Hugging Face's logo

nllg
/
clf-canine-m

Text Classification
Transformers
PyTorch
canine
Inference Endpoints
Model card Files Community
Jonas Belouadi commited on
Commit
633f7d6
•
1 Parent(s): b562eac

upload model files

Browse files
Files changed (9) hide show
  1. config.json +48 -0
  2. pytorch_model.bin +3 -0
  3. special_tokens_map.json +44 -0
  4. test-de_results.json +11 -0
  5. test-en_results.json +11 -0
  6. test_results.json +11 -0
  7. tokenizer_config.json +55 -0
  8. trainer_state.json +3189 -0
  9. training_args.bin +3 -0
config.json ADDED
@@ -0,0 +1,48 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "models/canine-c/meter/run-1/checkpoint-81700",
3
+ "architectures": [
4
+ "CanineForSequenceClassification"
5
+ ],
6
+ "attention_probs_dropout_prob": 0.1,
7
+ "bos_token_id": 57344,
8
+ "downsampling_rate": 4,
9
+ "eos_token_id": 57345,
10
+ "hidden_act": "gelu",
11
+ "hidden_dropout_prob": 0.1,
12
+ "hidden_size": 768,
13
+ "id2label": {
14
+ "0": "LABEL_0",
15
+ "1": "LABEL_1",
16
+ "2": "LABEL_2",
17
+ "3": "LABEL_3",
18
+ "4": "LABEL_4",
19
+ "5": "LABEL_5",
20
+ "6": "LABEL_6"
21
+ },
22
+ "initializer_range": 0.02,
23
+ "intermediate_size": 3072,
24
+ "label2id": {
25
+ "LABEL_0": 0,
26
+ "LABEL_1": 1,
27
+ "LABEL_2": 2,
28
+ "LABEL_3": 3,
29
+ "LABEL_4": 4,
30
+ "LABEL_5": 5,
31
+ "LABEL_6": 6
32
+ },
33
+ "layer_norm_eps": 1e-12,
34
+ "local_transformer_stride": 128,
35
+ "max_position_embeddings": 16384,
36
+ "model_type": "canine",
37
+ "num_attention_heads": 12,
38
+ "num_hash_buckets": 16384,
39
+ "num_hash_functions": 8,
40
+ "num_hidden_layers": 12,
41
+ "pad_token_id": 0,
42
+ "problem_type": "single_label_classification",
43
+ "torch_dtype": "float32",
44
+ "transformers_version": "4.20.0",
45
+ "type_vocab_size": 16,
46
+ "upsampling_kernel_size": 4,
47
+ "use_cache": true
48
+ }
pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ca2394e51b4ca39724bece986bd937e2a2ae3d8ad430ef265fdbba62aa76a328
3
+ size 528568015
special_tokens_map.json ADDED
@@ -0,0 +1,44 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "",
4
+ "lstrip": false,
5
+ "normalized": true,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "cls_token": {
10
+ "content": "",
11
+ "lstrip": false,
12
+ "normalized": true,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "eos_token": {
17
+ "content": "",
18
+ "lstrip": false,
19
+ "normalized": true,
20
+ "rstrip": false,
21
+ "single_word": false
22
+ },
23
+ "mask_token": {
24
+ "content": "",
25
+ "lstrip": true,
26
+ "normalized": true,
27
+ "rstrip": false,
28
+ "single_word": false
29
+ },
30
+ "pad_token": {
31
+ "content": "\u0000",
32
+ "lstrip": false,
33
+ "normalized": true,
34
+ "rstrip": false,
35
+ "single_word": false
36
+ },
37
+ "sep_token": {
38
+ "content": "",
39
+ "lstrip": false,
40
+ "normalized": true,
41
+ "rstrip": false,
42
+ "single_word": false
43
+ }
44
+ }
test-de_results.json ADDED
@@ -0,0 +1,11 @@
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 100.0,
3
+ "test_accuracy": 0.7777777777777778,
4
+ "test_f1": 0.53644033718575,
5
+ "test_loss": 2.2133560180664062,
6
+ "test_precision": 0.5642564642564641,
7
+ "test_recall": 0.5213481314360918,
8
+ "test_runtime": 0.2798,
9
+ "test_samples_per_second": 611.236,
10
+ "test_steps_per_second": 78.639
11
+ }
test-en_results.json ADDED
@@ -0,0 +1,11 @@
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 100.0,
3
+ "test_accuracy": 0.6580310880829016,
4
+ "test_f1": 0.45326536391956956,
5
+ "test_loss": 3.5779941082000732,
6
+ "test_precision": 0.48238947630540546,
7
+ "test_recall": 0.4350463560944953,
8
+ "test_runtime": 0.3171,
9
+ "test_samples_per_second": 608.634,
10
+ "test_steps_per_second": 78.839
11
+ }
test_results.json ADDED
@@ -0,0 +1,11 @@
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 100.0,
3
+ "test_accuracy": 0.7087912087912088,
4
+ "test_f1": 0.5848789835776328,
5
+ "test_loss": 2.9805569648742676,
6
+ "test_precision": 0.6035925656852257,
7
+ "test_recall": 0.5702147059289916,
8
+ "test_runtime": 0.58,
9
+ "test_samples_per_second": 627.64,
10
+ "test_steps_per_second": 79.317
11
+ }
tokenizer_config.json ADDED
@@ -0,0 +1,55 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_prefix_space": false,
3
+ "bos_token": {
4
+ "__type": "AddedToken",
5
+ "content": "",
6
+ "lstrip": false,
7
+ "normalized": true,
8
+ "rstrip": false,
9
+ "single_word": false
10
+ },
11
+ "cls_token": {
12
+ "__type": "AddedToken",
13
+ "content": "",
14
+ "lstrip": false,
15
+ "normalized": true,
16
+ "rstrip": false,
17
+ "single_word": false
18
+ },
19
+ "eos_token": {
20
+ "__type": "AddedToken",
21
+ "content": "",
22
+ "lstrip": false,
23
+ "normalized": true,
24
+ "rstrip": false,
25
+ "single_word": false
26
+ },
27
+ "mask_token": {
28
+ "__type": "AddedToken",
29
+ "content": "",
30
+ "lstrip": true,
31
+ "normalized": true,
32
+ "rstrip": false,
33
+ "single_word": false
34
+ },
35
+ "model_max_length": 2048,
36
+ "name_or_path": "google/canine-c",
37
+ "pad_token": {
38
+ "__type": "AddedToken",
39
+ "content": "\u0000",
40
+ "lstrip": false,
41
+ "normalized": true,
42
+ "rstrip": false,
43
+ "single_word": false
44
+ },
45
+ "sep_token": {
46
+ "__type": "AddedToken",
47
+ "content": "",
48
+ "lstrip": false,
49
+ "normalized": true,
50
+ "rstrip": false,
51
+ "single_word": false
52
+ },
53
+ "special_tokens_map_file": "/home/jbelouadi/.cache/huggingface/transformers/f8be97736b4765e3e8d559b4e7d4f1f531b1621150e5344b600684a27bc84e38.ab71f530366fe02e2834427e7b90198bfd0d573bc4279bfafdb2b95fe2b46dde",
54
+ "tokenizer_class": "CanineTokenizer"
55
+ }
trainer_state.json ADDED
@@ -0,0 +1,3189 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 100.0,
5
+ "global_step": 81700,
6
+ "is_hyper_param_search": true,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.0,
12
+ "learning_rate": 6.119951040391677e-09,
13
+ "loss": 1.9236,
14
+ "step": 1
15
+ },
16
+ {
17
+ "epoch": 0.31,
18
+ "learning_rate": 1.5299877600979192e-06,
19
+ "loss": 1.7998,
20
+ "step": 250
21
+ },
22
+ {
23
+ "epoch": 0.61,
24
+ "learning_rate": 3.0599755201958383e-06,
25
+ "loss": 1.4687,
26
+ "step": 500
27
+ },
28
+ {
29
+ "epoch": 0.92,
30
+ "learning_rate": 4.589963280293758e-06,
31
+ "loss": 1.3699,
32
+ "step": 750
33
+ },
34
+ {
35
+ "epoch": 1.0,
36
+ "eval_accuracy": 0.5730027548209367,
37
+ "eval_f1": 0.10407805854390793,
38
+ "eval_loss": 1.3314192295074463,
39
+ "eval_precision": 0.08185753640299095,
40
+ "eval_recall": 0.14285714285714285,
41
+ "eval_runtime": 0.5773,
42
+ "eval_samples_per_second": 628.824,
43
+ "eval_steps_per_second": 79.686,
44
+ "step": 817
45
+ },
46
+ {
47
+ "epoch": 1.22,
48
+ "learning_rate": 6.119951040391677e-06,
49
+ "loss": 1.3287,
50
+ "step": 1000
51
+ },
52
+ {
53
+ "epoch": 1.53,
54
+ "learning_rate": 7.649938800489596e-06,
55
+ "loss": 1.3002,
56
+ "step": 1250
57
+ },
58
+ {
59
+ "epoch": 1.84,
60
+ "learning_rate": 9.179926560587515e-06,
61
+ "loss": 1.258,
62
+ "step": 1500
63
+ },
64
+ {
65
+ "epoch": 2.0,
66
+ "eval_accuracy": 0.6060606060606061,
67
+ "eval_f1": 0.183279210886012,
68
+ "eval_loss": 1.1635808944702148,
69
+ "eval_precision": 0.15974025974025974,
70
+ "eval_recall": 0.21892762410003788,
71
+ "eval_runtime": 0.5777,
72
+ "eval_samples_per_second": 628.372,
73
+ "eval_steps_per_second": 79.628,
74
+ "step": 1634
75
+ },
76
+ {
77
+ "epoch": 2.14,
78
+ "learning_rate": 1.0709914320685436e-05,
79
+ "loss": 1.1884,
80
+ "step": 1750
81
+ },
82
+ {
83
+ "epoch": 2.45,
84
+ "learning_rate": 1.2239902080783353e-05,
85
+ "loss": 1.1126,
86
+ "step": 2000
87
+ },
88
+ {
89
+ "epoch": 2.75,
90
+ "learning_rate": 1.3769889840881272e-05,
91
+ "loss": 1.1123,
92
+ "step": 2250
93
+ },
94
+ {
95
+ "epoch": 3.0,
96
+ "eval_accuracy": 0.6143250688705234,
97
+ "eval_f1": 0.342984622186972,
98
+ "eval_loss": 1.053414225578308,
99
+ "eval_precision": 0.31294195637816785,
100
+ "eval_recall": 0.3835992619859615,
101
+ "eval_runtime": 0.5781,
102
+ "eval_samples_per_second": 627.939,
103
+ "eval_steps_per_second": 79.574,
104
+ "step": 2451
105
+ },
106
+ {
107
+ "epoch": 3.06,
108
+ "learning_rate": 1.5299877600979193e-05,
109
+ "loss": 1.0996,
110
+ "step": 2500
111
+ },
112
+ {
113
+ "epoch": 3.37,
114
+ "learning_rate": 1.6829865361077112e-05,
115
+ "loss": 0.9999,
116
+ "step": 2750
117
+ },
118
+ {
119
+ "epoch": 3.67,
120
+ "learning_rate": 1.835985312117503e-05,
121
+ "loss": 1.0141,
122
+ "step": 3000
123
+ },
124
+ {
125
+ "epoch": 3.98,
126
+ "learning_rate": 1.988984088127295e-05,
127
+ "loss": 0.993,
128
+ "step": 3250
129
+ },
130
+ {
131
+ "epoch": 4.0,
132
+ "eval_accuracy": 0.6528925619834711,
133
+ "eval_f1": 0.377005484318265,
134
+ "eval_loss": 1.0341299772262573,
135
+ "eval_precision": 0.4038621706705789,
136
+ "eval_recall": 0.4236673117522871,
137
+ "eval_runtime": 0.5774,
138
+ "eval_samples_per_second": 628.627,
139
+ "eval_steps_per_second": 79.661,
140
+ "step": 3268
141
+ },
142
+ {
143
+ "epoch": 4.28,
144
+ "learning_rate": 2.1419828641370872e-05,
145
+ "loss": 0.8827,
146
+ "step": 3500
147
+ },
148
+ {
149
+ "epoch": 4.59,
150
+ "learning_rate": 2.294981640146879e-05,
151
+ "loss": 0.9059,
152
+ "step": 3750
153
+ },
154
+ {
155
+ "epoch": 4.9,
156
+ "learning_rate": 2.4479804161566707e-05,
157
+ "loss": 0.8739,
158
+ "step": 4000
159
+ },
160
+ {
161
+ "epoch": 5.0,
162
+ "eval_accuracy": 0.6611570247933884,
163
+ "eval_f1": 0.3730793016507302,
164
+ "eval_loss": 0.9818496704101562,
165
+ "eval_precision": 0.48520306297883803,
166
+ "eval_recall": 0.39812740967297616,
167
+ "eval_runtime": 0.5781,
168
+ "eval_samples_per_second": 627.881,
169
+ "eval_steps_per_second": 79.566,
170
+ "step": 4085
171
+ },
172
+ {
173
+ "epoch": 5.2,
174
+ "learning_rate": 2.600979192166463e-05,
175
+ "loss": 0.8024,
176
+ "step": 4250
177
+ },
178
+ {
179
+ "epoch": 5.51,
180
+ "learning_rate": 2.7539779681762544e-05,
181
+ "loss": 0.8021,
182
+ "step": 4500
183
+ },
184
+ {
185
+ "epoch": 5.81,
186
+ "learning_rate": 2.9069767441860467e-05,
187
+ "loss": 0.8587,
188
+ "step": 4750
189
+ },
190
+ {
191
+ "epoch": 6.0,
192
+ "eval_accuracy": 0.6446280991735537,
193
+ "eval_f1": 0.372084738159454,
194
+ "eval_loss": 1.1154053211212158,
195
+ "eval_precision": 0.4065005040497793,
196
+ "eval_recall": 0.38243464936814686,
197
+ "eval_runtime": 0.5773,
198
+ "eval_samples_per_second": 628.792,
199
+ "eval_steps_per_second": 79.682,
200
+ "step": 4902
201
+ },
202
+ {
203
+ "epoch": 6.12,
204
+ "learning_rate": 3.0599755201958386e-05,
205
+ "loss": 0.7567,
206
+ "step": 5000
207
+ },
208
+ {
209
+ "epoch": 6.43,
210
+ "learning_rate": 3.2123623011015916e-05,
211
+ "loss": 0.6544,
212
+ "step": 5250
213
+ },
214
+ {
215
+ "epoch": 6.73,
216
+ "learning_rate": 3.365361077111383e-05,
217
+ "loss": 0.7251,
218
+ "step": 5500
219
+ },
220
+ {
221
+ "epoch": 7.0,
222
+ "eval_accuracy": 0.650137741046832,
223
