Commit
•
fe8825e
1
Parent(s):
afef5e6
Training in progress, epoch 20, checkpoint
Browse files- checkpoint-1620/config.json +57 -0
- checkpoint-1620/merges.txt +0 -0
- checkpoint-1620/model.safetensors +3 -0
- checkpoint-1620/optimizer.pt +3 -0
- checkpoint-1620/rng_state.pth +3 -0
- checkpoint-1620/scheduler.pt +3 -0
- checkpoint-1620/special_tokens_map.json +15 -0
- checkpoint-1620/tokenizer.json +0 -0
- checkpoint-1620/tokenizer_config.json +57 -0
- checkpoint-1620/trainer_state.json +942 -0
- checkpoint-1620/training_args.bin +3 -0
- checkpoint-1620/vocab.json +0 -0
checkpoint-1620/config.json
ADDED
|
@@ -0,0 +1,57 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "allenai/longformer-base-4096",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"LongformerForTokenClassification"
|
| 5 |
+
],
|
| 6 |
+
"attention_mode": "longformer",
|
| 7 |
+
"attention_probs_dropout_prob": 0.1,
|
| 8 |
+
"attention_window": [
|
| 9 |
+
700,
|
| 10 |
+
700,
|
| 11 |
+
700,
|
| 12 |
+
700,
|
| 13 |
+
700,
|
| 14 |
+
700,
|
| 15 |
+
700,
|
| 16 |
+
700,
|
| 17 |
+
700,
|
| 18 |
+
700,
|
| 19 |
+
700,
|
| 20 |
+
700
|
| 21 |
+
],
|
| 22 |
+
"bos_token_id": 0,
|
| 23 |
+
"eos_token_id": 2,
|
| 24 |
+
"gradient_checkpointing": false,
|
| 25 |
+
"hidden_act": "gelu",
|
| 26 |
+
"hidden_dropout_prob": 0.1,
|
| 27 |
+
"hidden_size": 768,
|
| 28 |
+
"id2label": {
|
| 29 |
+
"0": "O",
|
| 30 |
+
"1": "X_placeholder_X",
|
| 31 |
+
"2": "MajorClaim",
|
| 32 |
+
"3": "Claim",
|
| 33 |
+
"4": "Premise"
|
| 34 |
+
},
|
| 35 |
+
"ignore_attention_mask": false,
|
| 36 |
+
"initializer_range": 0.02,
|
| 37 |
+
"intermediate_size": 3072,
|
| 38 |
+
"label2id": {
|
| 39 |
+
"Claim": 3,
|
| 40 |
+
"MajorClaim": 2,
|
| 41 |
+
"O": 0,
|
| 42 |
+
"Premise": 4,
|
| 43 |
+
"X_placeholder_X": 1
|
| 44 |
+
},
|
| 45 |
+
"layer_norm_eps": 1e-05,
|
| 46 |
+
"max_position_embeddings": 4098,
|
| 47 |
+
"model_type": "longformer",
|
| 48 |
+
"num_attention_heads": 12,
|
| 49 |
+
"num_hidden_layers": 12,
|
| 50 |
+
"onnx_export": false,
|
| 51 |
+
"pad_token_id": 1,
|
| 52 |
+
"sep_token_id": 2,
|
| 53 |
+
"torch_dtype": "float32",
|
| 54 |
+
"transformers_version": "4.38.2",
|
| 55 |
+
"type_vocab_size": 1,
|
| 56 |
+
"vocab_size": 50265
|
| 57 |
+
}
|
checkpoint-1620/merges.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
checkpoint-1620/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:08be877f465e33400fd7e4c41aa043eda8c45d21b07ed3aa85558e1a4400900d
|
| 3 |
+
size 592324828
|
checkpoint-1620/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ada1314de39735cf23eab5bbd59e5da39913aea615e944d092ec73096644a54c
|
| 3 |
+
size 1014657786
|
checkpoint-1620/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f553c1fb23cbb8baae54ab2a9d56e54b26fd49c15995479fbc3f4c04df1c4989
|
| 3 |
+
size 14244
|
checkpoint-1620/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:62fa97b6a73c620ff6f69ca713c8a45c421ed3fdb8b0992517caf27356998de4
|
| 3 |
+
size 1064
|
checkpoint-1620/special_tokens_map.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token": "<s>",
|
| 3 |
+
"cls_token": "<s>",
|
| 4 |
+
"eos_token": "</s>",
|
| 5 |
+
"mask_token": {
|
| 6 |
+
"content": "<mask>",
|
| 7 |
+
"lstrip": true,
|
| 8 |
+
"normalized": false,
|
| 9 |
+
"rstrip": false,
|
| 10 |
+
"single_word": false
|
| 11 |
+
},
|
| 12 |
+
"pad_token": "<pad>",
|
| 13 |
+
"sep_token": "</s>",
|
| 14 |
+
"unk_token": "<unk>"
|
| 15 |
+
}
|
checkpoint-1620/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
checkpoint-1620/tokenizer_config.json
ADDED
|
@@ -0,0 +1,57 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_prefix_space": true,
|
| 3 |
+
"added_tokens_decoder": {
|
| 4 |
+
"0": {
|
| 5 |
+
"content": "<s>",
|
| 6 |
+
"lstrip": false,
|
| 7 |
+
"normalized": true,
|
| 8 |
+
"rstrip": false,
|
| 9 |
+
"single_word": false,
|
| 10 |
+
"special": true
|
| 11 |
+
},
|
| 12 |
+
"1": {
|
| 13 |
+
"content": "<pad>",
|
| 14 |
+
"lstrip": false,
|
| 15 |
+
"normalized": true,
|
| 16 |
+
"rstrip": false,
|
| 17 |
+
"single_word": false,
|
| 18 |
+
"special": true
|
| 19 |
+
},
|
| 20 |
+
"2": {
|
| 21 |
+
"content": "</s>",
|
| 22 |
+
"lstrip": false,
|
| 23 |
+
"normalized": true,
|
| 24 |
+
"rstrip": false,
|
| 25 |
+
"single_word": false,
|
| 26 |
+
"special": true
|
| 27 |
+
},
|
| 28 |
+
"3": {
|
| 29 |
+
"content": "<unk>",
|
| 30 |
+
"lstrip": false,
|
| 31 |
+
"normalized": true,
|
| 32 |
+
"rstrip": false,
|
| 33 |
+
"single_word": false,
|
| 34 |
+
"special": true
|
| 35 |
+
},
|
| 36 |
+
"50264": {
|
| 37 |
+
"content": "<mask>",
|
| 38 |
+
"lstrip": true,
|
| 39 |
+
"normalized": false,
|
| 40 |
+
"rstrip": false,
|
| 41 |
+
