Hugging Face's logo

nllg
/
clf-canine-r

Text Classification
Transformers
PyTorch
canine
Inference Endpoints
Model card Files Community
1
potamides commited on
Commit
8aceabe
·
1 Parent(s): 53632cf

upload model files

Browse files
Files changed (9) hide show
  1. config.json +30 -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 +4389 -0
  9. training_args.bin +3 -0
config.json ADDED
@@ -0,0 +1,30 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "models/canine-c/rhyme/run-3/checkpoint-176630",
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
+ "initializer_range": 0.02,
14
+ "intermediate_size": 3072,
15
+ "layer_norm_eps": 1e-12,
16
+ "local_transformer_stride": 128,
17
+ "max_position_embeddings": 16384,
18
+ "model_type": "canine",
19
+ "num_attention_heads": 12,
20
+ "num_hash_buckets": 16384,
21
+ "num_hash_functions": 8,
22
+ "num_hidden_layers": 12,
23
+ "pad_token_id": 0,
24
+ "problem_type": "single_label_classification",
25
+ "torch_dtype": "float32",
26
+ "transformers_version": "4.20.0",
27
+ "type_vocab_size": 16,
28
+ "upsampling_kernel_size": 4,
29
+ "use_cache": true
30
+ }
pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3e790d15594741d39d159f7a4ff23a3f52186af8d8aada9cfb18f85238bca0cd
3
+ size 528552655
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": 10.0,
3
+ "test_accuracy": 0.976890756302521,
4
+ "test_f1": 0.9768760599208592,
5
+ "test_loss": 0.13919147849082947,
6
+ "test_precision": 0.9768518671855593,
7
+ "test_recall": 0.9769024124441441,
8
+ "test_runtime": 3.0985,
9
+ "test_samples_per_second": 614.49,
10
+ "test_steps_per_second": 76.811
11
+ }
test-en_results.json ADDED
@@ -0,0 +1,11 @@
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 10.0,
3
+ "test_accuracy": 0.9804910864446686,
4
+ "test_f1": 0.9804910842374603,
5
+ "test_loss": 0.1231788620352745,
6
+ "test_precision": 0.9805811071604438,
7
+ "test_recall": 0.9805750169721656,
8
+ "test_runtime": 9.6545,
9
+ "test_samples_per_second": 615.882,
10
+ "test_steps_per_second": 77.063
11
+ }
test_results.json ADDED
@@ -0,0 +1,11 @@
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 10.0,
3
+ "test_accuracy": 0.9805095541401274,
4
+ "test_f1": 0.9805084530800466,
5
+ "test_loss": 0.12526775896549225,
6
+ "test_precision": 0.9806296026776948,
7
+ "test_recall": 0.980513638495975,
8
+ "test_runtime": 12.818,
9
+ "test_samples_per_second": 612.418,
10
+ "test_steps_per_second": 76.611
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 +1,4389 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 10.0,
5
+ "global_step": 176630,
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": 2.8307761988337205e-09,
13
+ "loss": 0.7015,
14
+ "step": 1
15
+ },
16
+ {
17
+ "epoch": 0.01,
18
+ "learning_rate": 7.0769404970843e-07,
19
+ "loss": 0.6985,
20
+ "step": 250
21
+ },
22
+ {
23
+ "epoch": 0.03,
24
+ "learning_rate": 1.41538809941686e-06,
25
+ "loss": 0.6961,
26
+ "step": 500
27
+ },
28
+ {
29
+ "epoch": 0.04,
30
+ "learning_rate": 2.12308214912529e-06,
31
+ "loss": 0.6965,
32
+ "step": 750
33
+ },
34
+ {
35
+ "epoch": 0.06,
36
+ "learning_rate": 2.83077619883372e-06,
37
+ "loss": 0.6961,
38
+ "step": 1000
39
+ },
40
+ {
41
+ "epoch": 0.07,
42
+ "learning_rate": 3.538470248542151e-06,
43
+ "loss": 0.6945,
44
+ "step": 1250
45
+ },
46
+ {
47
+ "epoch": 0.08,
48
+ "learning_rate": 4.24616429825058e-06,
49
+ "loss": 0.6941,
50
+ "step": 1500
51
+ },
52
+ {
53
+ "epoch": 0.1,
54
+ "learning_rate": 4.953858347959011e-06,
55
+ "loss": 0.6953,
56
+ "step": 1750
57
+ },
58
+ {
59
+ "epoch": 0.11,
60
+ "learning_rate": 5.66155239766744e-06,
61
+ "loss": 0.6938,
62
+ "step": 2000
63
+ },
64
+ {
65
+ "epoch": 0.13,
66
+ "learning_rate": 6.36924644737587e-06,
67
+ "loss": 0.6939,
68
+ "step": 2250
69
+ },
70
+ {
71
+ "epoch": 0.14,
72
+ "learning_rate": 7.076940497084302e-06,
73
+ "loss": 0.6943,
74
+ "step": 2500
75
+ },
76
+ {
77
+ "epoch": 0.16,
78
+ "learning_rate": 7.78463454679273e-06,
79
+ "loss": 0.694,
80
+ "step": 2750
81
+ },
82
+ {
83
+ "epoch": 0.17,
84
+ "learning_rate": 8.49232859650116e-06,
85
+ "loss": 0.695,
86
+ "step": 3000
87
+ },
88
+ {
89
+ "epoch": 0.18,
90
+ "learning_rate": 9.200022646209591e-06,
91
+ "loss": 0.6937,
92
+ "step": 3250
93
+ },
94
+ {
95
+ "epoch": 0.2,
96
+ "learning_rate": 9.907716695918021e-06,
97
+ "loss": 0.6944,
98
+ "step": 3500
99
+ },
100
+ {
101
+ "epoch": 0.21,
102
+ "learning_rate": 1.0615410745626452e-05,
103
+ "loss": 0.6954,
104
+ "step": 3750
105
+ },
106
+ {
107
+ "epoch": 0.23,
108
+ "learning_rate": 1.132310479533488e-05,
109
+ "loss": 0.6955,
110
+ "step": 4000
111
+ },
112
+ {
113
+ "epoch": 0.24,
114
+ "learning_rate": 1.2030798845043312e-05,
115
+ "loss": 0.6959,
116
+ "step": 4250
117
+ },
118
+ {
119
+ "epoch": 0.25,
120
+ "learning_rate": 1.273849289475174e-05,
121
+ "loss": 0.6926,
122
+ "step": 4500
123
+ },
124
+ {
125
+ "epoch": 0.27,
126
+ "learning_rate": 1.3446186944460171e-05,
127
+ "loss": 0.6946,
128
+ "step": 4750
129
+ },
130
+ {
131
+ "epoch": 0.28,
132
+ "learning_rate": 1.4153880994168603e-05,
133
+ "loss": 0.6934,
134
+ "step": 5000
135
+ },
136
+ {
137
+ "epoch": 0.3,
138
+ "learning_rate": 1.4861575043877032e-05,
139
+ "loss": 0.6956,
140
+ "step": 5250
141
+ },
142
+ {
143
+ "epoch": 0.31,
144
+ "learning_rate": 1.556926909358546e-05,
145
+ "loss": 0.6936,
146
+ "step": 5500
147
+ },
148
+ {
149
+ "epoch": 0.33,
150
+ "learning_rate": 1.627696314329389e-05,
151
+ "loss": 0.6944,
152
+ "step": 5750
153
+ },
154
+ {
155
+ "epoch": 0.34,
156
+ "learning_rate": 1.698465719300232e-05,
157
+ "loss": 0.6927,
158
+ "step": 6000
159
+ },
160
+ {
161
+ "epoch": 0.35,
162
+ "learning_rate": 1.769235124271075e-05,
163
+ "loss": 0.6959,
164
+ "step": 6250
165
+ },
166
+ {
167
+ "epoch": 0.37,
168
+ "learning_rate": 1.8400045292419182e-05,
169
+ "loss": 0.6954,
170
+ "step": 6500
171
+ },
172
+ {
173
+ "epoch": 0.38,
174
+ "learning_rate": 1.9107739342127612e-05,
175
+ "loss": 0.6931,
176
+ "step": 6750
177
+ },
178
+ {
179
+ "epoch": 0.4,
180
+ "learning_rate": 1.9815433391836042e-05,
181
+ "loss": 0.6948,
182
+ "step": 7000
183
+ },
184
+ {
185
+ "epoch": 0.41,
186
+ "learning_rate": 2.0523127441544473e-05,
187
+ "loss": 0.694,
188
+ "step": 7250
189
+ },
190
+ {
191
+ "epoch": 0.42,
192
+ "learning_rate": 2.1230821491252903e-05,
193
+ "loss": 0.6955,
194
+ "step": 7500
195
+ },
196
+ {
197
+ "epoch": 0.44,
198
+ "learning_rate": 2.1938515540961333e-05,
199
+ "loss": 0.6939,
200
+ "step": 7750
201
+ },
202
+ {
203
+ "epoch": 0.45,
204
+ "learning_rate": 2.264620959066976e-05,
205
+ "loss": 0.6944,
206
+ "step": 8000
207
+ },
208
+ {
209
+ "epoch": 0.47,
210
+ "learning_rate": 2.3353903640378194e-05,
211
+ "loss": 0.695,
212
+ "step": 8250
213
+ },
214
+ {
215
+ "epoch": 0.48,
216
+ "learning_rate": 2.4061597690086624e-05,
217
+ "loss": 0.6968,
218
+ "step": 8500
219
+ },
220
+ {
221
+ "epoch": 0.5,
222
+ "learning_rate": 2.476929173979505e-05,
223
+ "loss": 0.6958,
224
+ "step": 8750
225
+ },
226
+ {
227
+ "epoch": 0.51,
228
+ "learning_rate": 2.547698578950348e-05,
229
+ "loss": 0.6948,
230
+ "step": 9000
231
+ },
232
+ {
233
+ "epoch": 0.52,
234
+ "learning_rate": 2.6184679839211912e-05,
235
+ "loss": 0.6948,
236
+ "step": 9250
237
+ },
238
+ {
239
+ "epoch": 0.54,
240
+ "learning_rate": 2.6892373888920342e-05,
241
+ "loss": 0.6965,
242
+ "step": 9500
243
+ },
244
+ {
245
+ "epoch": 0.55,
246
+ "learning_rate": 2.760006793862877e-05,
247
+ "loss": 0.6935,
248
+ "step": 9750
249
+ },
250
+ {
251
+ "epoch": 0.57,
252
+ "learning_rate": 2.8307761988337206e-05,
253
+ "loss": 0.6949,
254
+ "step": 10000
255
+ },
256
+ {
257
+ "epoch": 0.58,
258
+ "learning_rate": 2.9015456038045637e-05,
259
+ "loss": 0.6949,
260
+ "step": 10250
261
+ },
262
+ {
263
+ "epoch": 0.59,
264
+ "learning_rate": 2.9723150087754064e-05,
265
+ "loss": 0.6946,
266
+ "step": 10500
267
+ },
268
+ {
269
+ "epoch": 0.61,
270
+ "learning_rate": 3.0430844137462494e-05,
271
+ "loss": 0.6944,
272
+ "step": 10750
273
+ },
274
+ {
275
+ "epoch": 0.62,
276
+ "learning_rate": 3.113853818717092e-05,
277
+ "loss": 0.694,
278
+ "step": 11000
279
+ },
280
+ {
281
+ "epoch": 0.64,
282
+ "learning_rate": 3.184623223687935e-05,
283
+ "loss": 0.6934,
284
+ "step": 11250
285
+ },
286
+ {
287
+ "epoch": 0.65,
288
+ "learning_rate": 3.255109551038895e-05,
289
+ "loss": 0.6955,
290
+ "step": 11500
291
+ },
292
+ {
293
+ "epoch": 0.67,
294
+ "learning_rate": 3.325878956009738e-05,
295
+ "loss": 0.6953,
296
+ "step": 11750
297
+ },
298
+ {
299
+ "epoch": 0.68,
300
+ "learning_rate": 3.396648360980581e-05,
301
+ "loss": 0.6957,
302
+ "step": 12000
303
+ },
304
+ {
305
+ "epoch": 0.69,
306
+ "learning_rate": 3.4674177659514237e-05,
307
+ "loss": 0.6939,
308
+ "step": 12250
309
+ },
310
+ {
311
+ "epoch": 0.71,
312
+ "learning_rate": 3.538187170922267e-05,
313
+ "loss": 0.6956,
314
+ "step": 12500
315
+ },
316
+ {
317
+ "epoch": 0.72,
318
+ "learning_rate": 3.60895657589311e-05,
319
+ "loss": 0.6948,
320
+ "step": 12750
321
+ },
322
+ {
323
+ "epoch": 0.74,
324
+ "learning_rate": 3.679725980863953e-05,
325
+ "loss": 0.6936,
326
+ "step": 13000
327
+ },
328
+ {
329
+ "epoch": 0.75,
330
+ "learning_rate": 3.7504953858347965e-05,
331
+ "loss": 0.6949,
332
+ "step": 13250
333
+ },
334
+ {
335
+ "epoch": 0.76,
336
+ "learning_rate": 3.820981713185756e-05,
337
+ "loss": 0.6941,
338
+ "step": 13500
339
+ },
340
+ {
341
+ "epoch": 0.78,
342
+ "learning_rate": 3.891751118156598e-05,
343
+ "loss": 0.6941,
344
+ "step": 13750
345
+ },
346
+ {
347
+ "epoch": 0.79,
348
+ "learning_rate": 3.962520523127441e-05,
349
+ "loss": 0.695,
350
+ "step": 14000
351
+ },
352
+ {
353
+ "epoch": 0.81,
354
+ "learning_rate": 4.033289928098285e-05,
355
+ "loss": 0.6957,
356
+ "step": 14250
357
+ },
358
+ {
359
+ "epoch": 0.82,
360
+ "learning_rate": 4.104059333069128e-05,
361
+ "loss": 0.6949,
362
+ "step": 14500
363
+ },
364
+ {
365
+ "epoch": 0.84,
366
+ "learning_rate": 4.174828738039971e-05,
367
+ "loss": 0.6945,
368
+ "step": 14750
369
+ },
370
+ {
371
+ "epoch": 0.85,
372
+ "learning_rate": 4.245598143010814e-05,
373
+ "loss": 0.6943,
374
+ "step": 15000
375
+ },
376
+ {
377
+ "epoch": 0.86,
378
+ "learning_rate": 4.3163675479816565e-05,
379
+ "loss": 0.6952,
380
+ "step": 15250
381
+ },
382
+ {
383
+ "epoch": 0.88,
384
+ "learning_rate": 4.3871369529524995e-05,
385
+ "loss": 0.6939,
386
+ "step": 15500
387
+ },
388
+ {
389
+ "epoch": 0.89,
390
+ "learning_rate": 4.4579063579233425e-05,
391
+ "loss": 0.6952,
392
+ "step": 15750
393
+ },
394
+ {
395
+ "epoch": 0.91,
396
+ "learning_rate": 4.5283926852743026e-05,
397
+ "loss": 0.6946,
398
+ "step": 16000
399
+ },
400
+ {
401
+ "epoch": 0.92,
402
+ "learning_rate": 4.599162090245146e-05,
403
+ "loss": 0.6939,
404
+ "step": 16250
405
+ },
406
+ {
407
+ "epoch": 0.93,
408
+ "learning_rate": 4.669931495215989e-05,
409
+ "loss": 0.695,
410
+ "step": 16500
411
+ },
412
+ {
413
+ "epoch": 0.95,
414
+ "learning_rate": 4.740700900186831e-05,
415
+ "loss": 0.6945,
416
+ "step": 16750
417
+ },
418
+ {
419
+ "epoch": 0.96,
420
+ "learning_rate": 4.811470305157674e-05,
421
+ "loss": 0.6938,
422
+ "step": 17000
423
+ },
424
+ {
425
+ "epoch": 0.98,
426
+ "learning_rate": 4.882239710128517e-05,
427
+ "loss": 0.6946,
428
+ "step": 17250
429
+ },
430
+ {
431
+ "epoch": 0.99,
432
+ "learning_rate": 4.953009115099361e-05,
433
+ "loss": 0.6943,
434
+ "step": 17500
435
+ },
436
+ {
437
+ "epoch": 1.0,
438
+ "eval_accuracy": 0.5002547770700637,
439
+ "eval_f1": 0.33344654835696697,
440
+ "eval_loss": 0.6931106448173523,
441
+ "eval_precision": 0.25012738853503186,
442
+ "eval_recall": 0.5,
443
+ "eval_runtime": 12.7276,
444
+ "eval_samples_per_second": 616.771,
445
+ "eval_steps_per_second": 77.155,
446
+ "step": 17663
447
+ },
448
+ {
449
+ "epoch": 1.0,
450
+ "learning_rate": 4.999996555277244e-05,
451
+ "loss": 0.6939,
452
+ "step": 17750
453
+ },
454
+ {
455
+ "epoch": 1.02,
456
+ "learning_rate": 4.999945538806009e-05,
457
+ "loss": 0.6931,
458
+ "step": 18000
459
+ },
460
+ {
461
+ "epoch": 1.03,
462
+ "learning_rate": 4.999834068779779e-05,
463
+ "loss": 0.6939,
464
+ "step": 18250
465
+ },
466
+ {
467
+ "epoch": 1.05,
468
+ "learning_rate": 4.9996612525797716e-05,
469
+ "loss": 0.6938,
470
+ "step": 18500
471
+ },
472
+ {
473
+ "epoch": 1.06,
474
+ "learning_rate": 4.999427420049964e-05,
475
+ "loss": 0.6948,
476
+ "step": 18750
477
+ },
478
+ {
479
+ "epoch": 1.08,
480
