Hugging Face's logo

nancyH
/
full_output

Model card Files
xet
 Community
nancyH commited on
Commit
ffc70f8
·
verified ·
1 Parent(s): a15f393

Delete demo_human_or_worm

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. demo_human_or_worm/base_3072/checkpoint-2600/config.json +0 -27
  2. demo_human_or_worm/base_3072/checkpoint-2600/model.safetensors +0 -3
  3. demo_human_or_worm/base_3072/checkpoint-2600/optimizer.pt +0 -3
  4. demo_human_or_worm/base_3072/checkpoint-2600/rng_state.pth +0 -3
  5. demo_human_or_worm/base_3072/checkpoint-2600/scheduler.pt +0 -3
  6. demo_human_or_worm/base_3072/checkpoint-2600/special_tokens_map.json +0 -7
  7. demo_human_or_worm/base_3072/checkpoint-2600/tokenizer.json +0 -0
  8. demo_human_or_worm/base_3072/checkpoint-2600/tokenizer_config.json +0 -56
  9. demo_human_or_worm/base_3072/checkpoint-2600/trainer_state.json +0 -188
  10. demo_human_or_worm/base_3072/checkpoint-2600/training_args.bin +0 -3
  11. demo_human_or_worm/base_3072/checkpoint-2800/config.json +0 -27
  12. demo_human_or_worm/base_3072/checkpoint-2800/model.safetensors +0 -3
  13. demo_human_or_worm/base_3072/checkpoint-2800/optimizer.pt +0 -3
  14. demo_human_or_worm/base_3072/checkpoint-2800/rng_state.pth +0 -3
  15. demo_human_or_worm/base_3072/checkpoint-2800/scheduler.pt +0 -3
  16. demo_human_or_worm/base_3072/checkpoint-2800/special_tokens_map.json +0 -7
  17. demo_human_or_worm/base_3072/checkpoint-2800/tokenizer.json +0 -0
  18. demo_human_or_worm/base_3072/checkpoint-2800/tokenizer_config.json +0 -56
  19. demo_human_or_worm/base_3072/checkpoint-2800/trainer_state.json +0 -201
  20. demo_human_or_worm/base_3072/checkpoint-2800/training_args.bin +0 -3
  21. demo_human_or_worm/base_3072/checkpoint-3000/config.json +0 -27
  22. demo_human_or_worm/base_3072/checkpoint-3000/model.safetensors +0 -3
  23. demo_human_or_worm/base_3072/checkpoint-3000/optimizer.pt +0 -3
  24. demo_human_or_worm/base_3072/checkpoint-3000/rng_state.pth +0 -3
  25. demo_human_or_worm/base_3072/checkpoint-3000/scheduler.pt +0 -3
  26. demo_human_or_worm/base_3072/checkpoint-3000/special_tokens_map.json +0 -7
  27. demo_human_or_worm/base_3072/checkpoint-3000/tokenizer.json +0 -0
  28. demo_human_or_worm/base_3072/checkpoint-3000/tokenizer_config.json +0 -56
  29. demo_human_or_worm/base_3072/checkpoint-3000/trainer_state.json +0 -214
  30. demo_human_or_worm/base_3072/checkpoint-3000/training_args.bin +0 -3
  31. demo_human_or_worm/base_3072/results/base_3072_hg38_BPE_3e-5_demo_human_or_worm_seed42/eval_results.json +0 -1