+ "eval_f1": 0.47516667580958355,
224
+ "eval_loss": 1.0870000123977661,
225
+ "eval_precision": 0.4639813382108336,
226
+ "eval_recall": 0.4947985272800543,
227
+ "eval_runtime": 0.5771,
228
+ "eval_samples_per_second": 628.974,
229
+ "eval_steps_per_second": 79.705,
230
+ "step": 5719
231
+ },
232
+ {
233
+ "epoch": 7.04,
234
+ "learning_rate": 3.5183598531211754e-05,
235
+ "loss": 0.6829,
236
+ "step": 5750
237
+ },
238
+ {
239
+ "epoch": 7.34,
240
+ "learning_rate": 3.6707466340269284e-05,
241
+ "loss": 0.5838,
242
+ "step": 6000
243
+ },
244
+ {
245
+ "epoch": 7.65,
246
+ "learning_rate": 3.82374541003672e-05,
247
+ "loss": 0.6713,
248
+ "step": 6250
249
+ },
250
+ {
251
+ "epoch": 7.96,
252
+ "learning_rate": 3.9767441860465115e-05,
253
+ "loss": 0.6269,
254
+ "step": 6500
255
+ },
256
+ {
257
+ "epoch": 8.0,
258
+ "eval_accuracy": 0.5950413223140496,
259
+ "eval_f1": 0.44966469663834185,
260
+ "eval_loss": 1.2386085987091064,
261
+ "eval_precision": 0.4427425690783476,
262
+ "eval_recall": 0.5007950793049315,
263
+ "eval_runtime": 0.5812,
264
+ "eval_samples_per_second": 624.61,
265
+ "eval_steps_per_second": 79.152,
266
+ "step": 6536
267
+ },
268
+ {
269
+ "epoch": 8.26,
270
+ "learning_rate": 4.129742962056304e-05,
271
+ "loss": 0.5744,
272
+ "step": 6750
273
+ },
274
+ {
275
+ "epoch": 8.57,
276
+ "learning_rate": 4.282741738066095e-05,
277
+ "loss": 0.5549,
278
+ "step": 7000
279
+ },
280
+ {
281
+ "epoch": 8.87,
282
+ "learning_rate": 4.435740514075888e-05,
283
+ "loss": 0.6049,
284
+ "step": 7250
285
+ },
286
+ {
287
+ "epoch": 9.0,
288
+ "eval_accuracy": 0.6198347107438017,
289
+ "eval_f1": 0.32356296980065924,
290
+ "eval_loss": 1.368962287902832,
291
+ "eval_precision": 0.47826317360800125,
292
+ "eval_recall": 0.30602226522793025,
293
+ "eval_runtime": 0.5769,
294
+ "eval_samples_per_second": 629.173,
295
+ "eval_steps_per_second": 79.73,
296
+ "step": 7353
297
+ },
298
+ {
299
+ "epoch": 9.18,
300
+ "learning_rate": 4.58873929008568e-05,
301
+ "loss": 0.5767,
302
+ "step": 7500
303
+ },
304
+ {
305
+ "epoch": 9.49,
306
+ "learning_rate": 4.741738066095471e-05,
307
+ "loss": 0.5394,
308
+ "step": 7750
309
+ },
310
+ {
311
+ "epoch": 9.79,
312
+ "learning_rate": 4.894124847001224e-05,
313
+ "loss": 0.5507,
314
+ "step": 8000
315
+ },
316
+ {
317
+ "epoch": 10.0,
318
+ "eval_accuracy": 0.6446280991735537,
319
+ "eval_f1": 0.4573642555682361,
320
+ "eval_loss": 1.2706860303878784,
321
+ "eval_precision": 0.4887882409493839,
322
+ "eval_recall": 0.4411244714508261,
323
+ "eval_runtime": 0.5758,
324
+ "eval_samples_per_second": 630.429,
325
+ "eval_steps_per_second": 79.889,
326
+ "step": 8170
327
+ },
328
+ {
329
+ "epoch": 10.1,
330
+ "learning_rate": 4.99998647112497e-05,
331
+ "loss": 0.609,
332
+ "step": 8250
333
+ },
334
+ {
335
+ "epoch": 10.4,
336
+ "learning_rate": 4.999756011662697e-05,
337
+ "loss": 0.5261,
338
+ "step": 8500
339
+ },
340
+ {
341
+ "epoch": 10.71,
342
+ "learning_rate": 4.9992429864131085e-05,
343
+ "loss": 0.5329,
344
+ "step": 8750
345
+ },
346
+ {
347
+ "epoch": 11.0,
348
+ "eval_accuracy": 0.6611570247933884,
349
+ "eval_f1": 0.44773949967842547,
350
+ "eval_loss": 1.4297670125961304,
351
+ "eval_precision": 0.4707209760487202,
352
+ "eval_recall": 0.4446715025532759,
353
+ "eval_runtime": 0.5769,
354
+ "eval_samples_per_second": 629.193,
355
+ "eval_steps_per_second": 79.732,
356
+ "step": 8987
357
+ },
358
+ {
359
+ "epoch": 11.02,
360
+ "learning_rate": 4.998443333261708e-05,
361
+ "loss": 0.5434,
362
+ "step": 9000
363
+ },
364
+ {
365
+ "epoch": 11.32,
366
+ "learning_rate": 4.997358633418236e-05,
367
+ "loss": 0.4591,
368
+ "step": 9250
369
+ },
370
+ {
371
+ "epoch": 11.63,
372
+ "learning_rate": 4.9959890106357945e-05,
373
+ "loss": 0.5308,
374
+ "step": 9500
375
+ },
376
+ {
377
+ "epoch": 11.93,
378
+ "learning_rate": 4.9943346211742556e-05,
379
+ "loss": 0.5027,
380
+ "step": 9750
381
+ },
382
+ {
383
+ "epoch": 12.0,
384
+ "eval_accuracy": 0.6776859504132231,
385
+ "eval_f1": 0.4263136100758106,
386
+ "eval_loss": 1.5769587755203247,
387
+ "eval_precision": 0.4877654234638867,
388
+ "eval_recall": 0.42436502673571636,
389
+ "eval_runtime": 0.5768,
390
+ "eval_samples_per_second": 629.365,
391
+ "eval_steps_per_second": 79.754,
392
+ "step": 9804
393
+ },
394
+ {
395
+ "epoch": 12.24,
396
+ "learning_rate": 4.9923956537824456e-05,
397
+ "loss": 0.4841,
398
+ "step": 10000
399
+ },
400
+ {
401
+ "epoch": 12.55,
402
+ "learning_rate": 4.990172329676604e-05,
403
+ "loss": 0.4372,
404
+ "step": 10250
405
+ },
406
+ {
407
+ "epoch": 12.85,
408
+ "learning_rate": 4.987664902515149e-05,
409
+ "loss": 0.4622,
410
+ "step": 10500
411
+ },
412
+ {
413
+ "epoch": 13.0,
414
+ "eval_accuracy": 0.7052341597796143,
415
+ "eval_f1": 0.4098788238632856,
416
+ "eval_loss": 1.5736510753631592,
417
+ "eval_precision": 0.4891951861291893,
418
+ "eval_recall": 0.38998470741081576,
419
+ "eval_runtime": 0.5769,
420
+ "eval_samples_per_second": 629.25,
421
+ "eval_steps_per_second": 79.74,
422
+ "step": 10621
423
+ },
424
+ {
425
+ "epoch": 13.16,
426
+ "learning_rate": 4.984873658369735e-05,
427
+ "loss": 0.4393,
428
+ "step": 10750
429
+ },
430
+ {
431
+ "epoch": 13.46,
432
+ "learning_rate": 4.981798915692615e-05,
433
+ "loss": 0.4071,
434
+ "step": 11000
435
+ },
436
+ {
437
+ "epoch": 13.77,
438
+ "learning_rate": 4.978441025280311e-05,
439
+ "loss": 0.4388,
440
+ "step": 11250
441
+ },
442
+ {
443
+ "epoch": 14.0,
444
+ "eval_accuracy": 0.6942148760330579,
445
+ "eval_f1": 0.4961367511874214,
446
+ "eval_loss": 1.7623515129089355,
447
+ "eval_precision": 0.5284679001386023,
448
+ "eval_recall": 0.4817265964741334,
449
+ "eval_runtime": 0.5819,
450
+ "eval_samples_per_second": 623.821,
451
+ "eval_steps_per_second": 79.052,
452
+ "step": 11438
453
+ },
454
+ {
455
+ "epoch": 14.08,
456
+ "learning_rate": 4.9748003702335874e-05,
457
+ "loss": 0.361,
458
+ "step": 11500
459
+ },
460
+ {
461
+ "epoch": 14.38,
462
+ "learning_rate": 4.970893619788296e-05,
463
+ "loss": 0.3574,
464
+ "step": 11750
465
+ },
466
+ {
467
+ "epoch": 14.69,
468
+ "learning_rate": 4.966689840442375e-05,
469
+ "loss": 0.3927,
470
+ "step": 12000
471
+ },
472
+ {
473
+ "epoch": 14.99,
474
+ "learning_rate": 4.9622046371513854e-05,
475
+ "loss": 0.3973,
476
+ "step": 12250
477
+ },
478
+ {
479
+ "epoch": 15.0,
480
+ "eval_accuracy": 0.650137741046832,
481
+ "eval_f1": 0.45953933518622214,
482
+ "eval_loss": 1.8296639919281006,
483
+ "eval_precision": 0.547006976703725,
484
+ "eval_recall": 0.4533570015097109,
485
+ "eval_runtime": 0.5782,
486
+ "eval_samples_per_second": 627.762,
487
+ "eval_steps_per_second": 79.551,
488
+ "step": 12255
489
+ },
490
+ {
491
+ "epoch": 15.3,
492
+ "learning_rate": 4.957438521630898e-05,
493
+ "loss": 0.3236,
494
+ "step": 12500
495
+ },
496
+ {
497
+ "epoch": 15.61,
498
+ "learning_rate": 4.952392037645683e-05,
499
+ "loss": 0.3474,
500
+ "step": 12750
501
+ },
502
+ {
503
+ "epoch": 15.91,
504
+ "learning_rate": 4.94706576094767e-05,
505
+ "loss": 0.3578,
506
+ "step": 13000
507
+ },
508
+ {
509
+ "epoch": 16.0,
510
+ "eval_accuracy": 0.7079889807162535,
511
+ "eval_f1": 0.49300722979484873,
512
+ "eval_loss": 1.4506721496582031,
513
+ "eval_precision": 0.5450297500169623,
514
+ "eval_recall": 0.4661049985865257,
515
+ "eval_runtime": 0.5767,
516
+ "eval_samples_per_second": 629.403,
517
+ "eval_steps_per_second": 79.759,
518
+ "step": 13072
519
+ },
520
+ {
521
+ "epoch": 16.22,
522
+ "learning_rate": 4.9414602992102564e-05,
523
+ "loss": 0.3665,
524
+ "step": 13250
525
+ },
526
+ {
527
+ "epoch": 16.52,
528
+ "learning_rate": 4.9355762919589846e-05,
529
+ "loss": 0.3197,
530
+ "step": 13500
531
+ },
532
+ {
533
+ "epoch": 16.83,
534
+ "learning_rate": 4.929414410498574e-05,
535
+ "loss": 0.3632,
536
+ "step": 13750
537
+ },
538
+ {
539
+ "epoch": 17.0,
540
+ "eval_accuracy": 0.6336088154269972,
541
+ "eval_f1": 0.44717986010161176,
542
+ "eval_loss": 1.922100305557251,
543
+ "eval_precision": 0.45028944394750214,
544
+ "eval_recall": 0.46335067846767364,
545
+ "eval_runtime": 0.5779,
546
+ "eval_samples_per_second": 628.173,
547
+ "eval_steps_per_second": 79.603,
548
+ "step": 13889
549
+ },
550
+ {
551
+ "epoch": 17.14,
552
+ "learning_rate": 4.922975357836337e-05,
553
+ "loss": 0.3885,
554
+ "step": 14000
555
+ },
556
+ {
557
+ "epoch": 17.44,
558
+ "learning_rate": 4.916259868601966e-05,
559
+ "loss": 0.3569,
560
+ "step": 14250
561
+ },
562
+ {
563
+ "epoch": 17.75,
564
+ "learning_rate": 4.909268708963725e-05,
565
+ "loss": 0.3409,
566
+ "step": 14500
567
+ },
568
+ {
569
+ "epoch": 18.0,
570
+ "eval_accuracy": 0.6859504132231405,
571
+ "eval_f1": 0.50463296007239,
572
+ "eval_loss": 1.6839758157730103,
573
+ "eval_precision": 0.532328893998914,
574
+ "eval_recall": 0.4880845242606326,
575
+ "eval_runtime": 0.5777,
576
+ "eval_samples_per_second": 628.346,
577
+ "eval_steps_per_second": 79.625,
578
+ "step": 14706
579
+ },
580
+ {
581
+ "epoch": 18.05,
582
+ "learning_rate": 4.902002676541038e-05,
583
+ "loss": 0.3504,
584
+ "step": 14750
585
+ },
586
+ {
587
+ "epoch": 18.36,
588
+ "learning_rate": 4.894462600313483e-05,
589
+ "loss": 0.3144,
590
+ "step": 15000
591
+ },
592
+ {
593
+ "epoch": 18.67,
594
+ "learning_rate": 4.8866493405262234e-05,
595
+ "loss": 0.3445,
596
+ "step": 15250
597
+ },
598
+ {
599
+ "epoch": 18.97,
600
+ "learning_rate": 4.878563788591853e-05,
601
+ "loss": 0.2723,
602
+ "step": 15500
603
+ },
604
+ {
605
+ "epoch": 19.0,
606
+ "eval_accuracy": 0.696969696969697,
607
+ "eval_f1": 0.5104610371804872,
608
+ "eval_loss": 1.8228646516799927,
609
+ "eval_precision": 0.5547209922405293,
610
+ "eval_recall": 0.49795298092711887,
611
+ "eval_runtime": 0.5785,
612
+ "eval_samples_per_second": 627.463,
613
+ "eval_steps_per_second": 79.513,
614
+ "step": 15523
615
+ },
616
+ {
617
+ "epoch": 19.28,
618
+ "learning_rate": 4.8702068669887026e-05,
619
+ "loss": 0.2524,
620
+ "step": 15750
621
+ },
622
+ {
623
+ "epoch": 19.58,
624
+ "learning_rate": 4.8615795291555885e-05,
625
+ "loss": 0.2412,
626
+ "step": 16000
627
+ },
628
+ {
629
+ "epoch": 19.89,
630
+ "learning_rate": 4.852682759383042e-05,
631
+ "loss": 0.3177,
632
+ "step": 16250
633
+ },
634
+ {
635
+ "epoch": 20.0,
636
+ "eval_accuracy": 0.6914600550964187,
637
+ "eval_f1": 0.4887299967298263,
638
+ "eval_loss": 1.9973198175430298,
639
+ "eval_precision": 0.6312488850962071,
640
+ "eval_recall": 0.4400535390991056,
641
+ "eval_runtime": 0.5767,
642
+ "eval_samples_per_second": 629.418,
643
+ "eval_steps_per_second": 79.761,
644
+ "step": 16340
645
+ },
646
+ {
647
+ "epoch": 20.2,
648
+ "learning_rate": 4.84355476675853e-05,
649
+ "loss": 0.2544,
650
+ "step": 16500
651
+ },
652
+ {
653
+ "epoch": 20.5,
654
+ "learning_rate": 4.8341232761821675e-05,
655
+ "loss": 0.2752,
656
+ "step": 16750
657
+ },
658
+ {
659
+ "epoch": 20.81,
660
+ "learning_rate": 4.824425486142494e-05,
661
+ "loss": 0.3368,