+ "learning_rate": 4.999132576898172e-05,
481
+ "loss": 0.6935,
482
+ "step": 19000
483
+ },
484
+ {
485
+ "epoch": 1.09,
486
+ "learning_rate": 4.998776730321471e-05,
487
+ "loss": 0.6945,
488
+ "step": 19250
489
+ },
490
+ {
491
+ "epoch": 1.1,
492
+ "learning_rate": 4.9983598890060156e-05,
493
+ "loss": 0.6949,
494
+ "step": 19500
495
+ },
496
+ {
497
+ "epoch": 1.12,
498
+ "learning_rate": 4.997882063126838e-05,
499
+ "loss": 0.6947,
500
+ "step": 19750
501
+ },
502
+ {
503
+ "epoch": 1.13,
504
+ "learning_rate": 4.997343264347589e-05,
505
+ "loss": 0.6935,
506
+ "step": 20000
507
+ },
508
+ {
509
+ "epoch": 1.15,
510
+ "learning_rate": 4.996743505820262e-05,
511
+ "loss": 0.6941,
512
+ "step": 20250
513
+ },
514
+ {
515
+ "epoch": 1.16,
516
+ "learning_rate": 4.996082802184866e-05,
517
+ "loss": 0.6947,
518
+ "step": 20500
519
+ },
520
+ {
521
+ "epoch": 1.17,
522
+ "learning_rate": 4.9953641774472115e-05,
523
+ "loss": 0.6937,
524
+ "step": 20750
525
+ },
526
+ {
527
+ "epoch": 1.19,
528
+ "learning_rate": 4.9945818770743544e-05,
529
+ "loss": 0.6958,
530
+ "step": 21000
531
+ },
532
+ {
533
+ "epoch": 1.2,
534
+ "learning_rate": 4.99373868435844e-05,
535
+ "loss": 0.6946,
536
+ "step": 21250
537
+ },
538
+ {
539
+ "epoch": 1.22,
540
+ "learning_rate": 4.992834619881665e-05,
541
+ "loss": 0.6937,
542
+ "step": 21500
543
+ },
544
+ {
545
+ "epoch": 1.23,
546
+ "learning_rate": 4.991869705712099e-05,
547
+ "loss": 0.6952,
548
+ "step": 21750
549
+ },
550
+ {
551
+ "epoch": 1.25,
552
+ "learning_rate": 4.990843965403141e-05,
553
+ "loss": 0.6938,
554
+ "step": 22000
555
+ },
556
+ {
557
+ "epoch": 1.26,
558
+ "learning_rate": 4.989757423992949e-05,
559
+ "loss": 0.6927,
560
+ "step": 22250
561
+ },
562
+ {
563
+ "epoch": 1.27,
564
+ "learning_rate": 4.9886101080038236e-05,
565
+ "loss": 0.6954,
566
+ "step": 22500
567
+ },
568
+ {
569
+ "epoch": 1.29,
570
+ "learning_rate": 4.987402045441564e-05,
571
+ "loss": 0.6933,
572
+ "step": 22750
573
+ },
574
+ {
575
+ "epoch": 1.3,
576
+ "learning_rate": 4.986138461821248e-05,
577
+ "loss": 0.6934,
578
+ "step": 23000
579
+ },
580
+ {
581
+ "epoch": 1.32,
582
+ "learning_rate": 4.9848092387414506e-05,
583
+ "loss": 0.6949,
584
+ "step": 23250
585
+ },
586
+ {
587
+ "epoch": 1.33,
588
+ "learning_rate": 4.983419361867128e-05,
589
+ "loss": 0.6947,
590
+ "step": 23500
591
+ },
592
+ {
593
+ "epoch": 1.34,
594
+ "learning_rate": 4.9819688651249495e-05,
595
+ "loss": 0.6951,
596
+ "step": 23750
597
+ },
598
+ {
599
+ "epoch": 1.36,
600
+ "learning_rate": 4.9804577839213096e-05,
601
+ "loss": 0.694,
602
+ "step": 24000
603
+ },
604
+ {
605
+ "epoch": 1.37,
606
+ "learning_rate": 4.978886155141456e-05,
607
+ "loss": 0.6941,
608
+ "step": 24250
609
+ },
610
+ {
611
+ "epoch": 1.39,
612
+ "learning_rate": 4.977254017148597e-05,
613
+ "loss": 0.6942,
614
+ "step": 24500
615
+ },
616
+ {
617
+ "epoch": 1.4,
618
+ "learning_rate": 4.9755614097829575e-05,
619
+ "loss": 0.6931,
620
+ "step": 24750
621
+ },
622
+ {
623
+ "epoch": 1.42,
624
+ "learning_rate": 4.9738083743608114e-05,
625
+ "loss": 0.6939,
626
+ "step": 25000
627
+ },
628
+ {
629
+ "epoch": 1.43,
630
+ "learning_rate": 4.9719949536734725e-05,
631
+ "loss": 0.6946,
632
+ "step": 25250
633
+ },
634
+ {
635
+ "epoch": 1.44,
636
+ "learning_rate": 4.97012880717214e-05,
637
+ "loss": 0.695,
638
+ "step": 25500
639
+ },
640
+ {
641
+ "epoch": 1.46,
642
+ "learning_rate": 4.968194991311238e-05,
643
+ "loss": 0.6935,
644
+ "step": 25750
645
+ },
646
+ {
647
+ "epoch": 1.47,
648
+ "learning_rate": 4.9662009272069223e-05,
649
+ "loss": 0.6946,
650
+ "step": 26000
651
+ },
652
+ {
653
+ "epoch": 1.49,
654
+ "learning_rate": 4.964146663533976e-05,
655
+ "loss": 0.6943,
656
+ "step": 26250
657
+ },
658
+ {
659
+ "epoch": 1.5,
660
+ "learning_rate": 4.962032250436647e-05,
661
+ "loss": 0.6942,
662
+ "step": 26500
663
+ },
664
+ {
665
+ "epoch": 1.51,
666
+ "learning_rate": 4.959857739527419e-05,
667
+ "loss": 0.6938,
668
+ "step": 26750
669
+ },
670
+ {
671
+ "epoch": 1.53,
672
+ "learning_rate": 4.957623183885755e-05,
673
+ "loss": 0.6942,
674
+ "step": 27000
675
+ },
676
+ {
677
+ "epoch": 1.54,
678
+ "learning_rate": 4.9553286380567973e-05,
679
+ "loss": 0.6947,
680
+ "step": 27250
681
+ },
682
+ {
683
+ "epoch": 1.56,
684
+ "learning_rate": 4.952983695283269e-05,
685
+ "loss": 0.6946,
686
+ "step": 27500
687
+ },
688
+ {
689
+ "epoch": 1.57,
690
+ "learning_rate": 4.9505695779615944e-05,
691
+ "loss": 0.6945,
692
+ "step": 27750
693
+ },
694
+ {
695
+ "epoch": 1.59,
696
+ "learning_rate": 4.9480956426300165e-05,
697
+ "loss": 0.6937,
698
+ "step": 28000
699
+ },
700
+ {
701
+ "epoch": 1.6,
702
+ "learning_rate": 4.945561949676898e-05,
703
+ "loss": 0.6935,
704
+ "step": 28250
705
+ },
706
+ {
707
+ "epoch": 1.61,
708
+ "learning_rate": 4.9429685609492773e-05,
709
+ "loss": 0.6943,
710
+ "step": 28500
711
+ },
712
+ {
713
+ "epoch": 1.63,
714
+ "learning_rate": 4.940315539751357e-05,
715
+ "loss": 0.6942,
716
+ "step": 28750
717
+ },
718
+ {
719
+ "epoch": 1.64,
720
+ "learning_rate": 4.9376029508429584e-05,
721
+ "loss": 0.6938,
722
+ "step": 29000
723
+ },
724
+ {
725
+ "epoch": 1.66,
726
+ "learning_rate": 4.934830860437941e-05,
727
+ "loss": 0.6941,
728
+ "step": 29250
729
+ },
730
+ {
731
+ "epoch": 1.67,
732
+ "learning_rate": 4.932010780600598e-05,
733
+ "loss": 0.6943,
734
+ "step": 29500
735
+ },
736
+ {
737
+ "epoch": 1.68,
738
+ "learning_rate": 4.92912012897121e-05,
739
+ "loss": 0.694,
740
+ "step": 29750
741
+ },
742
+ {
743
+ "epoch": 1.7,
744
+ "learning_rate": 4.926170182909523e-05,
745
+ "loss": 0.6943,
746
+ "step": 30000
747
+ },
748
+ {
749
+ "epoch": 1.71,
750
+ "learning_rate": 4.923161014423248e-05,
751
+ "loss": 0.6941,
752
+ "step": 30250
753
+ },
754
+ {
755
+ "epoch": 1.73,
756
+ "learning_rate": 4.920092696965702e-05,
757
+ "loss": 0.6947,
758
+ "step": 30500
759
+ },
760
+ {
761
+ "epoch": 1.74,
762
+ "learning_rate": 4.916965305434024e-05,
763
+ "loss": 0.6942,
764
+ "step": 30750
765
+ },
766
+ {
767
+ "epoch": 1.76,
768
+ "learning_rate": 4.913778916167339e-05,
769
+ "loss": 0.6941,
770
+ "step": 31000
771
+ },
772
+ {
773
+ "epoch": 1.77,
774
+ "learning_rate": 4.910533606944895e-05,
775
+ "loss": 0.6942,
776
+ "step": 31250
777
+ },
778
+ {
779
+ "epoch": 1.78,
780
+ "learning_rate": 4.907242790688348e-05,
781
+ "loss": 0.6942,
782
+ "step": 31500
783
+ },
784
+ {
785
+ "epoch": 1.8,
786
+ "learning_rate": 4.903880115520887e-05,
787
+ "loss": 0.6939,
788
+ "step": 31750
789
+ },
790
+ {
791
+ "epoch": 1.81,
792
+ "learning_rate": 4.900458762025803e-05,
793
+ "loss": 0.6933,
794
+ "step": 32000
795
+ },
796
+ {
797
+ "epoch": 1.83,
798
+ "learning_rate": 4.8969788137177854e-05,
799
+ "loss": 0.6943,
800
+ "step": 32250
801
+ },
802
+ {
803
+ "epoch": 1.84,
804
+ "learning_rate": 4.893440355541813e-05,
805
+ "loss": 0.6938,
806
+ "step": 32500
807
+ },
808
+ {
809
+ "epoch": 1.85,
810
+ "learning_rate": 4.889843473871082e-05,
811
+ "loss": 0.6947,
812
+ "step": 32750
813
+ },
814
+ {
815
+ "epoch": 1.87,
816
+ "learning_rate": 4.8861882565048975e-05,
817
+ "loss": 0.6939,
818
+ "step": 33000
819
+ },
820
+ {
821
+ "epoch": 1.88,
822
+ "learning_rate": 4.882474792666527e-05,
823
+ "loss": 0.694,
824
+ "step": 33250
825
+ },
826
+ {
827
+ "epoch": 1.9,
828
+ "learning_rate": 4.878703173001027e-05,
829
+ "loss": 0.693,
830
+ "step": 33500
831
+ },
832
+ {
833
+ "epoch": 1.91,
834
+ "learning_rate": 4.87488892384633e-05,
835
+ "loss": 0.6939,
836
+ "step": 33750
837
+ },
838
+ {
839
+ "epoch": 1.92,
840
+ "learning_rate": 4.871001501830819e-05,
841
+ "loss": 0.6947,
842
+ "step": 34000
843
+ },
844
+ {
845
+ "epoch": 1.94,
846
+ "learning_rate": 4.867056204049366e-05,
847
+ "loss": 0.6938,
848
+ "step": 34250
849
+ },
850
+ {
851
+ "epoch": 1.95,
852
+ "learning_rate": 4.863053126806055e-05,
853
+ "loss": 0.6932,
854
+ "step": 34500
855
+ },
856
+ {
857
+ "epoch": 1.97,
858
+ "learning_rate": 4.85899236781536e-05,
859
+ "loss": 0.6932,
860
+ "step": 34750
861
+ },
862
+ {
863
+ "epoch": 1.98,
864
+ "learning_rate": 4.854874026199756e-05,
865
+ "loss": 0.694,
866
+ "step": 35000
867
+ },
868
+ {
869
+ "epoch": 2.0,
870
+ "learning_rate": 4.8506982024873006e-05,
871
+ "loss": 0.6941,
872
+ "step": 35250
873
+ },
874
+ {
875
+ "epoch": 2.0,
876
+ "eval_accuracy": 0.4997452229299363,
877
+ "eval_f1": 0.3332200798437102,
878
+ "eval_loss": 0.6948533058166504,
879
+ "eval_precision": 0.24987261146496814,
880
+ "eval_recall": 0.5,
881
+ "eval_runtime": 12.6064,
882
+ "eval_samples_per_second": 622.697,
883
+ "eval_steps_per_second": 77.897,
884
+ "step": 35326
885
+ },
886
+ {
887
+ "epoch": 2.01,
888
+ "learning_rate": 4.846464998609178e-05,
889
+ "loss": 0.6946,
890
+ "step": 35500
891
+ },
892
+ {
893
+ "epoch": 2.02,
894
+ "learning_rate": 4.842174517897218e-05,
895
+ "loss": 0.6937,
896
+ "step": 35750
897
+ },
898
+ {
899
+ "epoch": 2.04,
900
+ "learning_rate": 4.8378443694392686e-05,
901
+ "loss": 0.6938,
902
+ "step": 36000
903
+ },
904
+ {
905
+ "epoch": 2.05,
906
+ "learning_rate": 4.833439878695692e-05,
907
+ "loss": 0.6936,
908
+ "step": 36250
909
+ },
910
+ {
911
+ "epoch": 2.07,
912
+ "learning_rate": 4.8289784290593636e-05,
913
+ "loss": 0.6934,
914
+ "step": 36500
915
+ },
916
+ {
917
+ "epoch": 2.08,
918
+ "learning_rate": 4.824460129433551e-05,
919
+ "loss": 0.6939,
920
+ "step": 36750
921
+ },
922
+ {
923
+ "epoch": 2.09,
924
+ "learning_rate": 4.819885090109222e-05,
925
+ "loss": 0.6936,
926
+ "step": 37000
927
+ },
928
+ {
929
+ "epoch": 2.11,
930
+ "learning_rate": 4.815253422762353e-05,
931
+ "loss": 0.6931,
932
+ "step": 37250
933
+ },
934
+ {
935
+ "epoch": 2.12,
936
+ "learning_rate": 4.8105652404512013e-05,
937
+ "loss": 0.6941,
938
+ "step": 37500
939
+ },
940
+ {
941
+ "epoch": 2.14,
942
+ "learning_rate": 4.8058206576135415e-05,
943
+ "loss": 0.6935,
944
+ "step": 37750
945
+ },
946
+ {
947
+ "epoch": 2.15,
948
+ "learning_rate": 4.801039105498368e-05,
949
+ "loss": 0.694,
950
+ "step": 38000
951
+ },
952
+ {
953
+ "epoch": 2.17,
954
+ "learning_rate": 4.796182294859949e-05,
955
+ "loss": 0.6936,
956
+ "step": 38250
957
+ },
958
+ {
959
+ "epoch": 2.18,
960
+ "learning_rate": 4.791269434780415e-05,
961
+ "loss": 0.6933,
962
+ "step": 38500
963
+ },
964
+ {
965
+ "epoch": 2.19,
966
+ "learning_rate": 4.7863006451818936e-05,
967
+ "loss": 0.6936,
968
+ "step": 38750
969
+ },
970
+ {
971
+ "epoch": 2.21,
972
+ "learning_rate": 4.781276047351739e-05,
973
+ "loss": 0.6932,
974
+ "step": 39000
975
+ },
976
+ {
977
+ "epoch": 2.22,
978
+ "learning_rate": 4.7761957639395794e-05,
979
+ "loss": 0.6943,
980
+ "step": 39250
981
+ },
982
+ {
983
+ "epoch": 2.24,
984
+ "learning_rate": 4.771059918954316e-05,
985
+ "loss": 0.6933,
986
+ "step": 39500
987
+ },
988
+ {
989
+ "epoch": 2.25,
990
+ "learning_rate": 4.7658686377610994e-05,
991
+ "loss": 0.6943,
992
+ "step": 39750
993
+ },
994
+ {
995
+ "epoch": 2.26,
996
+ "learning_rate": 4.7606220470782714e-05,
997
+ "loss": 0.6942,
998
+ "step": 40000
999
+ },
1000
+ {
1001
+ "epoch": 2.28,
1002
+ "learning_rate": 4.755341591813002e-05,
1003
+ "loss": 0.6933,
1004
+ "step": 40250
1005
+ },
1006
+ {
1007
+ "epoch": 2.29,
1008
+ "learning_rate": 4.749984987651673e-05,
1009
+ "loss": 0.6943,
1010
+ "step": 40500
1011
+ },
1012
+ {
1013
+ "epoch": 2.31,
1014
+ "learning_rate": 4.744573461718082e-05,
1015
+ "loss": 0.6934,
1016
+ "step": 40750
1017
+ },
1018
+ {
1019
+ "epoch": 2.32,
1020
+ "learning_rate": 4.739107146106705e-05,
1021
+ "loss": 0.6933,
1022
+ "step": 41000
1023
+ },
1024
+ {
1025
+ "epoch": 2.34,
1026
+ "learning_rate": 4.733586174249431e-05,
1027
+ "loss": 0.6938,
1028
+ "step": 41250
1029
+ },
1030
+ {
1031
+ "epoch": 2.35,
1032
+ "learning_rate": 4.728010680912296e-05,
1033
+ "loss": 0.6937,
1034
+ "step": 41500
1035
+ },
1036
+ {
1037
+ "epoch": 2.36,
1038
+ "learning_rate": 4.722380802192197e-05,
1039
+ "loss": 0.6935,
1040
+ "step": 41750
1041
+ },
1042
+ {
1043
+ "epoch": 2.38,
1044
+ "learning_rate": 4.716696675513569e-05,
1045
+ "loss": 0.694,
1046
+ "step": 42000
1047
+ },
1048
+ {
1049
+ "epoch": 2.39,
1050
+ "learning_rate": 4.710958439625032e-05,
1051
+ "loss": 0.6945,
1052
+ "step": 42250
1053
+ },
1054
+ {
1055
+ "epoch": 2.41,
1056
+ "learning_rate": 4.7051895107356976e-05,
1057
+ "loss": 0.694,
1058
+ "step": 42500
1059
+ },
1060
+ {
1061
+ "epoch": 2.42,
1062
+ "learning_rate": 4.6993436929806e-05,
1063
+ "loss": 0.694,
