upload checkpoint
Browse files- config.json +56 -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.json +0 -0
- tokenizer_config.json +14 -0
- trainer_state.json +1078 -0
- training_args.bin +3 -0
- vocab.txt +0 -0
config.json
ADDED
|
@@ -0,0 +1,56 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "google/electra-large-discriminator",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"ElectraForTokenClassification"
|
| 5 |
+
],
|
| 6 |
+
"attention_probs_dropout_prob": 0.1,
|
| 7 |
+
"classifier_dropout": null,
|
| 8 |
+
"embedding_size": 1024,
|
| 9 |
+
"hidden_act": "gelu",
|
| 10 |
+
"hidden_dropout_prob": 0.1,
|
| 11 |
+
"hidden_size": 1024,
|
| 12 |
+
"id2label": {
|
| 13 |
+
"0": "B-Value",
|
| 14 |
+
"1": "I-Value",
|
| 15 |
+
"2": "B-Rhetorical",
|
| 16 |
+
"3": "I-Rhetorical",
|
| 17 |
+
"4": "B-Policy",
|
| 18 |
+
"5": "I-Policy",
|
| 19 |
+
"6": "B-Testimony",
|
| 20 |
+
"7": "I-Testimony",
|
| 21 |
+
"8": "B-Fact",
|
| 22 |
+
"9": "I-Fact",
|
| 23 |
+
"10": "O"
|
| 24 |
+
},
|
| 25 |
+
"initializer_range": 0.02,
|
| 26 |
+
"intermediate_size": 4096,
|
| 27 |
+
"label2id": {
|
| 28 |
+
"B-Fact": 8,
|
| 29 |
+
"B-Policy": 4,
|
| 30 |
+
"B-Rhetorical": 2,
|
| 31 |
+
"B-Testimony": 6,
|
| 32 |
+
"B-Value": 0,
|
| 33 |
+
"I-Fact": 9,
|
| 34 |
+
"I-Policy": 5,
|
| 35 |
+
"I-Rhetorical": 3,
|
| 36 |
+
"I-Testimony": 7,
|
| 37 |
+
"I-Value": 1,
|
| 38 |
+
"O": 10
|
| 39 |
+
},
|
| 40 |
+
"layer_norm_eps": 1e-12,
|
| 41 |
+
"max_position_embeddings": 512,
|
| 42 |
+
"model_type": "electra",
|
| 43 |
+
"num_attention_heads": 16,
|
| 44 |
+
"num_hidden_layers": 24,
|
| 45 |
+
"pad_token_id": 0,
|
| 46 |
+
"position_embedding_type": "absolute",
|
| 47 |
+
"summary_activation": "gelu",
|
| 48 |
+
"summary_last_dropout": 0.1,
|
| 49 |
+
"summary_type": "first",
|
| 50 |
+
"summary_use_proj": true,
|
| 51 |
+
"torch_dtype": "float32",
|
| 52 |
+
"transformers_version": "4.28.0",
|
| 53 |
+
"type_vocab_size": 2,
|
| 54 |
+
"use_cache": true,
|
| 55 |
+
"vocab_size": 30522
|
| 56 |
+
}
|
optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4124981aff2725e0f7473627166c089a1d840dca77bc075515b263a99c235949
|
| 3 |
+
size 2673058117
|
pytorch_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9b517abc163f44b6d10d19ef623748445806502b8c38cd831c6609867d79bea8
|
| 3 |
+
size 1336553581
|
rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6cfe0e350a2f4a45becf30b20f10937f77ff255d808315a9d2e604eee24faafb
|
| 3 |
+
size 14575
|
scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:26fda3babc655c4d4878776d8c39929f31b08be1444c387e1824c93bcb6713ff
|
| 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.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1,14 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_prefix_space": true,
|
| 3 |
+
"clean_up_tokenization_spaces": true,
|
| 4 |
+
"cls_token": "[CLS]",
|
| 5 |
+
"do_lower_case": false,
|
| 6 |
+
"mask_token": "[MASK]",
|
| 7 |
+
"model_max_length": 512,
|
| 8 |
+
"pad_token": "[PAD]",
|
| 9 |
+
"sep_token": "[SEP]",
|
| 10 |
+
"strip_accents": null,
|
| 11 |
+
"tokenize_chinese_chars": true,
|
| 12 |
+
"tokenizer_class": "ElectraTokenizer",
|
| 13 |
+
"unk_token": "[UNK]"
|
| 14 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1078 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 86.0,
|
| 5 |
+
"global_step": 2752,
|
| 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 |
+
"eval_accuracy": 0.6175584724649537,
|
| 13 |
+
"eval_f1": 0.0769764216366158,
|
| 14 |
+
"eval_loss": 1.250605583190918,
|
| 15 |
+
"eval_precision": 0.06442251886244922,
|
| 16 |
+
"eval_recall": 0.09560723514211886,
|
| 17 |
+
"eval_runtime": 3.5402,
|
| 18 |
+
"eval_samples_per_second": 26.553,
|
| 19 |
+
"eval_steps_per_second": 3.39,
|
| 20 |
+
"step": 32
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 2.0,
|
| 24 |
+
"eval_accuracy": 0.6520655932770609,
|
| 25 |
+
"eval_f1": 0.18170731707317075,
|
| 26 |
+
"eval_loss": 1.0688396692276,
|
| 27 |
+
"eval_precision": 0.14063237376120813,
|
| 28 |
+
"eval_recall": 0.2566752799310939,
|
| 29 |
+
"eval_runtime": 3.2469,
|
| 30 |
+
"eval_samples_per_second": 28.95,
|
| 31 |
+
"eval_steps_per_second": 3.696,
|
| 32 |
+
"step": 64
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 3.0,
|
| 36 |
+
"eval_accuracy": 0.733239132859852,
|
| 37 |
+
"eval_f1": 0.26462128475551294,
|
| 38 |
+
"eval_loss": 0.8916715383529663,
|
| 39 |
+
"eval_precision": 0.21036585365853658,
|
| 40 |
+
"eval_recall": 0.35658914728682173,
|
| 41 |
+
"eval_runtime": 3.5217,
|
| 42 |
+
"eval_samples_per_second": 26.692,
|
| 43 |
+
"eval_steps_per_second": 3.407,
|
| 44 |
+
"step": 96
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 4.0,
|
| 48 |
+
"eval_accuracy": 0.7502695868813446,
|
| 49 |
+
"eval_f1": 0.31227821149751595,
|
| 50 |
+
"eval_loss": 0.8893136382102966,
|
| 51 |
+
"eval_precision": 0.2655401327700664,
|
| 52 |
+
"eval_recall": 0.3789836347975883,
|
| 53 |
+
"eval_runtime": 3.1549,
|
| 54 |
+
"eval_samples_per_second": 29.795,
|
| 55 |
+
"eval_steps_per_second": 3.804,
|
| 56 |
+
"step": 128
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 5.0,
|
| 60 |
+
"eval_accuracy": 0.7515338563938572,
|
| 61 |
+
"eval_f1": 0.3259361997226075,
|
| 62 |
+
"eval_loss": 0.8177661299705505,
|
| 63 |
+
"eval_precision": 0.2727800348229832,
|
| 64 |
+
"eval_recall": 0.40482342807924204,