662
+ "step": 17000
663
+ },
664
+ {
665
+ "epoch": 21.0,
666
+ "eval_accuracy": 0.6831955922865014,
667
+ "eval_f1": 0.45711959899817944,
668
+ "eval_loss": 1.6511964797973633,
669
+ "eval_precision": 0.5667705375640166,
670
+ "eval_recall": 0.4405123543670341,
671
+ "eval_runtime": 0.5771,
672
+ "eval_samples_per_second": 629.043,
673
+ "eval_steps_per_second": 79.713,
674
+ "step": 17157
675
+ },
676
+ {
677
+ "epoch": 21.11,
678
+ "learning_rate": 4.81446250305763e-05,
679
+ "loss": 0.2763,
680
+ "step": 17250
681
+ },
682
+ {
683
+ "epoch": 21.42,
684
+ "learning_rate": 4.804235463601491e-05,
685
+ "loss": 0.2232,
686
+ "step": 17500
687
+ },
688
+ {
689
+ "epoch": 21.73,
690
+ "learning_rate": 4.793745534574111e-05,
691
+ "loss": 0.2307,
692
+ "step": 17750
693
+ },
694
+ {
695
+ "epoch": 22.0,
696
+ "eval_accuracy": 0.696969696969697,
697
+ "eval_f1": 0.4706521650995364,
698
+ "eval_loss": 1.9451159238815308,
699
+ "eval_precision": 0.6334310394975756,
700
+ "eval_recall": 0.4397407703134304,
701
+ "eval_runtime": 0.5762,
702
+ "eval_samples_per_second": 630.015,
703
+ "eval_steps_per_second": 79.837,
704
+ "step": 17974
705
+ },
706
+ {
707
+ "epoch": 22.03,
708
+ "learning_rate": 4.782993912768523e-05,
709
+ "loss": 0.2376,
710
+ "step": 18000
711
+ },
712
+ {
713
+ "epoch": 22.34,
714
+ "learning_rate": 4.7719818248342104e-05,
715
+ "loss": 0.2096,
716
+ "step": 18250
717
+ },
718
+ {
719
+ "epoch": 22.64,
720
+ "learning_rate": 4.760756126979301e-05,
721
+ "loss": 0.225,
722
+ "step": 18500
723
+ },
724
+ {
725
+ "epoch": 22.95,
726
+ "learning_rate": 4.7492279345533425e-05,
727
+ "loss": 0.259,
728
+ "step": 18750
729
+ },
730
+ {
731
+ "epoch": 23.0,
732
+ "eval_accuracy": 0.6887052341597796,
733
+ "eval_f1": 0.4849675763556479,
734
+ "eval_loss": 2.131509304046631,
735
+ "eval_precision": 0.6931105898377018,
736
+ "eval_recall": 0.424335591884853,
737
+ "eval_runtime": 0.5766,
738
+ "eval_samples_per_second": 629.576,
739
+ "eval_steps_per_second": 79.781,
740
+ "step": 18791
741
+ },
742
+ {
743
+ "epoch": 23.26,
744
+ "learning_rate": 4.737443128349549e-05,
745
+ "loss": 0.2214,
746
+ "step": 19000
747
+ },
748
+ {
749
+ "epoch": 23.56,
750
+ "learning_rate": 4.7254030528931215e-05,
751
+ "loss": 0.2338,
752
+ "step": 19250
753
+ },
754
+ {
755
+ "epoch": 23.87,
756
+ "learning_rate": 4.7131090818328586e-05,
757
+ "loss": 0.2387,
758
+ "step": 19500
759
+ },
760
+ {
761
+ "epoch": 24.0,
762
+ "eval_accuracy": 0.7024793388429752,
763
+ "eval_f1": 0.5396161727004175,
764
+ "eval_loss": 1.6646941900253296,
765
+ "eval_precision": 0.5457638718508284,
766
+ "eval_recall": 0.5445803409781244,
767
+ "eval_runtime": 0.577,
768
+ "eval_samples_per_second": 629.121,
769
+ "eval_steps_per_second": 79.723,
770
+ "step": 19608
771
+ },
772
+ {
773
+ "epoch": 24.17,
774
+ "learning_rate": 4.700562617784434e-05,
775
+ "loss": 0.2034,
776
+ "step": 19750
777
+ },
778
+ {
779
+ "epoch": 24.48,
780
+ "learning_rate": 4.687765092170378e-05,
781
+ "loss": 0.1752,
782
+ "step": 20000
783
+ },
784
+ {
785
+ "epoch": 24.79,
786
+ "learning_rate": 4.6747179650567614e-05,
787
+ "loss": 0.2089,
788
+ "step": 20250
789
+ },
790
+ {
791
+ "epoch": 25.0,
792
+ "eval_accuracy": 0.6721763085399449,
793
+ "eval_f1": 0.4391025799309439,
794
+ "eval_loss": 2.24601674079895,
795
+ "eval_precision": 0.4972889051140895,
796
+ "eval_recall": 0.428331664230679,
797
+ "eval_runtime": 0.5771,
798
+ "eval_samples_per_second": 629.0,
799
+ "eval_steps_per_second": 79.708,
800
+ "step": 20425
801
+ },
802
+ {
803
+ "epoch": 25.09,
804
+ "learning_rate": 4.661422724986621e-05,
805
+ "loss": 0.2451,
806
+ "step": 20500
807
+ },
808
+ {
809
+ "epoch": 25.4,
810
+ "learning_rate": 4.647880888810129e-05,
811
+ "loss": 0.1822,
812
+ "step": 20750
813
+ },
814
+ {
815
+ "epoch": 25.7,
816
+ "learning_rate": 4.634094001511539e-05,
817
+ "loss": 0.2117,
818
+ "step": 21000
819
+ },
820
+ {
821
+ "epoch": 26.0,
822
+ "eval_accuracy": 0.6721763085399449,
823
+ "eval_f1": 0.5210259674763756,
824
+ "eval_loss": 2.145235300064087,
825
+ "eval_precision": 0.5857309337602864,
826
+ "eval_recall": 0.48917402725900266,
827
+ "eval_runtime": 0.5764,
828
+ "eval_samples_per_second": 629.763,
829
+ "eval_steps_per_second": 79.805,
830
+ "step": 21242
831
+ },
832
+ {
833
+ "epoch": 26.01,
834
+ "learning_rate": 4.620063636032916e-05,
835
+ "loss": 0.2111,
836
+ "step": 21250
837
+ },
838
+ {
839
+ "epoch": 26.32,
840
+ "learning_rate": 4.605791393094685e-05,
841
+ "loss": 0.1778,
842
+ "step": 21500
843
+ },
844
+ {
845
+ "epoch": 26.62,
846
+ "learning_rate": 4.591278901012997e-05,
847
+ "loss": 0.2428,
848
+ "step": 21750
849
+ },
850
+ {
851
+ "epoch": 26.93,
852
+ "learning_rate": 4.576587292912579e-05,
853
+ "loss": 0.2081,
854
+ "step": 22000
855
+ },
856
+ {
857
+ "epoch": 27.0,
858
+ "eval_accuracy": 0.6749311294765841,
859
+ "eval_f1": 0.43598115798812087,
860
+ "eval_loss": 2.0688838958740234,
861
+ "eval_precision": 0.470814340268371,
862
+ "eval_recall": 0.4225478100170218,
863
+ "eval_runtime": 0.5781,
864
+ "eval_samples_per_second": 627.962,
865
+ "eval_steps_per_second": 79.576,
866
+ "step": 22059
867
+ },
868
+ {
869
+ "epoch": 27.23,
870
+ "learning_rate": 4.561600241196927e-05,
871
+ "loss": 0.1615,
872
+ "step": 22250
873
+ },
874
+ {
875
+ "epoch": 27.54,
876
+ "learning_rate": 4.546377982093819e-05,
877
+ "loss": 0.2104,
878
+ "step": 22500
879
+ },
880
+ {
881
+ "epoch": 27.85,
882
+ "learning_rate": 4.530922252306483e-05,
883
+ "loss": 0.1858,
884
+ "step": 22750
885
+ },
886
+ {
887
+ "epoch": 28.0,
888
+ "eval_accuracy": 0.6584022038567493,
889
+ "eval_f1": 0.44101738523719697,
890
+ "eval_loss": 2.2427115440368652,
891
+ "eval_precision": 0.5339727994900408,
892
+ "eval_recall": 0.41865492881502736,
893
+ "eval_runtime": 0.5762,
894
+ "eval_samples_per_second": 630.011,
895
+ "eval_steps_per_second": 79.836,
896
+ "step": 22876
897
+ },
898
+ {
899
+ "epoch": 28.15,
900
+ "learning_rate": 4.5152348151747534e-05,
901
+ "loss": 0.1984,
902
+ "step": 23000
903
+ },
904
+ {
905
+ "epoch": 28.46,
906
+ "learning_rate": 4.499317460473887e-05,
907
+ "loss": 0.2388,
908
+ "step": 23250
909
+ },
910
+ {
911
+ "epoch": 28.76,
912
+ "learning_rate": 4.483172004210372e-05,
913
+ "loss": 0.1924,
914
+ "step": 23500
915
+ },
916
+ {
917
+ "epoch": 29.0,
918
+ "eval_accuracy": 0.6831955922865014,
919
+ "eval_f1": 0.4384895896523804,
920
+ "eval_loss": 2.0502805709838867,
921
+ "eval_precision": 0.6932407169452727,
922
+ "eval_recall": 0.39043344791497503,
923
+ "eval_runtime": 0.5764,
924
+ "eval_samples_per_second": 629.797,
925
+ "eval_steps_per_second": 79.809,
926
+ "step": 23693
927
+ },
928
+ {
929
+ "epoch": 29.07,
930
+ "learning_rate": 4.46680028841474e-05,
931
+ "loss": 0.1952,
932
+ "step": 23750
933
+ },
934
+ {
935
+ "epoch": 29.38,
936
+ "learning_rate": 4.450204180931408e-05,
937
+ "loss": 0.1551,
938
+ "step": 24000
939
+ },
940
+ {
941
+ "epoch": 29.68,
942
+ "learning_rate": 4.433453290310271e-05,
943
+ "loss": 0.1758,
944
+ "step": 24250
945
+ },
946
+ {
947
+ "epoch": 29.99,
948
+ "learning_rate": 4.4164835737020996e-05,
949
+ "loss": 0.151,
950
+ "step": 24500
951
+ },
952
+ {
953
+ "epoch": 30.0,
954
+ "eval_accuracy": 0.6804407713498623,
955
+ "eval_f1": 0.4790753635178762,
956
+ "eval_loss": 2.1431803703308105,
957
+ "eval_precision": 0.5843463803065593,
958
+ "eval_recall": 0.43816301719626843,
959
+ "eval_runtime": 0.5767,
960
+ "eval_samples_per_second": 629.492,
961
+ "eval_steps_per_second": 79.77,
962
+ "step": 24510
963
+ },
964
+ {
965
+ "epoch": 30.29,
966
+ "learning_rate": 4.399227494450688e-05,
967
+ "loss": 0.1475,
968
+ "step": 24750
969
+ },
970
+ {
971
+ "epoch": 30.6,
972
+ "learning_rate": 4.381754732872496e-05,
973
+ "loss": 0.1696,
974
+ "step": 25000
975
+ },
976
+ {
977
+ "epoch": 30.91,
978
+ "learning_rate": 4.3640672824299424e-05,
979
+ "loss": 0.1978,
980
+ "step": 25250
981
+ },
982
+ {
983
+ "epoch": 31.0,
984
+ "eval_accuracy": 0.6859504132231405,
985
+ "eval_f1": 0.42594599441215986,
986
+ "eval_loss": 2.2966179847717285,
987
+ "eval_precision": 0.5307515065579581,
988
+ "eval_recall": 0.41329913928436096,
989
+ "eval_runtime": 0.5775,
990
+ "eval_samples_per_second": 628.614,
991
+ "eval_steps_per_second": 79.659,
992
+ "step": 25327
993
+ },
994
+ {
995
+ "epoch": 31.21,
996
+ "learning_rate": 4.346167161079236e-05,
997
+ "loss": 0.167,
998
+ "step": 25500
999
+ },
1000
+ {
1001
+ "epoch": 31.52,
1002
+ "learning_rate": 4.328056411040151e-05,
1003
+ "loss": 0.1383,
1004
+ "step": 25750
1005
+ },
1006
+ {
1007
+ "epoch": 31.82,
1008
+ "learning_rate": 4.309737098563029e-05,
1009
+ "loss": 0.1922,
1010
+ "step": 26000
1011
+ },
1012
+ {
1013
+ "epoch": 32.0,
1014
+ "eval_accuracy": 0.6831955922865014,
1015
+ "eval_f1": 0.485788264860417,
1016
+ "eval_loss": 2.0524113178253174,
1017
+ "eval_precision": 0.5313209433300384,
1018
+ "eval_recall": 0.4710779846863592,
1019
+ "eval_runtime": 0.577,
1020
+ "eval_samples_per_second": 629.091,
1021
+ "eval_steps_per_second": 79.719,
1022
+ "step": 26144
1023
+ },
1024
+ {
1025
+ "epoch": 32.13,
1026
+ "learning_rate": 4.29121131369304e-05,
1027
+ "loss": 0.199,
1028
+ "step": 26250
1029
+ },
1030
+ {
1031
+ "epoch": 32.44,
1032
+ "learning_rate": 4.272481170031731e-05,
1033
+ "loss": 0.1964,
1034
+ "step": 26500
1035
+ },
1036
+ {
1037
+ "epoch": 32.74,
1038
+ "learning_rate": 4.253548804495887e-05,
1039
+ "loss": 0.1876,
1040
+ "step": 26750
1041
+ },
1042
+ {
1043
+ "epoch": 33.0,
1044
+ "eval_accuracy": 0.6859504132231405,
1045
+ "eval_f1": 0.45277059579119083,
1046
+ "eval_loss": 2.3500843048095703,
1047
+ "eval_precision": 0.6191480424964332,
1048
+ "eval_recall": 0.4147965198457809,
1049
+ "eval_runtime": 0.5777,
1050
+ "eval_samples_per_second": 628.4,
1051
+ "eval_steps_per_second": 79.632,
1052
+ "step": 26961
1053
+ },
1054
+ {
1055
+ "epoch": 33.05,
1056
+ "learning_rate": 4.234416377073727e-05,
1057
+ "loss": 0.1475,
1058
+ "step": 27000
1059
+ },
1060
+ {
1061
+ "epoch": 33.35,
1062
+ "learning_rate": 4.215086070578477e-05,
1063
+ "loss": 0.143,
1064
+ "step": 27250
1065
+ },
1066
+ {
1067
+ "epoch": 33.66,
1068
+ "learning_rate": 4.195560090399327e-05,
1069
+ "loss": 0.2021,
1070
+ "step": 27500
1071
+ },
1072
+ {
1073
+ "epoch": 33.97,
1074
+ "learning_rate": 4.175840664249825e-05,
1075
+ "loss": 0.1649,
1076
+ "step": 27750
1077
+ },
1078
+ {
1079
+ "epoch": 34.0,
1080
+ "eval_accuracy": 0.6584022038567493,
1081
+ "eval_f1": 0.4327576346908582,
1082
+ "eval_loss": 2.3630762100219727,
1083
+ "eval_precision": 0.46295462482075056,
1084
+ "eval_recall": 0.4282663662281889,
1085
+ "eval_runtime": 0.5765,
1086
+ "eval_samples_per_second": 629.661,
1087
+ "eval_steps_per_second": 79.792,
1088
+ "step": 27778
1089
+ },
1090
+ {
1091
+ "epoch": 34.27,
1092
+ "learning_rate": 4.1559300419137124e-05,
1093
+ "loss": 0.1333,