"single_word": false,
|
| 42 |
+
"special": true
|
| 43 |
+
}
|
| 44 |
+
},
|
| 45 |
+
"bos_token": "<s>",
|
| 46 |
+
"clean_up_tokenization_spaces": true,
|
| 47 |
+
"cls_token": "<s>",
|
| 48 |
+
"eos_token": "</s>",
|
| 49 |
+
"errors": "replace",
|
| 50 |
+
"mask_token": "<mask>",
|
| 51 |
+
"model_max_length": 4096,
|
| 52 |
+
"pad_token": "<pad>",
|
| 53 |
+
"sep_token": "</s>",
|
| 54 |
+
"tokenizer_class": "LongformerTokenizer",
|
| 55 |
+
"trim_offsets": true,
|
| 56 |
+
"unk_token": "<unk>"
|
| 57 |
+
}
|
checkpoint-1620/trainer_state.json
ADDED
|
@@ -0,0 +1,942 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 20.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 1620,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 1.0,
|
| 13 |
+
"eval_Claim": {
|
| 14 |
+
"f1-score": 0.5724454649827784,
|
| 15 |
+
"precision": 0.6002889477486154,
|
| 16 |
+
"recall": 0.5470704410796576,
|
| 17 |
+
"support": 4557.0
|
| 18 |
+
},
|
| 19 |
+
"eval_MajorClaim": {
|
| 20 |
+
"f1-score": 0.7860447185813415,
|
| 21 |
+
"precision": 0.698526892771497,
|
| 22 |
+
"recall": 0.8986337593653592,
|
| 23 |
+
"support": 2269.0
|
| 24 |
+
},
|
| 25 |
+
"eval_O": {
|
| 26 |
+
"f1-score": 0.9987820830889982,
|
| 27 |
+
"precision": 0.9975671292124707,
|
| 28 |
+
"recall": 1.0,
|
| 29 |
+
"support": 11071.0
|
| 30 |
+
},
|
| 31 |
+
"eval_Premise": {
|
| 32 |
+
"f1-score": 0.8965445389824622,
|
| 33 |
+
"precision": 0.9051258677512096,
|
| 34 |
+
"recall": 0.8881243979633962,
|
| 35 |
+
"support": 14534.0
|
| 36 |
+
},
|
| 37 |
+
"eval_accuracy": 0.8791279948197712,
|
| 38 |
+
"eval_loss": 0.3157936632633209,
|
| 39 |
+
"eval_macro avg": {
|
| 40 |
+
"f1-score": 0.8134542014088951,
|
| 41 |
+
"precision": 0.8003772093709481,
|
| 42 |
+
"recall": 0.8334571496021033,
|
| 43 |
+
"support": 32431.0
|
| 44 |
+
},
|
| 45 |
+
"eval_runtime": 4.8184,
|
| 46 |
+
"eval_samples_per_second": 16.603,
|
| 47 |
+
"eval_steps_per_second": 2.075,
|
| 48 |
+
"eval_weighted avg": {
|
| 49 |
+
"f1-score": 0.8781740995293698,
|
| 50 |
+
"precision": 0.8793943851252293,
|
| 51 |
+
"recall": 0.8791279948197712,
|
| 52 |
+
"support": 32431.0
|
| 53 |
+
},
|
| 54 |
+
"step": 81
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 2.0,
|
| 58 |
+
"eval_Claim": {
|
| 59 |
+
"f1-score": 0.6090644692195831,
|
| 60 |
+
"precision": 0.680108254397835,
|
| 61 |
+
"recall": 0.5514592933947773,
|
| 62 |
+
"support": 4557.0
|
| 63 |
+
},
|
| 64 |
+
"eval_MajorClaim": {
|
| 65 |
+
"f1-score": 0.8585858585858586,
|
| 66 |
+
"precision": 0.8960229995208433,
|
| 67 |
+
"recall": 0.8241516086381666,
|
| 68 |
+
"support": 2269.0
|
| 69 |
+
},
|
| 70 |
+
"eval_O": {
|
| 71 |
+
"f1-score": 0.999458141425088,
|
| 72 |
+
"precision": 0.9992776523702032,
|
| 73 |
+
"recall": 0.9996386956914461,
|
| 74 |
+
"support": 11071.0
|
| 75 |
+
},
|
| 76 |
+
"eval_Premise": {
|
| 77 |
+
"f1-score": 0.9105885478942475,
|
| 78 |
+
"precision": 0.8801849235905997,
|
| 79 |
+
"recall": 0.9431677445988716,
|
| 80 |
+
"support": 14534.0
|
| 81 |
+
},
|
| 82 |
+
"eval_accuracy": 0.8990780426135487,
|
| 83 |
+
"eval_loss": 0.24451124668121338,
|
| 84 |
+
"eval_macro avg": {
|
| 85 |
+
"f1-score": 0.8444242542811943,
|
| 86 |
+
"precision": 0.8638984574698704,
|
| 87 |
+
"recall": 0.8296043355808155,
|
| 88 |
+
"support": 32431.0
|
| 89 |
+
},
|
| 90 |
+
"eval_runtime": 4.8529,
|
| 91 |
+
"eval_samples_per_second": 16.485,
|
| 92 |
+
"eval_steps_per_second": 2.061,
|
| 93 |
+
"eval_weighted avg": {
|
| 94 |
+
"f1-score": 0.8949194640367888,
|
| 95 |
+
"precision": 0.8938342965082798,
|
| 96 |
+
"recall": 0.8990780426135487,
|
| 97 |
+
"support": 32431.0
|
| 98 |
+
},
|
| 99 |
+
"step": 162
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 3.0,
|
| 103 |
+
"eval_Claim": {
|
| 104 |
+
"f1-score": 0.6542320859402667,
|
| 105 |
+
"precision": 0.7079713847726111,
|
| 106 |
+
"recall": 0.60807548825982,
|
| 107 |
+
"support": 4557.0
|
| 108 |
+
},
|
| 109 |
+
"eval_MajorClaim": {
|
| 110 |
+
"f1-score": 0.8915187376725839,
|
| 111 |
+
"precision": 0.8866608544027899,
|
| 112 |
+
"recall": 0.8964301454385192,
|
| 113 |
+
"support": 2269.0
|
| 114 |
+
},
|
| 115 |
+
"eval_O": {
|
| 116 |
+
"f1-score": 0.9995936611133684,
|
| 117 |
+
"precision": 0.9992778479870013,
|
| 118 |
+
"recall": 0.9999096739228616,
|
| 119 |
+
"support": 11071.0
|
| 120 |
+
},
|
| 121 |
+
"eval_Premise": {
|
| 122 |
+
"f1-score": 0.9157990498332153,
|
| 123 |
+
"precision": 0.8973258501155497,
|
| 124 |
+
"recall": 0.935048850970139,
|
| 125 |
+
"support": 14534.0
|
| 126 |
+
},
|
| 127 |
+
"eval_accuracy": 0.9085442940396534,
|
| 128 |
+
"eval_loss": 0.22665414214134216,