|
| 65 |
+
"eval_runtime": 3.1577,
|
| 66 |
+
"eval_samples_per_second": 29.768,
|
| 67 |
+
"eval_steps_per_second": 3.8,
|
| 68 |
+
"step": 160
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 6.0,
|
| 72 |
+
"eval_accuracy": 0.7685271260179228,
|
| 73 |
+
"eval_f1": 0.3471241170534813,
|
| 74 |
+
"eval_loss": 0.818898618221283,
|
| 75 |
+
"eval_precision": 0.2847682119205298,
|
| 76 |
+
"eval_recall": 0.4444444444444444,
|
| 77 |
+
"eval_runtime": 3.5275,
|
| 78 |
+
"eval_samples_per_second": 26.648,
|
| 79 |
+
"eval_steps_per_second": 3.402,
|
| 80 |
+
"step": 192
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 7.0,
|
| 84 |
+
"eval_accuracy": 0.7693451827613134,
|
| 85 |
+
"eval_f1": 0.39416058394160586,
|
| 86 |
+
"eval_loss": 0.8939387798309326,
|
| 87 |
+
"eval_precision": 0.34198860037998735,
|
| 88 |
+
"eval_recall": 0.46511627906976744,
|
| 89 |
+
"eval_runtime": 4.3875,
|
| 90 |
+
"eval_samples_per_second": 21.424,
|
| 91 |
+
"eval_steps_per_second": 2.735,
|
| 92 |
+
"step": 224
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 8.0,
|
| 96 |
+
"eval_accuracy": 0.7565537500464805,
|
| 97 |
+
"eval_f1": 0.3823943661971831,
|
| 98 |
+
"eval_loss": 0.8892739415168762,
|
| 99 |
+
"eval_precision": 0.323406789755807,
|
| 100 |
+
"eval_recall": 0.46770025839793283,
|
| 101 |
+
"eval_runtime": 3.5968,
|
| 102 |
+
"eval_samples_per_second": 26.135,
|
| 103 |
+
"eval_steps_per_second": 3.336,
|
| 104 |
+
"step": 256
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 9.0,
|
| 108 |
+
"eval_accuracy": 0.7693451827613134,
|
| 109 |
+
"eval_f1": 0.41654247391952315,
|
| 110 |
+
"eval_loss": 0.9633013010025024,
|
| 111 |
+
"eval_precision": 0.3670387393302692,
|
| 112 |
+
"eval_recall": 0.48148148148148145,
|
| 113 |
+
"eval_runtime": 3.1245,
|
| 114 |
+
"eval_samples_per_second": 30.085,
|
| 115 |
+
"eval_steps_per_second": 3.841,
|
| 116 |
+
"step": 288
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 10.0,
|
| 120 |
+
"eval_accuracy": 0.765478005428922,
|
| 121 |
+
"eval_f1": 0.3988900450919181,
|
| 122 |
+
"eval_loss": 0.9256911873817444,
|
| 123 |
+
"eval_precision": 0.3339140534262485,
|
| 124 |
+
"eval_recall": 0.49526270456503013,
|
| 125 |
+
"eval_runtime": 3.0684,
|
| 126 |
+
"eval_samples_per_second": 30.635,
|
| 127 |
+
"eval_steps_per_second": 3.911,
|
| 128 |
+
"step": 320
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 11.0,
|
| 132 |
+
"eval_accuracy": 0.7651061614546536,
|
| 133 |
+
"eval_f1": 0.4374537379718727,
|
| 134 |
+
"eval_loss": 1.0472618341445923,
|
| 135 |
+
"eval_precision": 0.3835171966255678,
|
| 136 |
+
"eval_recall": 0.5090439276485789,
|
| 137 |
+
"eval_runtime": 15.8515,
|
| 138 |
+
"eval_samples_per_second": 5.93,
|
| 139 |
+
"eval_steps_per_second": 0.757,
|
| 140 |
+
"step": 352
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 12.0,
|
| 144 |
+
"eval_accuracy": 0.7513107500092961,
|
| 145 |
+
"eval_f1": 0.4116819141449683,
|
| 146 |
+
"eval_loss": 1.0901930332183838,
|
| 147 |
+
"eval_precision": 0.34800713860797144,
|
| 148 |
+
"eval_recall": 0.5038759689922481,
|
| 149 |
+
"eval_runtime": 3.3377,
|
| 150 |
+
"eval_samples_per_second": 28.163,
|
| 151 |
+
"eval_steps_per_second": 3.595,
|
| 152 |
+
"step": 384
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 13.0,
|
| 156 |
+
"eval_accuracy": 0.7735842040679731,
|
| 157 |
+
"eval_f1": 0.4424198250728863,
|
| 158 |
+
"eval_loss": 1.0164769887924194,
|
| 159 |
+
"eval_precision": 0.38344914718888184,
|
| 160 |
+
"eval_recall": 0.5228251507321274,
|
| 161 |
+
"eval_runtime": 3.1559,
|
| 162 |
+
"eval_samples_per_second": 29.785,
|
| 163 |
+
"eval_steps_per_second": 3.802,
|
| 164 |
+
"step": 416
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 14.0,
|
| 168 |
+
"eval_accuracy": 0.7737329416576805,
|
| 169 |
+
"eval_f1": 0.46205860255447034,
|
| 170 |
+
"eval_loss": 1.0768243074417114,
|
| 171 |
+
"eval_precision": 0.4097268487674883,
|
| 172 |
+
"eval_recall": 0.5297157622739018,
|
| 173 |
+
"eval_runtime": 3.0915,
|
| 174 |
+
"eval_samples_per_second": 30.406,
|
| 175 |
+
"eval_steps_per_second": 3.882,
|
| 176 |
+
"step": 448
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 15.0,
|
| 180 |
+
"eval_accuracy": 0.7817647715018778,
|
| 181 |
+
"eval_f1": 0.4762264150943396,
|
| 182 |
+
"eval_loss": 1.1149170398712158,
|
| 183 |
+
"eval_precision": 0.4237743451981195,
|
| 184 |
+
"eval_recall": 0.5434969853574505,
|
| 185 |
+
"eval_runtime": 3.2505,
|
| 186 |
+
"eval_samples_per_second": 28.919,
|
| 187 |
+
"eval_steps_per_second": 3.692,
|
| 188 |
+
"step": 480
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 15.62,
|
| 192 |
+
"learning_rate": 1.6875e-05,
|
| 193 |
+
"loss": 0.4087,
|
| 194 |
+
"step": 500
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 16.0,
|
| 198 |
+
"eval_accuracy": 0.7796452608485479,
|
| 199 |
+
"eval_f1": 0.4698749526335733,
|
| 200 |
+
"eval_loss": 1.1298694610595703,
|
| 201 |
+
"eval_precision": 0.41948579161028415,
|
| 202 |
+
"eval_recall": 0.5340223944875108,
|
| 203 |
+
"eval_runtime": 3.0471,
|
| 204 |
+
"eval_samples_per_second": 30.849,
|
| 205 |
+
"eval_steps_per_second": 3.938,
|
| 206 |
+
"step": 512
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 17.0,
|
| 210 |
+
"eval_accuracy": 0.7883464098464285,
|
| 211 |
+
"eval_f1": 0.4673040152963671,
|
| 212 |
+