1064
+ "step": 42750
1065
+ },
1066
+ {
1067
+ "epoch": 2.43,
1068
+ "learning_rate": 4.69344418959911e-05,
1069
+ "loss": 0.6934,
1070
+ "step": 43000
1071
+ },
1072
+ {
1073
+ "epoch": 2.45,
1074
+ "learning_rate": 4.687491144597158e-05,
1075
+ "loss": 0.6942,
1076
+ "step": 43250
1077
+ },
1078
+ {
1079
+ "epoch": 2.46,
1080
+ "learning_rate": 4.6814847032876174e-05,
1081
+ "loss": 0.6937,
1082
+ "step": 43500
1083
+ },
1084
+ {
1085
+ "epoch": 2.48,
1086
+ "learning_rate": 4.675425012286756e-05,
1087
+ "loss": 0.6932,
1088
+ "step": 43750
1089
+ },
1090
+ {
1091
+ "epoch": 2.49,
1092
+ "learning_rate": 4.6693122195106574e-05,
1093
+ "loss": 0.6942,
1094
+ "step": 44000
1095
+ },
1096
+ {
1097
+ "epoch": 2.51,
1098
+ "learning_rate": 4.663146474171613e-05,
1099
+ "loss": 0.6933,
1100
+ "step": 44250
1101
+ },
1102
+ {
1103
+ "epoch": 2.52,
1104
+ "learning_rate": 4.656952905945011e-05,
1105
+ "loss": 0.694,
1106
+ "step": 44500
1107
+ },
1108
+ {
1109
+ "epoch": 2.53,
1110
+ "learning_rate": 4.650681918580503e-05,
1111
+ "loss": 0.6939,
1112
+ "step": 44750
1113
+ },
1114
+ {
1115
+ "epoch": 2.55,
1116
+ "learning_rate": 4.644358433415702e-05,
1117
+ "loss": 0.6935,
1118
+ "step": 45000
1119
+ },
1120
+ {
1121
+ "epoch": 2.56,
1122
+ "learning_rate": 4.6379826048058654e-05,
1123
+ "loss": 0.6942,
1124
+ "step": 45250
1125
+ },
1126
+ {
1127
+ "epoch": 2.58,
1128
+ "learning_rate": 4.6315545883839474e-05,
1129
+ "loss": 0.6939,
1130
+ "step": 45500
1131
+ },
1132
+ {
1133
+ "epoch": 2.59,
1134
+ "learning_rate": 4.625074541056795e-05,
1135
+ "loss": 0.693,
1136
+ "step": 45750
1137
+ },
1138
+ {
1139
+ "epoch": 2.6,
1140
+ "learning_rate": 4.618542621001324e-05,
1141
+ "loss": 0.6941,
1142
+ "step": 46000
1143
+ },
1144
+ {
1145
+ "epoch": 2.62,
1146
+ "learning_rate": 4.611958987660653e-05,
1147
+ "loss": 0.6938,
1148
+ "step": 46250
1149
+ },
1150
+ {
1151
+ "epoch": 2.63,
1152
+ "learning_rate": 4.6053238017402155e-05,
1153
+ "loss": 0.6932,
1154
+ "step": 46500
1155
+ },
1156
+ {
1157
+ "epoch": 2.65,
1158
+ "learning_rate": 4.598664073664183e-05,
1159
+ "loss": 0.6938,
1160
+ "step": 46750
1161
+ },
1162
+ {
1163
+ "epoch": 2.66,
1164
+ "learning_rate": 4.591926474312919e-05,
1165
+ "loss": 0.694,
1166
+ "step": 47000
1167
+ },
1168
+ {
1169
+ "epoch": 2.68,
1170
+ "learning_rate": 4.585137811372323e-05,
1171
+ "loss": 0.6938,
1172
+ "step": 47250
1173
+ },
1174
+ {
1175
+ "epoch": 2.69,
1176
+ "learning_rate": 4.5782982505525664e-05,
1177
+ "loss": 0.6936,
1178
+ "step": 47500
1179
+ },
1180
+ {
1181
+ "epoch": 2.7,
1182
+ "learning_rate": 4.5714079588062306e-05,
1183
+ "loss": 0.6938,
1184
+ "step": 47750
1185
+ },
1186
+ {
1187
+ "epoch": 2.72,
1188
+ "learning_rate": 4.5644671043242314e-05,
1189
+ "loss": 0.6938,
1190
+ "step": 48000
1191
+ },
1192
+ {
1193
+ "epoch": 2.73,
1194
+ "learning_rate": 4.557475856531712e-05,
1195
+ "loss": 0.6943,
1196
+ "step": 48250
1197
+ },
1198
+ {
1199
+ "epoch": 2.75,
1200
+ "learning_rate": 4.550434386083909e-05,
1201
+ "loss": 0.693,
1202
+ "step": 48500
1203
+ },
1204
+ {
1205
+ "epoch": 2.76,
1206
+ "learning_rate": 4.543371330418998e-05,
1207
+ "loss": 0.6936,
1208
+ "step": 48750
1209
+ },
1210
+ {
1211
+ "epoch": 2.77,
1212
+ "learning_rate": 4.5362301306900956e-05,
1213
+ "loss": 0.6924,
1214
+ "step": 49000
1215
+ },
1216
+ {
1217
+ "epoch": 2.79,
1218
+ "learning_rate": 4.5290392269106644e-05,
1219
+ "loss": 0.6936,
1220
+ "step": 49250
1221
+ },
1222
+ {
1223
+ "epoch": 2.8,
1224
+ "learning_rate": 4.521798794609512e-05,
1225
+ "loss": 0.6941,
1226
+ "step": 49500
1227
+ },
1228
+ {
1229
+ "epoch": 2.82,
1230
+ "learning_rate": 4.514509010524428e-05,
1231
+ "loss": 0.693,
1232
+ "step": 49750
1233
+ },
1234
+ {
1235
+ "epoch": 2.83,
1236
+ "learning_rate": 4.507170052597872e-05,
1237
+ "loss": 0.6943,
1238
+ "step": 50000
1239
+ },
1240
+ {
1241
+ "epoch": 2.84,
1242
+ "learning_rate": 4.499782099972628e-05,
1243
+ "loss": 0.6936,
1244
+ "step": 50250
1245
+ },
1246
+ {
1247
+ "epoch": 2.86,
1248
+ "learning_rate": 4.4923453329874334e-05,
1249
+ "loss": 0.6938,
1250
+ "step": 50500
1251
+ },
1252
+ {
1253
+ "epoch": 2.87,
1254
+ "learning_rate": 4.484859933172575e-05,
1255
+ "loss": 0.694,
1256
+ "step": 50750
1257
+ },
1258
+ {
1259
+ "epoch": 2.89,
1260
+ "learning_rate": 4.477356314914456e-05,
1261
+ "loss": 0.6938,
1262
+ "step": 51000
1263
+ },
1264
+ {
1265
+ "epoch": 2.9,
1266
+ "learning_rate": 4.469774391472113e-05,
1267
+ "loss": 0.6933,
1268
+ "step": 51250
1269
+ },
1270
+ {
1271
+ "epoch": 2.92,
1272
+ "learning_rate": 4.4621443861531634e-05,
1273
+ "loss": 0.6936,
1274
+ "step": 51500
1275
+ },
1276
+ {
1277
+ "epoch": 2.93,
1278
+ "learning_rate": 4.4544664852048143e-05,
1279
+ "loss": 0.6932,
1280
+ "step": 51750
1281
+ },
1282
+ {
1283
+ "epoch": 2.94,
1284
+ "learning_rate": 4.446740876043395e-05,
1285
+ "loss": 0.693,
1286
+ "step": 52000
1287
+ },
1288
+ {
1289
+ "epoch": 2.96,
1290
+ "learning_rate": 4.438967747249785e-05,
1291
+ "loss": 0.6944,
1292
+ "step": 52250
1293
+ },
1294
+ {
1295
+ "epoch": 2.97,
1296
+ "learning_rate": 4.4311472885648076e-05,
1297
+ "loss": 0.694,
1298
+ "step": 52500
1299
+ },
1300
+ {
1301
+ "epoch": 2.99,
1302
+ "learning_rate": 4.423279690884606e-05,
1303
+ "loss": 0.6931,
1304
+ "step": 52750
1305
+ },
1306
+ {
1307
+ "epoch": 3.0,
1308
+ "eval_accuracy": 0.4997452229299363,
1309
+ "eval_f1": 0.3332200798437102,
1310
+ "eval_loss": 0.6931188702583313,
1311
+ "eval_precision": 0.24987261146496814,
1312
+ "eval_recall": 0.5,
1313
+ "eval_runtime": 12.4472,
1314
+ "eval_samples_per_second": 630.665,
1315
+ "eval_steps_per_second": 78.893,
1316
+ "step": 52989
1317
+ },
1318
+ {
1319
+ "epoch": 3.0,
1320
+ "learning_rate": 4.4153651462559756e-05,
1321
+ "loss": 0.6939,
1322
+ "step": 53000
1323
+ },
1324
+ {
1325
+ "epoch": 3.01,
1326
+ "learning_rate": 4.407403847871679e-05,
1327
+ "loss": 0.6935,
1328
+ "step": 53250
1329
+ },
1330
+ {
1331
+ "epoch": 3.03,
1332
+ "learning_rate": 4.399395990065732e-05,
1333
+ "loss": 0.693,
1334
+ "step": 53500
1335
+ },
1336
+ {
1337
+ "epoch": 3.04,
1338
+ "learning_rate": 4.3913740772924885e-05,
1339
+ "loss": 0.6932,
1340
+ "step": 53750
1341
+ },
1342
+ {
1343
+ "epoch": 3.06,
1344
+ "learning_rate": 4.383273872462815e-05,
1345
+ "loss": 0.6931,
1346
+ "step": 54000
1347
+ },
1348
+ {
1349
+ "epoch": 3.07,
1350
+ "learning_rate": 4.3751276972203183e-05,
1351
+ "loss": 0.6941,
1352
+ "step": 54250
1353
+ },
1354
+ {
1355
+ "epoch": 3.09,
1356
+ "learning_rate": 4.366935750411825e-05,
1357
+ "loss": 0.6938,
1358
+ "step": 54500
1359
+ },
1360
+ {
1361
+ "epoch": 3.1,
1362
+ "learning_rate": 4.3586982320014426e-05,
1363
+ "loss": 0.6942,
1364
+ "step": 54750
1365
+ },
1366
+ {
1367
+ "epoch": 3.11,
1368
+ "learning_rate": 4.3504153430656725e-05,
1369
+ "loss": 0.694,
1370
+ "step": 55000
1371
+ },
1372
+ {
1373
+ "epoch": 3.13,
1374
+ "learning_rate": 4.3420872857885045e-05,
1375
+ "loss": 0.693,
1376
+ "step": 55250
1377
+ },
1378
+ {
1379
+ "epoch": 3.14,
1380
+ "learning_rate": 4.333714263456478e-05,
1381
+ "loss": 0.6935,
1382
+ "step": 55500
1383
+ },
1384
+ {
1385
+ "epoch": 3.16,
1386
+ "learning_rate": 4.325330240477026e-05,
1387
+ "loss": 0.693,
1388
+ "step": 55750
1389
+ },
1390
+ {
1391
+ "epoch": 3.17,
1392
+ "learning_rate": 4.3168680800902995e-05,
1393
+ "loss": 0.6936,
1394
+ "step": 56000
1395
+ },
1396
+ {
1397
+ "epoch": 3.18,
1398
+ "learning_rate": 4.308361570245477e-05,
1399
+ "loss": 0.6926,
1400
+ "step": 56250
1401
+ },
1402
+ {
1403
+ "epoch": 3.2,
1404
+ "learning_rate": 4.299810918585097e-05,
1405
+ "loss": 0.6945,
1406
+ "step": 56500
1407
+ },
1408
+ {
1409
+ "epoch": 3.21,
1410
+ "learning_rate": 4.2912163338291946e-05,
1411
+ "loss": 0.6929,
1412
+ "step": 56750
1413
+ },
1414
+ {
1415
+ "epoch": 3.23,
1416
+ "learning_rate": 4.2825780257702033e-05,
1417
+ "loss": 0.6947,
1418
+ "step": 57000
1419
+ },
1420
+ {
1421
+ "epoch": 3.24,
1422
+ "learning_rate": 4.2738962052678356e-05,
1423
+ "loss": 0.6936,
1424
+ "step": 57250
1425
+ },
1426
+ {
1427
+ "epoch": 3.26,
1428
+ "learning_rate": 4.265171084243936e-05,
1429
+ "loss": 0.6934,
1430
+ "step": 57500
1431
+ },
1432
+ {
1433
+ "epoch": 3.27,
1434
+ "learning_rate": 4.256402875677308e-05,
1435
+ "loss": 0.694,
1436
+ "step": 57750
1437
+ },
1438
+ {
1439
+ "epoch": 3.28,
1440
+ "learning_rate": 4.247591793598513e-05,
1441
+ "loss": 0.6935,
1442
+ "step": 58000
1443
+ },
1444
+ {
1445
+ "epoch": 3.3,
1446
+ "learning_rate": 4.238773552736224e-05,
1447
+ "loss": 0.6933,
1448
+ "step": 58250
1449
+ },
1450
+ {
1451
+ "epoch": 3.31,
1452
+ "learning_rate": 4.2298775392430656e-05,
1453
+ "loss": 0.6934,
1454
+ "step": 58500
1455
+ },
1456
+ {
1457
+ "epoch": 3.33,
1458
+ "learning_rate": 4.220939299716943e-05,
1459
+ "loss": 0.6939,
1460
+ "step": 58750
1461
+ },
1462
+ {
1463
+ "epoch": 3.34,
1464
+ "learning_rate": 4.211959052338851e-05,
1465
+ "loss": 0.6933,
1466
+ "step": 59000
1467
+ },
1468
+ {
1469
+ "epoch": 3.35,
1470
+ "learning_rate": 4.202937016315183e-05,
1471
+ "loss": 0.6934,
1472
+ "step": 59250
1473
+ },
1474
+ {
1475
+ "epoch": 3.37,
1476
+ "learning_rate": 4.1938734118723924e-05,
1477
+ "loss": 0.6935,
1478
+ "step": 59500
1479
+ },
1480
+ {
1481
+ "epoch": 3.38,
1482
+ "learning_rate": 4.184768460251607e-05,
1483
+ "loss": 0.6932,
1484
+ "step": 59750
1485
+ },
1486
+ {
1487
+ "epoch": 3.4,
1488
+ "learning_rate": 4.1756223837032336e-05,
1489
+ "loss": 0.6939,
1490
+ "step": 60000
1491
+ },
1492
+ {
1493
+ "epoch": 3.41,
1494
+ "learning_rate": 4.166435405481533e-05,
1495
+ "loss": 0.6935,
1496
+ "step": 60250
1497
+ },
1498
+ {
1499
+ "epoch": 3.43,
1500
+ "learning_rate": 4.1572077498391684e-05,
1501
+ "loss": 0.6938,
1502
+ "step": 60500
1503
+ },
1504
+ {
1505
+ "epoch": 3.44,
1506
+ "learning_rate": 4.147976794734224e-05,
1507
+ "loss": 0.6936,
1508
+ "step": 60750
1509
+ },
1510
+ {
1511
+ "epoch": 3.45,
1512
+ "learning_rate": 4.138668621426548e-05,
1513
+ "loss": 0.6937,
1514
+ "step": 61000
1515
+ },
1516
+ {
1517
+ "epoch": 3.47,
1518
+ "learning_rate": 4.129320448480947e-05,
1519
+ "loss": 0.6935,
1520
+ "step": 61250
1521
+ },
1522
+ {
1523
+ "epoch": 3.48,
1524
+ "learning_rate": 4.1199325040848224e-05,
1525
+ "loss": 0.6932,
1526
+ "step": 61500
1527
+ },
1528
+ {
1529
+ "epoch": 3.5,
1530
+ "learning_rate": 4.11050501739639e-05,
1531
+ "loss": 0.6934,
1532
+ "step": 61750
1533
+ },
1534
+ {
1535
+ "epoch": 3.51,
1536
+ "learning_rate": 4.101038218539085e-05,
1537
+ "loss": 0.6934,
1538
+ "step": 62000
1539
+ },
1540
+ {
1541
+ "epoch": 3.52,
1542
+ "learning_rate": 4.091532338595949e-05,
1543
+ "loss": 0.6933,
1544
+ "step": 62250
1545
+ },
1546
+ {
1547
+ "epoch": 3.54,
1548
+ "learning_rate": 4.081987609603982e-05,
1549
+ "loss": 0.6933,
1550
+ "step": 62500
1551
+ },
1552
+ {
1553
+ "epoch": 3.55,
1554
+ "learning_rate": 4.072404264548489e-05,
1555
+ "loss": 0.694,
1556
+ "step": 62750
1557
+ },
1558
+ {
1559
+ "epoch": 3.57,
1560
+ "learning_rate": 4.0628211004123955e-05,
1561
+ "loss": 0.6937,
1562
+ "step": 63000
1563
+ },
1564
+ {
1565
+ "epoch": 3.58,
1566
+ "learning_rate": 4.053161378070489e-05,
1567
+ "loss": 0.6933,
1568
+ "step": 63250
1569
+ },
1570
+ {
1571
+ "epoch": 3.6,
1572
+ "learning_rate": 4.043463743308738e-05,
1573
+ "loss": 0.6936,
1574
+ "step": 63500
1575
+ },
1576
+ {
1577
+ "epoch": 3.61,
1578
+ "learning_rate": 4.03372843284485e-05,
1579
+ "loss": 0.6938,
1580
+ "step": 63750
1581
+ },
1582
+ {
1583
+ "epoch": 3.62,
1584
+ "learning_rate": 4.023955684316192e-05,
1585
+ "loss": 0.6936,
1586
+ "step": 64000
1587
+ },
1588
+ {
1589
+ "epoch": 3.64,
1590
+ "learning_rate": 4.014145736273984e-05,
1591
+ "loss": 0.6935,
1592
+ "step": 64250
1593
+ },
1594
+ {
1595
+ "epoch": 3.65,
1596
+ "learning_rate": 4.004298828177483e-05,
1597
+ "loss": 0.6931,
1598
+ "step": 64500
1599
+ },
1600
+ {
1601
+ "epoch": 3.67,
1602
+ "learning_rate": 3.9944152003881354e-05,
1603
+ "loss": 0.6921,
1604
+ "step": 64750
1605
+ },
1606
+ {
1607
+ "epoch": 3.68,
1608
+ "learning_rate": 3.984534846933014e-05,
1609
+ "loss": 0.6939,
1610
+ "step": 65000
1611
+ },
1612
+ {
1613
+ "epoch": 3.69,
1614
+ "learning_rate": 3.974578648883321e-05,
1615
+ "loss": 0.6935,
1616
+ "step": 65250
1617
+ },
1618
+ {
1619
+ "epoch": 3.71,
1620
+ "learning_rate": 3.9645864566055856e-05,
1621
+ "loss": 0.6934,
1622
+ "step": 65500
1623
+ },
1624
+ {
1625