  32. demo_human_or_worm/base_4096/checkpoint-2600/config.json +0 -27
  33. demo_human_or_worm/base_4096/checkpoint-2600/model.safetensors +0 -3
  34. demo_human_or_worm/base_4096/checkpoint-2600/optimizer.pt +0 -3
  35. demo_human_or_worm/base_4096/checkpoint-2600/rng_state.pth +0 -3
  36. demo_human_or_worm/base_4096/checkpoint-2600/scheduler.pt +0 -3
  37. demo_human_or_worm/base_4096/checkpoint-2600/special_tokens_map.json +0 -7
  38. demo_human_or_worm/base_4096/checkpoint-2600/tokenizer.json +0 -0
  39. demo_human_or_worm/base_4096/checkpoint-2600/tokenizer_config.json +0 -56
  40. demo_human_or_worm/base_4096/checkpoint-2600/trainer_state.json +0 -188
  41. demo_human_or_worm/base_4096/checkpoint-2600/training_args.bin +0 -3
  42. demo_human_or_worm/base_4096/checkpoint-2800/config.json +0 -27
  43. demo_human_or_worm/base_4096/checkpoint-2800/model.safetensors +0 -3
  44. demo_human_or_worm/base_4096/checkpoint-2800/optimizer.pt +0 -3
  45. demo_human_or_worm/base_4096/checkpoint-2800/rng_state.pth +0 -3
  46. demo_human_or_worm/base_4096/checkpoint-2800/scheduler.pt +0 -3
  47. demo_human_or_worm/base_4096/checkpoint-2800/special_tokens_map.json +0 -7
  48. demo_human_or_worm/base_4096/checkpoint-2800/tokenizer.json +0 -0
  49. demo_human_or_worm/base_4096/checkpoint-2800/tokenizer_config.json +0 -56
  50. demo_human_or_worm/base_4096/checkpoint-2800/trainer_state.json +0 -201
demo_human_or_worm/base_3072/checkpoint-2600/config.json DELETED
@@ -1,27 +0,0 @@
1
- {
2
- "_name_or_path": "/root/NaN/dna-tokenizer/pretrain/models/base_3072/checkpoint-100000",
3
- "architectures": [
4
- "BertForSequenceClassification"
5
- ],
6
- "attention_probs_dropout_prob": 0.1,
7
- "classifier_dropout": null,
8
- "hidden_act": "gelu",
9
- "hidden_dropout_prob": 0.1,
10
- "hidden_size": 768,
11
- "initializer_range": 0.02,
12
- "intermediate_size": 3072,
13
- "layer_norm_eps": 1e-12,
14
- "max_length": 512,
15
- "max_position_embeddings": 512,
16
- "model_type": "bert",
17
- "num_attention_heads": 12,
18
- "num_hidden_layers": 12,
19
- "pad_token_id": 0,
20
- "position_embedding_type": "absolute",
21
- "problem_type": "single_label_classification",
22
- "torch_dtype": "float32",
23
- "transformers_version": "4.35.2",
24
- "type_vocab_size": 2,
25
- "use_cache": true,
26
- "vocab_size": 3072
27
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
demo_human_or_worm/base_3072/checkpoint-2600/model.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:f3b3c5a359d14ae581419d5632e2bdeb058afbaf7802f5e349fe2bf0b47f3c19
3
- size 353632152
 