"eval_loss": 1.1149996519088745,
|
| 213 |
+
"eval_precision": 0.4202200825309491,
|
| 214 |
+
"eval_recall": 0.5262704565030146,
|
| 215 |
+
"eval_runtime": 3.1689,
|
| 216 |
+
"eval_samples_per_second": 29.664,
|
| 217 |
+
"eval_steps_per_second": 3.787,
|
| 218 |
+
"step": 544
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 18.0,
|
| 222 |
+
"eval_accuracy": 0.7801286580150969,
|
| 223 |
+
"eval_f1": 0.45127436281859074,
|
| 224 |
+
"eval_loss": 1.1859441995620728,
|
| 225 |
+
"eval_precision": 0.39946914399469147,
|
| 226 |
+
"eval_recall": 0.5185185185185185,
|
| 227 |
+
"eval_runtime": 7.3417,
|
| 228 |
+
"eval_samples_per_second": 12.803,
|
| 229 |
+
"eval_steps_per_second": 1.634,
|
| 230 |
+
"step": 576
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 19.0,
|
| 234 |
+
"eval_accuracy": 0.7609043245454208,
|
| 235 |
+
"eval_f1": 0.4414210128495843,
|
| 236 |
+
"eval_loss": 1.2266000509262085,
|
| 237 |
+
"eval_precision": 0.39326599326599326,
|
| 238 |
+
"eval_recall": 0.5030146425495263,
|
| 239 |
+
"eval_runtime": 3.4668,
|
| 240 |
+
"eval_samples_per_second": 27.114,
|
| 241 |
+
"eval_steps_per_second": 3.461,
|
| 242 |
+
"step": 608
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 20.0,
|
| 246 |
+
"eval_accuracy": 0.7837355445655003,
|
| 247 |
+
"eval_f1": 0.4707198806415516,
|
| 248 |
+
"eval_loss": 1.1946580410003662,
|
| 249 |
+
"eval_precision": 0.4151315789473684,
|
| 250 |
+
"eval_recall": 0.5434969853574505,
|
| 251 |
+
"eval_runtime": 3.1596,
|
| 252 |
+
"eval_samples_per_second": 29.75,
|
| 253 |
+
"eval_steps_per_second": 3.798,
|
| 254 |
+
"step": 640
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 21.0,
|
| 258 |
+
"eval_accuracy": 0.7890900977949652,
|
| 259 |
+
"eval_f1": 0.4974698326196964,
|
| 260 |
+
"eval_loss": 1.2461358308792114,
|
| 261 |
+
"eval_precision": 0.4538352272727273,
|
| 262 |
+
"eval_recall": 0.5503875968992248,
|
| 263 |
+
"eval_runtime": 3.1747,
|
| 264 |
+
"eval_samples_per_second": 29.609,
|
| 265 |
+
"eval_steps_per_second": 3.78,
|
| 266 |
+
"step": 672
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 22.0,
|
| 270 |
+
"eval_accuracy": 0.7804261331945116,
|
| 271 |
+
"eval_f1": 0.4870506378044067,
|
| 272 |
+
"eval_loss": 1.2504911422729492,
|
| 273 |
+
"eval_precision": 0.4417952314165498,
|
| 274 |
+
"eval_recall": 0.5426356589147286,
|
| 275 |
+
"eval_runtime": 3.1491,
|
| 276 |
+
"eval_samples_per_second": 29.849,
|
| 277 |
+
"eval_steps_per_second": 3.811,
|
| 278 |
+
"step": 704
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 23.0,
|
| 282 |
+
"eval_accuracy": 0.7844420481166102,
|
| 283 |
+
"eval_f1": 0.4932249322493225,
|
| 284 |
+
"eval_loss": 1.287372350692749,
|
| 285 |
+
"eval_precision": 0.4479606188466948,
|
| 286 |
+
"eval_recall": 0.5486649440137812,
|
| 287 |
+
"eval_runtime": 3.7236,
|
| 288 |
+
"eval_samples_per_second": 25.244,
|
| 289 |
+
"eval_steps_per_second": 3.223,
|
| 290 |
+
"step": 736
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 24.0,
|
| 294 |
+
"eval_accuracy": 0.7877886438850259,
|
| 295 |
+
"eval_f1": 0.49941107184923444,
|
| 296 |
+
"eval_loss": 1.277133584022522,
|
| 297 |
+
"eval_precision": 0.4588744588744589,
|
| 298 |
+
"eval_recall": 0.5478036175710594,
|
| 299 |
+
"eval_runtime": 3.2052,
|
| 300 |
+
"eval_samples_per_second": 29.328,
|
| 301 |
+
"eval_steps_per_second": 3.744,
|
| 302 |
+
"step": 768
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 25.0,
|
| 306 |
+
"eval_accuracy": 0.7892760197820994,
|
| 307 |
+
"eval_f1": 0.4747201852566576,
|
| 308 |
+
"eval_loss": 1.2719131708145142,
|
| 309 |
+
"eval_precision": 0.43006993006993005,
|
| 310 |
+
"eval_recall": 0.5297157622739018,
|
| 311 |
+
"eval_runtime": 3.0524,
|
| 312 |
+
"eval_samples_per_second": 30.795,
|
| 313 |
+
"eval_steps_per_second": 3.931,
|
| 314 |
+
"step": 800
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 26.0,
|
| 318 |
+
"eval_accuracy": 0.7896478637563679,
|
| 319 |
+
"eval_f1": 0.49407265774378584,
|
| 320 |
+
"eval_loss": 1.2584308385849,
|
| 321 |
+
"eval_precision": 0.4442916093535076,
|
| 322 |
+
"eval_recall": 0.5564168819982773,
|
| 323 |
+
"eval_runtime": 3.7701,
|
| 324 |
+
"eval_samples_per_second": 24.933,
|
| 325 |
+
"eval_steps_per_second": 3.183,
|
| 326 |
+
"step": 832
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 27.0,
|
| 330 |
+
"eval_accuracy": 0.7868590339493549,
|
| 331 |
+
"eval_f1": 0.4920077972709551,
|
| 332 |
+
"eval_loss": 1.3040637969970703,
|
| 333 |
+
"eval_precision": 0.4494301994301994,
|
| 334 |
+
"eval_recall": 0.5434969853574505,
|
| 335 |
+
"eval_runtime": 3.4169,
|
| 336 |
+
"eval_samples_per_second": 27.51,
|
| 337 |
+
"eval_steps_per_second": 3.512,
|
| 338 |
+
"step": 864
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 28.0,
|
| 342 |
+
"eval_accuracy": 0.7880861190644406,
|
| 343 |
+
"eval_f1": 0.4916158536585366,
|
| 344 |
+
"eval_loss": 1.3579275608062744,
|
| 345 |
+
"eval_precision": 0.4408749145591251,
|
| 346 |
+
"eval_recall": 0.5555555555555556,
|
| 347 |
+
"eval_runtime": 3.4487,
|
| 348 |
+
"eval_samples_per_second": 27.256,
|
| 349 |
+
"eval_steps_per_second": 3.48,
|
| 350 |
+
"step": 896
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 29.0,
|
| 354 |
+
"eval_accuracy": 0.7839958353474882,
|
| 355 |
+
"eval_f1": 0.501557632398754,
|
| 356 |
+