|
| 129 |
+
"eval_macro avg": {
|
| 130 |
+
"f1-score": 0.8652858836398586,
|
| 131 |
+
"precision": 0.872808984319488,
|
| 132 |
+
"recall": 0.8598660396478349,
|
| 133 |
+
"support": 32431.0
|
| 134 |
+
},
|
| 135 |
+
"eval_runtime": 4.8494,
|
| 136 |
+
"eval_samples_per_second": 16.497,
|
| 137 |
+
"eval_steps_per_second": 2.062,
|
| 138 |
+
"eval_weighted avg": {
|
| 139 |
+
"f1-score": 0.9059516032151627,
|
| 140 |
+
"precision": 0.9047762338408377,
|
| 141 |
+
"recall": 0.9085442940396534,
|
| 142 |
+
"support": 32431.0
|
| 143 |
+
},
|
| 144 |
+
"step": 243
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 4.0,
|
| 148 |
+
"eval_Claim": {
|
| 149 |
+
"f1-score": 0.6861642294713161,
|
| 150 |
+
"precision": 0.6424741478360781,
|
| 151 |
+
"recall": 0.7362299758613122,
|
| 152 |
+
"support": 4557.0
|
| 153 |
+
},
|
| 154 |
+
"eval_MajorClaim": {
|
| 155 |
+
"f1-score": 0.8995515695067264,
|
| 156 |
+
"precision": 0.9155636695572797,
|
| 157 |
+
"recall": 0.884089907448215,
|
| 158 |
+
"support": 2269.0
|
| 159 |
+
},
|
| 160 |
+
"eval_O": {
|
| 161 |
+
"f1-score": 0.9976481230212574,
|
| 162 |
+
"precision": 0.9990941208442794,
|
| 163 |
+
"recall": 0.9962063047601842,
|
| 164 |
+
"support": 11071.0
|
| 165 |
+
},
|
| 166 |
+
"eval_Premise": {
|
| 167 |
+
"f1-score": 0.9062532879739067,
|
| 168 |
+
"precision": 0.9242435081193219,
|
| 169 |
+
"recall": 0.8889500481629283,
|
| 170 |
+
"support": 14534.0
|
| 171 |
+
},
|
| 172 |
+
"eval_accuracy": 0.9037649162838025,
|
| 173 |
+
"eval_loss": 0.2452327460050583,
|
| 174 |
+
"eval_macro avg": {
|
| 175 |
+
"f1-score": 0.8724043024933017,
|
| 176 |
+
"precision": 0.8703438615892398,
|
| 177 |
+
"recall": 0.8763690590581599,
|
| 178 |
+
"support": 32431.0
|
| 179 |
+
},
|
| 180 |
+
"eval_runtime": 4.8393,
|
| 181 |
+
"eval_samples_per_second": 16.531,
|
| 182 |
+
"eval_steps_per_second": 2.066,
|
| 183 |
+
"eval_weighted avg": {
|
| 184 |
+
"f1-score": 0.9060584182508294,
|
| 185 |
+
"precision": 0.9095955973231696,
|
| 186 |
+
"recall": 0.9037649162838025,
|
| 187 |
+
"support": 32431.0
|
| 188 |
+
},
|
| 189 |
+
"step": 324
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 5.0,
|
| 193 |
+
"eval_Claim": {
|
| 194 |
+
"f1-score": 0.6640249759846301,
|
| 195 |
+
"precision": 0.7332272606735614,
|
| 196 |
+
"recall": 0.6067588325652842,
|
| 197 |
+
"support": 4557.0
|
| 198 |
+
},
|
| 199 |
+
"eval_MajorClaim": {
|
| 200 |
+
"f1-score": 0.8848007008322383,
|
| 201 |
+
"precision": 0.8794079233783195,
|
| 202 |
+
"recall": 0.8902600264433671,
|
| 203 |
+
"support": 2269.0
|
| 204 |
+
},
|
| 205 |
+
"eval_O": {
|
| 206 |
+
"f1-score": 0.999367774566474,
|
| 207 |
+
"precision": 0.9992775219001174,
|
| 208 |
+
"recall": 0.9994580435371692,
|
| 209 |
+
"support": 11071.0
|
| 210 |
+
},
|
| 211 |
+
"eval_Premise": {
|
| 212 |
+
"f1-score": 0.9201314377682404,
|
| 213 |
+
"precision": 0.8973839110529758,
|
| 214 |
+
"recall": 0.9440621989816981,
|
| 215 |
+
"support": 14534.0
|
| 216 |
+
},
|
| 217 |
+
"eval_accuracy": 0.9118127717307515,
|
| 218 |
+
"eval_loss": 0.28411003947257996,
|
| 219 |
+
"eval_macro avg": {
|
| 220 |
+
"f1-score": 0.8670812222878956,
|
| 221 |
+
"precision": 0.8773241542512435,
|
| 222 |
+
"recall": 0.8601347753818797,
|
| 223 |
+
"support": 32431.0
|
| 224 |
+
},
|
| 225 |
+
"eval_runtime": 4.8228,
|
| 226 |
+
"eval_samples_per_second": 16.588,
|
| 227 |
+
"eval_steps_per_second": 2.073,
|
| 228 |
+
"eval_weighted avg": {
|
| 229 |
+
"f1-score": 0.9087220731552942,
|
| 230 |
+
"precision": 0.9078434958291443,
|
| 231 |
+
"recall": 0.9118127717307515,
|
| 232 |
+
"support": 32431.0
|
| 233 |
+
},
|
| 234 |
+
"step": 405
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 6.0,
|
| 238 |
+
"eval_Claim": {
|
| 239 |
+
"f1-score": 0.6496087442553721,
|
| 240 |
+
"precision": 0.7484258729250143,
|
| 241 |
+
"recall": 0.5738424402018872,
|
| 242 |
+
"support": 4557.0
|
| 243 |
+
},
|
| 244 |
+
"eval_MajorClaim": {
|
| 245 |
+
"f1-score": 0.8865800865800866,
|
| 246 |
+
"precision": 0.8711186729051468,
|
| 247 |
+
"recall": 0.9026002644336713,
|
| 248 |
+
"support": 2269.0
|
| 249 |
+
},
|
| 250 |
+
"eval_O": {
|
| 251 |
+
"f1-score": 0.9986892655367232,
|
| 252 |
+
"precision": 0.9994572100597069,
|
| 253 |
+
"recall": 0.9979225002258152,
|
| 254 |
+
"support": 11071.0
|
| 255 |
+
},
|
| 256 |
+
"eval_Premise": {
|
| 257 |
+
"f1-score": 0.9191778088205947,
|
| 258 |
+
"precision": 0.889647180015452,
|
| 259 |
+
"recall": 0.9507362047612495,
|
| 260 |
+
"support": 14534.0
|
| 261 |
+
},
|
| 262 |
+
"eval_accuracy": 0.9105177145323918,
|
| 263 |
+
"eval_loss": 0.3082066774368286,
|
| 264 |
+