1094
+ "step": 28000
1095
+ },
1096
+ {
1097
+ "epoch": 34.58,
1098
+ "learning_rate": 4.135830494988252e-05,
1099
+ "loss": 0.1386,
1100
+ "step": 28250
1101
+ },
1102
+ {
1103
+ "epoch": 34.88,
1104
+ "learning_rate": 4.1155443166250585e-05,
1105
+ "loss": 0.2109,
1106
+ "step": 28500
1107
+ },
1108
+ {
1109
+ "epoch": 35.0,
1110
+ "eval_accuracy": 0.6749311294765841,
1111
+ "eval_f1": 0.4902477696595343,
1112
+ "eval_loss": 2.1397335529327393,
1113
+ "eval_precision": 0.5198444762520393,
1114
+ "eval_recall": 0.49638105463228616,
1115
+ "eval_runtime": 0.578,
1116
+ "eval_samples_per_second": 628.013,
1117
+ "eval_steps_per_second": 79.583,
1118
+ "step": 28595
1119
+ },
1120
+ {
1121
+ "epoch": 35.19,
1122
+ "learning_rate": 4.095073821268473e-05,
1123
+ "loss": 0.1839,
1124
+ "step": 28750
1125
+ },
1126
+ {
1127
+ "epoch": 35.5,
1128
+ "learning_rate": 4.0744213443915125e-05,
1129
+ "loss": 0.1558,
1130
+ "step": 29000
1131
+ },
1132
+ {
1133
+ "epoch": 35.8,
1134
+ "learning_rate": 4.0536729253084094e-05,
1135
+ "loss": 0.127,
1136
+ "step": 29250
1137
+ },
1138
+ {
1139
+ "epoch": 36.0,
1140
+ "eval_accuracy": 0.6694214876033058,
1141
+ "eval_f1": 0.4276520300779697,
1142
+ "eval_loss": 2.290356159210205,
1143
+ "eval_precision": 0.5064721567763604,
1144
+ "eval_recall": 0.40090887150616217,
1145
+ "eval_runtime": 0.5778,
1146
+ "eval_samples_per_second": 628.209,
1147
+ "eval_steps_per_second": 79.608,
1148
+ "step": 29412
1149
+ },
1150
+ {
1151
+ "epoch": 36.11,
1152
+ "learning_rate": 4.032748663336774e-05,
1153
+ "loss": 0.2138,
1154
+ "step": 29500
1155
+ },
1156
+ {
1157
+ "epoch": 36.41,
1158
+ "learning_rate": 4.011565850261534e-05,
1159
+ "loss": 0.1587,
1160
+ "step": 29750
1161
+ },
1162
+ {
1163
+ "epoch": 36.72,
1164
+ "learning_rate": 3.990210583066674e-05,
1165
+ "loss": 0.1653,
1166
+ "step": 30000
1167
+ },
1168
+ {
1169
+ "epoch": 37.0,
1170
+ "eval_accuracy": 0.6831955922865014,
1171
+ "eval_f1": 0.5118761031802896,
1172
+ "eval_loss": 2.097479820251465,
1173
+ "eval_precision": 0.558650914911419,
1174
+ "eval_recall": 0.4861388167114768,
1175
+ "eval_runtime": 0.5771,
1176
+ "eval_samples_per_second": 629.03,
1177
+ "eval_steps_per_second": 79.712,
1178
+ "step": 30229
1179
+ },
1180
+ {
1181
+ "epoch": 37.03,
1182
+ "learning_rate": 3.968685298168575e-05,
1183
+ "loss": 0.1729,
1184
+ "step": 30250
1185
+ },
1186
+ {
1187
+ "epoch": 37.33,
1188
+ "learning_rate": 3.946992451380895e-05,
1189
+ "loss": 0.1092,
1190
+ "step": 30500
1191
+ },
1192
+ {
1193
+ "epoch": 37.64,
1194
+ "learning_rate": 3.9251345176343774e-05,
1195
+ "loss": 0.1322,
1196
+ "step": 30750
1197
+ },
1198
+ {
1199
+ "epoch": 37.94,
1200
+ "learning_rate": 3.9031139906944916e-05,
1201
+ "loss": 0.1626,
1202
+ "step": 31000
1203
+ },
1204
+ {
1205
+ "epoch": 38.0,
1206
+ "eval_accuracy": 0.6666666666666666,
1207
+ "eval_f1": 0.4688775335224489,
1208
+ "eval_loss": 2.4606480598449707,
1209
+ "eval_precision": 0.4988188909516061,
1210
+ "eval_recall": 0.45583700987026105,
1211
+ "eval_runtime": 0.5758,
1212
+ "eval_samples_per_second": 630.398,
1213
+ "eval_steps_per_second": 79.885,
1214
+ "step": 31046
1215
+ },
1216
+ {
1217
+ "epoch": 38.25,
1218
+ "learning_rate": 3.880933382876915e-05,
1219
+ "loss": 0.1369,
1220
+ "step": 31250
1221
+ },
1222
+ {
1223
+ "epoch": 38.56,
1224
+ "learning_rate": 3.8585952247609145e-05,
1225
+ "loss": 0.1002,
1226
+ "step": 31500
1227
+ },
1228
+ {
1229
+ "epoch": 38.86,
1230
+ "learning_rate": 3.836102064900617e-05,
1231
+ "loss": 0.1454,
1232
+ "step": 31750
1233
+ },
1234
+ {
1235
+ "epoch": 39.0,
1236
+ "eval_accuracy": 0.6914600550964187,
1237
+ "eval_f1": 0.5079144154287897,
1238
+ "eval_loss": 2.172344446182251,
1239
+ "eval_precision": 0.5805565750272889,
1240
+ "eval_recall": 0.4738757992144692,
1241
+ "eval_runtime": 0.5781,
1242
+ "eval_samples_per_second": 627.915,
1243
+ "eval_steps_per_second": 79.571,
1244
+ "step": 31863
1245
+ },
1246
+ {
1247
+ "epoch": 39.17,
1248
+ "learning_rate": 3.8134564695342595e-05,
1249
+ "loss": 0.158,
1250
+ "step": 32000
1251
+ },
1252
+ {
1253
+ "epoch": 39.47,
1254
+ "learning_rate": 3.790661022291403e-05,
1255
+ "loss": 0.1078,
1256
+ "step": 32250
1257
+ },
1258
+ {
1259
+ "epoch": 39.78,
1260
+ "learning_rate": 3.767718323898163e-05,
1261
+ "loss": 0.1206,
1262
+ "step": 32500
1263
+ },
1264
+ {
1265
+ "epoch": 40.0,
1266
+ "eval_accuracy": 0.6942148760330579,
1267
+ "eval_f1": 0.48536884689746623,
1268
+ "eval_loss": 2.1714024543762207,
1269
+ "eval_precision": 0.5719430108062781,
1270
+ "eval_recall": 0.4474171388272374,
1271
+ "eval_runtime": 0.5772,
1272
+ "eval_samples_per_second": 628.944,
1273
+ "eval_steps_per_second": 79.701,
1274
+ "step": 32680
1275
+ },
1276
+ {
1277
+ "epoch": 40.09,
1278
+ "learning_rate": 3.744630991880502e-05,
1279
+ "loss": 0.1158,
1280
+ "step": 32750
1281
+ },
1282
+ {
1283
+ "epoch": 40.39,
1284
+ "learning_rate": 3.7214016602655875e-05,
1285
+ "loss": 0.115,
1286
+ "step": 33000
1287
+ },
1288
+ {
1289
+ "epoch": 40.7,
1290
+ "learning_rate": 3.698032979281283e-05,
1291
+ "loss": 0.1092,
1292
+ "step": 33250
1293
+ },
1294
+ {
1295
+ "epoch": 41.0,
1296
+ "eval_accuracy": 0.6914600550964187,
1297
+ "eval_f1": 0.4980899463610581,
1298
+ "eval_loss": 2.2357773780822754,
1299
+ "eval_precision": 0.5844052283360602,
1300
+ "eval_recall": 0.46764869298736295,
1301
+ "eval_runtime": 0.5777,
1302
+ "eval_samples_per_second": 628.373,
1303
+ "eval_steps_per_second": 79.629,
1304
+ "step": 33497
1305
+ },
1306
+ {
1307
+ "epoch": 41.0,
1308
+ "learning_rate": 3.6745276150537816e-05,
1309
+ "loss": 0.1262,
1310
+ "step": 33500
1311
+ },
1312
+ {
1313
+ "epoch": 41.31,
1314
+ "learning_rate": 3.650983070128024e-05,
1315
+ "loss": 0.1034,
1316
+ "step": 33750
1317
+ },
1318
+ {
1319
+ "epoch": 41.62,
1320
+ "learning_rate": 3.6272129196890194e-05,
1321
+ "loss": 0.0885,
1322
+ "step": 34000
1323
+ },
1324
+ {
1325
+ "epoch": 41.92,
1326
+ "learning_rate": 3.603410013431832e-05,
1327
+ "loss": 0.1156,
1328
+ "step": 34250
1329
+ },
1330
+ {
1331
+ "epoch": 42.0,
1332
+ "eval_accuracy": 0.6584022038567493,
1333
+ "eval_f1": 0.4280488256791789,
1334
+ "eval_loss": 2.558166265487671,
1335
+ "eval_precision": 0.45057717568928324,
1336
+ "eval_recall": 0.416898799749785,
1337
+ "eval_runtime": 0.5768,
1338
+ "eval_samples_per_second": 629.389,
1339
+ "eval_steps_per_second": 79.757,
1340
+ "step": 34314
1341
+ },
1342
+ {
1343
+ "epoch": 42.23,
1344
+ "learning_rate": 3.579385880846232e-05,
1345
+ "loss": 0.1169,
1346
+ "step": 34500
1347
+ },
1348
+ {
1349
+ "epoch": 42.53,
1350
+ "learning_rate": 3.5552386014312834e-05,
1351
+ "loss": 0.099,
1352
+ "step": 34750
1353
+ },
1354
+ {
1355
+ "epoch": 42.84,
1356
+ "learning_rate": 3.530970930143252e-05,
1357
+ "loss": 0.093,
1358
+ "step": 35000
1359
+ },
1360
+ {
1361
+ "epoch": 43.0,
1362
+ "eval_accuracy": 0.628099173553719,
1363
+ "eval_f1": 0.4313278010079715,
1364
+ "eval_loss": 2.627509355545044,
1365
+ "eval_precision": 0.5279577964214083,
1366
+ "eval_recall": 0.40927325616365023,
1367
+ "eval_runtime": 0.576,
1368
+ "eval_samples_per_second": 630.259,
1369
+ "eval_steps_per_second": 79.868,
1370
+ "step": 35131
1371
+ },
1372
+ {
1373
+ "epoch": 43.15,
1374
+ "learning_rate": 3.5065856356738794e-05,
1375
+ "loss": 0.1584,
1376
+ "step": 35250
1377
+ },
1378
+ {
1379
+ "epoch": 43.45,
1380
+ "learning_rate": 3.482085500134504e-05,
1381
+ "loss": 0.0967,
1382
+ "step": 35500
1383
+ },
1384
+ {
1385
+ "epoch": 43.76,
1386
+ "learning_rate": 3.4574733187386486e-05,
1387
+ "loss": 0.1484,
1388
+ "step": 35750
1389
+ },
1390
+ {
1391
+ "epoch": 44.0,
1392
+ "eval_accuracy": 0.6694214876033058,
1393
+ "eval_f1": 0.43577912693695875,
1394
+ "eval_loss": 2.5017268657684326,
1395
+ "eval_precision": 0.5655394346579834,
1396
+ "eval_recall": 0.3971485787064605,
1397
+ "eval_runtime": 0.5768,
1398
+ "eval_samples_per_second": 629.305,
1399
+ "eval_steps_per_second": 79.747,
1400
+ "step": 35948
1401
+ },
1402
+ {
1403
+ "epoch": 44.06,
1404
+ "learning_rate": 3.432751899483116e-05,
1405
+ "loss": 0.1379,
1406
+ "step": 36000
1407
+ },
1408
+ {
1409
+ "epoch": 44.37,
1410
+ "learning_rate": 3.4079240628276246e-05,
1411
+ "loss": 0.0877,
1412
+ "step": 36250
1413
+ },
1414
+ {
1415
+ "epoch": 44.68,
1416
+ "learning_rate": 3.382992641373025e-05,
1417
+ "loss": 0.1466,
1418
+ "step": 36500
1419
+ },
1420
+ {
1421
+ "epoch": 44.98,
1422
+ "learning_rate": 3.357960479538127e-05,
1423
+ "loss": 0.1411,
1424
+ "step": 36750
1425
+ },
1426
+ {
1427
+ "epoch": 45.0,
1428
+ "eval_accuracy": 0.6914600550964187,
1429
+ "eval_f1": 0.5234198940288918,
1430
+ "eval_loss": 2.3302204608917236,
1431
+ "eval_precision": 0.5942265966479041,
1432
+ "eval_recall": 0.5061625600726586,
1433
+ "eval_runtime": 0.5778,
1434
+ "eval_samples_per_second": 628.231,
1435
+ "eval_steps_per_second": 79.611,
1436
+ "step": 36765
1437
+ },
1438
+ {
1439
+ "epoch": 45.29,
1440
+ "learning_rate": 3.332830433235184e-05,
1441
+ "loss": 0.0849,
1442
+ "step": 37000
1443
+ },
1444
+ {
1445
+ "epoch": 45.59,
1446
+ "learning_rate": 3.307605369544058e-05,
1447
+ "loss": 0.126,
1448
+ "step": 37250
1449
+ },
1450
+ {
1451
+ "epoch": 45.9,
1452
+ "learning_rate": 3.28228816638512e-05,
1453
+ "loss": 0.1003,
1454
+ "step": 37500
1455
+ },
1456
+ {
1457
+ "epoch": 46.0,
1458
+ "eval_accuracy": 0.696969696969697,
1459
+ "eval_f1": 0.5745871087976351,
1460
+ "eval_loss": 2.3890013694763184,
1461
+ "eval_precision": 0.6828113242342491,
1462
+ "eval_recall": 0.5128949833691213,
1463
+ "eval_runtime": 0.577,
1464
+ "eval_samples_per_second": 629.101,
1465
+ "eval_steps_per_second": 79.721,
1466
+ "step": 37582
1467
+ },
1468
+ {
1469
+ "epoch": 46.21,
1470
+ "learning_rate": 3.256881712190906e-05,
1471
+ "loss": 0.11,
1472
+ "step": 37750
1473
+ },
1474
+ {
1475
+ "epoch": 46.51,
1476
+ "learning_rate": 3.231388905576575e-05,
1477
+ "loss": 0.1225,
1478
+ "step": 38000
1479
+ },
1480
+ {
1481
+ "epoch": 46.82,
1482
+ "learning_rate": 3.205915122367602e-05,
1483
+ "loss": 0.1245,
1484
+ "step": 38250
1485
+ },
1486
+ {
1487
+ "epoch": 47.0,
1488
+ "eval_accuracy": 0.6804407713498623,
1489
+ "eval_f1": 0.5184040661055641,
1490
+ "eval_loss": 2.339553117752075,
1491
+ "eval_precision": 0.6566468598124048,
1492
+ "eval_recall": 0.48311657554268395,
1493
+ "eval_runtime": 0.5772,
1494
+ "eval_samples_per_second": 628.951,
1495
+ "eval_steps_per_second": 79.702,
1496
+ "step": 38399
1497
+ },
1498
+ {
1499
+ "epoch": 47.12,
1500
+ "learning_rate": 3.180258662113338e-05,
1501
+ "loss": 0.1008,
1502
+ "step": 38500
1503
+ },
1504
+ {
1505
+ "epoch": 47.43,
1506
+ "learning_rate": 3.154524591341034e-05,
1507
+ "loss": 0.121,
1508
+ "step": 38750
1509
+ },
1510
+ {
1511
+ "epoch": 47.74,
1512
+ "learning_rate": 3.128715846043534e-05,
1513
+ "loss": 0.0998,
1514
+ "step": 39000
1515
+ },
1516
+ {
1517
+ "epoch": 48.0,