"eval_loss": 1.3749990463256836,
|
| 357 |
+
"eval_precision": 0.4577114427860697,
|
| 358 |
+
"eval_recall": 0.5546942291128337,
|
| 359 |
+
"eval_runtime": 3.131,
|
| 360 |
+
"eval_samples_per_second": 30.022,
|
| 361 |
+
"eval_steps_per_second": 3.833,
|
| 362 |
+
"step": 928
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 30.0,
|
| 366 |
+
"eval_accuracy": 0.7871936935261964,
|
| 367 |
+
"eval_f1": 0.4903883876029816,
|
| 368 |
+
"eval_loss": 1.3733103275299072,
|
| 369 |
+
"eval_precision": 0.45028818443804036,
|
| 370 |
+
"eval_recall": 0.5383290267011197,
|
| 371 |
+
"eval_runtime": 3.4968,
|
| 372 |
+
"eval_samples_per_second": 26.882,
|
| 373 |
+
"eval_steps_per_second": 3.432,
|
| 374 |
+
"step": 960
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 31.0,
|
| 378 |
+
"eval_accuracy": 0.7819135090915852,
|
| 379 |
+
"eval_f1": 0.5099260412611912,
|
| 380 |
+
"eval_loss": 1.395755410194397,
|
| 381 |
+
"eval_precision": 0.46519886363636365,
|
| 382 |
+
"eval_recall": 0.5641688199827735,
|
| 383 |
+
"eval_runtime": 3.2027,
|
| 384 |
+
"eval_samples_per_second": 29.351,
|
| 385 |
+
"eval_steps_per_second": 3.747,
|
| 386 |
+
"step": 992
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 31.25,
|
| 390 |
+
"learning_rate": 1.375e-05,
|
| 391 |
+
"loss": 0.0138,
|
| 392 |
+
"step": 1000
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 32.0,
|
| 396 |
+
"eval_accuracy": 0.7822481686684267,
|
| 397 |
+
"eval_f1": 0.49709639953542384,
|
| 398 |
+
"eval_loss": 1.4415804147720337,
|
| 399 |
+
"eval_precision": 0.45147679324894513,
|
| 400 |
+
"eval_recall": 0.5529715762273901,
|
| 401 |
+
"eval_runtime": 3.081,
|
| 402 |
+
"eval_samples_per_second": 30.51,
|
| 403 |
+
"eval_steps_per_second": 3.895,
|
| 404 |
+
"step": 1024
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 33.0,
|
| 408 |
+
"eval_accuracy": 0.7877886438850259,
|
| 409 |
+
"eval_f1": 0.49035187287173665,
|
| 410 |
+
"eval_loss": 1.384263277053833,
|
| 411 |
+
"eval_precision": 0.43724696356275305,
|
| 412 |
+
"eval_recall": 0.5581395348837209,
|
| 413 |
+
"eval_runtime": 3.7036,
|
| 414 |
+
"eval_samples_per_second": 25.381,
|
| 415 |
+
"eval_steps_per_second": 3.24,
|
| 416 |
+
"step": 1056
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 34.0,
|
| 420 |
+
"eval_accuracy": 0.7796080764511211,
|
| 421 |
+
"eval_f1": 0.490521327014218,
|
| 422 |
+
"eval_loss": 1.4162341356277466,
|
| 423 |
+
"eval_precision": 0.45295404814004375,
|
| 424 |
+
"eval_recall": 0.5348837209302325,
|
| 425 |
+
"eval_runtime": 3.1016,
|
| 426 |
+
"eval_samples_per_second": 30.307,
|
| 427 |
+
"eval_steps_per_second": 3.869,
|
| 428 |
+
"step": 1088
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 35.0,
|
| 432 |
+
"eval_accuracy": 0.7906146580894656,
|
| 433 |
+
"eval_f1": 0.5060240963855421,
|
| 434 |
+
"eval_loss": 1.380549669265747,
|
| 435 |
+
"eval_precision": 0.46104815864022664,
|
| 436 |
+
"eval_recall": 0.5607235142118863,
|
| 437 |
+
"eval_runtime": 3.5552,
|
| 438 |
+
"eval_samples_per_second": 26.44,
|
| 439 |
+
"eval_steps_per_second": 3.375,
|
| 440 |
+
"step": 1120
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 36.0,
|
| 444 |
+
"eval_accuracy": 0.7868590339493549,
|
| 445 |
+
"eval_f1": 0.4811210587777345,
|
| 446 |
+
"eval_loss": 1.4221726655960083,
|
| 447 |
+
"eval_precision": 0.43892045454545453,
|
| 448 |
+
"eval_recall": 0.5322997416020672,
|
| 449 |
+
"eval_runtime": 3.1728,
|
| 450 |
+
"eval_samples_per_second": 29.627,
|
| 451 |
+
"eval_steps_per_second": 3.782,
|
| 452 |
+
"step": 1152
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 37.0,
|
| 456 |
+
"eval_accuracy": 0.7899453389357826,
|
| 457 |
+
"eval_f1": 0.502147598594299,
|
| 458 |
+
"eval_loss": 1.407759189605713,
|
| 459 |
+
"eval_precision": 0.4592857142857143,
|
| 460 |
+
"eval_recall": 0.553832902670112,
|
| 461 |
+
"eval_runtime": 3.1234,
|
| 462 |
+
"eval_samples_per_second": 30.096,
|
| 463 |
+
"eval_steps_per_second": 3.842,
|
| 464 |
+
"step": 1184
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 38.0,
|
| 468 |
+
"eval_accuracy": 0.7908377644740266,
|
| 469 |
+
"eval_f1": 0.4872389791183295,
|
| 470 |
+
"eval_loss": 1.4327783584594727,
|
| 471 |
+
"eval_precision": 0.4421052631578947,
|
| 472 |
+
"eval_recall": 0.5426356589147286,
|
| 473 |
+
"eval_runtime": 3.409,
|
| 474 |
+
"eval_samples_per_second": 27.574,
|
| 475 |
+
"eval_steps_per_second": 3.52,
|
| 476 |
+
"step": 1216
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 39.0,
|
| 480 |
+
"eval_accuracy": 0.7851113672702934,
|
| 481 |
+
"eval_f1": 0.4913227921326649,
|
| 482 |
+
"eval_loss": 1.4441440105438232,
|
| 483 |
+
"eval_precision": 0.4448324022346369,
|
| 484 |
+
"eval_recall": 0.5486649440137812,
|
| 485 |
+
"eval_runtime": 3.2678,
|
| 486 |
+
"eval_samples_per_second": 28.766,
|
| 487 |
+
"eval_steps_per_second": 3.672,
|
| 488 |
+
"step": 1248
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 40.0,
|
| 492 |
+
"eval_accuracy": 0.7912839772431488,
|
| 493 |
+
"eval_f1": 0.5036864571206829,
|
| 494 |
+
"eval_loss": 1.4089561700820923,
|
| 495 |
+
"eval_precision": 0.4583333333333333,
|
| 496 |
+
"eval_recall": 0.5590008613264428,
|
| 497 |
+
"eval_runtime": 3.1449,
|
| 498 |
+
"eval_samples_per_second": 29.89,
|
| 499 |
+
"eval_steps_per_second": 3.816,
|
| 500 |
+
"step": 1280
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 41.0,
