
| 1523 |
+
{
|
| 1524 |
+
"epoch": 89.83,
|
| 1525 |
+
"learning_rate": 5.085877862595421e-07,
|
| 1526 |
+
"loss": 0.0055,
|
| 1527 |
+
"step": 47070
|
| 1528 |
+
},
|
| 1529 |
+
{
|
| 1530 |
+
"epoch": 90.0,
|
| 1531 |
+
"eval_accuracy": 0.5423037716615698,
|
| 1532 |
+
"eval_f1": 0.705531914893617,
|
| 1533 |
+
"eval_loss": 1.0747418403625488,
|
| 1534 |
+
"eval_roc_auc": 0.8082004085286085,
|
| 1535 |
+
"eval_runtime": 10.0179,
|
| 1536 |
+
"eval_samples_per_second": 97.924,
|
| 1537 |
+
"eval_steps_per_second": 6.588,
|
| 1538 |
+
"step": 47160
|
| 1539 |
+
},
|
| 1540 |
+
{
|
| 1541 |
+
"epoch": 90.83,
|
| 1542 |
+
"learning_rate": 4.586832061068703e-07,
|
| 1543 |
+
"loss": 0.0041,
|
| 1544 |
+
"step": 47593
|
| 1545 |
+
},
|
| 1546 |
+
{
|
| 1547 |
+
"epoch": 91.0,
|
| 1548 |
+
"eval_accuracy": 0.5524974515800204,
|
| 1549 |
+
"eval_f1": 0.7061855670103093,
|
| 1550 |
+
"eval_loss": 1.077362060546875,
|
| 1551 |
+
"eval_roc_auc": 0.8072014642766979,
|
| 1552 |
+
"eval_runtime": 9.974,
|
| 1553 |
+
"eval_samples_per_second": 98.356,
|
| 1554 |
+
"eval_steps_per_second": 6.617,
|
| 1555 |
+
"step": 47684
|
| 1556 |
+
},
|
| 1557 |
+
{
|
| 1558 |
+
"epoch": 91.82,
|
| 1559 |
+
"learning_rate": 4.0877862595419847e-07,
|
| 1560 |
+
"loss": 0.0051,
|
| 1561 |
+
"step": 48116
|
| 1562 |
+
},
|
| 1563 |
+
{
|
| 1564 |
+
"epoch": 92.0,
|
| 1565 |
+
"eval_accuracy": 0.5565749235474006,
|
| 1566 |
+
"eval_f1": 0.706792777300086,
|
| 1567 |
+
"eval_loss": 1.0821139812469482,
|
| 1568 |
+
"eval_roc_auc": 0.8074688439558423,
|
| 1569 |
+
"eval_runtime": 9.9576,
|
| 1570 |
+
"eval_samples_per_second": 98.517,
|
| 1571 |
+
"eval_steps_per_second": 6.628,
|
| 1572 |
+
"step": 48208
|
| 1573 |
+
},
|
| 1574 |
+
{
|
| 1575 |
+
"epoch": 92.82,
|
| 1576 |
+
"learning_rate": 3.588740458015268e-07,
|
| 1577 |
+
"loss": 0.0053,
|
| 1578 |
+
"step": 48639
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 93.0,
|
| 1582 |
+
"eval_accuracy": 0.5535168195718655,
|
| 1583 |
+
"eval_f1": 0.7076526225279449,
|
| 1584 |
+
"eval_loss": 1.0813453197479248,
|
| 1585 |
+
"eval_roc_auc": 0.8080317183447706,
|
| 1586 |
+
"eval_runtime": 10.0101,
|
| 1587 |
+
"eval_samples_per_second": 98.001,
|
| 1588 |
+
"eval_steps_per_second": 6.593,
|
| 1589 |
+
"step": 48732
|
| 1590 |
+
},
|
| 1591 |
+
{
|
| 1592 |
+
"epoch": 93.82,
|
| 1593 |
+
"learning_rate": 3.08969465648855e-07,
|
| 1594 |
+
"loss": 0.0052,
|
| 1595 |
+
"step": 49162
|
| 1596 |
+
},
|
| 1597 |
+
{
|
| 1598 |
+
"epoch": 94.0,
|
| 1599 |
+
"eval_accuracy": 0.5494393476044852,
|
| 1600 |
+
"eval_f1": 0.7078364565587735,
|
| 1601 |
+
"eval_loss": 1.0872832536697388,
|
| 1602 |
+
"eval_roc_auc": 0.8095935369856098,
|
| 1603 |
+
"eval_runtime": 10.0385,
|
| 1604 |
+
"eval_samples_per_second": 97.724,
|
| 1605 |
+
"eval_steps_per_second": 6.575,
|
| 1606 |
+
"step": 49256
|
| 1607 |
+
},
|
| 1608 |
+
{
|
| 1609 |
+
"epoch": 94.82,
|
| 1610 |
+
"learning_rate": 2.5906488549618325e-07,
|
| 1611 |
+
"loss": 0.0049,
|
| 1612 |
+
"step": 49685
|
| 1613 |
+
},
|
| 1614 |
+
{
|
| 1615 |
+
"epoch": 95.0,
|
| 1616 |
+
"eval_accuracy": 0.5504587155963303,
|
| 1617 |
+
"eval_f1": 0.7095681625740897,
|
| 1618 |
+
"eval_loss": 1.0950355529785156,
|
| 1619 |
+
"eval_roc_auc": 0.8116619999540979,
|
| 1620 |
+
"eval_runtime": 10.0471,
|
| 1621 |
+
"eval_samples_per_second": 97.64,
|
| 1622 |