"eval_macro avg": {
|
| 265 |
+
"f1-score": 0.8635139762981943,
|
| 266 |
+
"precision": 0.87716223397633,
|
| 267 |
+
"recall": 0.8562753524056559,
|
| 268 |
+
"support": 32431.0
|
| 269 |
+
},
|
| 270 |
+
"eval_runtime": 4.8389,
|
| 271 |
+
"eval_samples_per_second": 16.533,
|
| 272 |
+
"eval_steps_per_second": 2.067,
|
| 273 |
+
"eval_weighted avg": {
|
| 274 |
+
"f1-score": 0.9061618943658084,
|
| 275 |
+
"precision": 0.9059932736781678,
|
| 276 |
+
"recall": 0.9105177145323918,
|
| 277 |
+
"support": 32431.0
|
| 278 |
+
},
|
| 279 |
+
"step": 486
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 6.17,
|
| 283 |
+
"grad_norm": 0.8612700700759888,
|
| 284 |
+
"learning_rate": 1.7530864197530865e-05,
|
| 285 |
+
"loss": 0.2223,
|
| 286 |
+
"step": 500
|
| 287 |
+
},
|
| 288 |
+
{
|
| 289 |
+
"epoch": 7.0,
|
| 290 |
+
"eval_Claim": {
|
| 291 |
+
"f1-score": 0.6977381768334475,
|
| 292 |
+
"precision": 0.629950495049505,
|
| 293 |
+
"recall": 0.7818740399385561,
|
| 294 |
+
"support": 4557.0
|
| 295 |
+
},
|
| 296 |
+
"eval_MajorClaim": {
|
| 297 |
+
"f1-score": 0.8892209178228387,
|
| 298 |
+
"precision": 0.8621688741721855,
|
| 299 |
+
"recall": 0.9180255619215514,
|
| 300 |
+
"support": 2269.0
|
| 301 |
+
},
|
| 302 |
+
"eval_O": {
|
| 303 |
+
"f1-score": 0.9988698521766648,
|
| 304 |
+
"precision": 0.9998190045248869,
|
| 305 |
+
"recall": 0.9979225002258152,
|
| 306 |
+
"support": 11071.0
|
| 307 |
+
},
|
| 308 |
+
"eval_Premise": {
|
| 309 |
+
"f1-score": 0.9034227633516503,
|
| 310 |
+
"precision": 0.9449996243143738,
|
| 311 |
+
"recall": 0.8653502132929682,
|
| 312 |
+
"support": 14534.0
|
| 313 |
+
},
|
| 314 |
+
"eval_accuracy": 0.9025623631710401,
|
| 315 |
+
"eval_loss": 0.3512111008167267,
|
| 316 |
+
"eval_macro avg": {
|
| 317 |
+
"f1-score": 0.8723129275461503,
|
| 318 |
+
"precision": 0.8592344995152377,
|
| 319 |
+
"recall": 0.8907930788447227,
|
| 320 |
+
"support": 32431.0
|
| 321 |
+
},
|
| 322 |
+
"eval_runtime": 4.8174,
|
| 323 |
+
"eval_samples_per_second": 16.607,
|
| 324 |
+
"eval_steps_per_second": 2.076,
|
| 325 |
+
"eval_weighted avg": {
|
| 326 |
+
"f1-score": 0.9061105026169647,
|
| 327 |
+
"precision": 0.9136494810618672,
|
| 328 |
+
"recall": 0.9025623631710401,
|
| 329 |
+
"support": 32431.0
|
| 330 |
+
},
|
| 331 |
+
"step": 567
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 8.0,
|
| 335 |
+
"eval_Claim": {
|
| 336 |
+
"f1-score": 0.6987032799389779,
|
| 337 |
+
"precision": 0.6177710335525206,
|
| 338 |
+
"recall": 0.80403774412991,
|
| 339 |
+
"support": 4557.0
|
| 340 |
+
},
|
| 341 |
+
"eval_MajorClaim": {
|
| 342 |
+
"f1-score": 0.9015364639688379,
|
| 343 |
+
"precision": 0.8856292517006803,
|
| 344 |
+
"recall": 0.9180255619215514,
|
| 345 |
+
"support": 2269.0
|
| 346 |
+
},
|
| 347 |
+
"eval_O": {
|
| 348 |
+
"f1-score": 0.998915499322187,
|
| 349 |
+
"precision": 0.9994574554661362,
|
| 350 |
+
"recall": 0.9983741306115076,
|
| 351 |
+
"support": 11071.0
|
| 352 |
+
},
|
| 353 |
+
"eval_Premise": {
|
| 354 |
+
"f1-score": 0.8976577489773016,
|
| 355 |
+
"precision": 0.9472075788830315,
|
| 356 |
+
"recall": 0.8530342644832806,
|
| 357 |
+
"support": 14534.0
|
| 358 |
+
},
|
| 359 |
+
"eval_accuracy": 0.9003114304215103,
|
| 360 |
+
"eval_loss": 0.3705728054046631,
|
| 361 |
+
"eval_macro avg": {
|
| 362 |
+
"f1-score": 0.8742032480518261,
|
| 363 |
+
"precision": 0.8625163299005921,
|
| 364 |
+
"recall": 0.8933679252865624,
|
| 365 |
+
"support": 32431.0
|
| 366 |
+
},
|
| 367 |
+
"eval_runtime": 4.8542,
|
| 368 |
+
"eval_samples_per_second": 16.48,
|
| 369 |
+
"eval_steps_per_second": 2.06,
|
| 370 |
+
"eval_weighted avg": {
|
| 371 |
+
"f1-score": 0.9045397397569994,
|
| 372 |
+
"precision": 0.9144455555782818,
|
| 373 |
+
"recall": 0.9003114304215103,
|
| 374 |
+
"support": 32431.0
|
| 375 |
+
},
|
| 376 |
+
"step": 648
|
| 377 |
+
},
|
| 378 |
+
{
|
| 379 |
+
"epoch": 9.0,
|
| 380 |
+
"eval_Claim": {
|
| 381 |
+
"f1-score": 0.6657139454567107,
|
| 382 |
+
"precision": 0.7278645833333334,
|
| 383 |
+
"recall": 0.6133421110379635,
|
| 384 |
+
"support": 4557.0
|
| 385 |
+
},
|
| 386 |
+
"eval_MajorClaim": {
|
| 387 |
+
"f1-score": 0.8882521489971346,
|
| 388 |
+
"precision": 0.8884479717813051,
|
| 389 |
+
"recall": 0.8880564125165271,
|
| 390 |
+
"support": 2269.0
|
| 391 |
+
},
|
| 392 |
+
"eval_O": {
|
| 393 |
+
"f1-score": 0.9987339482727438,
|
| 394 |
+
"precision": 0.9999094612947035,
|
| 395 |
+
"recall": 0.9975611959172613,
|
| 396 |
+
"support": 11071.0
|
| 397 |
+
},
|
| 398 |
+
"eval_Premise": {
|
| 399 |
+
"f1-score": 0.9178854152690192,
|
| 400 |
+
"precision": 0.8955360649299646,
|
| 401 |
+