|
| 504 |
+
"eval_accuracy": 0.7902428141151973,
|
| 505 |
+
"eval_f1": 0.503858024691358,
|
| 506 |
+
"eval_loss": 1.4232546091079712,
|
| 507 |
+
"eval_precision": 0.45632424877707894,
|
| 508 |
+
"eval_recall": 0.5624461670973299,
|
| 509 |
+
"eval_runtime": 8.2173,
|
| 510 |
+
"eval_samples_per_second": 11.439,
|
| 511 |
+
"eval_steps_per_second": 1.46,
|
| 512 |
+
"step": 1312
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 42.0,
|
| 516 |
+
"eval_accuracy": 0.7909493176663073,
|
| 517 |
+
"eval_f1": 0.49484536082474234,
|
| 518 |
+
"eval_loss": 1.4232605695724487,
|
| 519 |
+
"eval_precision": 0.4444444444444444,
|
| 520 |
+
"eval_recall": 0.5581395348837209,
|
| 521 |
+
"eval_runtime": 3.0349,
|
| 522 |
+
"eval_samples_per_second": 30.973,
|
| 523 |
+
"eval_steps_per_second": 3.954,
|
| 524 |
+
"step": 1344
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 43.0,
|
| 528 |
+
"eval_accuracy": 0.7923995091659539,
|
| 529 |
+
"eval_f1": 0.5162324649298597,
|
| 530 |
+
"eval_loss": 1.4146913290023804,
|
| 531 |
+
"eval_precision": 0.4827586206896552,
|
| 532 |
+
"eval_recall": 0.5546942291128337,
|
| 533 |
+
"eval_runtime": 3.288,
|
| 534 |
+
"eval_samples_per_second": 28.589,
|
| 535 |
+
"eval_steps_per_second": 3.65,
|
| 536 |
+
"step": 1376
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 44.0,
|
| 540 |
+
"eval_accuracy": 0.7880117502695869,
|
| 541 |
+
"eval_f1": 0.4955684007707129,
|
| 542 |
+
"eval_loss": 1.4241943359375,
|
| 543 |
+
"eval_precision": 0.44839609483960946,
|
| 544 |
+
"eval_recall": 0.553832902670112,
|
| 545 |
+
"eval_runtime": 6.1717,
|
| 546 |
+
"eval_samples_per_second": 15.231,
|
| 547 |
+
"eval_steps_per_second": 1.944,
|
| 548 |
+
"step": 1408
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 45.0,
|
| 552 |
+
"eval_accuracy": 0.7880861190644406,
|
| 553 |
+
"eval_f1": 0.49018853405155827,
|
| 554 |
+
"eval_loss": 1.421156644821167,
|
| 555 |
+
"eval_precision": 0.4429763560500695,
|
| 556 |
+
"eval_recall": 0.5486649440137812,
|
| 557 |
+
"eval_runtime": 3.1226,
|
| 558 |
+
"eval_samples_per_second": 30.103,
|
| 559 |
+
"eval_steps_per_second": 3.843,
|
| 560 |
+
"step": 1440
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 46.0,
|
| 564 |
+
"eval_accuracy": 0.7936637786784665,
|
| 565 |
+
"eval_f1": 0.5089463220675944,
|
| 566 |
+
"eval_loss": 1.424131989479065,
|
| 567 |
+
"eval_precision": 0.4726735598227474,
|
| 568 |
+
"eval_recall": 0.5512489233419466,
|
| 569 |
+
"eval_runtime": 3.3347,
|
| 570 |
+
"eval_samples_per_second": 28.189,
|
| 571 |
+
"eval_steps_per_second": 3.599,
|
| 572 |
+
"step": 1472
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 46.88,
|
| 576 |
+
"learning_rate": 1.0625e-05,
|
| 577 |
+
"loss": 0.0034,
|
| 578 |
+
"step": 1500
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 47.0,
|
| 582 |
+
"eval_accuracy": 0.7858550552188301,
|
| 583 |
+
"eval_f1": 0.5065398335315101,
|
| 584 |
+
"eval_loss": 1.443221926689148,
|
| 585 |
+
"eval_precision": 0.46916299559471364,
|
| 586 |
+
"eval_recall": 0.5503875968992248,
|
| 587 |
+
"eval_runtime": 3.0632,
|
| 588 |
+
"eval_samples_per_second": 30.687,
|
| 589 |
+
"eval_steps_per_second": 3.917,
|
| 590 |
+
"step": 1504
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 48.0,
|
| 594 |
+
"eval_accuracy": 0.784739523296025,
|
| 595 |
+
"eval_f1": 0.49553398058252424,
|
| 596 |
+
"eval_loss": 1.4684823751449585,
|
| 597 |
+
"eval_precision": 0.4512022630834512,
|
| 598 |
+
"eval_recall": 0.549526270456503,
|
| 599 |
+
"eval_runtime": 3.2701,
|
| 600 |
+
"eval_samples_per_second": 28.745,
|
| 601 |
+
"eval_steps_per_second": 3.67,
|
| 602 |
+
"step": 1536
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 49.0,
|
| 606 |
+
"eval_accuracy": 0.7949280481909791,
|
| 607 |
+
"eval_f1": 0.5152113789016199,
|
| 608 |
+
"eval_loss": 1.4661859273910522,
|
| 609 |
+
"eval_precision": 0.4759124087591241,
|
| 610 |
+
"eval_recall": 0.5615848406546081,
|
| 611 |
+
"eval_runtime": 3.3592,
|
| 612 |
+
"eval_samples_per_second": 27.983,
|
| 613 |
+
"eval_steps_per_second": 3.572,
|
| 614 |
+
"step": 1568
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 50.0,
|
| 618 |
+
"eval_accuracy": 0.7894619417692337,
|
| 619 |
+
"eval_f1": 0.506033476060724,
|
| 620 |
+
"eval_loss": 1.5200412273406982,
|
| 621 |
+
"eval_precision": 0.4616477272727273,
|
| 622 |
+
"eval_recall": 0.5598621877691645,
|
| 623 |
+
"eval_runtime": 3.1539,
|
| 624 |
+
"eval_samples_per_second": 29.805,
|
| 625 |
+
"eval_steps_per_second": 3.805,
|
| 626 |
+
"step": 1600
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 51.0,
|
| 630 |
+
"eval_accuracy": 0.7907262112817461,
|
| 631 |
+
"eval_f1": 0.5102848101265823,
|
| 632 |
+
"eval_loss": 1.49446702003479,
|
| 633 |
+
"eval_precision": 0.4718361375274323,
|
| 634 |
+
"eval_recall": 0.5555555555555556,
|
| 635 |
+
"eval_runtime": 6.9973,
|
| 636 |
+
"eval_samples_per_second": 13.434,
|
| 637 |
+
"eval_steps_per_second": 1.715,
|
| 638 |
+
"step": 1632
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 52.0,
|
| 642 |
+
"eval_accuracy": 0.7916558212174172,
|
| 643 |
+
"eval_f1": 0.5182511030886483,
|
| 644 |
+
"eval_loss": 1.5040817260742188,
|
| 645 |
+
"eval_precision": 0.484984984984985,
|
| 646 |
+
"eval_recall": 0.5564168819982773,
|
| 647 |
+
"eval_runtime": 8.0818,
|
| 648 |