1518
+ "eval_accuracy": 0.696969696969697,
1519
+ "eval_f1": 0.5459744525861538,
1520
+ "eval_loss": 2.4267241954803467,
1521
+ "eval_precision": 0.6807294155120241,
1522
+ "eval_recall": 0.4855936141034663,
1523
+ "eval_runtime": 0.5758,
1524
+ "eval_samples_per_second": 630.432,
1525
+ "eval_steps_per_second": 79.889,
1526
+ "step": 39216
1527
+ },
1528
+ {
1529
+ "epoch": 48.04,
1530
+ "learning_rate": 3.102835370733277e-05,
1531
+ "loss": 0.0962,
1532
+ "step": 39250
1533
+ },
1534
+ {
1535
+ "epoch": 48.35,
1536
+ "learning_rate": 3.07688611810636e-05,
1537
+ "loss": 0.0778,
1538
+ "step": 39500
1539
+ },
1540
+ {
1541
+ "epoch": 48.65,
1542
+ "learning_rate": 3.0508710487056635e-05,
1543
+ "loss": 0.1145,
1544
+ "step": 39750
1545
+ },
1546
+ {
1547
+ "epoch": 48.96,
1548
+ "learning_rate": 3.0247931305830845e-05,
1549
+ "loss": 0.1048,
1550
+ "step": 40000
1551
+ },
1552
+ {
1553
+ "epoch": 49.0,
1554
+ "eval_accuracy": 0.6749311294765841,
1555
+ "eval_f1": 0.536908729765805,
1556
+ "eval_loss": 2.3678715229034424,
1557
+ "eval_precision": 0.6530412239759585,
1558
+ "eval_recall": 0.48379632135789763,
1559
+ "eval_runtime": 0.578,
1560
+ "eval_samples_per_second": 628.037,
1561
+ "eval_steps_per_second": 79.586,
1562
+ "step": 40033
1563
+ },
1564
+ {
1565
+ "epoch": 49.27,
1566
+ "learning_rate": 2.998655338960914e-05,
1567
+ "loss": 0.0751,
1568
+ "step": 40250
1569
+ },
1570
+ {
1571
+ "epoch": 49.57,
1572
+ "learning_rate": 2.9724606558923933e-05,
1573
+ "loss": 0.107,
1574
+ "step": 40500
1575
+ },
1576
+ {
1577
+ "epoch": 49.88,
1578
+ "learning_rate": 2.9462120699214922e-05,
1579
+ "loss": 0.0605,
1580
+ "step": 40750
1581
+ },
1582
+ {
1583
+ "epoch": 50.0,
1584
+ "eval_accuracy": 0.6859504132231405,
1585
+ "eval_f1": 0.5723258179466014,
1586
+ "eval_loss": 2.3588879108428955,
1587
+ "eval_precision": 0.6557361185869716,
1588
+ "eval_recall": 0.5364202334337802,
1589
+ "eval_runtime": 0.5756,
1590
+ "eval_samples_per_second": 630.622,
1591
+ "eval_steps_per_second": 79.913,
1592
+ "step": 40850
1593
+ },
1594
+ {
1595
+ "epoch": 50.18,
1596
+ "learning_rate": 2.9199125757419482e-05,
1597
+ "loss": 0.0718,
1598
+ "step": 41000
1599
+ },
1600
+ {
1601
+ "epoch": 50.49,
1602
+ "learning_rate": 2.8935651738556013e-05,
1603
+ "loss": 0.0661,
1604
+ "step": 41250
1605
+ },
1606
+ {
1607
+ "epoch": 50.8,
1608
+ "learning_rate": 2.8671728702300687e-05,
1609
+ "loss": 0.101,
1610
+ "step": 41500
1611
+ },
1612
+ {
1613
+ "epoch": 51.0,
1614
+ "eval_accuracy": 0.7024793388429752,
1615
+ "eval_f1": 0.5646342747625367,
1616
+ "eval_loss": 2.261863946914673,
1617
+ "eval_precision": 0.6370711351914359,
1618
+ "eval_recall": 0.5380352030290453,
1619
+ "eval_runtime": 0.5759,
1620
+ "eval_samples_per_second": 630.369,
1621
+ "eval_steps_per_second": 79.881,
1622
+ "step": 41667
1623
+ },
1624
+ {
1625
+ "epoch": 51.1,
1626
+ "learning_rate": 2.840738675955793e-05,
1627
+ "loss": 0.1039,
1628
+ "step": 41750
1629
+ },
1630
+ {
1631
+ "epoch": 51.41,
1632
+ "learning_rate": 2.814371572615747e-05,
1633
+ "loss": 0.0769,
1634
+ "step": 42000
1635
+ },
1636
+ {
1637
+ "epoch": 51.71,
1638
+ "learning_rate": 2.787862786484319e-05,
1639
+ "loss": 0.0858,
1640
+ "step": 42250
1641
+ },
1642
+ {
1643
+ "epoch": 52.0,
1644
+ "eval_accuracy": 0.6859504132231405,
1645
+ "eval_f1": 0.5195773061626721,
1646
+ "eval_loss": 2.5068295001983643,
1647
+ "eval_precision": 0.6063341583445594,
1648
+ "eval_recall": 0.4876037550298634,
1649
+ "eval_runtime": 0.5762,
1650
+ "eval_samples_per_second": 629.983,
1651
+ "eval_steps_per_second": 79.833,
1652
+ "step": 42484
1653
+ },
1654
+ {
1655
+ "epoch": 52.02,
1656
+ "learning_rate": 2.761321158169134e-05,
1657
+ "loss": 0.0844,
1658
+ "step": 42500
1659
+ },
1660
+ {
1661
+ "epoch": 52.33,
1662
+ "learning_rate": 2.734749715797047e-05,
1663
+ "loss": 0.069,
1664
+ "step": 42750
1665
+ },
1666
+ {
1667
+ "epoch": 52.63,
1668
+ "learning_rate": 2.7081514908963913e-05,
1669
+ "loss": 0.0784,
1670
+ "step": 43000
1671
+ },
1672
+ {
1673
+ "epoch": 52.94,
1674
+ "learning_rate": 2.681529518051109e-05,
1675
+ "loss": 0.0795,
1676
+ "step": 43250
1677
+ },
1678
+ {
1679
+ "epoch": 53.0,
1680
+ "eval_accuracy": 0.6887052341597796,
1681
+ "eval_f1": 0.4958589687437866,
1682
+ "eval_loss": 2.4998273849487305,
1683
+ "eval_precision": 0.6786654073384627,
1684
+ "eval_recall": 0.45129423122033957,
1685
+ "eval_runtime": 0.5744,
1686
+ "eval_samples_per_second": 631.991,
1687
+ "eval_steps_per_second": 80.087,
1688
+ "step": 43301
1689
+ },
1690
+ {
1691
+ "epoch": 53.24,
1692
+ "learning_rate": 2.6548868345545402e-05,
1693
+ "loss": 0.072,
1694
+ "step": 43500
1695
+ },
1696
+ {
1697
+ "epoch": 53.55,
1698
+ "learning_rate": 2.628226480062897e-05,
1699
+ "loss": 0.0401,
1700
+ "step": 43750
1701
+ },
1702
+ {
1703
+ "epoch": 53.86,
1704
+ "learning_rate": 2.6015514962484717e-05,
1705
+ "loss": 0.0674,
1706
+ "step": 44000
1707
+ },
1708
+ {
1709
+ "epoch": 54.0,
1710
+ "eval_accuracy": 0.6639118457300276,
1711
+ "eval_f1": 0.5109508473836721,
1712
+ "eval_loss": 2.704511880874634,
1713
+ "eval_precision": 0.6101414588534028,
1714
+ "eval_recall": 0.47100482987182496,
1715
+ "eval_runtime": 0.5772,
1716
+ "eval_samples_per_second": 628.872,
1717
+ "eval_steps_per_second": 79.692,
1718
+ "step": 44118
1719
+ },
1720
+ {
1721
+ "epoch": 54.16,
1722
+ "learning_rate": 2.574971691776212e-05,
1723
+ "loss": 0.0601,
1724
+ "step": 44250
1725
+ },
1726
+ {
1727
+ "epoch": 54.47,
1728
+ "learning_rate": 2.5482766087609973e-05,
1729
+ "loss": 0.0502,
1730
+ "step": 44500
1731
+ },
1732
+ {
1733
+ "epoch": 54.77,
1734
+ "learning_rate": 2.5215760178811658e-05,
1735
+ "loss": 0.1174,
1736
+ "step": 44750
1737
+ },
1738
+ {
1739
+ "epoch": 55.0,
1740
+ "eval_accuracy": 0.6859504132231405,
1741
+ "eval_f1": 0.516165985819526,
1742
+ "eval_loss": 2.3755171298980713,
1743
+ "eval_precision": 0.5969723691945914,
1744
+ "eval_recall": 0.4715294694358734,
1745
+ "eval_runtime": 0.5774,
1746
+ "eval_samples_per_second": 628.713,
1747
+ "eval_steps_per_second": 79.672,
1748
+ "step": 44935
1749
+ },
1750
+ {
1751
+ "epoch": 55.08,
1752
+ "learning_rate": 2.4948729653995663e-05,
1753
+ "loss": 0.0826,
1754
+ "step": 45000
1755
+ },
1756
+ {
1757
+ "epoch": 55.39,
1758
+ "learning_rate": 2.4681704978598928e-05,
1759
+ "loss": 0.0422,
1760
+ "step": 45250
1761
+ },
1762
+ {
1763
+ "epoch": 55.69,
1764
+ "learning_rate": 2.4414716617390998e-05,
1765
+ "loss": 0.059,
1766
+ "step": 45500
1767
+ },
1768
+ {
1769
+ "epoch": 56.0,
1770
+ "learning_rate": 2.414779503099838e-05,
1771
+ "loss": 0.0886,
1772
+ "step": 45750
1773
+ },
1774
+ {
1775
+ "epoch": 56.0,
1776
+ "eval_accuracy": 0.6804407713498623,
1777
+ "eval_f1": 0.47301725201698097,
1778
+ "eval_loss": 2.507392406463623,
1779
+ "eval_precision": 0.4981856952904839,
1780
+ "eval_recall": 0.4597725584486668,
1781
+ "eval_runtime": 0.5761,
1782
+ "eval_samples_per_second": 630.088,
1783
+ "eval_steps_per_second": 79.846,
1784
+ "step": 45752
1785
+ },
1786
+ {
1787
+ "epoch": 56.3,
1788
+ "learning_rate": 2.388097067242925e-05,
1789
+ "loss": 0.0528,
1790
+ "step": 46000
1791
+ },
1792
+ {
1793
+ "epoch": 56.61,
1794
+ "learning_rate": 2.3614273983599068e-05,
1795
+ "loss": 0.0643,
1796
+ "step": 46250
1797
+ },
1798
+ {
1799
+ "epoch": 56.92,
1800
+ "learning_rate": 2.334773539185752e-05,
1801
+ "loss": 0.058,
1802
+ "step": 46500
1803
+ },
1804
+ {
1805
+ "epoch": 57.0,
1806
+ "eval_accuracy": 0.6804407713498623,
1807
+ "eval_f1": 0.4807142977366666,
1808
+ "eval_loss": 2.5247209072113037,
1809
+ "eval_precision": 0.5200869236583522,
1810
+ "eval_recall": 0.45992668729737696,
1811
+ "eval_runtime": 0.5758,
1812
+ "eval_samples_per_second": 630.387,
1813
+ "eval_steps_per_second": 79.884,
1814
+ "step": 46569
1815
+ },
1816
+ {
1817
+ "epoch": 57.22,
1818
+ "learning_rate": 2.308138530651701e-05,
1819
+ "loss": 0.0823,
1820
+ "step": 46750
1821
+ },
1822
+ {
1823
+ "epoch": 57.53,
1824
+ "learning_rate": 2.2815254115383325e-05,
1825
+ "loss": 0.0428,
1826
+ "step": 47000
1827
+ },
1828
+ {
1829
+ "epoch": 57.83,
1830
+ "learning_rate": 2.254937218128868e-05,
1831
+ "loss": 0.0558,
1832
+ "step": 47250
1833
+ },
1834
+ {
1835
+ "epoch": 58.0,
1836
+ "eval_accuracy": 0.6914600550964187,
1837
+ "eval_f1": 0.4594613590535599,
1838
+ "eval_loss": 2.47005033493042,
1839
+ "eval_precision": 0.6139090044619999,
1840
+ "eval_recall": 0.4228677965439049,
1841
+ "eval_runtime": 0.5759,
1842
+ "eval_samples_per_second": 630.356,
1843
+ "eval_steps_per_second": 79.88,
1844
+ "step": 47386
1845
+ },
1846
+ {
1847
+ "epoch": 58.14,
1848
+ "learning_rate": 2.2283769838627582e-05,
1849
+ "loss": 0.0931,
1850
+ "step": 47500
1851
+ },
1852
+ {
1853
+ "epoch": 58.45,
1854
+ "learning_rate": 2.2018477389896086e-05,
1855
+ "loss": 0.0508,
1856
+ "step": 47750
1857
+ },
1858
+ {
1859
+ "epoch": 58.75,
1860
+ "learning_rate": 2.1753525102234484e-05,
1861
+ "loss": 0.0452,
1862
+ "step": 48000
1863
+ },
1864
+ {
1865
+ "epoch": 59.0,
1866
+ "eval_accuracy": 0.6942148760330579,
1867
+ "eval_f1": 0.46439631667557835,
1868
+ "eval_loss": 2.541719675064087,
1869
+ "eval_precision": 0.5814331265699046,
1870
+ "eval_recall": 0.43331322139819684,
1871
+ "eval_runtime": 0.5754,
1872
+ "eval_samples_per_second": 630.853,
1873
+ "eval_steps_per_second": 79.943,
1874
+ "step": 48203
1875
+ },
1876
+ {
1877
+ "epoch": 59.06,
1878
+ "learning_rate": 2.1488943203974225e-05,
1879
+ "loss": 0.0487,
1880
+ "step": 48250
1881
+ },
1882
+ {
1883
+ "epoch": 59.36,
1884
+ "learning_rate": 2.122581776857167e-05,
1885
+ "loss": 0.0777,
1886
+ "step": 48500
1887
+ },
1888
+ {
1889
+ "epoch": 59.67,
1890
+ "learning_rate": 2.0963119490709552e-05,
1891
+ "loss": 0.0656,
1892
+ "step": 48750
1893
+ },
1894
+ {
1895
+ "epoch": 59.98,
1896
+ "learning_rate": 2.0699825885128935e-05,
1897
+ "loss": 0.0589,
1898
+ "step": 49000
1899
+ },
1900
+ {
1901
+ "epoch": 60.0,
1902
+ "eval_accuracy": 0.6721763085399449,
1903
+ "eval_f1": 0.4951052024505615,
1904
+ "eval_loss": 2.631096363067627,
1905
+ "eval_precision": 0.5573207158969782,
1906
+ "eval_recall": 0.4686150056839712,
1907
+ "eval_runtime": 0.5753,
1908
+ "eval_samples_per_second": 630.965,
1909
+ "eval_steps_per_second": 79.957,
1910
+ "step": 49020
1911
+ },
1912
+ {
1913
+ "epoch": 60.28,
1914
+ "learning_rate": 2.0437022885194815e-05,
1915
+ "loss": 0.0345,
1916
+ "step": 49250
1917
+ },
1918
+ {
1919
+ "epoch": 60.59,
1920
+ "learning_rate": 2.017474047402702e-05,
1921
+ "loss": 0.0656,
1922
+ "step": 49500
1923
+ },
1924
+ {
1925
+ "epoch": 60.89,
1926
+ "learning_rate": 1.991300857535151e-05,
1927
+ "loss": 0.0993,
1928
+ "step": 49750
1929
+ },
1930
+ {
1931
+ "epoch": 61.0,
1932
+ "eval_accuracy": 0.6666666666666666,
1933
+ "eval_f1": 0.4534585450214669,
1934
+ "eval_loss": 2.720975160598755,
1935
+ "eval_precision": 0.5187203506875638,