+ "epoch": 3.72,
1626
+ "learning_rate": 3.954558514007616e-05,
1627
+ "loss": 0.6936,
1628
+ "step": 65750
1629
+ },
1630
+ {
1631
+ "epoch": 3.74,
1632
+ "learning_rate": 3.944495065869881e-05,
1633
+ "loss": 0.6933,
1634
+ "step": 66000
1635
+ },
1636
+ {
1637
+ "epoch": 3.75,
1638
+ "learning_rate": 3.934396357839535e-05,
1639
+ "loss": 0.6937,
1640
+ "step": 66250
1641
+ },
1642
+ {
1643
+ "epoch": 3.76,
1644
+ "learning_rate": 3.9242626364244185e-05,
1645
+ "loss": 0.6936,
1646
+ "step": 66500
1647
+ },
1648
+ {
1649
+ "epoch": 3.78,
1650
+ "learning_rate": 3.914094148987043e-05,
1651
+ "loss": 0.6931,
1652
+ "step": 66750
1653
+ },
1654
+ {
1655
+ "epoch": 3.79,
1656
+ "learning_rate": 3.9039320241892266e-05,
1657
+ "loss": 0.6938,
1658
+ "step": 67000
1659
+ },
1660
+ {
1661
+ "epoch": 3.81,
1662
+ "learning_rate": 3.89369488676115e-05,
1663
+ "loss": 0.6936,
1664
+ "step": 67250
1665
+ },
1666
+ {
1667
+ "epoch": 3.82,
1668
+ "learning_rate": 3.883423729464678e-05,
1669
+ "loss": 0.6926,
1670
+ "step": 67500
1671
+ },
1672
+ {
1673
+ "epoch": 3.84,
1674
+ "learning_rate": 3.8731188030171076e-05,
1675
+ "loss": 0.6938,
1676
+ "step": 67750
1677
+ },
1678
+ {
1679
+ "epoch": 3.85,
1680
+ "learning_rate": 3.862780358960041e-05,
1681
+ "loss": 0.6929,
1682
+ "step": 68000
1683
+ },
1684
+ {
1685
+ "epoch": 3.86,
1686
+ "learning_rate": 3.8524086496532375e-05,
1687
+ "loss": 0.6937,
1688
+ "step": 68250
1689
+ },
1690
+ {
1691
+ "epoch": 3.88,
1692
+ "learning_rate": 3.842003928268455e-05,
1693
+ "loss": 0.6935,
1694
+ "step": 68500
1695
+ },
1696
+ {
1697
+ "epoch": 3.89,
1698
+ "learning_rate": 3.831566448783271e-05,
1699
+ "loss": 0.6928,
1700
+ "step": 68750
1701
+ },
1702
+ {
1703
+ "epoch": 3.91,
1704
+ "learning_rate": 3.8210964659748866e-05,
1705
+ "loss": 0.694,
1706
+ "step": 69000
1707
+ },
1708
+ {
1709
+ "epoch": 3.92,
1710
+ "learning_rate": 3.810636308234164e-05,
1711
+ "loss": 0.6934,
1712
+ "step": 69250
1713
+ },
1714
+ {
1715
+ "epoch": 3.93,
1716
+ "learning_rate": 3.800102213732213e-05,
1717
+ "loss": 0.6937,
1718
+ "step": 69500
1719
+ },
1720
+ {
1721
+ "epoch": 3.95,
1722
+ "learning_rate": 3.789536383943985e-05,
1723
+ "loss": 0.6936,
1724
+ "step": 69750
1725
+ },
1726
+ {
1727
+ "epoch": 3.96,
1728
+ "learning_rate": 3.778939076779688e-05,
1729
+ "loss": 0.6934,
1730
+ "step": 70000
1731
+ },
1732
+ {
1733
+ "epoch": 3.98,
1734
+ "learning_rate": 3.768310550917889e-05,
1735
+ "loss": 0.6936,
1736
+ "step": 70250
1737
+ },
1738
+ {
1739
+ "epoch": 3.99,
1740
+ "learning_rate": 3.757651065799198e-05,
1741
+ "loss": 0.6936,
1742
+ "step": 70500
1743
+ },
1744
+ {
1745
+ "epoch": 4.0,
1746
+ "eval_accuracy": 0.5002547770700637,
1747
+ "eval_f1": 0.33344654835696697,
1748
+ "eval_loss": 0.6933583617210388,
1749
+ "eval_precision": 0.25012738853503186,
1750
+ "eval_recall": 0.5,
1751
+ "eval_runtime": 12.4824,
1752
+ "eval_samples_per_second": 628.887,
1753
+ "eval_steps_per_second": 78.671,
1754
+ "step": 70652
1755
+ },
1756
+ {
1757
+ "epoch": 4.01,
1758
+ "learning_rate": 3.7469608816199355e-05,
1759
+ "loss": 0.6932,
1760
+ "step": 70750
1761
+ },
1762
+ {
1763
+ "epoch": 4.02,
1764
+ "learning_rate": 3.7362402593257796e-05,
1765
+ "loss": 0.6939,
1766
+ "step": 71000
1767
+ },
1768
+ {
1769
+ "epoch": 4.03,
1770
+ "learning_rate": 3.725489460605399e-05,
1771
+ "loss": 0.6935,
1772
+ "step": 71250
1773
+ },
1774
+ {
1775
+ "epoch": 4.05,
1776
+ "learning_rate": 3.7147087478840654e-05,
1777
+ "loss": 0.693,
1778
+ "step": 71500
1779
+ },
1780
+ {
1781
+ "epoch": 4.06,
1782
+ "learning_rate": 3.7039416844865045e-05,
1783
+ "loss": 0.694,
1784
+ "step": 71750
1785
+ },
1786
+ {
1787
+ "epoch": 4.08,
1788
+ "learning_rate": 3.693102050974699e-05,
1789
+ "loss": 0.6933,
1790
+ "step": 72000
1791
+ },
1792
+ {
1793
+ "epoch": 4.09,
1794
+ "learning_rate": 3.682233294033409e-05,
1795
+ "loss": 0.6933,
1796
+ "step": 72250
1797
+ },
1798
+ {
1799
+ "epoch": 4.1,
1800
+ "learning_rate": 3.671335678967246e-05,
1801
+ "loss": 0.6933,
1802
+ "step": 72500
1803
+ },
1804
+ {
1805
+ "epoch": 4.12,
1806
+ "learning_rate": 3.6604094717852435e-05,
1807
+ "loss": 0.6935,
1808
+ "step": 72750
1809
+ },
1810
+ {
1811
+ "epoch": 4.13,
1812
+ "learning_rate": 3.649454939194364e-05,
1813
+ "loss": 0.6932,
1814
+ "step": 73000
1815
+ },
1816
+ {
1817
+ "epoch": 4.15,
1818
+ "learning_rate": 3.638472348592989e-05,
1819
+ "loss": 0.6934,
1820
+ "step": 73250
1821
+ },
1822
+ {
1823
+ "epoch": 4.16,
1824
+ "learning_rate": 3.627461968064393e-05,
1825
+ "loss": 0.6935,
1826
+ "step": 73500
1827
+ },
1828
+ {
1829
+ "epoch": 4.18,
1830
+ "learning_rate": 3.616424066370194e-05,
1831
+ "loss": 0.6929,
1832
+ "step": 73750
1833
+ },
1834
+ {
1835
+ "epoch": 4.19,
1836
+ "learning_rate": 3.605358912943803e-05,
1837
+ "loss": 0.6929,
1838
+ "step": 74000
1839
+ },
1840
+ {
1841
+ "epoch": 4.2,
1842
+ "learning_rate": 3.594311199812862e-05,
1843
+ "loss": 0.6937,
1844
+ "step": 74250
1845
+ },
1846
+ {
1847
+ "epoch": 4.22,
1848
+ "learning_rate": 3.5831924601798475e-05,
1849
+ "loss": 0.6934,
1850
+ "step": 74500
1851
+ },
1852
+ {
1853
+ "epoch": 4.23,
1854
+ "learning_rate": 3.5720472799928105e-05,
1855
+ "loss": 0.6937,
1856
+ "step": 74750
1857
+ },
1858
+ {
1859
+ "epoch": 4.25,
1860
+ "learning_rate": 3.560875931303811e-05,
1861
+ "loss": 0.6934,
1862
+ "step": 75000
1863
+ },
1864
+ {
1865
+ "epoch": 4.26,
1866
+ "learning_rate": 3.549678686803674e-05,
1867
+ "loss": 0.6936,
1868
+ "step": 75250
1869
+ },
1870
+ {
1871
+ "epoch": 4.27,
1872
+ "learning_rate": 3.5384558198153416e-05,
1873
+ "loss": 0.694,
1874
+ "step": 75500
1875
+ },
1876
+ {
1877
+ "epoch": 4.29,
1878
+ "learning_rate": 3.527207604287196e-05,
1879
+ "loss": 0.6934,
1880
+ "step": 75750
1881
+ },
1882
+ {
1883
+ "epoch": 4.3,
1884
+ "learning_rate": 3.51593431478637e-05,
1885
+ "loss": 0.693,
1886
+ "step": 76000
1887
+ },
1888
+ {
1889
+ "epoch": 4.32,
1890
+ "learning_rate": 3.504681467879116e-05,
1891
+ "loss": 0.6931,
1892
+ "step": 76250
1893
+ },
1894
+ {
1895
+ "epoch": 4.33,
1896
+ "learning_rate": 3.4933589541177024e-05,
1897
+ "loss": 0.6936,
1898
+ "step": 76500
1899
+ },
1900
+ {
1901
+ "epoch": 4.35,
1902
+ "learning_rate": 3.4820121926237256e-05,
1903
+ "loss": 0.6934,
1904
+ "step": 76750
1905
+ },
1906
+ {
1907
+ "epoch": 4.36,
1908
+ "learning_rate": 3.47064146036981e-05,
1909
+ "loss": 0.6937,
1910
+ "step": 77000
1911
+ },
1912
+ {
1913
+ "epoch": 4.37,
1914
+ "learning_rate": 3.4592470349137056e-05,
1915
+ "loss": 0.6934,
1916
+ "step": 77250
1917
+ },
1918
+ {
1919
+ "epoch": 4.39,
1920
+ "learning_rate": 3.447829194391507e-05,
1921
+ "loss": 0.6932,
1922
+ "step": 77500
1923
+ },
1924
+ {
1925
+ "epoch": 4.4,
1926
+ "learning_rate": 3.436388217510869e-05,
1927
+ "loss": 0.6937,
1928
+ "step": 77750
1929
+ },
1930
+ {
1931
+ "epoch": 4.42,
1932
+ "learning_rate": 3.4249243835441994e-05,
1933
+ "loss": 0.6928,
1934
+ "step": 78000
1935
+ },
1936
+ {
1937
+ "epoch": 4.43,
1938
+ "learning_rate": 3.413437972321844e-05,
1939
+ "loss": 0.692,
1940
+ "step": 78250
1941
+ },
1942
+ {
1943
+ "epoch": 4.44,
1944
+ "learning_rate": 3.401975343100874e-05,
1945
+ "loss": 0.694,
1946
+ "step": 78500
1947
+ },
1948
+ {
1949
+ "epoch": 4.46,
1950
+ "learning_rate": 3.390444706559248e-05,
1951
+ "loss": 0.693,
1952
+ "step": 78750
1953
+ },
1954
+ {
1955
+ "epoch": 4.47,
1956
+ "learning_rate": 3.3788923344053156e-05,
1957
+ "loss": 0.6936,
1958
+ "step": 79000
1959
+ },
1960
+ {
1961
+ "epoch": 4.49,
1962
+ "learning_rate": 3.367318508630627e-05,
1963
+ "loss": 0.6932,
1964
+ "step": 79250
1965
+ },
1966
+ {
1967
+ "epoch": 4.5,
1968
+ "learning_rate": 3.355723511750409e-05,
1969
+ "loss": 0.6933,
1970
+ "step": 79500
1971
+ },
1972
+ {
1973
+ "epoch": 4.52,
1974
+ "learning_rate": 3.3441076267966755e-05,
1975
+ "loss": 0.6935,
1976
+ "step": 79750
1977
+ },
1978
+ {
1979
+ "epoch": 4.53,
1980
+ "learning_rate": 3.3324711373113114e-05,
1981
+ "loss": 0.6932,
1982
+ "step": 80000
1983
+ },
1984
+ {
1985
+ "epoch": 4.54,
1986
+ "learning_rate": 3.320814327339158e-05,
1987
+ "loss": 0.693,
1988
+ "step": 80250
1989
+ },
1990
+ {
1991
+ "epoch": 4.56,
1992
+ "learning_rate": 3.30918422833875e-05,
1993
+ "loss": 0.6941,
1994
+ "step": 80500
1995
+ },
1996
+ {
1997
+ "epoch": 4.57,
1998
+ "learning_rate": 3.2974877099399125e-05,
1999
+ "loss": 0.6934,
2000
+ "step": 80750
2001
+ },
2002
+ {
2003
+ "epoch": 4.59,
2004
+ "learning_rate": 3.28577172499413e-05,
2005
+ "loss": 0.6932,
2006
+ "step": 81000
2007
+ },
2008
+ {
2009
+ "epoch": 4.6,
2010
+ "learning_rate": 3.274036559486714e-05,
2011
+ "loss": 0.6935,
2012
+ "step": 81250
2013
+ },
2014
+ {
2015
+ "epoch": 4.61,
2016
+ "learning_rate": 3.26228249987117e-05,
2017
+ "loss": 0.6931,
2018
+ "step": 81500
2019
+ },
2020
+ {
2021
+ "epoch": 4.63,
2022
+ "learning_rate": 3.250509833062205e-05,
2023
+ "loss": 0.6929,
2024
+ "step": 81750
2025
+ },
2026
+ {
2027
+ "epoch": 4.64,
2028
+ "learning_rate": 3.238718846428726e-05,
2029
+ "loss": 0.6932,
2030
+ "step": 82000
2031
+ },
2032
+ {
2033
+ "epoch": 4.66,
2034
+ "learning_rate": 3.226909827786824e-05,
2035
+ "loss": 0.6934,
2036
+ "step": 82250
2037
+ },
2038
+ {
2039
+ "epoch": 4.67,
2040
+ "learning_rate": 3.2150830653927466e-05,
2041
+ "loss": 0.6928,
2042
+ "step": 82500
2043
+ },
2044
+ {
2045
+ "epoch": 4.68,
2046
+ "learning_rate": 3.203238847935866e-05,
2047
+ "loss": 0.6936,
2048
+ "step": 82750
2049
+ },
2050
+ {
2051
+ "epoch": 4.7,
2052
+ "learning_rate": 3.191424943876217e-05,
2053
+ "loss": 0.6935,
2054
+ "step": 83000
2055
+ },
2056
+ {
2057
+ "epoch": 4.71,
2058
+ "learning_rate": 3.179546750987436e-05,
2059
+ "loss": 0.6934,
2060
+ "step": 83250
2061
+ },
2062
+ {
2063
+ "epoch": 4.73,
2064
+ "learning_rate": 3.1676519704715806e-05,
2065
+ "loss": 0.6934,
2066
+ "step": 83500
2067
+ },
2068
+ {
2069
+ "epoch": 4.74,
2070
+ "learning_rate": 3.155740892678332e-05,
2071
+ "loss": 0.6934,
2072
+ "step": 83750
2073
+ },
2074
+ {
2075
+ "epoch": 4.76,
2076
+ "learning_rate": 3.143813808355187e-05,
2077
+ "loss": 0.6932,
2078
+ "step": 84000
2079
+ },
2080
+ {
2081
+ "epoch": 4.77,
2082
+ "learning_rate": 3.1318710086403584e-05,
2083
+ "loss": 0.6935,
2084
+ "step": 84250
2085
+ },
2086
+ {
2087
+ "epoch": 4.78,
2088
+ "learning_rate": 3.1199127850556704e-05,
2089
+ "loss": 0.6935,
2090
+ "step": 84500
2091
+ },
2092
+ {
2093
+ "epoch": 4.8,
2094
+ "learning_rate": 3.10793942949944e-05,
2095
+ "loss": 0.6936,
2096
+ "step": 84750
2097
+ },
2098
+ {
2099
+ "epoch": 4.81,
2100
+ "learning_rate": 3.095951234239353e-05,
2101
+ "loss": 0.6932,
2102
+ "step": 85000
2103
+ },
2104
+ {
2105
+ "epoch": 4.83,
2106
+ "learning_rate": 3.083948491905332e-05,
2107
+ "loss": 0.6933,
2108
+ "step": 85250
2109
+ },
2110
+ {
2111
+ "epoch": 4.84,
2112
+ "learning_rate": 3.0719314954823894e-05,
2113
+ "loss": 0.693,
2114
+ "step": 85500
2115
+ },
2116
+ {
2117
+ "epoch": 4.85,
2118
+ "learning_rate": 3.059900538303479e-05,
2119
+ "loss": 0.6931,
2120
+ "step": 85750
2121
+ },
2122
+ {
2123
+ "epoch": 4.87,
2124
+ "learning_rate": 3.047855914042333e-05,
2125
+ "loss": 0.6938,
2126
+ "step": 86000
2127
+ },
2128
+ {
2129
+ "epoch": 4.88,
2130
+ "learning_rate": 3.0357979167062957e-05,
2131
+ "loss": 0.6933,
2132
+ "step": 86250
2133
+ },
2134
+ {
2135
+ "epoch": 4.9,
2136
+ "learning_rate": 3.0237751505958946e-05,
2137
+ "loss": 0.6935,
2138
+ "step": 86500
2139
+ },
2140
+ {
2141
+ "epoch": 4.91,
2142
+ "learning_rate": 3.0116913409796476e-05,
2143
+ "loss": 0.6926,
2144
+ "step": 86750
2145
+ },
2146
+ {
2147
+ "epoch": 4.93,
2148
+ "learning_rate": 2.999595041059932e-05,
2149
+ "loss": 0.6938,
2150
+ "step": 87000
2151
+ },
2152
+ {
2153
+ "epoch": 4.94,
2154
+ "learning_rate": 2.9874865461054845e-05,
2155
+ "loss": 0.6934,
2156
+ "step": 87250
2157
+ },
2158
+ {
2159
+ "epoch": 4.95,
2160
+ "learning_rate": 2.9753661516827242e-05,
2161
+ "loss": 0.6927,
2162
+ "step": 87500
2163
+ },
2164
+ {
2165
+ "epoch": 4.97,
2166
+ "learning_rate": 2.9632341536485315e-05,
2167
+ "loss": 0.6938,
2168
+ "step": 87750
2169
+ },
2170
+ {
2171
+ "epoch": 4.98,
2172
+ "learning_rate": 2.9510908481430306e-05,
2173
+ "loss": 0.6931,
2174
+ "step": 88000
2175
+ },
2176
+ {
2177
+ "epoch": 5.0,
2178
+ "learning_rate": 2.938936531582359e-05,
2179
+ "loss": 0.6934,
2180
+ "step": 88250
2181
+ },
2182
+ {
2183
+ "epoch": 5.0,
2184
+ "eval_accuracy": 0.4997452229299363,