"recall": 0.9413788358332187,
|
| 402 |
+
"support": 14534.0
|
| 403 |
+
},
|
| 404 |
+
"eval_accuracy": 0.9107335573987851,
|
| 405 |
+
"eval_loss": 0.41826051473617554,
|
| 406 |
+
"eval_macro avg": {
|
| 407 |
+
"f1-score": 0.8676463644989021,
|
| 408 |
+
"precision": 0.8779395203348266,
|
| 409 |
+
"recall": 0.8600846388262426,
|
| 410 |
+
"support": 32431.0
|
| 411 |
+
},
|
| 412 |
+
"eval_runtime": 4.809,
|
| 413 |
+
"eval_samples_per_second": 16.635,
|
| 414 |
+
"eval_steps_per_second": 2.079,
|
| 415 |
+
"eval_weighted avg": {
|
| 416 |
+
"f1-score": 0.9079779452489347,
|
| 417 |
+
"precision": 0.9071100542045434,
|
| 418 |
+
"recall": 0.9107335573987851,
|
| 419 |
+
"support": 32431.0
|
| 420 |
+
},
|
| 421 |
+
"step": 729
|
| 422 |
+
},
|
| 423 |
+
{
|
| 424 |
+
"epoch": 10.0,
|
| 425 |
+
"eval_Claim": {
|
| 426 |
+
"f1-score": 0.6738416222952305,
|
| 427 |
+
"precision": 0.6964871194379391,
|
| 428 |
+
"recall": 0.652622339258284,
|
| 429 |
+
"support": 4557.0
|
| 430 |
+
},
|
| 431 |
+
"eval_MajorClaim": {
|
| 432 |
+
"f1-score": 0.8951019550136641,
|
| 433 |
+
"precision": 0.8557073954983923,
|
| 434 |
+
"recall": 0.9382988100484795,
|
| 435 |
+
"support": 2269.0
|
| 436 |
+
},
|
| 437 |
+
"eval_O": {
|
| 438 |
+
"f1-score": 0.9989605459393501,
|
| 439 |
+
"precision": 0.9996382054992764,
|
| 440 |
+
"recall": 0.998283804534369,
|
| 441 |
+
"support": 11071.0
|
| 442 |
+
},
|
| 443 |
+
"eval_Premise": {
|
| 444 |
+
"f1-score": 0.9109121470961545,
|
| 445 |
+
"precision": 0.9083259218717931,
|
| 446 |
+
"recall": 0.9135131415990092,
|
| 447 |
+
"support": 14534.0
|
| 448 |
+
},
|
| 449 |
+
"eval_accuracy": 0.9075267490980852,
|
| 450 |
+
"eval_loss": 0.5459077954292297,
|
| 451 |
+
"eval_macro avg": {
|
| 452 |
+
"f1-score": 0.8697040675860997,
|
| 453 |
+
"precision": 0.8650396605768502,
|
| 454 |
+
"recall": 0.8756795238600354,
|
| 455 |
+
"support": 32431.0
|
| 456 |
+
},
|
| 457 |
+
"eval_runtime": 4.8478,
|
| 458 |
+
"eval_samples_per_second": 16.502,
|
| 459 |
+
"eval_steps_per_second": 2.063,
|
| 460 |
+
"eval_weighted avg": {
|
| 461 |
+
"f1-score": 0.9065515080853328,
|
| 462 |
+
"precision": 0.9060496255197702,
|
| 463 |
+
"recall": 0.9075267490980852,
|
| 464 |
+
"support": 32431.0
|
| 465 |
+
},
|
| 466 |
+
"step": 810
|
| 467 |
+
},
|
| 468 |
+
{
|
| 469 |
+
"epoch": 11.0,
|
| 470 |
+
"eval_Claim": {
|
| 471 |
+
"f1-score": 0.6581417175035196,
|
| 472 |
+
"precision": 0.7070834383665239,
|
| 473 |
+
"recall": 0.6155365371955234,
|
| 474 |
+
"support": 4557.0
|
| 475 |
+
},
|
| 476 |
+
"eval_MajorClaim": {
|
| 477 |
+
"f1-score": 0.881979417560762,
|
| 478 |
+
"precision": 0.876414273281114,
|
| 479 |
+
"recall": 0.8876156897311591,
|
| 480 |
+
"support": 2269.0
|
| 481 |
+
},
|
| 482 |
+
"eval_O": {
|
| 483 |
+
"f1-score": 0.9999548390010388,
|
| 484 |
+
"precision": 0.9999096820809249,
|
| 485 |
+
"recall": 1.0,
|
| 486 |
+
"support": 11071.0
|
| 487 |
+
},
|
| 488 |
+
"eval_Premise": {
|
| 489 |
+
"f1-score": 0.9173079519373565,
|
| 490 |
+
"precision": 0.9002915065588976,
|
| 491 |
+
"recall": 0.9349800467868447,
|
| 492 |
+
"support": 14534.0
|
| 493 |
+
},
|
| 494 |
+
"eval_accuracy": 0.9089759797724399,
|
| 495 |
+
"eval_loss": 0.5726844668388367,
|
| 496 |
+
"eval_macro avg": {
|
| 497 |
+
"f1-score": 0.8643459815006692,
|
| 498 |
+
"precision": 0.8709247250718651,
|
| 499 |
+
"recall": 0.8595330684283817,
|
| 500 |
+
"support": 32431.0
|
| 501 |
+
},
|
| 502 |
+
"eval_runtime": 4.8447,
|
| 503 |
+
"eval_samples_per_second": 16.513,
|
| 504 |
+
"eval_steps_per_second": 2.064,
|
| 505 |
+
"eval_weighted avg": {
|
| 506 |
+
"f1-score": 0.9066330640790278,
|
| 507 |
+
"precision": 0.9054793272287636,
|
| 508 |
+
"recall": 0.9089759797724399,
|
| 509 |
+
"support": 32431.0
|
| 510 |
+
},
|
| 511 |
+
"step": 891
|
| 512 |
+
},
|
| 513 |
+
{
|
| 514 |
+
"epoch": 12.0,
|
| 515 |
+
"eval_Claim": {
|
| 516 |
+
"f1-score": 0.7079179557694512,
|
| 517 |
+
"precision": 0.7407673860911271,
|
| 518 |
+
"recall": 0.6778582400702217,
|
| 519 |
+
"support": 4557.0
|
| 520 |
+
},
|
| 521 |
+
"eval_MajorClaim": {
|
| 522 |
+
"f1-score": 0.8947480331703169,
|
| 523 |
+
"precision": 0.8644207066557108,
|
| 524 |
+
"recall": 0.9272807404142794,
|
| 525 |
+
"support": 2269.0
|
| 526 |
+
},
|
| 527 |
+
"eval_O": {
|
| 528 |
+
"f1-score": 0.999186477447347,
|
| 529 |
+
"precision": 0.9999095431931253,
|
| 530 |
+
"recall": 0.998464456688646,
|
| 531 |
+
"support": 11071.0
|
| 532 |
+
},
|
| 533 |
+
"eval_Premise": {
|
| 534 |
+
"f1-score": 0.9243840851702722,
|
| 535 |
+
"precision": 0.9169374492282697,
|
| 536 |
+
"recall": 0.9319526627218935,
|
| 537 |
+