1936
+ "eval_recall": 0.423817756545589,
1937
+ "eval_runtime": 0.5751,
1938
+ "eval_samples_per_second": 631.214,
1939
+ "eval_steps_per_second": 79.989,
1940
+ "step": 49837
1941
+ },
1942
+ {
1943
+ "epoch": 61.2,
1944
+ "learning_rate": 1.9651857050086467e-05,
1945
+ "loss": 0.0831,
1946
+ "step": 50000
1947
+ },
1948
+ {
1949
+ "epoch": 61.51,
1950
+ "learning_rate": 1.9391315692935428e-05,
1951
+ "loss": 0.0378,
1952
+ "step": 50250
1953
+ },
1954
+ {
1955
+ "epoch": 61.81,
1956
+ "learning_rate": 1.9131414228988008e-05,
1957
+ "loss": 0.0795,
1958
+ "step": 50500
1959
+ },
1960
+ {
1961
+ "epoch": 62.0,
1962
+ "eval_accuracy": 0.6942148760330579,
1963
+ "eval_f1": 0.5159346012569708,
1964
+ "eval_loss": 2.37697696685791,
1965
+ "eval_precision": 0.6010404846231928,
1966
+ "eval_recall": 0.4778067239274136,
1967
+ "eval_runtime": 0.5752,
1968
+ "eval_samples_per_second": 631.136,
1969
+ "eval_steps_per_second": 79.979,
1970
+ "step": 50654
1971
+ },
1972
+ {
1973
+ "epoch": 62.12,
1974
+ "learning_rate": 1.8872182310328605e-05,
1975
+ "loss": 0.0624,
1976
+ "step": 50750
1977
+ },
1978
+ {
1979
+ "epoch": 62.42,
1980
+ "learning_rate": 1.861364951265337e-05,
1981
+ "loss": 0.0685,
1982
+ "step": 51000
1983
+ },
1984
+ {
1985
+ "epoch": 62.73,
1986
+ "learning_rate": 1.835584533189595e-05,
1987
+ "loss": 0.0438,
1988
+ "step": 51250
1989
+ },
1990
+ {
1991
+ "epoch": 63.0,
1992
+ "eval_accuracy": 0.6942148760330579,
1993
+ "eval_f1": 0.5006627978013612,
1994
+ "eval_loss": 2.5155351161956787,
1995
+ "eval_precision": 0.5974857177798354,
1996
+ "eval_recall": 0.4612218207353675,
1997
+ "eval_runtime": 0.5756,
1998
+ "eval_samples_per_second": 630.635,
1999
+ "eval_steps_per_second": 79.915,
2000
+ "step": 51471
2001
+ },
2002
+ {
2003
+ "epoch": 63.04,
2004
+ "learning_rate": 1.8098799180862294e-05,
2005
+ "loss": 0.0512,
2006
+ "step": 51500
2007
+ },
2008
+ {
2009
+ "epoch": 63.34,
2010
+ "learning_rate": 1.784254038587494e-05,
2011
+ "loss": 0.0443,
2012
+ "step": 51750
2013
+ },
2014
+ {
2015
+ "epoch": 63.65,
2016
+ "learning_rate": 1.758709818342722e-05,
2017
+ "loss": 0.0331,
2018
+ "step": 52000
2019
+ },
2020
+ {
2021
+ "epoch": 63.95,
2022
+ "learning_rate": 1.7332501716847632e-05,
2023
+ "loss": 0.0432,
2024
+ "step": 52250
2025
+ },
2026
+ {
2027
+ "epoch": 64.0,
2028
+ "eval_accuracy": 0.6859504132231405,
2029
+ "eval_f1": 0.4691190044905215,
2030
+ "eval_loss": 2.5460140705108643,
2031
+ "eval_precision": 0.533927553927554,
2032
+ "eval_recall": 0.44402096603451285,
2033
+ "eval_runtime": 0.5748,
2034
+ "eval_samples_per_second": 631.506,
2035
+ "eval_steps_per_second": 80.026,
2036
+ "step": 52288
2037
+ },
2038
+ {
2039
+ "epoch": 64.26,
2040
+ "learning_rate": 1.7078780032974923e-05,
2041
+ "loss": 0.0442,
2042
+ "step": 52500
2043
+ },
2044
+ {
2045
+ "epoch": 64.57,
2046
+ "learning_rate": 1.682596207884414e-05,
2047
+ "loss": 0.0281,
2048
+ "step": 52750
2049
+ },
2050
+ {
2051
+ "epoch": 64.87,
2052
+ "learning_rate": 1.657608800518716e-05,
2053
+ "loss": 0.0609,
2054
+ "step": 53000
2055
+ },
2056
+ {
2057
+ "epoch": 65.0,
2058
+ "eval_accuracy": 0.7052341597796143,
2059
+ "eval_f1": 0.5783901018472066,
2060
+ "eval_loss": 2.4328320026397705,
2061
+ "eval_precision": 0.6903590657995944,
2062
+ "eval_recall": 0.5364771859229988,
2063
+ "eval_runtime": 0.5756,
2064
+ "eval_samples_per_second": 630.632,
2065
+ "eval_steps_per_second": 79.915,
2066
+ "step": 53105
2067
+ },
2068
+ {
2069
+ "epoch": 65.18,
2070
+ "learning_rate": 1.6325156131696083e-05,
2071
+ "loss": 0.0475,
2072
+ "step": 53250
2073
+ },
2074
+ {
2075
+ "epoch": 65.48,
2076
+ "learning_rate": 1.6075213968684297e-05,
2077
+ "loss": 0.0447,
2078
+ "step": 53500
2079
+ },
2080
+ {
2081
+ "epoch": 65.79,
2082
+ "learning_rate": 1.5826290031982382e-05,
2083
+ "loss": 0.0426,
2084
+ "step": 53750
2085
+ },
2086
+ {
2087
+ "epoch": 66.0,
2088
+ "eval_accuracy": 0.7107438016528925,
2089
+ "eval_f1": 0.5337400275204507,
2090
+ "eval_loss": 2.376704216003418,
2091
+ "eval_precision": 0.6895915165326075,
2092
+ "eval_recall": 0.47468881009521396,
2093
+ "eval_runtime": 0.5754,
2094
+ "eval_samples_per_second": 630.914,
2095
+ "eval_steps_per_second": 79.95,
2096
+ "step": 53922
2097
+ },
2098
+ {
2099
+ "epoch": 66.1,
2100
+ "learning_rate": 1.5578412721251766e-05,
2101
+ "loss": 0.0541,
2102
+ "step": 54000
2103
+ },
2104
+ {
2105
+ "epoch": 66.4,
2106
+ "learning_rate": 1.533161031674459e-05,
2107
+ "loss": 0.0469,
2108
+ "step": 54250
2109
+ },
2110
+ {
2111
+ "epoch": 66.71,
2112
+ "learning_rate": 1.5085910976077283e-05,
2113
+ "loss": 0.0194,
2114
+ "step": 54500
2115
+ },
2116
+ {
2117
+ "epoch": 67.0,
2118
+ "eval_accuracy": 0.7079889807162535,
2119
+ "eval_f1": 0.5368473703383648,
2120
+ "eval_loss": 2.5384085178375244,
2121
+ "eval_precision": 0.6938002473716759,
2122
+ "eval_recall": 0.4738072682653963,
2123
+ "eval_runtime": 0.575,
2124
+ "eval_samples_per_second": 631.275,
2125
+ "eval_steps_per_second": 79.996,
2126
+ "step": 54739
2127
+ },
2128
+ {
2129
+ "epoch": 67.01,
2130
+ "learning_rate": 1.4841342731017988e-05,
2131
+ "loss": 0.0356,
2132
+ "step": 54750
2133
+ },
2134
+ {
2135
+ "epoch": 67.32,
2136
+ "learning_rate": 1.4598904775700689e-05,
2137
+ "loss": 0.0421,
2138
+ "step": 55000
2139
+ },
2140
+ {
2141
+ "epoch": 67.63,
2142
+ "learning_rate": 1.4356677495577313e-05,
2143
+ "loss": 0.0423,
2144
+ "step": 55250
2145
+ },
2146
+ {
2147
+ "epoch": 67.93,
2148
+ "learning_rate": 1.4115664509103169e-05,
2149
+ "loss": 0.0557,
2150
+ "step": 55500
2151
+ },
2152
+ {
2153
+ "epoch": 68.0,
2154
+ "eval_accuracy": 0.7162534435261708,
2155
+ "eval_f1": 0.5433935912807646,
2156
+ "eval_loss": 2.393681049346924,
2157
+ "eval_precision": 0.6931352645638361,
2158
+ "eval_recall": 0.4979390717383328,
2159
+ "eval_runtime": 0.575,
2160
+ "eval_samples_per_second": 631.299,
2161
+ "eval_steps_per_second": 79.999,
2162
+ "step": 55556
2163
+ },
2164
+ {
2165
+ "epoch": 68.24,
2166
+ "learning_rate": 1.3875893313381589e-05,
2167
+ "loss": 0.0432,
2168
+ "step": 55750
2169
+ },
2170
+ {
2171
+ "epoch": 68.54,
2172
+ "learning_rate": 1.3637391263840368e-05,
2173
+ "loss": 0.0429,
2174
+ "step": 56000
2175
+ },
2176
+ {
2177
+ "epoch": 68.85,
2178
+ "learning_rate": 1.3400185571110769e-05,
2179
+ "loss": 0.0466,
2180
+ "step": 56250
2181
+ },
2182
+ {
2183
+ "epoch": 69.0,
2184
+ "eval_accuracy": 0.7107438016528925,
2185
+ "eval_f1": 0.5718625572894164,
2186
+ "eval_loss": 2.403398275375366,
2187
+ "eval_precision": 0.6765347833670815,
2188
+ "eval_recall": 0.5215229283579037,
2189
+ "eval_runtime": 0.5744,
2190
+ "eval_samples_per_second": 631.944,
2191
+ "eval_steps_per_second": 80.081,
2192
+ "step": 56373
2193
+ },
2194
+ {
2195
+ "epoch": 69.16,
2196
+ "learning_rate": 1.316430329792307e-05,
2197
+ "loss": 0.0465,
2198
+ "step": 56500
2199
+ },
2200
+ {
2201
+ "epoch": 69.46,
2202
+ "learning_rate": 1.2929771356018988e-05,
2203
+ "loss": 0.0548,
2204
+ "step": 56750
2205
+ },
2206
+ {
2207
+ "epoch": 69.77,
2208
+ "learning_rate": 1.2696616503081343e-05,
2209
+ "loss": 0.0517,
2210
+ "step": 57000
2211
+ },
2212
+ {
2213
+ "epoch": 70.0,
2214
+ "eval_accuracy": 0.7052341597796143,
2215
+ "eval_f1": 0.5117271310290573,
2216
+ "eval_loss": 2.4453024864196777,
2217
+ "eval_precision": 0.7065175565175564,
2218
+ "eval_recall": 0.46373618855145954,
2219
+ "eval_runtime": 0.5763,
2220
+ "eval_samples_per_second": 629.826,
2221
+ "eval_steps_per_second": 79.813,
2222
+ "step": 57190
2223
+ },
2224
+ {
2225
+ "epoch": 70.07,
2226
+ "learning_rate": 1.2464865339681253e-05,
2227
+ "loss": 0.0226,
2228
+ "step": 57250
2229
+ },
2230
+ {
2231
+ "epoch": 70.38,
2232
+ "learning_rate": 1.2234544306243286e-05,
2233
+ "loss": 0.0292,
2234
+ "step": 57500
2235
+ },
2236
+ {
2237
+ "epoch": 70.69,
2238
+ "learning_rate": 1.2005679680028897e-05,
2239
+ "loss": 0.0368,
2240
+ "step": 57750
2241
+ },
2242
+ {
2243
+ "epoch": 70.99,
2244
+ "learning_rate": 1.177920411295463e-05,
2245
+ "loss": 0.0437,
2246
+ "step": 58000
2247
+ },
2248
+ {
2249
+ "epoch": 71.0,
2250
+ "eval_accuracy": 0.6887052341597796,
2251
+ "eval_f1": 0.4571751135533447,
2252
+ "eval_loss": 2.6550002098083496,
2253
+ "eval_precision": 0.6424798334634401,
2254
+ "eval_recall": 0.41016626367611586,
2255
+ "eval_runtime": 0.5744,
2256
+ "eval_samples_per_second": 632.012,
2257
+ "eval_steps_per_second": 80.09,
2258
+ "step": 58007
2259
+ },
2260
+ {
2261
+ "epoch": 71.3,
2262
+ "learning_rate": 1.1553324380057995e-05,
2263
+ "loss": 0.037,
2264
+ "step": 58250
2265
+ },
2266
+ {
2267
+ "epoch": 71.6,
2268
+ "learning_rate": 1.1328978774573311e-05,
2269
+ "loss": 0.0361,
2270
+ "step": 58500
2271
+ },
2272
+ {
2273
+ "epoch": 71.91,
2274
+ "learning_rate": 1.1106192892027184e-05,
2275
+ "loss": 0.0451,
2276
+ "step": 58750
2277
+ },
2278
+ {
2279
+ "epoch": 72.0,
2280
+ "eval_accuracy": 0.6942148760330579,
2281
+ "eval_f1": 0.476209725136688,
2282
+ "eval_loss": 2.6152384281158447,
2283
+ "eval_precision": 0.6680127132820342,
2284
+ "eval_recall": 0.4248004219371215,
2285
+ "eval_runtime": 0.5754,
2286
+ "eval_samples_per_second": 630.887,
2287
+ "eval_steps_per_second": 79.947,
2288
+ "step": 58824
2289
+ },
2290
+ {
2291
+ "epoch": 72.22,
2292
+ "learning_rate": 1.0884992149997797e-05,
2293
+ "loss": 0.0112,
2294
+ "step": 59000
2295
+ },
2296
+ {
2297
+ "epoch": 72.52,
2298
+ "learning_rate": 1.066540178521517e-05,
2299
+ "loss": 0.043,
2300
+ "step": 59250
2301
+ },
2302
+ {
2303
+ "epoch": 72.83,
2304
+ "learning_rate": 1.04474468506818e-05,
2305
+ "loss": 0.0294,
2306
+ "step": 59500
2307
+ },
2308
+ {
2309
+ "epoch": 73.0,
2310
+ "eval_accuracy": 0.6887052341597796,
2311
+ "eval_f1": 0.5392699243016044,
2312
+ "eval_loss": 2.5824713706970215,
2313
+ "eval_precision": 0.6700918129489558,
2314
+ "eval_recall": 0.4843369001004469,
2315
+ "eval_runtime": 0.5751,
2316
+ "eval_samples_per_second": 631.242,
2317
+ "eval_steps_per_second": 79.992,
2318
+ "step": 59641
2319
+ },
2320
+ {
2321
+ "epoch": 73.13,
2322
+ "learning_rate": 1.0231152212814419e-05,
2323
+ "loss": 0.0468,
2324
+ "step": 59750
2325
+ },
2326
+ {
2327
+ "epoch": 73.44,
2328
+ "learning_rate": 1.0016542548606983e-05,
2329
+ "loss": 0.0316,
2330
+ "step": 60000
2331
+ },
2332
+ {
2333
+ "epoch": 73.75,
2334
+ "learning_rate": 9.803642342815278e-06,
2335
+ "loss": 0.0429,
2336
+ "step": 60250
2337
+ },
2338
+ {
2339
+ "epoch": 74.0,
2340
+ "eval_accuracy": 0.696969696969697,
2341
+ "eval_f1": 0.48452204181281855,
2342
+ "eval_loss": 2.451073408126831,
2343
+ "eval_precision": 0.6746470741527906,
2344
+ "eval_recall": 0.4329222604161029,
2345
+ "eval_runtime": 0.5756,
2346
+ "eval_samples_per_second": 630.696,
2347
+ "eval_steps_per_second": 79.923,
2348
+ "step": 60458
2349
+ },
2350
+ {
2351
+ "epoch": 74.05,
2352
+ "learning_rate": 9.592475885163487e-06,
2353
+ "loss": 0.0236,
2354