2185
+ "eval_f1": 0.3332200798437102,
2186
+ "eval_loss": 0.6931193470954895,
2187
+ "eval_precision": 0.24987261146496814,
2188
+ "eval_recall": 0.5,
2189
+ "eval_runtime": 12.6829,
2190
+ "eval_samples_per_second": 618.944,
2191
+ "eval_steps_per_second": 77.427,
2192
+ "step": 88315
2193
+ },
2194
+ {
2195
+ "epoch": 5.01,
2196
+ "learning_rate": 2.926820181724712e-05,
2197
+ "loss": 0.6931,
2198
+ "step": 88500
2199
+ },
2200
+ {
2201
+ "epoch": 5.02,
2202
+ "learning_rate": 2.914644774447821e-05,
2203
+ "loss": 0.6929,
2204
+ "step": 88750
2205
+ },
2206
+ {
2207
+ "epoch": 5.04,
2208
+ "learning_rate": 2.9024592457585668e-05,
2209
+ "loss": 0.6936,
2210
+ "step": 89000
2211
+ },
2212
+ {
2213
+ "epoch": 5.05,
2214
+ "learning_rate": 2.8902638931037468e-05,
2215
+ "loss": 0.6931,
2216
+ "step": 89250
2217
+ },
2218
+ {
2219
+ "epoch": 5.07,
2220
+ "learning_rate": 2.878059014169962e-05,
2221
+ "loss": 0.6933,
2222
+ "step": 89500
2223
+ },
2224
+ {
2225
+ "epoch": 5.08,
2226
+ "learning_rate": 2.8658449068763467e-05,
2227
+ "loss": 0.6934,
2228
+ "step": 89750
2229
+ },
2230
+ {
2231
+ "epoch": 5.1,
2232
+ "learning_rate": 2.8536218693672996e-05,
2233
+ "loss": 0.6934,
2234
+ "step": 90000
2235
+ },
2236
+ {
2237
+ "epoch": 5.11,
2238
+ "learning_rate": 2.841390200005203e-05,
2239
+ "loss": 0.6932,
2240
+ "step": 90250
2241
+ },
2242
+ {
2243
+ "epoch": 5.12,
2244
+ "learning_rate": 2.8291501973631418e-05,
2245
+ "loss": 0.6935,
2246
+ "step": 90500
2247
+ },
2248
+ {
2249
+ "epoch": 5.14,
2250
+ "learning_rate": 2.8169511679747517e-05,
2251
+ "loss": 0.6933,
2252
+ "step": 90750
2253
+ },
2254
+ {
2255
+ "epoch": 5.15,
2256
+ "learning_rate": 2.8046954256446972e-05,
2257
+ "loss": 0.6932,
2258
+ "step": 91000
2259
+ },
2260
+ {
2261
+ "epoch": 5.17,
2262
+ "learning_rate": 2.7924322457482288e-05,
2263
+ "loss": 0.6935,
2264
+ "step": 91250
2265
+ },
2266
+ {
2267
+ "epoch": 5.18,
2268
+ "learning_rate": 2.780161927627598e-05,
2269
+ "loss": 0.6932,
2270
+ "step": 91500
2271
+ },
2272
+ {
2273
+ "epoch": 5.19,
2274
+ "learning_rate": 2.7678847707993006e-05,
2275
+ "loss": 0.6935,
2276
+ "step": 91750
2277
+ },
2278
+ {
2279
+ "epoch": 5.21,
2280
+ "learning_rate": 2.7556010749467626e-05,
2281
+ "loss": 0.693,
2282
+ "step": 92000
2283
+ },
2284
+ {
2285
+ "epoch": 5.22,
2286
+ "learning_rate": 2.7433111399130273e-05,
2287
+ "loss": 0.6933,
2288
+ "step": 92250
2289
+ },
2290
+ {
2291
+ "epoch": 5.24,
2292
+ "learning_rate": 2.7310152656934363e-05,
2293
+ "loss": 0.6931,
2294
+ "step": 92500
2295
+ },
2296
+ {
2297
+ "epoch": 5.25,
2298
+ "learning_rate": 2.7187629693164157e-05,
2299
+ "loss": 0.6934,
2300
+ "step": 92750
2301
+ },
2302
+ {
2303
+ "epoch": 5.27,
2304
+ "learning_rate": 2.7064561380404474e-05,
2305
+ "loss": 0.6933,
2306
+ "step": 93000
2307
+ },
2308
+ {
2309
+ "epoch": 5.28,
2310
+ "learning_rate": 2.6941442672033014e-05,
2311
+ "loss": 0.6934,
2312
+ "step": 93250
2313
+ },
2314
+ {
2315
+ "epoch": 5.29,
2316
+ "learning_rate": 2.6818276573357664e-05,
2317
+ "loss": 0.6932,
2318
+ "step": 93500
2319
+ },
2320
+ {
2321
+ "epoch": 5.31,
2322
+ "learning_rate": 2.6695066090843123e-05,
2323
+ "loss": 0.6933,
2324
+ "step": 93750
2325
+ },
2326
+ {
2327
+ "epoch": 5.32,
2328
+ "learning_rate": 2.657181423203749e-05,
2329
+ "loss": 0.6934,
2330
+ "step": 94000
2331
+ },
2332
+ {
2333
+ "epoch": 5.34,
2334
+ "learning_rate": 2.6448524005498838e-05,
2335
+ "loss": 0.6933,
2336
+ "step": 94250
2337
+ },
2338
+ {
2339
+ "epoch": 5.35,
2340
+ "learning_rate": 2.6325198420721808e-05,
2341
+ "loss": 0.6933,
2342
+ "step": 94500
2343
+ },
2344
+ {
2345
+ "epoch": 5.36,
2346
+ "learning_rate": 2.6201840488064118e-05,
2347
+ "loss": 0.6934,
2348
+ "step": 94750
2349
+ },
2350
+ {
2351
+ "epoch": 5.38,
2352
+ "learning_rate": 2.6078453218673098e-05,
2353
+ "loss": 0.6932,
2354
+ "step": 95000
2355
+ },
2356
+ {
2357
+ "epoch": 5.39,
2358
+ "learning_rate": 2.595553332723595e-05,
2359
+ "loss": 0.6936,
2360
+ "step": 95250
2361
+ },
2362
+ {
2363
+ "epoch": 5.41,
2364
+ "learning_rate": 2.5832096507858917e-05,
2365
+ "loss": 0.6928,
2366
+ "step": 95500
2367
+ },
2368
+ {
2369
+ "epoch": 5.42,
2370
+ "learning_rate": 2.5708639377139736e-05,
2371
+ "loss": 0.6933,
2372
+ "step": 95750
2373
+ },
2374
+ {
2375
+ "epoch": 5.44,
2376
+ "learning_rate": 2.558516494864715e-05,
2377
+ "loss": 0.6932,
2378
+ "step": 96000
2379
+ },
2380
+ {
2381
+ "epoch": 5.45,
2382
+ "learning_rate": 2.546167623637212e-05,
2383
+ "loss": 0.6933,
2384
+ "step": 96250
2385
+ },
2386
+ {
2387
+ "epoch": 5.46,
2388
+ "learning_rate": 2.5338176254654305e-05,
2389
+ "loss": 0.6934,
2390
+ "step": 96500
2391
+ },
2392
+ {
2393
+ "epoch": 5.48,
2394
+ "learning_rate": 2.5214668018108413e-05,
2395
+ "loss": 0.6933,
2396
+ "step": 96750
2397
+ },
2398
+ {
2399
+ "epoch": 5.49,
2400
+ "learning_rate": 2.509115454155066e-05,
2401
+ "loss": 0.6934,
2402
+ "step": 97000
2403
+ },
2404
+ {
2405
+ "epoch": 5.51,
2406
+ "learning_rate": 2.4968132903168124e-05,
2407
+ "loss": 0.6929,
2408
+ "step": 97250
2409
+ },
2410
+ {
2411
+ "epoch": 5.52,
2412
+ "learning_rate": 2.4844617982307774e-05,
2413
+ "loss": 0.6936,
2414
+ "step": 97500
2415
+ },
2416
+ {
2417
+ "epoch": 5.53,
2418
+ "learning_rate": 2.4721106854297538e-05,
2419
+ "loss": 0.6932,
2420
+ "step": 97750
2421
+ },
2422
+ {
2423
+ "epoch": 5.55,
2424
+ "learning_rate": 2.4597602534024195e-05,
2425
+ "loss": 0.6935,
2426
+ "step": 98000
2427
+ },
2428
+ {
2429
+ "epoch": 5.56,
2430
+ "learning_rate": 2.447410803620838e-05,
2431
+ "loss": 0.6931,
2432
+ "step": 98250
2433
+ },
2434
+ {
2435
+ "epoch": 5.58,
2436
+ "learning_rate": 2.4350626375330963e-05,
2437
+ "loss": 0.6931,
2438
+ "step": 98500
2439
+ },
2440
+ {
2441
+ "epoch": 5.59,
2442
+ "learning_rate": 2.4227160565559448e-05,
2443
+ "loss": 0.6933,
2444
+ "step": 98750
2445
+ },
2446
+ {
2447
+ "epoch": 5.6,
2448
+ "learning_rate": 2.410371362067444e-05,
2449
+ "loss": 0.6933,
2450
+ "step": 99000
2451
+ },
2452
+ {
2453
+ "epoch": 5.62,
2454
+ "learning_rate": 2.398078220668819e-05,
2455
+ "loss": 0.6934,
2456
+ "step": 99250
2457
+ },
2458
+ {
2459
+ "epoch": 5.63,
2460
+ "learning_rate": 2.385738192543777e-05,
2461
+ "loss": 0.6934,
2462
+ "step": 99500
2463
+ },
2464
+ {
2465
+ "epoch": 5.65,
2466
+ "learning_rate": 2.3734009535311082e-05,
2467
+ "loss": 0.693,
2468
+ "step": 99750
2469
+ },
2470
+ {
2471
+ "epoch": 5.66,
2472
+ "learning_rate": 2.3610668047808355e-05,
2473
+ "loss": 0.6936,
2474
+ "step": 100000
2475
+ },
2476
+ {
2477
+ "epoch": 5.68,
2478
+ "learning_rate": 2.3487360473675493e-05,
2479
+ "loss": 0.693,
2480
+ "step": 100250
2481
+ },
2482
+ {
2483
+ "epoch": 5.69,
2484
+ "learning_rate": 2.3364089822830585e-05,
2485
+ "loss": 0.6934,
2486
+ "step": 100500
2487
+ },
2488
+ {
2489
+ "epoch": 5.7,
2490
+ "learning_rate": 2.3240859104290418e-05,
2491
+ "loss": 0.6932,
2492
+ "step": 100750
2493
+ },
2494
+ {
2495
+ "epoch": 5.72,
2496
+ "learning_rate": 2.311767132609704e-05,
2497
+ "loss": 0.6932,
2498
+ "step": 101000
2499
+ },
2500
+ {
2501
+ "epoch": 5.73,
2502
+ "learning_rate": 2.2994529495244343e-05,
2503
+ "loss": 0.6934,
2504
+ "step": 101250
2505
+ },
2506
+ {
2507
+ "epoch": 5.75,
2508
+ "learning_rate": 2.287192888761777e-05,
2509
+ "loss": 0.6934,
2510
+ "step": 101500
2511
+ },
2512
+ {
2513
+ "epoch": 5.76,
2514
+ "learning_rate": 2.2748887754053225e-05,
2515
+ "loss": 0.6932,
2516
+ "step": 101750
2517
+ },
2518
+ {
2519
+ "epoch": 5.77,
2520
+ "learning_rate": 2.2625901569777145e-05,
2521
+ "loss": 0.693,
2522
+ "step": 102000
2523
+ },
2524
+ {
2525
+ "epoch": 5.79,
2526
+ "learning_rate": 2.2502973336862524e-05,
2527
+ "loss": 0.6933,
2528
+ "step": 102250
2529
+ },
2530
+ {
2531
+ "epoch": 5.8,
2532
+ "learning_rate": 2.23801060559678e-05,
2533
+ "loss": 0.6935,
2534
+ "step": 102500
2535
+ },
2536
+ {
2537
+ "epoch": 5.82,
2538
+ "learning_rate": 2.2257302726263566e-05,
2539
+ "loss": 0.6931,
2540
+ "step": 102750
2541
+ },
2542
+ {
2543
+ "epoch": 5.83,
2544
+ "learning_rate": 2.213456634535938e-05,
2545
+ "loss": 0.6933,
2546
+ "step": 103000
2547
+ },
2548
+ {
2549
+ "epoch": 5.85,
2550
+ "learning_rate": 2.201189990923061e-05,
2551
+ "loss": 0.6936,
2552
+ "step": 103250
2553
+ },
2554
+ {
2555
+ "epoch": 5.86,
2556
+ "learning_rate": 2.1889796636872324e-05,
2557
+ "loss": 0.6934,
2558
+ "step": 103500
2559
+ },
2560
+ {
2561
+ "epoch": 5.87,
2562
+ "learning_rate": 2.176727876163321e-05,
2563
+ "loss": 0.6929,
2564
+ "step": 103750
2565
+ },
2566
+ {
2567
+ "epoch": 5.89,
2568
+ "learning_rate": 2.1644839796600443e-05,
2569
+ "loss": 0.6935,
2570
+ "step": 104000
2571
+ },
2572
+ {
2573
+ "epoch": 5.9,
2574
+ "learning_rate": 2.152248273048949e-05,
2575
+ "loss": 0.6932,
2576
+ "step": 104250
2577
+ },
2578
+ {
2579
+ "epoch": 5.92,
2580
+ "learning_rate": 2.1400210550016697e-05,
2581
+ "loss": 0.6931,
2582
+ "step": 104500
2583
+ },
2584
+ {
2585
+ "epoch": 5.93,
2586
+ "learning_rate": 2.127802623982636e-05,
2587
+ "loss": 0.6932,
2588
+ "step": 104750
2589
+ },
2590
+ {
2591
+ "epoch": 5.94,
2592
+ "learning_rate": 2.1155932782417855e-05,
2593
+ "loss": 0.6933,
2594
+ "step": 105000
2595
+ },
2596
+ {
2597
+ "epoch": 5.96,
2598
+ "learning_rate": 2.1033933158072878e-05,
2599
+ "loss": 0.6928,
2600
+ "step": 105250
2601
+ },
2602
+ {
2603
+ "epoch": 5.97,
2604
+ "learning_rate": 2.0912030344782672e-05,
2605
+ "loss": 0.6935,
2606
+ "step": 105500
2607
+ },
2608
+ {
2609
+ "epoch": 5.99,
2610
+ "learning_rate": 2.0790714327565363e-05,
2611
+ "loss": 0.6932,
2612
+ "step": 105750
2613
+ },
2614
+ {
2615
+ "epoch": 6.0,
2616
+ "eval_accuracy": 0.5002547770700637,
2617
+ "eval_f1": 0.33344654835696697,
2618
+ "eval_loss": 0.6931106448173523,
2619
+ "eval_precision": 0.25012738853503186,
2620
+ "eval_recall": 0.5,
2621
+ "eval_runtime": 12.652,
2622
+ "eval_samples_per_second": 620.456,
2623
+ "eval_steps_per_second": 77.616,
2624
+ "step": 105978
2625
+ },
2626
+ {
2627
+ "epoch": 6.0,
2628
+ "learning_rate": 2.066901364387442e-05,
2629
+ "loss": 0.6934,
2630
+ "step": 106000
2631
+ },
2632
+ {
2633
+ "epoch": 6.02,
2634
+ "learning_rate": 2.0547418678865e-05,
2635
+ "loss": 0.693,
2636
+ "step": 106250
2637
+ },
2638
+ {
2639
+ "epoch": 6.03,
2640
+ "learning_rate": 2.0425932400650658e-05,
2641
+ "loss": 0.6935,
2642
+ "step": 106500
2643
+ },
2644
+ {
2645
+ "epoch": 6.04,
2646
+ "learning_rate": 2.0304557774691947e-05,
2647
+ "loss": 0.6935,
2648
+ "step": 106750
2649
+ },
2650
+ {
2651
+ "epoch": 6.06,
2652
+ "learning_rate": 2.0183297763724e-05,
2653
+ "loss": 0.693,
2654
+ "step": 107000
2655
+ },
2656
+ {
2657
+ "epoch": 6.07,
2658
+ "learning_rate": 2.006215532768421e-05,
2659
+ "loss": 0.6932,
2660
+ "step": 107250
2661
+ },
2662
+ {
2663
+ "epoch": 6.09,
2664
+ "learning_rate": 1.9941133423640003e-05,
2665
+ "loss": 0.6933,
2666
+ "step": 107500
2667
+ },
2668
+ {
2669
+ "epoch": 6.1,
2670
+ "learning_rate": 1.9820235005716594e-05,
2671
+ "loss": 0.6924,
2672
+ "step": 107750
2673
+ },
2674
+ {
2675
+ "epoch": 6.11,
2676
+ "learning_rate": 1.9699945857176588e-05,
2677
+ "loss": 0.6935,
2678
+ "step": 108000
2679
+ },
2680
+ {
2681
+ "epoch": 6.13,
2682
+ "learning_rate": 1.9579302738332077e-05,
2683
+ "loss": 0.6935,
2684
+ "step": 108250
2685
+ },
2686
+ {
2687
+ "epoch": 6.14,
2688
+ "learning_rate": 1.9458791937837308e-05,
2689
+ "loss": 0.6932,
2690
+ "step": 108500
2691
+ },
2692
+ {
2693
+ "epoch": 6.16,
2694
+ "learning_rate": 1.9338416397341575e-05,
2695
+ "loss": 0.6932,
2696
+ "step": 108750
2697
+ },
2698
+ {
2699
+ "epoch": 6.17,
2700
+ "learning_rate": 1.92181790551925e-05,
2701
+ "loss": 0.6933,
2702
+ "step": 109000
2703
+ },
2704
+ {
2705
+ "epoch": 6.19,
2706
+ "learning_rate": 1.9098082846364272e-05,
2707
+ "loss": 0.6932,
2708
+ "step": 109250
2709
+ },
2710
+ {
2711
+ "epoch": 6.2,
2712
+ "learning_rate": 1.8978130702386082e-05,
2713
+ "loss": 0.6932,
2714
+ "step": 109500
2715
+ },
2716
+ {
2717
+ "epoch": 6.21,
2718
+ "learning_rate": 1.8858325551270503e-05,
2719
+ "loss": 0.6933,
2720
+ "step": 109750
2721
+ },
2722
+ {
2723
+ "epoch": 6.23,
2724
+ "learning_rate": 1.8739148635869862e-05,
2725
+ "loss": 0.6933,
2726
+ "step": 110000
2727
+ },
2728
+ {
2729
+ "epoch": 6.24,
2730
+ "learning_rate": 1.8619645622928097e-05,
2731
+ "loss": 0.6932,
2732
+ "step": 110250
2733
+ },
2734
+ {
2735
+ "epoch": 6.26,
2736
+ "learning_rate": 1.8500298353412176e-05,