"support": 14534.0
|
| 538 |
+
},
|
| 539 |
+
"eval_accuracy": 0.9186272393697388,
|
| 540 |
+
"eval_loss": 0.5660926699638367,
|
| 541 |
+
"eval_macro avg": {
|
| 542 |
+
"f1-score": 0.8815591378893468,
|
| 543 |
+
"precision": 0.8805087712920582,
|
| 544 |
+
"recall": 0.8838890249737601,
|
| 545 |
+
"support": 32431.0
|
| 546 |
+
},
|
| 547 |
+
"eval_runtime": 4.8659,
|
| 548 |
+
"eval_samples_per_second": 16.441,
|
| 549 |
+
"eval_steps_per_second": 2.055,
|
| 550 |
+
"eval_weighted avg": {
|
| 551 |
+
"f1-score": 0.917429533390557,
|
| 552 |
+
"precision": 0.9168331226787283,
|
| 553 |
+
"recall": 0.9186272393697388,
|
| 554 |
+
"support": 32431.0
|
| 555 |
+
},
|
| 556 |
+
"step": 972
|
| 557 |
+
},
|
| 558 |
+
{
|
| 559 |
+
"epoch": 12.35,
|
| 560 |
+
"grad_norm": 2.9101791381835938,
|
| 561 |
+
"learning_rate": 1.506172839506173e-05,
|
| 562 |
+
"loss": 0.0294,
|
| 563 |
+
"step": 1000
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 13.0,
|
| 567 |
+
"eval_Claim": {
|
| 568 |
+
"f1-score": 0.6744648496900223,
|
| 569 |
+
"precision": 0.7221943887775552,
|
| 570 |
+
"recall": 0.6326530612244898,
|
| 571 |
+
"support": 4557.0
|
| 572 |
+
},
|
| 573 |
+
"eval_MajorClaim": {
|
| 574 |
+
"f1-score": 0.9044719714465672,
|
| 575 |
+
"precision": 0.863672814755413,
|
| 576 |
+
"recall": 0.9493168796826796,
|
| 577 |
+
"support": 2269.0
|
| 578 |
+
},
|
| 579 |
+
"eval_O": {
|
| 580 |
+
"f1-score": 0.9995934041111363,
|
| 581 |
+
"precision": 0.9999096167751266,
|
| 582 |
+
"recall": 0.9992773913828923,
|
| 583 |
+
"support": 11071.0
|
| 584 |
+
},
|
| 585 |
+
"eval_Premise": {
|
| 586 |
+
"f1-score": 0.9161992180860105,
|
| 587 |
+
"precision": 0.9055171023452725,
|
| 588 |
+
"recall": 0.9271363698912893,
|
| 589 |
+
"support": 14534.0
|
| 590 |
+
},
|
| 591 |
+
"eval_accuracy": 0.9119361105115475,
|
| 592 |
+
"eval_loss": 0.5764271020889282,
|
| 593 |
+
"eval_macro avg": {
|
| 594 |
+
"f1-score": 0.8736823608334341,
|
| 595 |
+
"precision": 0.8728234806633418,
|
| 596 |
+
"recall": 0.8770959255453377,
|
| 597 |
+
"support": 32431.0
|
| 598 |
+
},
|
| 599 |
+
"eval_runtime": 4.8358,
|
| 600 |
+
"eval_samples_per_second": 16.543,
|
| 601 |
+
"eval_steps_per_second": 2.068,
|
| 602 |
+
"eval_weighted avg": {
|
| 603 |
+
"f1-score": 0.9098800911419986,
|
| 604 |
+
"precision": 0.909053016531805,
|
| 605 |
+
"recall": 0.9119361105115475,
|
| 606 |
+
"support": 32431.0
|
| 607 |
+
},
|
| 608 |
+
"step": 1053
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 14.0,
|
| 612 |
+
"eval_Claim": {
|
| 613 |
+
"f1-score": 0.7047413793103449,
|
| 614 |
+
"precision": 0.6923565530383231,
|
| 615 |
+
"recall": 0.717577353522054,
|
| 616 |
+
"support": 4557.0
|
| 617 |
+
},
|
| 618 |
+
"eval_MajorClaim": {
|
| 619 |
+
"f1-score": 0.9112001687407719,
|
| 620 |
+
"precision": 0.8737864077669902,
|
| 621 |
+
"recall": 0.9519612163948876,
|
| 622 |
+
"support": 2269.0
|
| 623 |
+
},
|
| 624 |
+
"eval_O": {
|
| 625 |
+
"f1-score": 0.9992316384180792,
|
| 626 |
+
"precision": 1.0,
|
| 627 |
+
"recall": 0.998464456688646,
|
| 628 |
+
"support": 11071.0
|
| 629 |
+
},
|
| 630 |
+
"eval_Premise": {
|
| 631 |
+
"f1-score": 0.9145424153781864,
|
| 632 |
+
"precision": 0.9258919757439007,
|
| 633 |
+
"recall": 0.903467730838035,
|
| 634 |
+
"support": 14534.0
|
| 635 |
+
},
|
| 636 |
+
"eval_accuracy": 0.9131694983195091,
|
| 637 |
+
"eval_loss": 0.5468704700469971,
|
| 638 |
+
"eval_macro avg": {
|
| 639 |
+
"f1-score": 0.8824289004618456,
|
| 640 |
+
"precision": 0.8730087341373035,
|
| 641 |
+
"recall": 0.8928676893609057,
|
| 642 |
+
"support": 32431.0
|
| 643 |
+
},
|
| 644 |
+
"eval_runtime": 4.8411,
|
| 645 |
+
"eval_samples_per_second": 16.525,
|
| 646 |
+
"eval_steps_per_second": 2.066,
|
| 647 |
+
"eval_weighted avg": {
|
| 648 |
+
"f1-score": 0.9137390947680665,
|
| 649 |
+
"precision": 0.9147298617643856,
|
| 650 |
+
"recall": 0.9131694983195091,
|
| 651 |
+
"support": 32431.0
|
| 652 |
+
},
|
| 653 |
+
"step": 1134
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 15.0,
|
| 657 |
+
"eval_Claim": {
|
| 658 |
+
"f1-score": 0.6868638486408813,
|
| 659 |
+
"precision": 0.7559304164470216,
|
| 660 |
+
"recall": 0.6293614219881501,
|
| 661 |
+
"support": 4557.0
|
| 662 |
+
},
|
| 663 |
+
"eval_MajorClaim": {
|
| 664 |
+
"f1-score": 0.9058546000428909,
|
| 665 |
+
"precision": 0.8822055137844611,
|
| 666 |
+
"recall": 0.9308065226972234,
|
| 667 |
+
"support": 2269.0
|
| 668 |
+
},
|
| 669 |
+
"eval_O": {
|
| 670 |
+
"f1-score": 0.9999548349216385,
|
| 671 |
+
"precision": 1.0,
|
| 672 |
+
"recall": 0.9999096739228616,
|
| 673 |
+
"support": 11071.0
|
| 674 |
+
},
|
| 675 |
+