+ "step": 60500
2355
+ },
2356
+ {
2357
+ "epoch": 74.36,
2358
+ "learning_rate": 9.383067267572908e-06,
2359
+ "loss": 0.0221,
2360
+ "step": 60750
2361
+ },
2362
+ {
2363
+ "epoch": 74.66,
2364
+ "learning_rate": 9.17544038141342e-06,
2365
+ "loss": 0.0313,
2366
+ "step": 61000
2367
+ },
2368
+ {
2369
+ "epoch": 74.97,
2370
+ "learning_rate": 8.969618914777637e-06,
2371
+ "loss": 0.0389,
2372
+ "step": 61250
2373
+ },
2374
+ {
2375
+ "epoch": 75.0,
2376
+ "eval_accuracy": 0.6749311294765841,
2377
+ "eval_f1": 0.47978279633622645,
2378
+ "eval_loss": 2.7327232360839844,
2379
+ "eval_precision": 0.5541808071284741,
2380
+ "eval_recall": 0.44874471450826137,
2381
+ "eval_runtime": 0.5752,
2382
+ "eval_samples_per_second": 631.071,
2383
+ "eval_steps_per_second": 79.97,
2384
+ "step": 61275
2385
+ },
2386
+ {
2387
+ "epoch": 75.28,
2388
+ "learning_rate": 8.765626349778362e-06,
2389
+ "loss": 0.0227,
2390
+ "step": 61500
2391
+ },
2392
+ {
2393
+ "epoch": 75.58,
2394
+ "learning_rate": 8.563485959869575e-06,
2395
+ "loss": 0.0303,
2396
+ "step": 61750
2397
+ },
2398
+ {
2399
+ "epoch": 75.89,
2400
+ "learning_rate": 8.363220807191058e-06,
2401
+ "loss": 0.0305,
2402
+ "step": 62000
2403
+ },
2404
+ {
2405
+ "epoch": 76.0,
2406
+ "eval_accuracy": 0.7024793388429752,
2407
+ "eval_f1": 0.5534820298880379,
2408
+ "eval_loss": 2.4534895420074463,
2409
+ "eval_precision": 0.6256500777723188,
2410
+ "eval_recall": 0.5284146306621677,
2411
+ "eval_runtime": 0.5743,
2412
+ "eval_samples_per_second": 632.097,
2413
+ "eval_steps_per_second": 80.1,
2414
+ "step": 62092
2415
+ },
2416
+ {
2417
+ "epoch": 76.19,
2418
+ "learning_rate": 8.164853739937368e-06,
2419
+ "loss": 0.034,
2420
+ "step": 62250
2421
+ },
2422
+ {
2423
+ "epoch": 76.5,
2424
+ "learning_rate": 7.968407389751003e-06,
2425
+ "loss": 0.034,
2426
+ "step": 62500
2427
+ },
2428
+ {
2429
+ "epoch": 76.81,
2430
+ "learning_rate": 7.773904169140392e-06,
2431
+ "loss": 0.023,
2432
+ "step": 62750
2433
+ },
2434
+ {
2435
+ "epoch": 77.0,
2436
+ "eval_accuracy": 0.7024793388429752,
2437
+ "eval_f1": 0.4687311706012172,
2438
+ "eval_loss": 2.6240921020507812,
2439
+ "eval_precision": 0.5393879300200679,
2440
+ "eval_recall": 0.4422954747770019,
2441
+ "eval_runtime": 0.5739,
2442
+ "eval_samples_per_second": 632.494,
2443
+ "eval_steps_per_second": 80.151,
2444
+ "step": 62909
2445
+ },
2446
+ {
2447
+ "epoch": 77.11,
2448
+ "learning_rate": 7.581366268922896e-06,
2449
+ "loss": 0.0132,
2450
+ "step": 63000
2451
+ },
2452
+ {
2453
+ "epoch": 77.42,
2454
+ "learning_rate": 7.3908156556929705e-06,
2455
+ "loss": 0.0201,
2456
+ "step": 63250
2457
+ },
2458
+ {
2459
+ "epoch": 77.72,
2460
+ "learning_rate": 7.20227406931612e-06,
2461
+ "loss": 0.0305,
2462
+ "step": 63500
2463
+ },
2464
+ {
2465
+ "epoch": 78.0,
2466
+ "eval_accuracy": 0.6942148760330579,
2467
+ "eval_f1": 0.4474972103847281,
2468
+ "eval_loss": 2.737638235092163,
2469
+ "eval_precision": 0.539260468004995,
2470
+ "eval_recall": 0.407899366643209,
2471
+ "eval_runtime": 0.5754,
2472
+ "eval_samples_per_second": 630.913,
2473
+ "eval_steps_per_second": 79.95,
2474
+ "step": 63726
2475
+ },
2476
+ {
2477
+ "epoch": 78.03,
2478
+ "learning_rate": 7.01576302044851e-06,
2479
+ "loss": 0.0258,
2480
+ "step": 63750
2481
+ },
2482
+ {
2483
+ "epoch": 78.34,
2484
+ "learning_rate": 6.831303788082866e-06,
2485
+ "loss": 0.0212,
2486
+ "step": 64000
2487
+ },
2488
+ {
2489
+ "epoch": 78.64,
2490
+ "learning_rate": 6.6489174171207504e-06,
2491
+ "loss": 0.0178,
2492
+ "step": 64250
2493
+ },
2494
+ {
2495
+ "epoch": 78.95,
2496
+ "learning_rate": 6.469341688844232e-06,
2497
+ "loss": 0.0278,
2498
+ "step": 64500
2499
+ },
2500
+ {
2501
+ "epoch": 79.0,
2502
+ "eval_accuracy": 0.7052341597796143,
2503
+ "eval_f1": 0.5369051843353192,
2504
+ "eval_loss": 2.641106605529785,
2505
+ "eval_precision": 0.6364117161569525,
2506
+ "eval_recall": 0.48457279242377765,
2507
+ "eval_runtime": 0.5738,
2508
+ "eval_samples_per_second": 632.582,
2509
+ "eval_steps_per_second": 80.162,
2510
+ "step": 64543
2511
+ },
2512
+ {
2513
+ "epoch": 79.25,
2514
+ "learning_rate": 6.291154729438112e-06,
2515
+ "loss": 0.0211,
2516
+ "step": 64750
2517
+ },
2518
+ {
2519
+ "epoch": 79.56,
2520
+ "learning_rate": 6.115102256888608e-06,
2521
+ "loss": 0.027,
2522
+ "step": 65000
2523
+ },
2524
+ {
2525
+ "epoch": 79.87,
2526
+ "learning_rate": 5.9412043569724075e-06,
2527
+ "loss": 0.0245,
2528
+ "step": 65250
2529
+ },
2530
+ {
2531
+ "epoch": 80.0,
2532
+ "eval_accuracy": 0.6887052341597796,
2533
+ "eval_f1": 0.5139284663519239,
2534
+ "eval_loss": 2.8429274559020996,
2535
+ "eval_precision": 0.6295462275834203,
2536
+ "eval_recall": 0.46902355991025946,
2537
+ "eval_runtime": 0.5746,
2538
+ "eval_samples_per_second": 631.731,
2539
+ "eval_steps_per_second": 80.054,
2540
+ "step": 65360
2541
+ },
2542
+ {
2543
+ "epoch": 80.17,
2544
+ "learning_rate": 5.769480869651678e-06,
2545
+ "loss": 0.0177,
2546
+ "step": 65500
2547
+ },
2548
+ {
2549
+ "epoch": 80.48,
2550
+ "learning_rate": 5.599951386810407e-06,
2551
+ "loss": 0.0229,
2552
+ "step": 65750
2553
+ },
2554
+ {
2555
+ "epoch": 80.78,
2556
+ "learning_rate": 5.432635250019285e-06,
2557
+ "loss": 0.0227,
2558
+ "step": 66000
2559
+ },
2560
+ {
2561
+ "epoch": 81.0,
2562
+ "eval_accuracy": 0.7245179063360881,
2563
+ "eval_f1": 0.5777058546571006,
2564
+ "eval_loss": 2.6588237285614014,
2565
+ "eval_precision": 0.6858633718409165,
2566
+ "eval_recall": 0.5245856565437846,
2567
+ "eval_runtime": 0.5738,
2568
+ "eval_samples_per_second": 632.57,
2569
+ "eval_steps_per_second": 80.16,
2570
+ "step": 66177
2571
+ },
2572
+ {
2573
+ "epoch": 81.09,
2574
+ "learning_rate": 5.26755154832895e-06,
2575
+ "loss": 0.0261,
2576
+ "step": 66250
2577
+ },
2578
+ {
2579
+ "epoch": 81.4,
2580
+ "learning_rate": 5.1047191160921495e-06,
2581
+ "loss": 0.0191,
2582
+ "step": 66500
2583
+ },
2584
+ {
2585
+ "epoch": 81.7,
2586
+ "learning_rate": 4.9441565308149724e-06,
2587
+ "loss": 0.0226,
2588
+ "step": 66750
2589
+ },
2590
+ {
2591
+ "epoch": 82.0,
2592
+ "eval_accuracy": 0.7024793388429752,
2593
+ "eval_f1": 0.5153196455189113,
2594
+ "eval_loss": 2.683469533920288,
2595
+ "eval_precision": 0.5957972186543615,
2596
+ "eval_recall": 0.47307357735313404,
2597
+ "eval_runtime": 0.5753,
2598
+ "eval_samples_per_second": 630.959,
2599
+ "eval_steps_per_second": 79.956,
2600
+ "step": 66994
2601
+ },
2602
+ {
2603
+ "epoch": 82.01,
2604
+ "learning_rate": 4.785882111037252e-06,
2605
+ "loss": 0.0164,
2606
+ "step": 67000
2607
+ },
2608
+ {
2609
+ "epoch": 82.31,
2610
+ "learning_rate": 4.629913914242723e-06,
2611
+ "loss": 0.0207,
2612
+ "step": 67250
2613
+ },
2614
+ {
2615
+ "epoch": 82.62,
2616
+ "learning_rate": 4.4762697347987634e-06,
2617
+ "loss": 0.0211,
2618
+ "step": 67500
2619
+ },
2620
+ {
2621
+ "epoch": 82.93,
2622
+ "learning_rate": 4.324967101926272e-06,
2623
+ "loss": 0.0224,
2624
+ "step": 67750
2625
+ },
2626
+ {
2627
+ "epoch": 83.0,
2628
+ "eval_accuracy": 0.699724517906336,
2629
+ "eval_f1": 0.5510071098243216,
2630
+ "eval_loss": 2.7483408451080322,
2631
+ "eval_precision": 0.6299525870954442,
2632
+ "eval_recall": 0.5086697101475426,
2633
+ "eval_runtime": 0.5745,
2634
+ "eval_samples_per_second": 631.809,
2635
+ "eval_steps_per_second": 80.064,
2636
+ "step": 67811
2637
+ },
2638
+ {
2639
+ "epoch": 83.23,
2640
+ "learning_rate": 4.176023277699789e-06,
2641
+ "loss": 0.027,
2642
+ "step": 68000
2643
+ },
2644
+ {
2645
+ "epoch": 83.54,
2646
+ "learning_rate": 4.029455255077999e-06,
2647
+ "loss": 0.0154,
2648
+ "step": 68250
2649
+ },
2650
+ {
2651
+ "epoch": 83.84,
2652
+ "learning_rate": 3.8852797559650935e-06,
2653
+ "loss": 0.0195,
2654
+ "step": 68500
2655
+ },
2656
+ {
2657
+ "epoch": 84.0,
2658
+ "eval_accuracy": 0.699724517906336,
2659
+ "eval_f1": 0.52634249226563,
2660
+ "eval_loss": 2.7812142372131348,
2661
+ "eval_precision": 0.5949473728885494,
2662
+ "eval_recall": 0.49608820981973695,
2663
+ "eval_runtime": 0.5756,
2664
+ "eval_samples_per_second": 630.699,
2665
+ "eval_steps_per_second": 79.923,
2666
+ "step": 68628
2667
+ },
2668
+ {
2669
+ "epoch": 84.15,
2670
+ "learning_rate": 3.7440754752084e-06,
2671
+ "loss": 0.012,
2672
+ "step": 68750
2673
+ },
2674
+ {
2675
+ "epoch": 84.46,
2676
+ "learning_rate": 3.604724362660877e-06,
2677
+ "loss": 0.0262,
2678
+ "step": 69000
2679
+ },
2680
+ {
2681
+ "epoch": 84.76,
2682
+ "learning_rate": 3.467814231049432e-06,
2683
+ "loss": 0.0106,
2684
+ "step": 69250
2685
+ },
2686
+ {
2687
+ "epoch": 85.0,
2688
+ "eval_accuracy": 0.6942148760330579,
2689
+ "eval_f1": 0.5085355239316488,
2690
+ "eval_loss": 2.8171160221099854,
2691
+ "eval_precision": 0.6049143980994077,
2692
+ "eval_recall": 0.4679249219581732,
2693
+ "eval_runtime": 0.5729,
2694
+ "eval_samples_per_second": 633.612,
2695
+ "eval_steps_per_second": 80.292,
2696
+ "step": 69445
2697
+ },
2698
+ {
2699
+ "epoch": 85.07,
2700
+ "learning_rate": 3.33336070041218e-06,
2701
+ "loss": 0.0258,
2702
+ "step": 69500
2703
+ },
2704
+ {
2705
+ "epoch": 85.37,
2706
+ "learning_rate": 3.2013791105143466e-06,
2707
+ "loss": 0.0118,
2708
+ "step": 69750
2709
+ },
2710
+ {
2711
+ "epoch": 85.68,
2712
+ "learning_rate": 3.071884519098131e-06,
2713
+ "loss": 0.0217,
2714
+ "step": 70000
2715
+ },
2716
+ {
2717
+ "epoch": 85.99,
2718
+ "learning_rate": 2.9448917001647703e-06,
2719
+ "loss": 0.0155,
2720
+ "step": 70250
2721
+ },
2722
+ {
2723
+ "epoch": 86.0,
2724
+ "eval_accuracy": 0.6887052341597796,
2725
+ "eval_f1": 0.4962240507440637,
2726
+ "eval_loss": 2.8894941806793213,
2727
+ "eval_precision": 0.5999622071050642,
2728
+ "eval_recall": 0.45233290929103737,
2729
+ "eval_runtime": 0.575,
2730
+ "eval_samples_per_second": 631.309,
2731
+ "eval_steps_per_second": 80.001,
2732
+ "step": 70262
2733
+ },
2734
+ {
2735
+ "epoch": 86.29,
2736
+ "learning_rate": 2.820415142289015e-06,
2737
+ "loss": 0.0196,
2738
+ "step": 70500
2739
+ },
2740
+ {
2741
+ "epoch": 86.6,
2742
+ "learning_rate": 2.698951772130373e-06,
2743
+ "loss": 0.0064,
2744
+ "step": 70750
2745
+ },
2746
+ {
2747
+ "epoch": 86.9,
2748
+ "learning_rate": 2.5795398473242555e-06,
2749
+ "loss": 0.0111,
2750
+ "step": 71000
2751
+ },
2752
+ {
2753
+ "epoch": 87.0,
2754
+ "eval_accuracy": 0.6859504132231405,
2755
+ "eval_f1": 0.4869283447078967,
2756
+ "eval_loss": 2.887157678604126,
2757
+ "eval_precision": 0.596748772605851,
2758
+ "eval_recall": 0.44054806714905237,
2759
+ "eval_runtime": 0.575,
2760
+ "eval_samples_per_second": 631.314,
2761
+ "eval_steps_per_second": 80.001,
2762
+ "step": 71079
2763
+ },
2764
+ {
2765
+ "epoch": 87.21,
2766
+ "learning_rate": 2.462685866465117e-06,
2767
+ "loss": 0.0174,
2768
+ "step": 71250
2769
+ },
2770
+ {
2771
+ "epoch": 87.52,
2772
+ "learning_rate": 2.3484031613905387e-06,
2773
+ "loss": 0.0105,
2774
+ "step": 71500
2775
+ },
2776
+ {
2777
+ "epoch": 87.82,
2778