2737
+ "loss": 0.6933,
2738
+ "step": 110500
2739
+ },
2740
+ {
2741
+ "epoch": 6.27,
2742
+ "learning_rate": 1.838110974056977e-05,
2743
+ "loss": 0.6933,
2744
+ "step": 110750
2745
+ },
2746
+ {
2747
+ "epoch": 6.28,
2748
+ "learning_rate": 1.826208269377578e-05,
2749
+ "loss": 0.6933,
2750
+ "step": 111000
2751
+ },
2752
+ {
2753
+ "epoch": 6.3,
2754
+ "learning_rate": 1.8143220118461316e-05,
2755
+ "loss": 0.6932,
2756
+ "step": 111250
2757
+ },
2758
+ {
2759
+ "epoch": 6.31,
2760
+ "learning_rate": 1.802452491604275e-05,
2761
+ "loss": 0.6931,
2762
+ "step": 111500
2763
+ },
2764
+ {
2765
+ "epoch": 6.33,
2766
+ "learning_rate": 1.790599998385092e-05,
2767
+ "loss": 0.6935,
2768
+ "step": 111750
2769
+ },
2770
+ {
2771
+ "epoch": 6.34,
2772
+ "learning_rate": 1.778764821506038e-05,
2773
+ "loss": 0.6934,
2774
+ "step": 112000
2775
+ },
2776
+ {
2777
+ "epoch": 6.36,
2778
+ "learning_rate": 1.7669944846963703e-05,
2779
+ "loss": 0.6933,
2780
+ "step": 112250
2781
+ },
2782
+ {
2783
+ "epoch": 6.37,
2784
+ "learning_rate": 1.75519473460329e-05,
2785
+ "loss": 0.6933,
2786
+ "step": 112500
2787
+ },
2788
+ {
2789
+ "epoch": 6.38,
2790
+ "learning_rate": 1.7434131650871434e-05,
2791
+ "loss": 0.6933,
2792
+ "step": 112750
2793
+ },
2794
+ {
2795
+ "epoch": 6.4,
2796
+ "learning_rate": 1.7316500637341497e-05,
2797
+ "loss": 0.6934,
2798
+ "step": 113000
2799
+ },
2800
+ {
2801
+ "epoch": 6.41,
2802
+ "learning_rate": 1.7199057176797255e-05,
2803
+ "loss": 0.693,
2804
+ "step": 113250
2805
+ },
2806
+ {
2807
+ "epoch": 6.43,
2808
+ "learning_rate": 1.7081804136014705e-05,
2809
+ "loss": 0.6934,
2810
+ "step": 113500
2811
+ },
2812
+ {
2813
+ "epoch": 6.44,
2814
+ "learning_rate": 1.696474437712175e-05,
2815
+ "loss": 0.6932,
2816
+ "step": 113750
2817
+ },
2818
+ {
2819
+ "epoch": 6.45,
2820
+ "learning_rate": 1.684788075752831e-05,
2821
+ "loss": 0.6933,
2822
+ "step": 114000
2823
+ },
2824
+ {
2825
+ "epoch": 6.47,
2826
+ "learning_rate": 1.6731216129856575e-05,
2827
+ "loss": 0.6928,
2828
+ "step": 114250
2829
+ },
2830
+ {
2831
+ "epoch": 6.48,
2832
+ "learning_rate": 1.6614753341871385e-05,
2833
+ "loss": 0.6933,
2834
+ "step": 114500
2835
+ },
2836
+ {
2837
+ "epoch": 6.5,
2838
+ "learning_rate": 1.6498959857342135e-05,
2839
+ "loss": 0.6934,
2840
+ "step": 114750
2841
+ },
2842
+ {
2843
+ "epoch": 6.51,
2844
+ "learning_rate": 1.6382908436519833e-05,
2845
+ "loss": 0.6933,
2846
+ "step": 115000
2847
+ },
2848
+ {
2849
+ "epoch": 6.52,
2850
+ "learning_rate": 1.6267067357518934e-05,
2851
+ "loss": 0.6932,
2852
+ "step": 115250
2853
+ },
2854
+ {
2855
+ "epoch": 6.54,
2856
+ "learning_rate": 1.615143944800157e-05,
2857
+ "loss": 0.6932,
2858
+ "step": 115500
2859
+ },
2860
+ {
2861
+ "epoch": 6.55,
2862
+ "learning_rate": 1.6036027530426446e-05,
2863
+ "loss": 0.6932,
2864
+ "step": 115750
2865
+ },
2866
+ {
2867
+ "epoch": 6.57,
2868
+ "learning_rate": 1.592083442197995e-05,
2869
+ "loss": 0.693,
2870
+ "step": 116000
2871
+ },
2872
+ {
2873
+ "epoch": 6.58,
2874
+ "learning_rate": 1.5805862934507337e-05,
2875
+ "loss": 0.6932,
2876
+ "step": 116250
2877
+ },
2878
+ {
2879
+ "epoch": 6.6,
2880
+ "learning_rate": 1.5691115874444174e-05,
2881
+ "loss": 0.6931,
2882
+ "step": 116500
2883
+ },
2884
+ {
2885
+ "epoch": 6.61,
2886
+ "learning_rate": 1.5576596042747766e-05,
2887
+ "loss": 0.6928,
2888
+ "step": 116750
2889
+ },
2890
+ {
2891
+ "epoch": 6.62,
2892
+ "learning_rate": 1.546230623482881e-05,
2893
+ "loss": 0.6937,
2894
+ "step": 117000
2895
+ },
2896
+ {
2897
+ "epoch": 6.64,
2898
+ "learning_rate": 1.534870500100411e-05,
2899
+ "loss": 0.6932,
2900
+ "step": 117250
2901
+ },
2902
+ {
2903
+ "epoch": 6.65,
2904
+ "learning_rate": 1.5234882656415627e-05,
2905
+ "loss": 0.6931,
2906
+ "step": 117500
2907
+ },
2908
+ {
2909
+ "epoch": 6.67,
2910
+ "learning_rate": 1.5121298676773482e-05,
2911
+ "loss": 0.6931,
2912
+ "step": 117750
2913
+ },
2914
+ {
2915
+ "epoch": 6.68,
2916
+ "learning_rate": 1.5007955834644389e-05,
2917
+ "loss": 0.6932,
2918
+ "step": 118000
2919
+ },
2920
+ {
2921
+ "epoch": 6.69,
2922
+ "learning_rate": 1.4894856896708926e-05,
2923
+ "loss": 0.6933,
2924
+ "step": 118250
2925
+ },
2926
+ {
2927
+ "epoch": 6.71,
2928
+ "learning_rate": 1.478200462369401e-05,
2929
+ "loss": 0.6935,
2930
+ "step": 118500
2931
+ },
2932
+ {
2933
+ "epoch": 6.72,
2934
+ "learning_rate": 1.4669401770305513e-05,
2935
+ "loss": 0.6933,
2936
+ "step": 118750
2937
+ },
2938
+ {
2939
+ "epoch": 6.74,
2940
+ "learning_rate": 1.4557051085160978e-05,
2941
+ "loss": 0.6932,
2942
+ "step": 119000
2943
+ },
2944
+ {
2945
+ "epoch": 6.75,
2946
+ "learning_rate": 1.4445403182409694e-05,
2947
+ "loss": 0.6931,
2948
+ "step": 119250
2949
+ },
2950
+ {
2951
+ "epoch": 6.77,
2952
+ "learning_rate": 1.4333564018887296e-05,
2953
+ "loss": 0.6931,
2954
+ "step": 119500
2955
+ },
2956
+ {
2957
+ "epoch": 6.78,
2958
+ "learning_rate": 1.422198522135445e-05,
2959
+ "loss": 0.6932,
2960
+ "step": 119750
2961
+ },
2962
+ {
2963
+ "epoch": 6.79,
2964
+ "learning_rate": 1.4110669513431698e-05,
2965
+ "loss": 0.6931,
2966
+ "step": 120000
2967
+ },
2968
+ {
2969
+ "epoch": 6.81,
2970
+ "learning_rate": 1.39996196123176e-05,
2971
+ "loss": 0.6929,
2972
+ "step": 120250
2973
+ },
2974
+ {
2975
+ "epoch": 6.82,
2976
+ "learning_rate": 1.3888838228722412e-05,
2977
+ "loss": 0.6934,
2978
+ "step": 120500
2979
+ },
2980
+ {
2981
+ "epoch": 6.84,
2982
+ "learning_rate": 1.377832806680191e-05,
2983
+ "loss": 0.6932,
2984
+ "step": 120750
2985
+ },
2986
+ {
2987
+ "epoch": 6.85,
2988
+ "learning_rate": 1.3668091824091411e-05,
2989
+ "loss": 0.6932,
2990
+ "step": 121000
2991
+ },
2992
+ {
2993
+ "epoch": 6.86,
2994
+ "learning_rate": 1.3558132191439892e-05,
2995
+ "loss": 0.6931,
2996
+ "step": 121250
2997
+ },
2998
+ {
2999
+ "epoch": 6.88,
3000
+ "learning_rate": 1.3448890014393772e-05,
3001
+ "loss": 0.6933,
3002
+ "step": 121500
3003
+ },
3004
+ {
3005
+ "epoch": 6.89,
3006
+ "learning_rate": 1.3339490514123848e-05,
3007
+ "loss": 0.6934,
3008
+ "step": 121750
3009
+ },
3010
+ {
3011
+ "epoch": 6.91,
3012
+ "learning_rate": 1.3230375645018111e-05,
3013
+ "loss": 0.6934,
3014
+ "step": 122000
3015
+ },
3016
+ {
3017
+ "epoch": 6.92,
3018
+ "learning_rate": 1.3121548070553007e-05,
3019
+ "loss": 0.6926,
3020
+ "step": 122250
3021
+ },
3022
+ {
3023
+ "epoch": 6.94,
3024
+ "learning_rate": 1.3013010447192154e-05,
3025
+ "loss": 0.6932,
3026
+ "step": 122500
3027
+ },
3028
+ {
3029
+ "epoch": 6.95,
3030
+ "learning_rate": 1.2904765424321502e-05,
3031
+ "loss": 0.6934,
3032
+ "step": 122750
3033
+ },
3034
+ {
3035
+ "epoch": 6.96,
3036
+ "learning_rate": 1.2796815644184693e-05,
3037
+ "loss": 0.6933,
3038
+ "step": 123000
3039
+ },
3040
+ {
3041
+ "epoch": 6.98,
3042
+ "learning_rate": 1.2689163741818505e-05,
3043
+ "loss": 0.6932,
3044
+ "step": 123250
3045
+ },
3046
+ {
3047
+ "epoch": 6.99,
3048
+ "learning_rate": 1.2581812344988603e-05,
3049
+ "loss": 0.6932,
3050
+ "step": 123500
3051
+ },
3052
+ {
3053
+ "epoch": 7.0,
3054
+ "eval_accuracy": 0.5002547770700637,
3055
+ "eval_f1": 0.33344654835696697,
3056
+ "eval_loss": 0.6931138634681702,
3057
+ "eval_precision": 0.25012738853503186,
3058
+ "eval_recall": 0.5,
3059
+ "eval_runtime": 12.5558,
3060
+ "eval_samples_per_second": 625.208,
3061
+ "eval_steps_per_second": 78.211,
3062
+ "step": 123641
3063
+ },
3064
+ {
3065
+ "epoch": 7.01,
3066
+ "learning_rate": 1.2475191659916228e-05,
3067
+ "loss": 0.6933,
3068
+ "step": 123750
3069
+ },
3070
+ {
3071
+ "epoch": 7.02,
3072
+ "learning_rate": 1.2368447899899433e-05,
3073
+ "loss": 0.6933,
3074
+ "step": 124000
3075
+ },
3076
+ {
3077
+ "epoch": 7.03,
3078
+ "learning_rate": 1.2262012474041122e-05,
3079
+ "loss": 0.6932,
3080
+ "step": 124250
3081
+ },
3082
+ {
3083
+ "epoch": 7.05,
3084
+ "learning_rate": 1.2155887980412958e-05,
3085
+ "loss": 0.693,
3086
+ "step": 124500
3087
+ },
3088
+ {
3089
+ "epoch": 7.06,
3090
+ "learning_rate": 1.2050077009496776e-05,
3091
+ "loss": 0.6934,
3092
+ "step": 124750
3093
+ },
3094
+ {
3095
+ "epoch": 7.08,
3096
+ "learning_rate": 1.1944582144121394e-05,
3097
+ "loss": 0.6933,
3098
+ "step": 125000
3099
+ },
3100
+ {
3101
+ "epoch": 7.09,
3102
+ "learning_rate": 1.1839405959399536e-05,
3103
+ "loss": 0.6931,
3104
+ "step": 125250
3105
+ },
3106
+ {
3107
+ "epoch": 7.11,
3108
+ "learning_rate": 1.1734551022664981e-05,
3109
+ "loss": 0.693,
3110
+ "step": 125500
3111
+ },
3112
+ {
3113
+ "epoch": 7.12,
3114
+ "learning_rate": 1.1630437369518046e-05,
3115
+ "loss": 0.6935,
3116
+ "step": 125750
3117
+ },
3118
+ {
3119
+ "epoch": 7.13,
3120
+ "learning_rate": 1.152623128882203e-05,
3121
+ "loss": 0.6933,
3122
+ "step": 126000
3123
+ },
3124
+ {
3125
+ "epoch": 7.15,
3126
+ "learning_rate": 1.1422354100656728e-05,
3127
+ "loss": 0.6931,
3128
+ "step": 126250
3129
+ },
3130
+ {
3131
+ "epoch": 7.16,
3132
+ "learning_rate": 1.1318808340647638e-05,
3133
+ "loss": 0.6927,
3134
+ "step": 126500
3135
+ },
3136
+ {
3137
+ "epoch": 7.18,
3138
+ "learning_rate": 1.1215596536330131e-05,
3139
+ "loss": 0.6935,
3140
+ "step": 126750
3141
+ },
3142
+ {
3143
+ "epoch": 7.19,
3144
+ "learning_rate": 1.1112721207087779e-05,
3145
+ "loss": 0.6933,
3146
+ "step": 127000
3147
+ },
3148
+ {
3149
+ "epoch": 7.2,
3150
+ "learning_rate": 1.1010184864090856e-05,
3151
+ "loss": 0.6933,
3152
+ "step": 127250
3153
+ },
3154
+ {
3155
+ "epoch": 7.22,
3156
+ "learning_rate": 1.0907990010235016e-05,
3157
+ "loss": 0.6933,
3158
+ "step": 127500
3159
+ },
3160
+ {
3161
+ "epoch": 7.23,
3162
+ "learning_rate": 1.080613914008024e-05,
3163
+ "loss": 0.6932,
3164
+ "step": 127750
3165
+ },
3166
+ {
3167
+ "epoch": 7.25,
3168
+ "learning_rate": 1.0704634739789915e-05,
3169
+ "loss": 0.6931,
3170
+ "step": 128000
3171
+ },
3172
+ {
3173
+ "epoch": 7.26,
3174
+ "learning_rate": 1.06038832105021e-05,
3175
+ "loss": 0.6933,
3176
+ "step": 128250
3177
+ },
3178
+ {
3179
+ "epoch": 7.28,
3180
+ "learning_rate": 1.0503077763966945e-05,
3181
+ "loss": 0.6933,
3182
+ "step": 128500
3183
+ },
3184
+ {
3185
+ "epoch": 7.29,
3186
+ "learning_rate": 1.0402626184975775e-05,
3187
+ "loss": 0.6933,
3188
+ "step": 128750
3189
+ },
3190
+ {
3191
+ "epoch": 7.3,
3192
+ "learning_rate": 1.0302530925535489e-05,
3193
+ "loss": 0.6933,
3194
+ "step": 129000
3195
+ },
3196
+ {
3197
+ "epoch": 7.32,
3198
+ "learning_rate": 1.0202794428955301e-05,
3199
+ "loss": 0.6932,
3200
+ "step": 129250
3201
+ },
3202
+ {
3203
+ "epoch": 7.33,
3204
+ "learning_rate": 1.0103419129787083e-05,
3205
+ "loss": 0.6932,
3206
+ "step": 129500
3207
+ },
3208
+ {
3209
+ "epoch": 7.35,
3210
+ "learning_rate": 1.0004407453765927e-05,
3211
+ "loss": 0.693,
3212
+ "step": 129750
3213
+ },
3214
+ {
3215
+ "epoch": 7.36,
3216
+ "learning_rate": 9.905761817750958e-06,
3217
+ "loss": 0.6934,
3218
+ "step": 130000
3219
+ },
3220
+ {
3221
+ "epoch": 7.37,
3222
+ "learning_rate": 9.807484629666289e-06,
3223
+ "loss": 0.6929,
3224
+ "step": 130250
3225
+ },
3226
+ {
3227
+ "epoch": 7.39,
3228
+ "learning_rate": 9.709969171909833e-06,
3229
+ "loss": 0.6931,
3230
+ "step": 130500
3231
+ },
3232
+ {
3233
+ "epoch": 7.4,
3234
+ "learning_rate": 9.612434569729081e-06,
3235
+ "loss": 0.6935,
3236
+ "step": 130750
3237
+ },
3238
+ {
3239
+ "epoch": 7.42,
3240
+ "learning_rate": 9.515275575549665e-06,
3241
+ "loss": 0.6932,
3242
+ "step": 131000
3243
+ },
3244
+ {
3245
+ "epoch": 7.43,
3246
+ "learning_rate": 9.418494561007033e-06,
3247
+ "loss": 0.6932,
3248
+ "step": 131250
3249
+ },
3250
+ {
3251
+ "epoch": 7.44,
3252
+ "learning_rate": 9.322093888510217e-06,
3253
+ "loss": 0.6933,
3254
+ "step": 131500
3255
+ },
3256
+ {
3257
+ "epoch": 7.46,
3258
+ "learning_rate": 9.226075911184137e-06,
3259
+ "loss": 0.6932,
3260
+ "step": 131750
3261
+ },
3262
+ {
3263
+ "epoch": 7.47,
3264
+ "learning_rate": 9.130442972812207e-06,
3265
+ "loss": 0.6933,
3266
+ "step": 132000
3267
+ },
3268
+ {
3269
+ "epoch": 7.49,
3270
+ "learning_rate": 9.0351974077791e-06,
3271
+ "loss": 0.6932,
3272
+ "step": 132250
3273
+ },
3274
+ {
3275
+ "epoch": 7.5,
3276
+ "learning_rate": 8.940720185129978e-06,
3277
+ "loss": 0.6932,
3278
+ "step": 132500
3279
+ },
3280
+ {
3281
+ "epoch": 7.52,
3282
+ "learning_rate": 8.846254759394507e-06,
3283
+ "loss": 0.6931,
3284
+ "step": 132750
3285
+ },
3286
+ {
3287
+ "epoch": 7.53,
3288
+ "learning_rate": 8.752183643986523e-06,
3289
+ "loss": 0.6932,
3290
+ "step": 133000
3291