"eval_Premise": {
|
| 676 |
+
"f1-score": 0.9218029420675262,
|
| 677 |
+
"precision": 0.9023924075660713,
|
| 678 |
+
"recall": 0.9420668776661622,
|
| 679 |
+
"support": 14534.0
|
| 680 |
+
},
|
| 681 |
+
"eval_accuracy": 0.9170855046097869,
|
| 682 |
+
"eval_loss": 0.6552095413208008,
|
| 683 |
+
"eval_macro avg": {
|
| 684 |
+
"f1-score": 0.8786190564182342,
|
| 685 |
+
"precision": 0.8851320844493885,
|
| 686 |
+
"recall": 0.8755361240685993,
|
| 687 |
+
"support": 32431.0
|
| 688 |
+
},
|
| 689 |
+
"eval_runtime": 4.8081,
|
| 690 |
+
"eval_samples_per_second": 16.639,
|
| 691 |
+
"eval_steps_per_second": 2.08,
|
| 692 |
+
"eval_weighted avg": {
|
| 693 |
+
"f1-score": 0.9143537536055225,
|
| 694 |
+
"precision": 0.9137205288178378,
|
| 695 |
+
"recall": 0.9170855046097869,
|
| 696 |
+
"support": 32431.0
|
| 697 |
+
},
|
| 698 |
+
"step": 1215
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 16.0,
|
| 702 |
+
"eval_Claim": {
|
| 703 |
+
"f1-score": 0.7261954261954263,
|
| 704 |
+
"precision": 0.6899071696622556,
|
| 705 |
+
"recall": 0.7665130568356375,
|
| 706 |
+
"support": 4557.0
|
| 707 |
+
},
|
| 708 |
+
"eval_MajorClaim": {
|
| 709 |
+
"f1-score": 0.9104052573932091,
|
| 710 |
+
"precision": 0.9050522648083623,
|
| 711 |
+
"recall": 0.9158219479947113,
|
| 712 |
+
"support": 2269.0
|
| 713 |
+
},
|
| 714 |
+
"eval_O": {
|
| 715 |
+
"f1-score": 0.9999548349216385,
|
| 716 |
+
"precision": 1.0,
|
| 717 |
+
"recall": 0.9999096739228616,
|
| 718 |
+
"support": 11071.0
|
| 719 |
+
},
|
| 720 |
+
"eval_Premise": {
|
| 721 |
+
"f1-score": 0.9186291000841041,
|
| 722 |
+
"precision": 0.9360805599200114,
|
| 723 |
+
"recall": 0.9018164304389706,
|
| 724 |
+
"support": 14534.0
|
| 725 |
+
},
|
| 726 |
+
"eval_accuracy": 0.9172705127809812,
|
| 727 |
+
"eval_loss": 0.604762613773346,
|
| 728 |
+
"eval_macro avg": {
|
| 729 |
+
"f1-score": 0.8887961546485946,
|
| 730 |
+
"precision": 0.8827599985976573,
|
| 731 |
+
"recall": 0.8960152772980452,
|
| 732 |
+
"support": 32431.0
|
| 733 |
+
},
|
| 734 |
+
"eval_runtime": 4.8523,
|
| 735 |
+
"eval_samples_per_second": 16.487,
|
| 736 |
+
"eval_steps_per_second": 2.061,
|
| 737 |
+
"eval_weighted avg": {
|
| 738 |
+
"f1-score": 0.918776399254959,
|
| 739 |
+
"precision": 0.92113920073012,
|
| 740 |
+
"recall": 0.9172705127809812,
|
| 741 |
+
"support": 32431.0
|
| 742 |
+
},
|
| 743 |
+
"step": 1296
|
| 744 |
+
},
|
| 745 |
+
{
|
| 746 |
+
"epoch": 17.0,
|
| 747 |
+
"eval_Claim": {
|
| 748 |
+
"f1-score": 0.6726977499420088,
|
| 749 |
+
"precision": 0.7134071340713407,
|
| 750 |
+
"recall": 0.6363835856923414,
|
| 751 |
+
"support": 4557.0
|
| 752 |
+
},
|
| 753 |
+
"eval_MajorClaim": {
|
| 754 |
+
"f1-score": 0.898166556218246,
|
| 755 |
+
"precision": 0.9003542958370239,
|
| 756 |
+
"recall": 0.8959894226531512,
|
| 757 |
+
"support": 2269.0
|
| 758 |
+
},
|
| 759 |
+
"eval_O": {
|
| 760 |
+
"f1-score": 0.9999548349216385,
|
| 761 |
+
"precision": 1.0,
|
| 762 |
+
"recall": 0.9999096739228616,
|
| 763 |
+
"support": 11071.0
|
| 764 |
+
},
|
| 765 |
+
"eval_Premise": {
|
| 766 |
+
"f1-score": 0.9172189909373732,
|
| 767 |
+
"precision": 0.9018486500864477,
|
| 768 |
+
"recall": 0.9331223338378973,
|
| 769 |
+
"support": 14534.0
|
| 770 |
+
},
|
| 771 |
+
"eval_accuracy": 0.9116277635595572,
|
| 772 |
+
"eval_loss": 0.6666390299797058,
|
| 773 |
+
"eval_macro avg": {
|
| 774 |
+
"f1-score": 0.8720095330048165,
|
| 775 |
+
"precision": 0.8789025199987031,
|
| 776 |
+
"recall": 0.8663512540265629,
|
| 777 |
+
"support": 32431.0
|
| 778 |
+
},
|
| 779 |
+
"eval_runtime": 4.8204,
|
| 780 |
+
"eval_samples_per_second": 16.596,
|
| 781 |
+
"eval_steps_per_second": 2.075,
|
| 782 |
+
"eval_weighted avg": {
|
| 783 |
+
"f1-score": 0.9097710324765249,
|
| 784 |
+
"precision": 0.9087714991080675,
|
| 785 |
+
"recall": 0.9116277635595572,
|
| 786 |
+
"support": 32431.0
|
| 787 |
+
},
|
| 788 |
+
"step": 1377
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 18.0,
|
| 792 |
+
"eval_Claim": {
|
| 793 |
+
"f1-score": 0.7242054854157598,
|
| 794 |
+
"precision": 0.7184193478730296,
|
| 795 |
+
"recall": 0.7300855826201448,
|
| 796 |
+
"support": 4557.0
|
| 797 |
+
},
|
| 798 |
+
"eval_MajorClaim": {
|
| 799 |
+
"f1-score": 0.91550204257149,
|
| 800 |
+
"precision": 0.8937867338371117,
|
| 801 |
+
"recall": 0.9382988100484795,
|
| 802 |
+
"support": 2269.0
|
| 803 |
+
},
|
| 804 |
+
"eval_O": {
|
| 805 |
+
"f1-score": 0.9999548349216385,
|
| 806 |
+
"precision": 1.0,
|
| 807 |
+
"recall": 0.9999096739228616,
|
| 808 |
+
"support": 11071.0
|
| 809 |
+
},
|
| 810 |
+
"eval_Premise": {
|
| 811 |
+
"f1-score": 0.9218198185721211,
|
| 812 |