+ "learning_rate": 2.2367047705819572e-06,
2779
+ "loss": 0.0097,
2780
+ "step": 71750
2781
+ },
2782
+ {
2783
+ "epoch": 88.0,
2784
+ "eval_accuracy": 0.6859504132231405,
2785
+ "eval_f1": 0.4890889302500872,
2786
+ "eval_loss": 2.9936180114746094,
2787
+ "eval_precision": 0.5928653178112443,
2788
+ "eval_recall": 0.44590630018585686,
2789
+ "eval_runtime": 0.5751,
2790
+ "eval_samples_per_second": 631.226,
2791
+ "eval_steps_per_second": 79.99,
2792
+ "step": 71896
2793
+ },
2794
+ {
2795
+ "epoch": 88.13,
2796
+ "learning_rate": 2.127603437677139e-06,
2797
+ "loss": 0.0083,
2798
+ "step": 72000
2799
+ },
2800
+ {
2801
+ "epoch": 88.43,
2802
+ "learning_rate": 2.0211116100162435e-06,
2803
+ "loss": 0.0124,
2804
+ "step": 72250
2805
+ },
2806
+ {
2807
+ "epoch": 88.74,
2808
+ "learning_rate": 1.9172414372217146e-06,
2809
+ "loss": 0.0086,
2810
+ "step": 72500
2811
+ },
2812
+ {
2813
+ "epoch": 89.0,
2814
+ "eval_accuracy": 0.6831955922865014,
2815
+ "eval_f1": 0.49100657070581877,
2816
+ "eval_loss": 3.0106709003448486,
2817
+ "eval_precision": 0.5958884973599873,
2818
+ "eval_recall": 0.44583543850785234,
2819
+ "eval_runtime": 0.5744,
2820
+ "eval_samples_per_second": 631.965,
2821
+ "eval_steps_per_second": 80.084,
2822
+ "step": 72713
2823
+ },
2824
+ {
2825
+ "epoch": 89.05,
2826
+ "learning_rate": 1.8160047698121518e-06,
2827
+ "loss": 0.0104,
2828
+ "step": 72750
2829
+ },
2830
+ {
2831
+ "epoch": 89.35,
2832
+ "learning_rate": 1.7178022403391475e-06,
2833
+ "loss": 0.0108,
2834
+ "step": 73000
2835
+ },
2836
+ {
2837
+ "epoch": 89.66,
2838
+ "learning_rate": 1.6218562848936775e-06,
2839
+ "loss": 0.0174,
2840
+ "step": 73250
2841
+ },
2842
+ {
2843
+ "epoch": 89.96,
2844
+ "learning_rate": 1.528577535241521e-06,
2845
+ "loss": 0.0072,
2846
+ "step": 73500
2847
+ },
2848
+ {
2849
+ "epoch": 90.0,
2850
+ "eval_accuracy": 0.6859504132231405,
2851
+ "eval_f1": 0.4893894643187819,
2852
+ "eval_loss": 3.0227720737457275,
2853
+ "eval_precision": 0.5889153512253815,
2854
+ "eval_recall": 0.4449414911853336,
2855
+ "eval_runtime": 0.5765,
2856
+ "eval_samples_per_second": 629.693,
2857
+ "eval_steps_per_second": 79.796,
2858
+ "step": 73530
2859
+ },
2860
+ {
2861
+ "epoch": 90.27,
2862
+ "learning_rate": 1.4383336890610749e-06,
2863
+ "loss": 0.0151,
2864
+ "step": 73750
2865
+ },
2866
+ {
2867
+ "epoch": 90.58,
2868
+ "learning_rate": 1.3504101989962132e-06,
2869
+ "loss": 0.0109,
2870
+ "step": 74000
2871
+ },
2872
+ {
2873
+ "epoch": 90.88,
2874
+ "learning_rate": 1.2651848839316887e-06,
2875
+ "loss": 0.0107,
2876
+ "step": 74250
2877
+ },
2878
+ {
2879
+ "epoch": 91.0,
2880
+ "eval_accuracy": 0.6887052341597796,
2881
+ "eval_f1": 0.49676183815019276,
2882
+ "eval_loss": 2.977621078491211,
2883
+ "eval_precision": 0.5930682617874735,
2884
+ "eval_recall": 0.45364507208719035,
2885
+ "eval_runtime": 0.5751,
2886
+ "eval_samples_per_second": 631.186,
2887
+ "eval_steps_per_second": 79.985,
2888
+ "step": 74347
2889
+ },
2890
+ {
2891
+ "epoch": 91.19,
2892
+ "learning_rate": 1.182667467199558e-06,
2893
+ "loss": 0.0081,
2894
+ "step": 74500
2895
+ },
2896
+ {
2897
+ "epoch": 91.49,
2898
+ "learning_rate": 1.1028673631885173e-06,
2899
+ "loss": 0.0136,
2900
+ "step": 74750
2901
+ },
2902
+ {
2903
+ "epoch": 91.8,
2904
+ "learning_rate": 1.0257936762698288e-06,
2905
+ "loss": 0.0105,
2906
+ "step": 75000
2907
+ },
2908
+ {
2909
+ "epoch": 92.0,
2910
+ "eval_accuracy": 0.6887052341597796,
2911
+ "eval_f1": 0.4999633782778992,
2912
+ "eval_loss": 3.0019350051879883,
2913
+ "eval_precision": 0.5966416670279332,
2914
+ "eval_recall": 0.4551455728179866,
2915
+ "eval_runtime": 0.5756,
2916
+ "eval_samples_per_second": 630.598,
2917
+ "eval_steps_per_second": 79.91,
2918
+ "step": 75164
2919
+ },
2920
+ {
2921
+ "epoch": 92.11,
2922
+ "learning_rate": 9.514551997585913e-07,
2923
+ "loss": 0.0061,
2924
+ "step": 75250
2925
+ },
2926
+ {
2927
+ "epoch": 92.41,
2928
+ "learning_rate": 8.798604149105355e-07,
2929
+ "loss": 0.0094,
2930
+ "step": 75500
2931
+ },
2932
+ {
2933
+ "epoch": 92.72,
2934
+ "learning_rate": 8.110174899543743e-07,
2935
+ "loss": 0.0138,
2936
+ "step": 75750
2937
+ },
2938
+ {
2939
+ "epoch": 93.0,
2940
+ "eval_accuracy": 0.6887052341597796,
2941
+ "eval_f1": 0.4999633782778992,
2942
+ "eval_loss": 3.0192649364471436,
2943
+ "eval_precision": 0.5966416670279332,
2944
+ "eval_recall": 0.4551455728179866,
2945
+ "eval_runtime": 0.5738,
2946
+ "eval_samples_per_second": 632.575,
2947
+ "eval_steps_per_second": 80.161,
2948
+ "step": 75981
2949
+ },
2950
+ {
2951
+ "epoch": 93.02,
2952
+ "learning_rate": 7.449342791599201e-07,
2953
+ "loss": 0.0106,
2954
+ "step": 76000
2955
+ },
2956
+ {
2957
+ "epoch": 93.33,
2958
+ "learning_rate": 6.81618321941968e-07,
2959
+ "loss": 0.0107,
2960
+ "step": 76250
2961
+ },
2962
+ {
2963
+ "epoch": 93.64,
2964
+ "learning_rate": 6.210768420001373e-07,
2965
+ "loss": 0.0081,
2966
+ "step": 76500
2967
+ },
2968
+ {
2969
+ "epoch": 93.94,
2970
+ "learning_rate": 5.633167464947242e-07,
2971
+ "loss": 0.0123,
2972
+ "step": 76750
2973
+ },
2974
+ {
2975
+ "epoch": 94.0,
2976
+ "eval_accuracy": 0.6859504132231405,
2977
+ "eval_f1": 0.48841087906946806,
2978
+ "eval_loss": 3.0272629261016846,
2979
+ "eval_precision": 0.5842731342731343,
2980
+ "eval_recall": 0.4449414911853336,
2981
+ "eval_runtime": 0.5754,
2982
+ "eval_samples_per_second": 630.882,
2983
+ "eval_steps_per_second": 79.946,
2984
+ "step": 76798
2985
+ },
2986
+ {
2987
+ "epoch": 94.25,
2988
+ "learning_rate": 5.08344625258661e-07,
2989
+ "loss": 0.0113,
2990
+ "step": 77000
2991
+ },
2992
+ {
2993
+ "epoch": 94.55,
2994
+ "learning_rate": 4.561667500456862e-07,
2995
+ "loss": 0.0064,
2996
+ "step": 77250
2997
+ },
2998
+ {
2999
+ "epoch": 94.86,
3000
+ "learning_rate": 4.067890738147978e-07,
3001
+ "loss": 0.0114,
3002
+ "step": 77500
3003
+ },
3004
+ {
3005
+ "epoch": 95.0,
3006
+ "eval_accuracy": 0.6859504132231405,
3007
+ "eval_f1": 0.49834117213459417,
3008
+ "eval_loss": 3.0193073749542236,
3009
+ "eval_precision": 0.592981803003125,
3010
+ "eval_recall": 0.45445875963117344,
3011
+ "eval_runtime": 0.5755,
3012
+ "eval_samples_per_second": 630.739,
3013
+ "eval_steps_per_second": 79.928,
3014
+ "step": 77615
3015
+ },
3016
+ {
3017
+ "epoch": 95.17,
3018
+ "learning_rate": 3.6021723005109365e-07,
3019
+ "loss": 0.013,
3020
+ "step": 77750
3021
+ },
3022
+ {
3023
+ "epoch": 95.47,
3024
+ "learning_rate": 3.1645653212303004e-07,
3025
+ "loss": 0.0127,
3026
+ "step": 78000
3027
+ },
3028
+ {
3029
+ "epoch": 95.78,
3030
+ "learning_rate": 2.755119726762373e-07,
3031
+ "loss": 0.0087,
3032
+ "step": 78250
3033
+ },
3034
+ {
3035
+ "epoch": 96.0,
3036
+ "eval_accuracy": 0.6859504132231405,
3037
+ "eval_f1": 0.4971705409799476,
3038
+ "eval_loss": 3.0252628326416016,
3039
+ "eval_precision": 0.5904316041796341,
3040
+ "eval_recall": 0.45445875963117344,
3041
+ "eval_runtime": 0.5748,
3042
+ "eval_samples_per_second": 631.48,
3043
+ "eval_steps_per_second": 80.022,
3044
+ "step": 78432
3045
+ },
3046
+ {
3047
+ "epoch": 96.08,
3048
+ "learning_rate": 2.3738822306390575e-07,
3049
+ "loss": 0.0076,
3050
+ "step": 78500
3051
+ },
3052
+ {
3053
+ "epoch": 96.39,
3054
+ "learning_rate": 2.0208963281382054e-07,
3055
+ "loss": 0.0059,
3056
+ "step": 78750
3057
+ },
3058
+ {
3059
+ "epoch": 96.7,
3060
+ "learning_rate": 1.6962022913215026e-07,
3061
+ "loss": 0.0112,
3062
+ "step": 79000
3063
+ },
3064
+ {
3065
+ "epoch": 97.0,
3066
+ "eval_accuracy": 0.6859504132231405,
3067
+ "eval_f1": 0.4971705409799476,
3068
+ "eval_loss": 3.0260462760925293,
3069
+ "eval_precision": 0.5904316041796341,
3070
+ "eval_recall": 0.45445875963117344,
3071
+ "eval_runtime": 0.575,
3072
+ "eval_samples_per_second": 631.313,
3073
+ "eval_steps_per_second": 80.001,
3074
+ "step": 79249
3075
+ },
3076
+ {
3077
+ "epoch": 97.0,
3078
+ "learning_rate": 1.399837164439588e-07,
3079
+ "loss": 0.0198,
3080
+ "step": 79250
3081
+ },
3082
+ {
3083
+ "epoch": 97.31,
3084
+ "learning_rate": 1.131834759705852e-07,
3085
+ "loss": 0.0108,
3086
+ "step": 79500
3087
+ },
3088
+ {
3089
+ "epoch": 97.61,
3090
+ "learning_rate": 8.922256534386886e-08,
3091
+ "loss": 0.0096,
3092
+ "step": 79750
3093
+ },
3094
+ {
3095
+ "epoch": 97.92,
3096
+ "learning_rate": 6.818252895382516e-08,
3097
+ "loss": 0.0154,
3098
+ "step": 80000
3099
+ },
3100
+ {
3101
+ "epoch": 98.0,
3102
+ "eval_accuracy": 0.6859504132231405,
3103
+ "eval_f1": 0.4971705409799476,
3104
+ "eval_loss": 3.032817840576172,
3105
+ "eval_precision": 0.5904316041796341,
3106
+ "eval_recall": 0.45445875963117344,
3107
+ "eval_runtime": 0.5745,
3108
+ "eval_samples_per_second": 631.834,
3109
+ "eval_steps_per_second": 80.067,
3110
+ "step": 80066
3111
+ },
3112
+ {
3113
+ "epoch": 98.23,
3114
+ "learning_rate": 4.989677258919745e-08,
3115
+ "loss": 0.0066,
3116
+ "step": 80250
3117
+ },
3118
+ {
3119
+ "epoch": 98.53,
3120
+ "learning_rate": 3.4457566433288015e-08,
3121
+ "loss": 0.0089,
3122
+ "step": 80500
3123
+ },
3124
+ {
3125
+ "epoch": 98.84,
3126
+ "learning_rate": 2.1866671940751205e-08,
3127
+ "loss": 0.0113,
3128
+ "step": 80750
3129
+ },
3130
+ {
3131
+ "epoch": 99.0,
3132
+ "eval_accuracy": 0.6859504132231405,
3133
+ "eval_f1": 0.4971705409799476,
3134
+ "eval_loss": 3.0352275371551514,
3135
+ "eval_precision": 0.5904316041796341,
3136
+ "eval_recall": 0.45445875963117344,
3137
+ "eval_runtime": 0.5755,
3138
+ "eval_samples_per_second": 630.795,
3139
+ "eval_steps_per_second": 79.935,
3140
+ "step": 80883
3141
+ },
3142
+ {
3143
+ "epoch": 99.14,
3144
+ "learning_rate": 1.212552560317659e-08,
3145
+ "loss": 0.0052,
3146
+ "step": 81000
3147
+ },
3148
+ {
3149
+ "epoch": 99.45,
3150
+ "learning_rate": 5.235238785186725e-09,
3151
+ "loss": 0.0124,
3152
+ "step": 81250
3153
+ },
3154
+ {
3155
+ "epoch": 99.76,
3156
+ "learning_rate": 1.1965975976552247e-09,
3157
+ "loss": 0.0094,
3158
+ "step": 81500
3159
+ },
3160
+ {
3161
+ "epoch": 100.0,
3162
+ "eval_accuracy": 0.6859504132231405,
3163
+ "eval_f1": 0.4971705409799476,
3164
+ "eval_loss": 3.034996271133423,
3165
+ "eval_precision": 0.5904316041796341,
3166
+ "eval_recall": 0.45445875963117344,
3167
+ "eval_runtime": 0.5749,
3168
+ "eval_samples_per_second": 631.447,
3169
+ "eval_steps_per_second": 80.018,
3170
+ "step": 81700
3171
+ },
3172
+ {
3173
+ "epoch": 100.0,
3174
+ "step": 81700,
3175
+ "total_flos": 2.247320628668568e+16,
3176
+ "train_loss": 0.19591666076364248,
3177
+ "train_runtime": 5839.9501,
3178
+ "train_samples_per_second": 111.885,
3179
+ "train_steps_per_second": 13.99
3180
+ }
3181
+ ],
3182
+ "max_steps": 81700,
3183
+ "num_train_epochs": 100,
3184
+ "total_flos": 2.247320628668568e+16,
3185
+ "trial_name": null,
3186
+ "trial_params": {
3187
+ "learning_rate": 5e-06
3188
+ }
3189
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2d122769bd01b7c3c05639627de322a35ada8628020b07b1f86a5177d7c27a3f
3
+ size 3375