+ },
3292
+ {
3293
+ "epoch": 7.54,
3294
+ "learning_rate": 8.658509135166829e-06,
3295
+ "loss": 0.6933,
3296
+ "step": 133250
3297
+ },
3298
+ {
3299
+ "epoch": 7.56,
3300
+ "learning_rate": 8.565233519515157e-06,
3301
+ "loss": 0.6932,
3302
+ "step": 133500
3303
+ },
3304
+ {
3305
+ "epoch": 7.57,
3306
+ "learning_rate": 8.472359073874304e-06,
3307
+ "loss": 0.693,
3308
+ "step": 133750
3309
+ },
3310
+ {
3311
+ "epoch": 7.59,
3312
+ "learning_rate": 8.379888065294575e-06,
3313
+ "loss": 0.6934,
3314
+ "step": 134000
3315
+ },
3316
+ {
3317
+ "epoch": 7.6,
3318
+ "learning_rate": 8.287822750978453e-06,
3319
+ "loss": 0.6932,
3320
+ "step": 134250
3321
+ },
3322
+ {
3323
+ "epoch": 7.61,
3324
+ "learning_rate": 8.196531192128304e-06,
3325
+ "loss": 0.6932,
3326
+ "step": 134500
3327
+ },
3328
+ {
3329
+ "epoch": 7.63,
3330
+ "learning_rate": 8.105282353121044e-06,
3331
+ "loss": 0.6931,
3332
+ "step": 134750
3333
+ },
3334
+ {
3335
+ "epoch": 7.64,
3336
+ "learning_rate": 8.014445911458718e-06,
3337
+ "loss": 0.6934,
3338
+ "step": 135000
3339
+ },
3340
+ {
3341
+ "epoch": 7.66,
3342
+ "learning_rate": 7.924024084444284e-06,
3343
+ "loss": 0.6932,
3344
+ "step": 135250
3345
+ },
3346
+ {
3347
+ "epoch": 7.67,
3348
+ "learning_rate": 7.834019079260019e-06,
3349
+ "loss": 0.6931,
3350
+ "step": 135500
3351
+ },
3352
+ {
3353
+ "epoch": 7.69,
3354
+ "learning_rate": 7.744433092913655e-06,
3355
+ "loss": 0.6931,
3356
+ "step": 135750
3357
+ },
3358
+ {
3359
+ "epoch": 7.7,
3360
+ "learning_rate": 7.655268312184721e-06,
3361
+ "loss": 0.693,
3362
+ "step": 136000
3363
+ },
3364
+ {
3365
+ "epoch": 7.71,
3366
+ "learning_rate": 7.5665269135712034e-06,
3367
+ "loss": 0.6934,
3368
+ "step": 136250
3369
+ },
3370
+ {
3371
+ "epoch": 7.73,
3372
+ "learning_rate": 7.478563476084949e-06,
3373
+ "loss": 0.6932,
3374
+ "step": 136500
3375
+ },
3376
+ {
3377
+ "epoch": 7.74,
3378
+ "learning_rate": 7.390673614707866e-06,
3379
+ "loss": 0.6931,
3380
+ "step": 136750
3381
+ },
3382
+ {
3383
+ "epoch": 7.76,
3384
+ "learning_rate": 7.3032135941602745e-06,
3385
+ "loss": 0.6932,
3386
+ "step": 137000
3387
+ },
3388
+ {
3389
+ "epoch": 7.77,
3390
+ "learning_rate": 7.216185549327245e-06,
3391
+ "loss": 0.6931,
3392
+ "step": 137250
3393
+ },
3394
+ {
3395
+ "epoch": 7.78,
3396
+ "learning_rate": 7.129591604549363e-06,
3397
+ "loss": 0.6931,
3398
+ "step": 137500
3399
+ },
3400
+ {
3401
+ "epoch": 7.8,
3402
+ "learning_rate": 7.043433873570918e-06,
3403
+ "loss": 0.6932,
3404
+ "step": 137750
3405
+ },
3406
+ {
3407
+ "epoch": 7.81,
3408
+ "learning_rate": 6.957714459488293e-06,
3409
+ "loss": 0.6934,
3410
+ "step": 138000
3411
+ },
3412
+ {
3413
+ "epoch": 7.83,
3414
+ "learning_rate": 6.87243545469862e-06,
3415
+ "loss": 0.693,
3416
+ "step": 138250
3417
+ },
3418
+ {
3419
+ "epoch": 7.84,
3420
+ "learning_rate": 6.787937402713973e-06,
3421
+ "loss": 0.6934,
3422
+ "step": 138500
3423
+ },
3424
+ {
3425
+ "epoch": 7.86,
3426
+ "learning_rate": 6.703543668291617e-06,
3427
+ "loss": 0.6932,
3428
+ "step": 138750
3429
+ },
3430
+ {
3431
+ "epoch": 7.87,
3432
+ "learning_rate": 6.619596547430409e-06,
3433
+ "loss": 0.693,
3434
+ "step": 139000
3435
+ },
3436
+ {
3437
+ "epoch": 7.88,
3438
+ "learning_rate": 6.536098089266093e-06,
3439
+ "loss": 0.6933,
3440
+ "step": 139250
3441
+ },
3442
+ {
3443
+ "epoch": 7.9,
3444
+ "learning_rate": 6.453050331982624e-06,
3445
+ "loss": 0.6933,
3446
+ "step": 139500
3447
+ },
3448
+ {
3449
+ "epoch": 7.91,
3450
+ "learning_rate": 6.370455302762429e-06,
3451
+ "loss": 0.6931,
3452
+ "step": 139750
3453
+ },
3454
+ {
3455
+ "epoch": 7.93,
3456
+ "learning_rate": 6.2883150177369095e-06,
3457
+ "loss": 0.6933,
3458
+ "step": 140000
3459
+ },
3460
+ {
3461
+ "epoch": 7.94,
3462
+ "learning_rate": 6.206631481937219e-06,
3463
+ "loss": 0.6933,
3464
+ "step": 140250
3465
+ },
3466
+ {
3467
+ "epoch": 7.95,
3468
+ "learning_rate": 6.125406689245361e-06,
3469
+ "loss": 0.693,
3470
+ "step": 140500
3471
+ },
3472
+ {
3473
+ "epoch": 7.97,
3474
+ "learning_rate": 6.04496475823072e-06,
3475
+ "loss": 0.6931,
3476
+ "step": 140750
3477
+ },
3478
+ {
3479
+ "epoch": 7.98,
3480
+ "learning_rate": 5.964661533859653e-06,
3481
+ "loss": 0.6934,
3482
+ "step": 141000
3483
+ },
3484
+ {
3485
+ "epoch": 8.0,
3486
+ "learning_rate": 5.884822959043998e-06,
3487
+ "loss": 0.6933,
3488
+ "step": 141250
3489
+ },
3490
+ {
3491
+ "epoch": 8.0,
3492
+ "eval_accuracy": 0.4997452229299363,
3493
+ "eval_f1": 0.3332200798437102,
3494
+ "eval_loss": 0.6931153535842896,
3495
+ "eval_precision": 0.24987261146496814,
3496
+ "eval_recall": 0.5,
3497
+ "eval_runtime": 12.5013,
3498
+ "eval_samples_per_second": 627.937,
3499
+ "eval_steps_per_second": 78.552,
3500
+ "step": 141304
3501
+ },
3502
+ {
3503
+ "epoch": 8.01,
3504
+ "learning_rate": 5.805450982630542e-06,
3505
+ "loss": 0.6931,
3506
+ "step": 141500
3507
+ },
3508
+ {
3509
+ "epoch": 8.03,
3510
+ "learning_rate": 5.726547542076485e-06,
3511
+ "loss": 0.6932,
3512
+ "step": 141750
3513
+ },
3514
+ {
3515
+ "epoch": 8.04,
3516
+ "learning_rate": 5.6481145634021515e-06,
3517
+ "loss": 0.6931,
3518
+ "step": 142000
3519
+ },
3520
+ {
3521
+ "epoch": 8.05,
3522
+ "learning_rate": 5.570153961143942e-06,
3523
+ "loss": 0.6932,
3524
+ "step": 142250
3525
+ },
3526
+ {
3527
+ "epoch": 8.07,
3528
+ "learning_rate": 5.492667638307647e-06,
3529
+ "loss": 0.6932,
3530
+ "step": 142500
3531
+ },
3532
+ {
3533
+ "epoch": 8.08,
3534
+ "learning_rate": 5.415964575902305e-06,
3535
+ "loss": 0.6933,
3536
+ "step": 142750
3537
+ },
3538
+ {
3539
+ "epoch": 8.1,
3540
+ "learning_rate": 5.339430558640929e-06,
3541
+ "loss": 0.693,
3542
+ "step": 143000
3543
+ },
3544
+ {
3545
+ "epoch": 8.11,
3546
+ "learning_rate": 5.263376452722673e-06,
3547
+ "loss": 0.6931,
3548
+ "step": 143250
3549
+ },
3550
+ {
3551
+ "epoch": 8.12,
3552
+ "learning_rate": 5.187804114616051e-06,
3553
+ "loss": 0.6933,
3554
+ "step": 143500
3555
+ },
3556
+ {
3557
+ "epoch": 8.14,
3558
+ "learning_rate": 5.112715389029707e-06,
3559
+ "loss": 0.6932,
3560
+ "step": 143750
3561
+ },
3562
+ {
3563
+ "epoch": 8.15,
3564
+ "learning_rate": 5.038112108867363e-06,
3565
+ "loss": 0.6931,
3566
+ "step": 144000
3567
+ },
3568
+ {
3569
+ "epoch": 8.17,
3570
+ "learning_rate": 4.963996095183115e-06,
3571
+ "loss": 0.6932,
3572
+ "step": 144250
3573
+ },
3574
+ {
3575
+ "epoch": 8.18,
3576
+ "learning_rate": 4.890369157136956e-06,
3577
+ "loss": 0.6931,
3578
+ "step": 144500
3579
+ },
3580
+ {
3581
+ "epoch": 8.2,
3582
+ "learning_rate": 4.817233091950621e-06,
3583
+ "loss": 0.6932,
3584
+ "step": 144750
3585
+ },
3586
+ {
3587
+ "epoch": 8.21,
3588
+ "learning_rate": 4.744879274763003e-06,
3589
+ "loss": 0.6934,
3590
+ "step": 145000
3591
+ },
3592
+ {
3593
+ "epoch": 8.22,
3594
+ "learning_rate": 4.672728317747921e-06,
3595
+ "loss": 0.6934,
3596
+ "step": 145250
3597
+ },
3598
+ {
3599
+ "epoch": 8.24,
3600
+ "learning_rate": 4.601073546170611e-06,
3601
+ "loss": 0.6932,
3602
+ "step": 145500
3603
+ },
3604
+ {
3605
+ "epoch": 8.25,
3606
+ "learning_rate": 4.529916709112531e-06,
3607
+ "loss": 0.693,
3608
+ "step": 145750
3609
+ },
3610
+ {
3611
+ "epoch": 8.27,
3612
+ "learning_rate": 4.459259543500649e-06,
3613
+ "loss": 0.6932,
3614
+ "step": 146000
3615
+ },
3616
+ {
3617
+ "epoch": 8.28,
3618
+ "learning_rate": 4.38910377406504e-06,
3619
+ "loss": 0.6932,
3620
+ "step": 146250
3621
+ },
3622
+ {
3623
+ "epoch": 8.29,
3624
+ "learning_rate": 4.319451113296763e-06,
3625
+ "loss": 0.6931,
3626
+ "step": 146500
3627
+ },
3628
+ {
3629
+ "epoch": 8.31,
3630
+ "learning_rate": 4.250303261406091e-06,
3631
+ "loss": 0.6933,
3632
+ "step": 146750
3633
+ },
3634
+ {
3635
+ "epoch": 8.32,
3636
+ "learning_rate": 4.1819354605351554e-06,
3637
+ "loss": 0.6934,
3638
+ "step": 147000
3639
+ },
3640
+ {
3641
+ "epoch": 8.34,
3642
+ "learning_rate": 4.113800241689689e-06,
3643
+ "loss": 0.6931,
3644
+ "step": 147250
3645
+ },
3646
+ {
3647
+ "epoch": 8.35,
3648
+ "learning_rate": 4.046174851626572e-06,
3649
+ "loss": 0.6932,
3650
+ "step": 147500
3651
+ },
3652
+ {
3653
+ "epoch": 8.36,
3654
+ "learning_rate": 3.979060941070722e-06,
3655
+ "loss": 0.6932,
3656
+ "step": 147750
3657
+ },
3658
+ {
3659
+ "epoch": 8.38,
3660
+ "learning_rate": 3.9124601482619205e-06,
3661
+ "loss": 0.6932,
3662
+ "step": 148000
3663
+ },
3664
+ {
3665
+ "epoch": 8.39,
3666
+ "learning_rate": 3.846374098914826e-06,
3667
+ "loss": 0.693,
3668
+ "step": 148250
3669
+ },
3670
+ {
3671
+ "epoch": 8.41,
3672
+ "learning_rate": 3.780804406179295e-06,
3673
+ "loss": 0.6931,
3674
+ "step": 148500
3675
+ },
3676
+ {
3677
+ "epoch": 8.42,
3678
+ "learning_rate": 3.715752670600986e-06,
3679
+ "loss": 0.6934,
3680
+ "step": 148750
3681
+ },
3682
+ {
3683
+ "epoch": 8.44,
3684
+ "learning_rate": 3.651220480082326e-06,
3685
+ "loss": 0.6932,
3686
+ "step": 149000
3687
+ },
3688
+ {
3689
+ "epoch": 8.45,
3690
+ "learning_rate": 3.587209409843728e-06,
3691
+ "loss": 0.6931,
3692
+ "step": 149250
3693
+ },
3694
+ {
3695
+ "epoch": 8.46,
3696
+ "learning_rate": 3.523721022385132e-06,
3697
+ "loss": 0.6932,
3698
+ "step": 149500
3699
+ },
3700
+ {
3701
+ "epoch": 8.48,
3702
+ "learning_rate": 3.4607568674478897e-06,
3703
+ "loss": 0.6933,
3704
+ "step": 149750
3705
+ },
3706
+ {
3707
+ "epoch": 8.49,
3708
+ "learning_rate": 3.3983184819769e-06,
3709
+ "loss": 0.6932,
3710
+ "step": 150000
3711
+ },
3712
+ {
3713
+ "epoch": 8.51,
3714
+ "learning_rate": 3.3366539820747027e-06,
3715
+ "loss": 0.6932,
3716
+ "step": 150250
3717
+ },
3718
+ {
3719
+ "epoch": 8.52,
3720
+ "learning_rate": 3.2752695767854113e-06,
3721
+ "loss": 0.6931,
3722
+ "step": 150500
3723
+ },
3724
+ {
3725
+ "epoch": 8.53,
3726
+ "learning_rate": 3.2144154686773565e-06,
3727
+ "loss": 0.6933,
3728
+ "step": 150750
3729
+ },
3730
+ {
3731
+ "epoch": 8.55,
3732
+ "learning_rate": 3.1540931431895514e-06,
3733
+ "loss": 0.6932,
3734
+ "step": 151000
3735
+ },
3736
+ {
3737
+ "epoch": 8.56,
3738
+ "learning_rate": 3.0943040727802797e-06,
3739
+ "loss": 0.6932,
3740
+ "step": 151250
3741
+ },
3742
+ {
3743
+ "epoch": 8.58,
3744
+ "learning_rate": 3.0350497168911334e-06,
3745
+ "loss": 0.6931,
3746
+ "step": 151500
3747
+ },
3748
+ {
3749
+ "epoch": 8.59,
3750
+ "learning_rate": 2.976331521911438e-06,
3751
+ "loss": 0.6931,
3752
+ "step": 151750
3753
+ },
3754
+ {
3755
+ "epoch": 8.61,
3756
+ "learning_rate": 2.918150921142901e-06,
3757
+ "loss": 0.6933,
3758
+ "step": 152000
3759
+ },
3760
+ {
3761
+ "epoch": 8.62,
3762
+ "learning_rate": 2.860509334764647e-06,
3763
+ "loss": 0.6932,
3764
+ "step": 152250
3765
+ },
3766
+ {
3767
+ "epoch": 8.63,
3768
+ "learning_rate": 2.8034081697985437e-06,
3769
+ "loss": 0.6931,
3770
+ "step": 152500
3771
+ },
3772
+ {
3773
+ "epoch": 8.65,
3774
+ "learning_rate": 2.746848820074854e-06,
3775
+ "loss": 0.6933,
3776
+ "step": 152750
3777
+ },
3778
+ {
3779
+ "epoch": 8.66,
3780
+ "learning_rate": 2.6910556469509853e-06,
3781
+ "loss": 0.6934,
3782
+ "step": 153000
3783
+ },
3784
+ {
3785
+ "epoch": 8.68,
3786
+ "learning_rate": 2.6355818753078243e-06,
3787
+ "loss": 0.6932,
3788
+ "step": 153250
3789
+ },
3790
+ {
3791
+ "epoch": 8.69,
3792
+ "learning_rate": 2.5806540155199815e-06,
3793
+ "loss": 0.6932,
3794
+ "step": 153500
3795
+ },
3796
+ {
3797
+ "epoch": 8.7,
3798
+ "learning_rate": 2.5262734083676936e-06,
3799
+ "loss": 0.6933,
3800
+ "step": 153750
3801
+ },
3802
+ {
3803
+ "epoch": 8.72,
3804
+ "learning_rate": 2.4724413812728525e-06,
3805
+ "loss": 0.6932,
3806
+ "step": 154000
3807
+ },
3808
+ {
3809
+ "epoch": 8.73,
3810
+ "learning_rate": 2.4191592482665893e-06,
3811
+ "loss": 0.693,
3812
+ "step": 154250
3813
+ },
3814
+ {
3815
+ "epoch": 8.75,
3816
+ "learning_rate": 2.3664283099572192e-06,
3817
+ "loss": 0.6933,
3818
+ "step": 154500
3819
+ },
3820
+ {
3821
+ "epoch": 8.76,
3822
+ "learning_rate": 2.3142498534984735e-06,
3823
+ "loss": 0.693,
3824
+ "step": 154750
3825
+ },
3826
+ {
3827
+ "epoch": 8.78,
3828
+ "learning_rate": 2.262625152558104e-06,
3829
+ "loss": 0.6932,
3830
+ "step": 155000
3831
+ },
3832
+ {
3833
+ "epoch": 8.79,
3834
+ "learning_rate": 2.211758638779948e-06,
3835
+ "loss": 0.6932,
3836
+ "step": 155250
3837
+ },
3838
+ {
3839
+ "epoch": 8.8,
3840
+ "learning_rate": 2.161242988266132e-06,
3841
+ "loss": 0.6932,
3842
+ "step": 155500
3843
+ },
3844
+ {
3845
+ "epoch": 8.82,
3846
+ "learning_rate": 2.111284828143695e-06,
3847
+ "loss": 0.6932,