+
"precision": 0.9277948146083078,
|
| 813 |
+
"recall": 0.9159212880143113,
|
| 814 |
+
"support": 14534.0
|
| 815 |
+
},
|
| 816 |
+
"eval_accuracy": 0.9200456353488946,
|
| 817 |
+
"eval_loss": 0.5898464918136597,
|
| 818 |
+
"eval_macro avg": {
|
| 819 |
+
"f1-score": 0.8903705453702524,
|
| 820 |
+
"precision": 0.8850002240796123,
|
| 821 |
+
"recall": 0.8960538386514494,
|
| 822 |
+
"support": 32431.0
|
| 823 |
+
},
|
| 824 |
+
"eval_runtime": 4.8354,
|
| 825 |
+
"eval_samples_per_second": 16.545,
|
| 826 |
+
"eval_steps_per_second": 2.068,
|
| 827 |
+
"eval_weighted avg": {
|
| 828 |
+
"f1-score": 0.9202833015380038,
|
| 829 |
+
"precision": 0.9206441029524512,
|
| 830 |
+
"recall": 0.9200456353488946,
|
| 831 |
+
"support": 32431.0
|
| 832 |
+
},
|
| 833 |
+
"step": 1458
|
| 834 |
+
},
|
| 835 |
+
{
|
| 836 |
+
"epoch": 18.52,
|
| 837 |
+
"grad_norm": 0.2594422399997711,
|
| 838 |
+
"learning_rate": 1.2592592592592593e-05,
|
| 839 |
+
"loss": 0.0093,
|
| 840 |
+
"step": 1500
|
| 841 |
+
},
|
| 842 |
+
{
|
| 843 |
+
"epoch": 19.0,
|
| 844 |
+
"eval_Claim": {
|
| 845 |
+
"f1-score": 0.7198604752561586,
|
| 846 |
+
"precision": 0.7151830192765866,
|
| 847 |
+
"recall": 0.7245995172262454,
|
| 848 |
+
"support": 4557.0
|
| 849 |
+
},
|
| 850 |
+
"eval_MajorClaim": {
|
| 851 |
+
"f1-score": 0.9025769956002515,
|
| 852 |
+
"precision": 0.860223642172524,
|
| 853 |
+
"recall": 0.9493168796826796,
|
| 854 |
+
"support": 2269.0
|
| 855 |
+
},
|
| 856 |
+
"eval_O": {
|
| 857 |
+
"f1-score": 0.9975098474215601,
|
| 858 |
+
"precision": 1.0,
|
| 859 |
+
"recall": 0.9950320657573841,
|
| 860 |
+
"support": 11071.0
|
| 861 |
+
},
|
| 862 |
+
"eval_Premise": {
|
| 863 |
+
"f1-score": 0.9212571111419453,
|
| 864 |
+
"precision": 0.9289911851126347,
|
| 865 |
+
"recall": 0.9136507499655979,
|
| 866 |
+
"support": 14534.0
|
| 867 |
+
},
|
| 868 |
+
"eval_accuracy": 0.9173630168665783,
|
| 869 |
+
"eval_loss": 0.6108261346817017,
|
| 870 |
+
"eval_macro avg": {
|
| 871 |
+
"f1-score": 0.8853011073549788,
|
| 872 |
+
"precision": 0.8760994616404363,
|
| 873 |
+
"recall": 0.8956498031579767,
|
| 874 |
+
"support": 32431.0
|
| 875 |
+
},
|
| 876 |
+
"eval_runtime": 4.8341,
|
| 877 |
+
"eval_samples_per_second": 16.549,
|
| 878 |
+
"eval_steps_per_second": 2.069,
|
| 879 |
+
"eval_weighted avg": {
|
| 880 |
+
"f1-score": 0.9176816552958715,
|
| 881 |
+
"precision": 0.9183773040411919,
|
| 882 |
+
"recall": 0.9173630168665783,
|
| 883 |
+
"support": 32431.0
|
| 884 |
+
},
|
| 885 |
+
"step": 1539
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 20.0,
|
| 889 |
+
"eval_Claim": {
|
| 890 |
+
"f1-score": 0.7150392937883518,
|
| 891 |
+
"precision": 0.7018174133558749,
|
| 892 |
+
"recall": 0.7287689269256089,
|
| 893 |
+
"support": 4557.0
|
| 894 |
+
},
|
| 895 |
+
"eval_MajorClaim": {
|
| 896 |
+
"f1-score": 0.8972902097902098,
|
| 897 |
+
"precision": 0.8899003034243607,
|
| 898 |
+
"recall": 0.9048038783605112,
|
| 899 |
+
"support": 2269.0
|
| 900 |
+
},
|
| 901 |
+
"eval_O": {
|
| 902 |
+
"f1-score": 0.9992316384180792,
|
| 903 |
+
"precision": 1.0,
|
| 904 |
+
"recall": 0.998464456688646,
|
| 905 |
+
"support": 11071.0
|
| 906 |
+
},
|
| 907 |
+
"eval_Premise": {
|
| 908 |
+
"f1-score": 0.9203380437794403,
|
| 909 |
+
"precision": 0.9266285395452644,
|
| 910 |
+
"recall": 0.9141323792486583,
|
| 911 |
+
"support": 14534.0
|
| 912 |
+
},
|
| 913 |
+
"eval_accuracy": 0.9162221331442139,
|
| 914 |
+
"eval_loss": 0.6273905038833618,
|
| 915 |
+
"eval_macro avg": {
|
| 916 |
+
"f1-score": 0.8829747964440202,
|
| 917 |
+
"precision": 0.879586564081375,
|
| 918 |
+
"recall": 0.886542410305856,
|
| 919 |
+
"support": 32431.0
|
| 920 |
+
},
|
| 921 |
+
"eval_runtime": 4.8321,
|
| 922 |
+
"eval_samples_per_second": 16.556,
|
| 923 |
+
"eval_steps_per_second": 2.069,
|
| 924 |
+
"eval_weighted avg": {
|
| 925 |
+
"f1-score": 0.9168102169228345,
|
| 926 |
+
"precision": 0.9175167258142971,
|
| 927 |
+
"recall": 0.9162221331442139,
|
| 928 |
+
"support": 32431.0
|
| 929 |
+
},
|
| 930 |
+
"step": 1620
|
| 931 |
+
}
|
| 932 |
+
],
|
| 933 |
+
"logging_steps": 500,
|
| 934 |
+
"max_steps": 4050,
|
| 935 |
+
"num_input_tokens_seen": 0,
|
| 936 |
+
"num_train_epochs": 50,
|
| 937 |
+
"save_steps": 500,
|
| 938 |
+
"total_flos": 2875774654536000.0,
|
| 939 |
+
"train_batch_size": 4,
|
| 940 |
+
"trial_name": null,
|
| 941 |
+
"trial_params": null
|
| 942 |
+
}
|
checkpoint-1620/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b2dd7b387b7349843b62a9d5b4facde1d14a6a35003d55f9e21a477105ecd1d3
|
| 3 |
+
size 4856
|
checkpoint-1620/vocab.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|