+
"eval_samples_per_second": 11.631,
|
| 649 |
+
"eval_steps_per_second": 1.485,
|
| 650 |
+
"step": 1664
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 53.0,
|
| 654 |
+
"eval_accuracy": 0.7928829063325029,
|
| 655 |
+
"eval_f1": 0.5159489633173845,
|
| 656 |
+
"eval_loss": 1.5629594326019287,
|
| 657 |
+
"eval_precision": 0.4803266518188567,
|
| 658 |
+
"eval_recall": 0.5572782084409992,
|
| 659 |
+
"eval_runtime": 3.423,
|
| 660 |
+
"eval_samples_per_second": 27.462,
|
| 661 |
+
"eval_steps_per_second": 3.506,
|
| 662 |
+
"step": 1696
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 54.0,
|
| 666 |
+
"eval_accuracy": 0.7840702041423419,
|
| 667 |
+
"eval_f1": 0.5010117361392149,
|
| 668 |
+
"eval_loss": 1.5059171915054321,
|
| 669 |
+
"eval_precision": 0.4725190839694656,
|
| 670 |
+
"eval_recall": 0.533161068044789,
|
| 671 |
+
"eval_runtime": 3.1466,
|
| 672 |
+
"eval_samples_per_second": 29.874,
|
| 673 |
+
"eval_steps_per_second": 3.814,
|
| 674 |
+
"step": 1728
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 55.0,
|
| 678 |
+
"eval_accuracy": 0.790763395679173,
|
| 679 |
+
"eval_f1": 0.5078616352201258,
|
| 680 |
+
"eval_loss": 1.5826448202133179,
|
| 681 |
+
"eval_precision": 0.4671005061460593,
|
| 682 |
+
"eval_recall": 0.5564168819982773,
|
| 683 |
+
"eval_runtime": 3.1003,
|
| 684 |
+
"eval_samples_per_second": 30.319,
|
| 685 |
+
"eval_steps_per_second": 3.871,
|
| 686 |
+
"step": 1760
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 56.0,
|
| 690 |
+
"eval_accuracy": 0.791246792845722,
|
| 691 |
+
"eval_f1": 0.520757758968158,
|
| 692 |
+
"eval_loss": 1.5211557149887085,
|
| 693 |
+
"eval_precision": 0.4893939393939394,
|
| 694 |
+
"eval_recall": 0.5564168819982773,
|
| 695 |
+
"eval_runtime": 3.5151,
|
| 696 |
+
"eval_samples_per_second": 26.742,
|
| 697 |
+
"eval_steps_per_second": 3.414,
|
| 698 |
+
"step": 1792
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 57.0,
|
| 702 |
+
"eval_accuracy": 0.7894619417692337,
|
| 703 |
+
"eval_f1": 0.5172550575168584,
|
| 704 |
+
"eval_loss": 1.5194625854492188,
|
| 705 |
+
"eval_precision": 0.47941176470588237,
|
| 706 |
+
"eval_recall": 0.5615848406546081,
|
| 707 |
+
"eval_runtime": 3.1721,
|
| 708 |
+
"eval_samples_per_second": 29.634,
|
| 709 |
+
"eval_steps_per_second": 3.783,
|
| 710 |
+
"step": 1824
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 58.0,
|
| 714 |
+
"eval_accuracy": 0.7864871899750865,
|
| 715 |
+
"eval_f1": 0.5015600624024961,
|
| 716 |
+
"eval_loss": 1.5568981170654297,
|
| 717 |
+
"eval_precision": 0.45830363506771205,
|
| 718 |
+
"eval_recall": 0.553832902670112,
|
| 719 |
+
"eval_runtime": 3.0367,
|
| 720 |
+
"eval_samples_per_second": 30.954,
|
| 721 |
+
"eval_steps_per_second": 3.952,
|
| 722 |
+
"step": 1856
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 59.0,
|
| 726 |
+
"eval_accuracy": 0.789610679358941,
|
| 727 |
+
"eval_f1": 0.5186953062848051,
|
| 728 |
+
"eval_loss": 1.5378462076187134,
|
| 729 |
+
"eval_precision": 0.4818920916481892,
|
| 730 |
+
"eval_recall": 0.5615848406546081,
|
| 731 |
+
"eval_runtime": 3.0368,
|
| 732 |
+
"eval_samples_per_second": 30.954,
|
| 733 |
+
"eval_steps_per_second": 3.952,
|
| 734 |
+
"step": 1888
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 60.0,
|
| 738 |
+
"eval_accuracy": 0.7838842821552077,
|
| 739 |
+
"eval_f1": 0.490257209664848,
|
| 740 |
+
"eval_loss": 1.5430467128753662,
|
| 741 |
+
"eval_precision": 0.44768683274021354,
|
| 742 |
+
"eval_recall": 0.5417743324720069,
|
| 743 |
+
"eval_runtime": 3.2599,
|
| 744 |
+
"eval_samples_per_second": 28.835,
|
| 745 |
+
"eval_steps_per_second": 3.681,
|
| 746 |
+
"step": 1920
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 61.0,
|
| 750 |
+
"eval_accuracy": 0.7890529133975384,
|
| 751 |
+
"eval_f1": 0.5021816739389131,
|
| 752 |
+
"eval_loss": 1.5567139387130737,
|
| 753 |
+
"eval_precision": 0.46544117647058825,
|
| 754 |
+
"eval_recall": 0.5452196382428941,
|
| 755 |
+
"eval_runtime": 3.4648,
|
| 756 |
+
"eval_samples_per_second": 27.13,
|
| 757 |
+
"eval_steps_per_second": 3.463,
|
| 758 |
+
"step": 1952
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 62.0,
|
| 762 |
+
"eval_accuracy": 0.7909865020637341,
|
| 763 |
+
"eval_f1": 0.5128824476650564,
|
| 764 |
+
"eval_loss": 1.5516618490219116,
|
| 765 |
+
"eval_precision": 0.48148148148148145,
|
| 766 |
+
"eval_recall": 0.5486649440137812,
|
| 767 |
+
"eval_runtime": 3.3346,
|
| 768 |
+
"eval_samples_per_second": 28.19,
|
| 769 |
+
"eval_steps_per_second": 3.599,
|
| 770 |
+
"step": 1984
|
| 771 |
+
},
|
| 772 |
+
{
|
| 773 |
+
"epoch": 62.5,
|
| 774 |
+
"learning_rate": 7.500000000000001e-06,
|
| 775 |
+
"loss": 0.0017,
|
| 776 |
+
"step": 2000
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 63.0,
|
| 780 |
+
"eval_accuracy": 0.7902056297177704,
|
| 781 |
+
"eval_f1": 0.508203281312525,
|
| 782 |
+
"eval_loss": 1.5522751808166504,
|
| 783 |
+
"eval_precision": 0.47458893871449925,
|
| 784 |
+
"eval_recall": 0.5469422911283376,
|
| 785 |
+
"eval_runtime": 3.095,
|
| 786 |
+
"eval_samples_per_second": 30.372,
|
| 787 |
+
"eval_steps_per_second": 3.877,
|
| 788 |
+
"step": 2016
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 64.0,
|
| 792 |
+
"eval_accuracy": 0.787044955936489,
|
| 793 |
+
"eval_f1": 0.4970691676436107,
|
| 794 |
+
"eval_loss": 1.5659950971603394,