3848
+ "step": 155750
3849
+ },
3850
+ {
3851
+ "epoch": 8.83,
3852
+ "learning_rate": 2.0618853778833075e-06,
3853
+ "loss": 0.6933,
3854
+ "step": 156000
3855
+ },
3856
+ {
3857
+ "epoch": 8.85,
3858
+ "learning_rate": 2.013045843317621e-06,
3859
+ "loss": 0.6933,
3860
+ "step": 156250
3861
+ },
3862
+ {
3863
+ "epoch": 8.86,
3864
+ "learning_rate": 1.964767416611829e-06,
3865
+ "loss": 0.6933,
3866
+ "step": 156500
3867
+ },
3868
+ {
3869
+ "epoch": 8.87,
3870
+ "learning_rate": 1.9170512762345825e-06,
3871
+ "loss": 0.6932,
3872
+ "step": 156750
3873
+ },
3874
+ {
3875
+ "epoch": 8.89,
3876
+ "learning_rate": 1.8698985869291973e-06,
3877
+ "loss": 0.6932,
3878
+ "step": 157000
3879
+ },
3880
+ {
3881
+ "epoch": 8.9,
3882
+ "learning_rate": 1.8234957258351364e-06,
3883
+ "loss": 0.6931,
3884
+ "step": 157250
3885
+ },
3886
+ {
3887
+ "epoch": 8.92,
3888
+ "learning_rate": 1.7774711126562565e-06,
3889
+ "loss": 0.6933,
3890
+ "step": 157500
3891
+ },
3892
+ {
3893
+ "epoch": 8.93,
3894
+ "learning_rate": 1.732013357678619e-06,
3895
+ "loss": 0.6933,
3896
+ "step": 157750
3897
+ },
3898
+ {
3899
+ "epoch": 8.95,
3900
+ "learning_rate": 1.6871235705187283e-06,
3901
+ "loss": 0.6931,
3902
+ "step": 158000
3903
+ },
3904
+ {
3905
+ "epoch": 8.96,
3906
+ "learning_rate": 1.6428028469290813e-06,
3907
+ "loss": 0.693,
3908
+ "step": 158250
3909
+ },
3910
+ {
3911
+ "epoch": 8.97,
3912
+ "learning_rate": 1.5990522687714172e-06,
3913
+ "loss": 0.693,
3914
+ "step": 158500
3915
+ },
3916
+ {
3917
+ "epoch": 8.99,
3918
+ "learning_rate": 1.5558729039903296e-06,
3919
+ "loss": 0.6933,
3920
+ "step": 158750
3921
+ },
3922
+ {
3923
+ "epoch": 9.0,
3924
+ "eval_accuracy": 0.4997452229299363,
3925
+ "eval_f1": 0.3332200798437102,
3926
+ "eval_loss": 0.6933611035346985,
3927
+ "eval_precision": 0.24987261146496814,
3928
+ "eval_recall": 0.5,
3929
+ "eval_runtime": 12.7446,
3930
+ "eval_samples_per_second": 615.949,
3931
+ "eval_steps_per_second": 77.052,
3932
+ "step": 158967
3933
+ },
3934
+ {
3935
+ "epoch": 9.0,
3936
+ "learning_rate": 1.5132658065871747e-06,
3937
+ "loss": 0.6932,
3938
+ "step": 159000
3939
+ },
3940
+ {
3941
+ "epoch": 9.02,
3942
+ "learning_rate": 1.471232016594365e-06,
3943
+ "loss": 0.6932,
3944
+ "step": 159250
3945
+ },
3946
+ {
3947
+ "epoch": 9.03,
3948
+ "learning_rate": 1.4299372524579662e-06,
3949
+ "loss": 0.6932,
3950
+ "step": 159500
3951
+ },
3952
+ {
3953
+ "epoch": 9.04,
3954
+ "learning_rate": 1.3890508380014433e-06,
3955
+ "loss": 0.6931,
3956
+ "step": 159750
3957
+ },
3958
+ {
3959
+ "epoch": 9.06,
3960
+ "learning_rate": 1.348740763022696e-06,
3961
+ "loss": 0.6931,
3962
+ "step": 160000
3963
+ },
3964
+ {
3965
+ "epoch": 9.07,
3966
+ "learning_rate": 1.3090080114841863e-06,
3967
+ "loss": 0.6934,
3968
+ "step": 160250
3969
+ },
3970
+ {
3971
+ "epoch": 9.09,
3972
+ "learning_rate": 1.269853553255998e-06,
3973
+ "loss": 0.6932,
3974
+ "step": 160500
3975
+ },
3976
+ {
3977
+ "epoch": 9.1,
3978
+ "learning_rate": 1.231278344092171e-06,
3979
+ "loss": 0.6933,
3980
+ "step": 160750
3981
+ },
3982
+ {
3983
+ "epoch": 9.12,
3984
+ "learning_rate": 1.1932833256073695e-06,
3985
+ "loss": 0.6931,
3986
+ "step": 161000
3987
+ },
3988
+ {
3989
+ "epoch": 9.13,
3990
+ "learning_rate": 1.155869425253886e-06,
3991
+ "loss": 0.6931,
3992
+ "step": 161250
3993
+ },
3994
+ {
3995
+ "epoch": 9.14,
3996
+ "learning_rate": 1.1191837231720042e-06,
3997
+ "loss": 0.6931,
3998
+ "step": 161500
3999
+ },
4000
+ {
4001
+ "epoch": 9.16,
4002
+ "learning_rate": 1.0829324511818617e-06,
4003
+ "loss": 0.6932,
4004
+ "step": 161750
4005
+ },
4006
+ {
4007
+ "epoch": 9.17,
4008
+ "learning_rate": 1.0472649909701676e-06,
4009
+ "loss": 0.6931,
4010
+ "step": 162000
4011
+ },
4012
+ {
4013
+ "epoch": 9.19,
4014
+ "learning_rate": 1.0121822131738956e-06,
4015
+ "loss": 0.6933,
4016
+ "step": 162250
4017
+ },
4018
+ {
4019
+ "epoch": 9.2,
4020
+ "learning_rate": 9.776849741580247e-07,
4021
+ "loss": 0.6932,
4022
+ "step": 162500
4023
+ },
4024
+ {
4025
+ "epoch": 9.21,
4026
+ "learning_rate": 9.437741159946139e-07,
4027
+ "loss": 0.6933,
4028
+ "step": 162750
4029
+ },
4030
+ {
4031
+ "epoch": 9.23,
4032
+ "learning_rate": 9.104504664422669e-07,
4033
+ "loss": 0.6932,
4034
+ "step": 163000
4035
+ },
4036
+ {
4037
+ "epoch": 9.24,
4038
+ "learning_rate": 8.777148389259194e-07,
4039
+ "loss": 0.6931,
4040
+ "step": 163250
4041
+ },
4042
+ {
4043
+ "epoch": 9.26,
4044
+ "learning_rate": 8.455680325169802e-07,
4045
+ "loss": 0.6933,
4046
+ "step": 163500
4047
+ },
4048
+ {
4049
+ "epoch": 9.27,
4050
+ "learning_rate": 8.140108319138362e-07,
4051
+ "loss": 0.6932,
4052
+ "step": 163750
4053
+ },
4054
+ {
4055
+ "epoch": 9.28,
4056
+ "learning_rate": 7.831666976848146e-07,
4057
+ "loss": 0.6932,
4058
+ "step": 164000
4059
+ },
4060
+ {
4061
+ "epoch": 9.3,
4062
+ "learning_rate": 7.527886391862193e-07,
4063
+ "loss": 0.6932,
4064
+ "step": 164250
4065
+ },
4066
+ {
4067
+ "epoch": 9.31,
4068
+ "learning_rate": 7.230024512235062e-07,
4069
+ "loss": 0.6932,
4070
+ "step": 164500
4071
+ },
4072
+ {
4073
+ "epoch": 9.33,
4074
+ "learning_rate": 6.938088608727439e-07,
4075
+ "loss": 0.6933,
4076
+ "step": 164750
4077
+ },
4078
+ {
4079
+ "epoch": 9.34,
4080
+ "learning_rate": 6.652085807447861e-07,
4081
+ "loss": 0.693,
4082
+ "step": 165000
4083
+ },
4084
+ {
4085
+ "epoch": 9.36,
4086
+ "learning_rate": 6.372023089678797e-07,
4087
+ "loss": 0.6932,
4088
+ "step": 165250
4089
+ },
4090
+ {
4091
+ "epoch": 9.37,
4092
+ "learning_rate": 6.097907291706201e-07,
4093
+ "loss": 0.693,
4094
+ "step": 165500
4095
+ },
4096
+ {
4097
+ "epoch": 9.38,
4098
+ "learning_rate": 5.829745104652673e-07,
4099
+ "loss": 0.6932,
4100
+ "step": 165750
4101
+ },
4102
+ {
4103
+ "epoch": 9.4,
4104
+ "learning_rate": 5.568580001272361e-07,
4105
+ "loss": 0.6932,
4106
+ "step": 166000
4107
+ },
4108
+ {
4109
+ "epoch": 9.41,
4110
+ "learning_rate": 5.31232064917378e-07,
4111
+ "loss": 0.6931,
4112
+ "step": 166250
4113
+ },
4114
+ {
4115
+ "epoch": 9.43,
4116
+ "learning_rate": 5.062034084038104e-07,
4117
+ "loss": 0.6932,
4118
+ "step": 166500
4119
+ },
4120
+ {
4121
+ "epoch": 9.44,
4122
+ "learning_rate": 4.817726415320162e-07,
4123
+ "loss": 0.6932,
4124
+ "step": 166750
4125
+ },
4126
+ {
4127
+ "epoch": 9.45,
4128
+ "learning_rate": 4.5794036065309466e-07,
4129
+ "loss": 0.6932,
4130
+ "step": 167000
4131
+ },
4132
+ {
4133
+ "epoch": 9.47,
4134
+ "learning_rate": 4.347071475091918e-07,
4135
+ "loss": 0.6932,
4136
+ "step": 167250
4137
+ },
4138
+ {
4139
+ "epoch": 9.48,
4140
+ "learning_rate": 4.120735692193151e-07,
4141
+ "loss": 0.6931,
4142
+ "step": 167500
4143
+ },
4144
+ {
4145
+ "epoch": 9.5,
4146
+ "learning_rate": 3.900401782654717e-07,
4147
+ "loss": 0.6932,
4148
+ "step": 167750
4149
+ },
4150
+ {
4151
+ "epoch": 9.51,
4152
+ "learning_rate": 3.686075124792021e-07,
4153
+ "loss": 0.6932,
4154
+ "step": 168000
4155
+ },
4156
+ {
4157
+ "epoch": 9.53,
4158
+ "learning_rate": 3.477760950284292e-07,
4159
+ "loss": 0.6932,
4160
+ "step": 168250
4161
+ },
4162
+ {
4163
+ "epoch": 9.54,
4164
+ "learning_rate": 3.2762615368826244e-07,
4165
+ "loss": 0.6932,
4166
+ "step": 168500
4167
+ },
4168
+ {
4169
+ "epoch": 9.55,
4170
+ "learning_rate": 3.0799633372767536e-07,
4171
+ "loss": 0.6931,
4172
+ "step": 168750
4173
+ },
4174
+ {
4175
+ "epoch": 9.57,
4176
+ "learning_rate": 2.889692416117279e-07,
4177
+ "loss": 0.6931,
4178
+ "step": 169000
4179
+ },
4180
+ {
4181
+ "epoch": 9.58,
4182
+ "learning_rate": 2.7054534178868607e-07,
4183
+ "loss": 0.6932,
4184
+ "step": 169250
4185
+ },
4186
+ {
4187
+ "epoch": 9.6,
4188
+ "learning_rate": 2.527250839829881e-07,
4189
+ "loss": 0.6932,
4190
+ "step": 169500
4191
+ },
4192
+ {
4193
+ "epoch": 9.61,
4194
+ "learning_rate": 2.3550890318425888e-07,
4195
+ "loss": 0.6932,
4196
+ "step": 169750
4197
+ },
4198
+ {
4199
+ "epoch": 9.62,
4200
+ "learning_rate": 2.1889721963671284e-07,
4201
+ "loss": 0.6931,
4202
+ "step": 170000
4203
+ },
4204
+ {
4205
+ "epoch": 9.64,
4206
+ "learning_rate": 2.0289043882887604e-07,
4207
+ "loss": 0.6932,
4208
+ "step": 170250
4209
+ },
4210
+ {
4211
+ "epoch": 9.65,
4212
+ "learning_rate": 1.8748895148369973e-07,
4213
+ "loss": 0.6933,
4214
+ "step": 170500
4215
+ },
4216
+ {
4217
+ "epoch": 9.67,
4218
+ "learning_rate": 1.727511098436996e-07,
4219
+ "loss": 0.6931,
4220
+ "step": 170750
4221
+ },
4222
+ {
4223
+ "epoch": 9.68,
4224
+ "learning_rate": 1.5855889766091536e-07,
4225
+ "loss": 0.6932,
4226
+ "step": 171000
4227
+ },
4228
+ {
4229
+ "epoch": 9.7,
4230
+ "learning_rate": 1.4497306106658893e-07,
4231
+ "loss": 0.693,
4232
+ "step": 171250
4233
+ },
4234
+ {
4235
+ "epoch": 9.71,
4236
+ "learning_rate": 1.3199393168881468e-07,
4237
+ "loss": 0.6932,
4238
+ "step": 171500
4239
+ },
4240
+ {
4241
+ "epoch": 9.72,
4242
+ "learning_rate": 1.1962182634605302e-07,
4243
+ "loss": 0.6932,
4244
+ "step": 171750
4245
+ },
4246
+ {
4247
+ "epoch": 9.74,
4248
+ "learning_rate": 1.0785704703941135e-07,
4249
+ "loss": 0.6932,
4250
+ "step": 172000
4251
+ },
4252
+ {
4253
+ "epoch": 9.75,
4254
+ "learning_rate": 9.669988094526128e-08,
4255
+ "loss": 0.6933,
4256
+ "step": 172250
4257
+ },
4258
+ {
4259
+ "epoch": 9.77,
4260
+ "learning_rate": 8.615060040823852e-08,
4261
+ "loss": 0.6931,
4262
+ "step": 172500
4263
+ },
4264
+ {
4265
+ "epoch": 9.78,
4266
+ "learning_rate": 7.620946293458442e-08,
4267
+ "loss": 0.6932,
4268
+ "step": 172750
4269
+ },
4270
+ {
4271
+ "epoch": 9.79,
4272
+ "learning_rate": 6.691282998163018e-08,
4273
+ "loss": 0.6931,
4274
+ "step": 173000
4275
+ },
4276
+ {
4277
+ "epoch": 9.81,
4278
+ "learning_rate": 5.8186256880568555e-08,
4279
+ "loss": 0.6933,
4280
+ "step": 173250
4281
+ },
4282
+ {
4283
+ "epoch": 9.82,
4284
+ "learning_rate": 5.006850944801467e-08,
4285
+ "loss": 0.6932,
4286
+ "step": 173500
4287
+ },
4288
+ {
4289
+ "epoch": 9.84,
4290
+ "learning_rate": 4.25597858368737e-08,
4291
+ "loss": 0.6934,
4292
+ "step": 173750
4293
+ },
4294
+ {
4295
+ "epoch": 9.85,
4296
+ "learning_rate": 3.5660269333887e-08,
4297
+ "loss": 0.6931,
4298
+ "step": 174000
4299
+ },
4300
+ {
4301
+ "epoch": 9.87,
4302
+ "learning_rate": 2.9370128355143966e-08,
4303
+ "loss": 0.6932,
4304
+ "step": 174250
4305
+ },
4306
+ {
4307
+ "epoch": 9.88,
4308
+ "learning_rate": 2.3689516441977034e-08,
4309
+ "loss": 0.6931,
4310
+ "step": 174500
4311
+ },
4312
+ {
4313
+ "epoch": 9.89,
4314
+ "learning_rate": 1.8618572257209087e-08,
4315
+ "loss": 0.6932,
4316
+ "step": 174750
4317
+ },
4318
+ {
4319
+ "epoch": 9.91,
4320
+ "learning_rate": 1.4174049338511186e-08,
4321
+ "loss": 0.6932,
4322
+ "step": 175000
4323
+ },
4324
+ {
4325
+ "epoch": 9.92,
4326
+ "learning_rate": 1.0320357269619218e-08,
4327
+ "loss": 0.6932,
4328
+ "step": 175250
4329
+ },
4330
+ {
4331
+ "epoch": 9.94,
4332
+ "learning_rate": 7.076659268151487e-09,
4333
+ "loss": 0.6932,
4334
+ "step": 175500
4335
+ },
4336
+ {
4337
+ "epoch": 9.95,
4338
+ "learning_rate": 4.443034512263689e-09,
4339
+ "loss": 0.6931,
4340
+ "step": 175750
4341
+ },
4342
+ {
4343
+ "epoch": 9.96,
4344
+ "learning_rate": 2.4195472883042914e-09,
4345
+ "loss": 0.6931,
4346
+ "step": 176000
4347
+ },
4348
+ {
4349
+ "epoch": 9.98,
4350
+ "learning_rate": 1.0062469892796466e-09,
4351
+ "loss": 0.6932,
4352
+ "step": 176250
4353
+ },
4354
+ {
4355
+ "epoch": 9.99,
4356
+ "learning_rate": 2.031681136188679e-10,
4357
+ "loss": 0.6932,
4358
+ "step": 176500
4359
+ },
4360
+ {
4361
+ "epoch": 10.0,
4362
+ "eval_accuracy": 0.4997452229299363,
4363
+ "eval_f1": 0.3332200798437102,
4364
+ "eval_loss": 0.693359375,
4365
+ "eval_precision": 0.24987261146496814,
4366
+ "eval_recall": 0.5,
4367
+ "eval_runtime": 12.7636,
4368
+ "eval_samples_per_second": 615.031,
4369
+ "eval_steps_per_second": 76.938,
4370
+ "step": 176630
4371
+ },
4372
+ {
4373
+ "epoch": 10.0,
4374
+ "step": 176630,
4375
+ "total_flos": 8.989707126226944e+16,
4376
+ "train_loss": 0.6935468676189722,
4377
+ "train_runtime": 13050.2295,
4378
+ "train_samples_per_second": 108.274,
4379
+ "train_steps_per_second": 13.535
4380
+ }
4381
+ ],
4382
+ "max_steps": 176630,
4383
+ "num_train_epochs": 10,
4384
+ "total_flos": 8.989707126226944e+16,
4385
+ "trial_name": null,
4386
+ "trial_params": {
4387
+ "learning_rate": 1e-05
4388
+ }
4389
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:80046edbe03039c83e89330dabc0823f42a46fce7fbf10cc447880992c78f30a
3
+ size 3375