+
"eval_steps_per_second": 6.569,
|
| 1623 |
+
"step": 49780
|
| 1624 |
+
},
|
| 1625 |
+
{
|
| 1626 |
+
"epoch": 95.82,
|
| 1627 |
+
"learning_rate": 2.0916030534351148e-07,
|
| 1628 |
+
"loss": 0.0055,
|
| 1629 |
+
"step": 50208
|
| 1630 |
+
},
|
| 1631 |
+
{
|
| 1632 |
+
"epoch": 96.0,
|
| 1633 |
+
"eval_accuracy": 0.54638124362895,
|
| 1634 |
+
"eval_f1": 0.7064846416382252,
|
| 1635 |
+
"eval_loss": 1.091185450553894,
|
| 1636 |
+
"eval_roc_auc": 0.8084396731771133,
|
| 1637 |
+
"eval_runtime": 10.0192,
|
| 1638 |
+
"eval_samples_per_second": 97.912,
|
| 1639 |
+
"eval_steps_per_second": 6.587,
|
| 1640 |
+
"step": 50304
|
| 1641 |
+
},
|
| 1642 |
+
{
|
| 1643 |
+
"epoch": 96.81,
|
| 1644 |
+
"learning_rate": 1.5925572519083971e-07,
|
| 1645 |
+
"loss": 0.0048,
|
| 1646 |
+
"step": 50731
|
| 1647 |
+
},
|
| 1648 |
+
{
|
| 1649 |
+
"epoch": 97.0,
|
| 1650 |
+
"eval_accuracy": 0.5555555555555556,
|
| 1651 |
+
"eval_f1": 0.70926243567753,
|
| 1652 |
+
"eval_loss": 1.0918930768966675,
|
| 1653 |
+
"eval_roc_auc": 0.809481076863051,
|
| 1654 |
+
"eval_runtime": 10.1024,
|
| 1655 |
+
"eval_samples_per_second": 97.106,
|
| 1656 |
+
"eval_steps_per_second": 6.533,
|
| 1657 |
+
"step": 50828
|
| 1658 |
+
},
|
| 1659 |
+
{
|
| 1660 |
+
"epoch": 97.81,
|
| 1661 |
+
"learning_rate": 1.0935114503816793e-07,
|
| 1662 |
+
"loss": 0.0052,
|
| 1663 |
+
"step": 51254
|
| 1664 |
+
},
|
| 1665 |
+
{
|
| 1666 |
+
"epoch": 98.0,
|
| 1667 |
+
"eval_accuracy": 0.5514780835881753,
|
| 1668 |
+
"eval_f1": 0.7080479452054794,
|
| 1669 |
+
"eval_loss": 1.0924346446990967,
|
| 1670 |
+
"eval_roc_auc": 0.8089463175047622,
|
| 1671 |
+
"eval_runtime": 10.042,
|
| 1672 |
+
"eval_samples_per_second": 97.689,
|
| 1673 |
+
"eval_steps_per_second": 6.572,
|
| 1674 |
+
"step": 51352
|
| 1675 |
+
},
|
| 1676 |
+
{
|
| 1677 |
+
"epoch": 98.81,
|
| 1678 |
+
"learning_rate": 5.9446564885496193e-08,
|
| 1679 |
+
"loss": 0.0041,
|
| 1680 |
+
"step": 51777
|
| 1681 |
+
},
|
| 1682 |
+
{
|
| 1683 |
+
"epoch": 99.0,
|
| 1684 |
+
"eval_accuracy": 0.5504587155963303,
|
| 1685 |
+
"eval_f1": 0.70824434002563,
|
| 1686 |
+
"eval_loss": 1.0939857959747314,
|
| 1687 |
+
"eval_roc_auc": 0.8094036170847583,
|
| 1688 |
+
"eval_runtime": 10.0515,
|
| 1689 |
+
"eval_samples_per_second": 97.598,
|
| 1690 |
+
"eval_steps_per_second": 6.566,
|
| 1691 |
+
"step": 51876
|
| 1692 |
+
},
|
| 1693 |
+
{
|
| 1694 |
+
"epoch": 99.81,
|
| 1695 |
+
"learning_rate": 9.541984732824428e-09,
|
| 1696 |
+
"loss": 0.0043,
|
| 1697 |
+
"step": 52300
|
| 1698 |
+
},
|
| 1699 |
+
{
|
| 1700 |
+
"epoch": 100.0,
|
| 1701 |
+
"eval_accuracy": 0.5514780835881753,
|
| 1702 |
+
"eval_f1": 0.7073378839590444,
|
| 1703 |
+
"eval_loss": 1.0944114923477173,
|
| 1704 |
+
"eval_roc_auc": 0.8090025475660416,
|
| 1705 |
+
"eval_runtime": 10.0289,
|
| 1706 |
+
"eval_samples_per_second": 97.818,
|
| 1707 |
+
"eval_steps_per_second": 6.581,
|
| 1708 |
+
"step": 52400
|
| 1709 |
+
}
|
| 1710 |
+
],
|
| 1711 |
+
"max_steps": 52400,
|
| 1712 |
+
"num_train_epochs": 100,
|
| 1713 |
+
"total_flos": 1.03939481527296e+17,
|
| 1714 |
+
"trial_name": null,
|
| 1715 |
+
"trial_params": null
|
| 1716 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cf631f4c0b74fa8d3b4cdb53d55bcf1b57bcba0abafbe6b794fc7456083a0dce
|
| 3 |
+
size 3323
|
vocab.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|