|
| 795 |
+
"eval_precision": 0.45493562231759654,
|
| 796 |
+
"eval_recall": 0.5478036175710594,
|
| 797 |
+
"eval_runtime": 3.7257,
|
| 798 |
+
"eval_samples_per_second": 25.23,
|
| 799 |
+
"eval_steps_per_second": 3.221,
|
| 800 |
+
"step": 2048
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 65.0,
|
| 804 |
+
"eval_accuracy": 0.791246792845722,
|
| 805 |
+
"eval_f1": 0.5023219814241486,
|
| 806 |
+
"eval_loss": 1.5194514989852905,
|
| 807 |
+
"eval_precision": 0.4560787069571328,
|
| 808 |
+
"eval_recall": 0.5590008613264428,
|
| 809 |
+
"eval_runtime": 3.1193,
|
| 810 |
+
"eval_samples_per_second": 30.135,
|
| 811 |
+
"eval_steps_per_second": 3.847,
|
| 812 |
+
"step": 2080
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 66.0,
|
| 816 |
+
"eval_accuracy": 0.7909493176663073,
|
| 817 |
+
"eval_f1": 0.504884720593982,
|
| 818 |
+
"eval_loss": 1.5329976081848145,
|
| 819 |
+
"eval_precision": 0.46208869814020026,
|
| 820 |
+
"eval_recall": 0.5564168819982773,
|
| 821 |
+
"eval_runtime": 3.2791,
|
| 822 |
+
"eval_samples_per_second": 28.666,
|
| 823 |
+
"eval_steps_per_second": 3.66,
|
| 824 |
+
"step": 2112
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 67.0,
|
| 828 |
+
"eval_accuracy": 0.7897222325512215,
|
| 829 |
+
"eval_f1": 0.5118296529968455,
|
| 830 |
+
"eval_loss": 1.5393718481063843,
|
| 831 |
+
"eval_precision": 0.472,
|
| 832 |
+
"eval_recall": 0.5590008613264428,
|
| 833 |
+
"eval_runtime": 3.3565,
|
| 834 |
+
"eval_samples_per_second": 28.006,
|
| 835 |
+
"eval_steps_per_second": 3.575,
|
| 836 |
+
"step": 2144
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 68.0,
|
| 840 |
+
"eval_accuracy": 0.7929572751273566,
|
| 841 |
+
"eval_f1": 0.5202220459952419,
|
| 842 |
+
"eval_loss": 1.5860378742218018,
|
| 843 |
+
"eval_precision": 0.4819985304922851,
|
| 844 |
+
"eval_recall": 0.5650301464254953,
|
| 845 |
+
"eval_runtime": 3.0372,
|
| 846 |
+
"eval_samples_per_second": 30.949,
|
| 847 |
+
"eval_steps_per_second": 3.951,
|
| 848 |
+
"step": 2176
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 69.0,
|
| 852 |
+
"eval_accuracy": 0.7916186368199903,
|
| 853 |
+
"eval_f1": 0.5208667736757624,
|
| 854 |
+
"eval_loss": 1.622610092163086,
|
| 855 |
+
"eval_precision": 0.48760330578512395,
|
| 856 |
+
"eval_recall": 0.5590008613264428,
|
| 857 |
+
"eval_runtime": 3.1363,
|
| 858 |
+
"eval_samples_per_second": 29.972,
|
| 859 |
+
"eval_steps_per_second": 3.826,
|
| 860 |
+
"step": 2208
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 70.0,
|
| 864 |
+
"eval_accuracy": 0.7935150410887591,
|
| 865 |
+
"eval_f1": 0.5205158264947246,
|
| 866 |
+
"eval_loss": 1.613077163696289,
|
| 867 |
+
"eval_precision": 0.47639484978540775,
|
| 868 |
+
"eval_recall": 0.5736434108527132,
|
| 869 |
+
"eval_runtime": 3.4181,
|
| 870 |
+
"eval_samples_per_second": 27.501,
|
| 871 |
+
"eval_steps_per_second": 3.511,
|
| 872 |
+
"step": 2240
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 71.0,
|
| 876 |
+
"eval_accuracy": 0.7962666864983453,
|
| 877 |
+
"eval_f1": 0.5341365461847388,
|
| 878 |
+
"eval_loss": 1.617754340171814,
|
| 879 |
+
"eval_precision": 0.5003762227238525,
|
| 880 |
+
"eval_recall": 0.5727820844099913,
|
| 881 |
+
"eval_runtime": 3.9965,
|
| 882 |
+
"eval_samples_per_second": 23.521,
|
| 883 |
+
"eval_steps_per_second": 3.003,
|
| 884 |
+
"step": 2272
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 72.0,
|
| 888 |
+
"eval_accuracy": 0.7960063957163575,
|
| 889 |
+
"eval_f1": 0.5309876049580169,
|
| 890 |
+
"eval_loss": 1.6110599040985107,
|
| 891 |
+
"eval_precision": 0.4955223880597015,
|
| 892 |
+
"eval_recall": 0.5719207579672696,
|
| 893 |
+
"eval_runtime": 3.0526,
|
| 894 |
+
"eval_samples_per_second": 30.794,
|
| 895 |
+
"eval_steps_per_second": 3.931,
|
| 896 |
+
"step": 2304
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 73.0,
|
| 900 |
+
"eval_accuracy": 0.7905774736920388,
|
| 901 |
+
"eval_f1": 0.5087025316455697,
|
| 902 |
+
"eval_loss": 1.5602927207946777,
|
| 903 |
+
"eval_precision": 0.4703730797366496,
|
| 904 |
+
"eval_recall": 0.553832902670112,
|
| 905 |
+
"eval_runtime": 3.378,
|
| 906 |
+
"eval_samples_per_second": 27.827,
|
| 907 |
+
"eval_steps_per_second": 3.552,
|
| 908 |
+
"step": 2336
|
| 909 |
+
},
|
| 910 |
+
{
|
| 911 |
+
"epoch": 74.0,
|
| 912 |
+
"eval_accuracy": 0.7928085375376492,
|
| 913 |
+
"eval_f1": 0.5181962025316456,
|
| 914 |
+
"eval_loss": 1.555016040802002,
|
| 915 |
+
"eval_precision": 0.479151426481346,
|
| 916 |
+
"eval_recall": 0.5641688199827735,
|
| 917 |
+
"eval_runtime": 3.0403,
|
| 918 |
+
"eval_samples_per_second": 30.918,
|
| 919 |
+
"eval_steps_per_second": 3.947,
|
| 920 |
+
"step": 2368
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 75.0,
|
| 924 |
+
"eval_accuracy": 0.7963782396906258,
|
| 925 |
+
"eval_f1": 0.5212933753943217,
|
| 926 |
+
"eval_loss": 1.5542960166931152,
|
| 927 |
+
"eval_precision": 0.4807272727272727,
|
| 928 |
+
"eval_recall": 0.5693367786391043,
|
| 929 |
+
"eval_runtime": 3.1539,
|
| 930 |
+
"eval_samples_per_second": 29.805,
|
| 931 |
+
"eval_steps_per_second": 3.805,
|
| 932 |
+
"step": 2400
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"epoch": 76.0,
|
| 936 |
+
"eval_accuracy": 0.7952998921652474,
|
| 937 |
+
"eval_f1": 0.5285996055226825,
|
| 938 |
+
"eval_loss": 1.560102939605713,
|
| 939 |
+
"eval_precision": 0.487627365356623,
|
| 940 |
+
"eval_recall": 0.5770887166236004,
|
| 941 |
+
"eval_runtime": 3.3272,
|
| 942 |
+
"eval_samples_per_second": 28.252,
|
| 943 |
+
"eval_steps_per_second": 3.607,
|
| 944 |
+
"step": 2432
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 77.0,
|
| 948 |
+
"eval_accuracy": 0.7926597999479419,
|
| 949 |
+
"eval_f1": 0.5247999999999999,
|
| 950 |
+
"eval_loss": 1.559518575668335,
|
| 951 |
+
"eval_precision": 0.4899178491411501,
|
| 952 |
+
"eval_recall": 0.5650301464254953,
|
| 953 |
+
"eval_runtime": 3.1588,
|
| 954 |
+
"eval_samples_per_second": 29.758,
|
| 955 |
+
"eval_steps_per_second": 3.799,
|
| 956 |
+
"step": 2464
|
| 957 |
+
},
|
| 958 |
+
{
|
| 959 |
+
"epoch": 78.0,
|
| 960 |
+
"eval_accuracy": 0.7940356226527349,
|
| 961 |
+
"eval_f1": 0.533066132264529,
|
| 962 |
+
"eval_loss": 1.5632045269012451,
|
| 963 |
+
"eval_precision": 0.4985007496251874,
|
| 964 |
+
"eval_recall": 0.5727820844099913,
|
| 965 |
+
"eval_runtime": 3.5803,
|
| 966 |
+
"eval_samples_per_second": 26.255,
|
| 967 |
+
"eval_steps_per_second": 3.352,
|
| 968 |
+
"step": 2496
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 78.12,
|
| 972 |
+
"learning_rate": 4.3750000000000005e-06,
|
| 973 |
+
"loss": 0.0011,
|
| 974 |
+
"step": 2500
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 79.0,
|
| 978 |
+
"eval_accuracy": 0.7931431971144908,
|
| 979 |
+
"eval_f1": 0.5266084193804608,
|
| 980 |
+
"eval_loss": 1.5693398714065552,
|
| 981 |
+
"eval_precision": 0.4885777450257922,
|
| 982 |
+
"eval_recall": 0.5710594315245479,
|
| 983 |
+
"eval_runtime": 3.2284,
|
| 984 |
+
"eval_samples_per_second": 29.117,
|
| 985 |
+
"eval_steps_per_second": 3.717,
|
| 986 |
+
"step": 2528
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 80.0,
|
| 990 |
+
"eval_accuracy": 0.7927341687427956,
|
| 991 |
+
"eval_f1": 0.5263575108997226,
|
| 992 |
+
"eval_loss": 1.5728816986083984,
|
| 993 |
+
"eval_precision": 0.48751835535976507,
|
| 994 |
+
"eval_recall": 0.5719207579672696,
|
| 995 |
+
"eval_runtime": 3.0942,
|
| 996 |
+
"eval_samples_per_second": 30.38,
|
| 997 |
+
"eval_steps_per_second": 3.878,
|
| 998 |
+
"step": 2560
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"epoch": 81.0,
|
| 1002 |
+
"eval_accuracy": 0.7929200907299297,
|
| 1003 |
+
"eval_f1": 0.5305313243457573,
|
| 1004 |
+
"eval_loss": 1.5773258209228516,
|
| 1005 |
+
"eval_precision": 0.49155033063923587,
|
| 1006 |
+
"eval_recall": 0.5762273901808785,
|
| 1007 |
+
"eval_runtime": 3.5248,
|
| 1008 |
+
"eval_samples_per_second": 26.668,
|
| 1009 |
+
"eval_steps_per_second": 3.404,
|
| 1010 |
+
"step": 2592
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 82.0,
|
| 1014 |
+
"eval_accuracy": 0.7932919347041981,
|
| 1015 |
+
"eval_f1": 0.5318382058470164,
|
| 1016 |
+
"eval_loss": 1.579202651977539,
|
| 1017 |
+
"eval_precision": 0.49700598802395207,
|
| 1018 |
+
"eval_recall": 0.5719207579672696,
|
| 1019 |
+
"eval_runtime": 3.0984,
|
| 1020 |
+
"eval_samples_per_second": 30.338,
|
| 1021 |
+
"eval_steps_per_second": 3.873,
|
| 1022 |
+
"step": 2624
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 83.0,
|
| 1026 |
+
"eval_accuracy": 0.793329119101625,
|
| 1027 |
+
"eval_f1": 0.5278780585639792,
|
| 1028 |
+
"eval_loss": 1.58578622341156,
|
| 1029 |
+
"eval_precision": 0.493993993993994,
|
| 1030 |
+
"eval_recall": 0.5667527993109388,
|
| 1031 |
+
"eval_runtime": 3.0353,
|
| 1032 |
+
"eval_samples_per_second": 30.969,
|
| 1033 |
+
"eval_steps_per_second": 3.954,
|
| 1034 |
+
"step": 2656
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 84.0,
|
| 1038 |
+
"eval_accuracy": 0.7925110623582344,
|
| 1039 |
+
"eval_f1": 0.528965241709948,
|
| 1040 |
+
"eval_loss": 1.584820032119751,
|
| 1041 |
+
"eval_precision": 0.4932935916542474,
|
| 1042 |
+
"eval_recall": 0.570198105081826,
|
| 1043 |
+
"eval_runtime": 3.3844,
|
| 1044 |
+
"eval_samples_per_second": 27.774,
|
| 1045 |
+
"eval_steps_per_second": 3.546,
|
| 1046 |
+
"step": 2688
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 85.0,
|
| 1050 |
+
"eval_accuracy": 0.7922135871788197,
|
| 1051 |
+
"eval_f1": 0.5333866453418633,
|
| 1052 |
+
"eval_loss": 1.5820199251174927,
|
| 1053 |
+
"eval_precision": 0.49776119402985075,
|
| 1054 |
+
"eval_recall": 0.5745047372954349,
|
| 1055 |
+
"eval_runtime": 3.2337,
|
| 1056 |
+
"eval_samples_per_second": 29.069,
|
| 1057 |
+
"eval_steps_per_second": 3.711,
|
| 1058 |
+
"step": 2720
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 86.0,
|
| 1062 |
+
"eval_accuracy": 0.7920648495891124,
|
| 1063 |
+
"eval_f1": 0.5324830609804703,
|
| 1064 |
+
"eval_loss": 1.5860569477081299,
|
| 1065 |
+
"eval_precision": 0.49554896142433236,
|
| 1066 |
+
"eval_recall": 0.5753660637381568,
|
| 1067 |
+
"eval_runtime": 3.1113,
|
| 1068 |
+
"eval_samples_per_second": 30.212,
|
| 1069 |
+
"eval_steps_per_second": 3.857,
|
| 1070 |
+
"step": 2752
|
| 1071 |
+
}
|
| 1072 |
+
],
|
| 1073 |
+
"max_steps": 3200,
|
| 1074 |
+
"num_train_epochs": 100,
|
| 1075 |
+
"total_flos": 1.8890572813041664e+16,
|
| 1076 |
+
"trial_name": null,
|
| 1077 |
+
"trial_params": null
|
| 1078 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fb1e434b531988ca004c6ae71f60286e85ab7b5eca3cbdd0940f28e887686c64
|
| 3 |
+
size 3643
|
vocab.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|