 
 
 
demo_human_or_worm/base_3072/checkpoint-2600/optimizer.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:9d55c3b0f6af5069592be7f29fbeb61f724008138058ab4827b6afb5c206a8da
3
- size 707385995
 
 
 
 
demo_human_or_worm/base_3072/checkpoint-2600/rng_state.pth DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:2378bcb27aed37ebdd3529313a295e8022b1fd7da7cb9b7ce5737ea8840f6e9b
3
- size 14709
 
 
 
 
demo_human_or_worm/base_3072/checkpoint-2600/scheduler.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:bddc3849c97826c2754f7b8e56907ee68f9747753bacae1ed49cffe8ff782a2b
3
- size 1465
 
 
 
 
demo_human_or_worm/base_3072/checkpoint-2600/special_tokens_map.json DELETED
@@ -1,7 +0,0 @@
1
- {
2
- "cls_token": "[CLS]",
3
- "mask_token": "[MASK]",
4
- "pad_token": "[PAD]",
5
- "sep_token": "[SEP]",
6
- "unk_token": "[UNK]"
7
- }
 
 
 
 
 
 
 
 
demo_human_or_worm/base_3072/checkpoint-2600/tokenizer.json DELETED
The diff for this file is too large to render. See raw diff
 
demo_human_or_worm/base_3072/checkpoint-2600/tokenizer_config.json DELETED
@@ -1,56 +0,0 @@
1
- {
2
- "added_tokens_decoder": {
3
- "0": {
4
- "content": "[PAD]",
5
- "lstrip": false,
6
- "normalized": false,
7
- "rstrip": false,
8
- "single_word": false,
9
- "special": true
10
- },
11
- "1": {
12
- "content": "[UNK]",
13
- "lstrip": false,
14
- "normalized": false,
15
- "rstrip": false,
16
- "single_word": false,
17
- "special": true
18
- },
19
- "2": {
20
- "content": "[CLS]",
21
- "lstrip": false,
22
- "normalized": false,
23
- "rstrip": false,
24
- "single_word": false,
25
- "special": true
26
- },
27
- "3": {
28
- "content": "[SEP]",
29
- "lstrip": false,
30
- "normalized": false,
31
- "rstrip": false,
32
- "single_word": false,
33
- "special": true
34
- },
35
- "4": {
36
- "content": "[MASK]",
37
- "lstrip": false,
38
- "normalized": false,
39
- "rstrip": false,
40
- "single_word": false,
41
- "special": true
42
- }
43
- },
44
- "cache_dir": null,
45
- "clean_up_tokenization_spaces": true,
46
- "cls_token": "[CLS]",
47
- "mask_token": "[MASK]",
48
- "model_max_length": 512,
49
- "pad_token": "[PAD]",
50
- "padding_side": "right",
51
- "sep_token": "[SEP]",
52
- "tokenizer_class": "PreTrainedTokenizerFast",
53
- "trust_remote_code": true,
54
- "unk_token": "[UNK]",
55
- "use_fast": true
56
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
demo_human_or_worm/base_3072/checkpoint-2600/trainer_state.json DELETED
@@ -1,188 +0,0 @@
1
- {
2
- "best_metric": null,
3
- "best_model_checkpoint": null,
4
- "epoch": 4.16,
5
- "eval_steps": 200,
6
- "global_step": 2600,
7
- "is_hyper_param_search": false,
8
- "is_local_process_zero": true,
9
- "is_world_process_zero": true,
10
- "log_history": [
11
- {
12
- "epoch": 0.32,
13
- "eval_accuracy": 0.9465,
14
- "eval_f1": 0.9464975982771866,
15
- "eval_loss": 0.1371922791004181,
16
- "eval_matthews_correlation": 0.8929953669388209,
17
- "eval_precision": 0.9464958884715917,
18
- "eval_recall": 0.9464994784744453,
19
- "eval_runtime": 2.4801,
20
- "eval_samples_per_second": 4032.121,
21
- "eval_steps_per_second": 31.854,
22
- "step": 200
23
- },
24
- {
25
- "epoch": 0.64,
26
- "eval_accuracy": 0.9463,
27
- "eval_f1": 0.9462799898334131,
28
- "eval_loss": 0.1387450248003006,
29
- "eval_matthews_correlation": 0.8938709006332453,
30
- "eval_precision": 0.9473898846305053,
31
- "eval_recall": 0.946481477592402,
32
- "eval_runtime": 2.462,
33
- "eval_samples_per_second": 4061.723,
34
- "eval_steps_per_second": 32.088,
35
- "step": 400
36
- },
37
- {
38
- "epoch": 0.96,
39
- "eval_accuracy": 0.9521,
40
- "eval_f1": 0.9520968570747926,
41
- "eval_loss": 0.1217288225889206,
42
- "eval_matthews_correlation": 0.904628043960211,
43
- "eval_precision": 0.9524247130775669,
44
- "eval_recall": 0.9522033579645404,
45
- "eval_runtime": 2.4642,
46
- "eval_samples_per_second": 4058.034,
47
- "eval_steps_per_second": 32.058,
48
- "step": 600
49
- },
50
- {
51
- "epoch": 1.28,
52
- "eval_accuracy": 0.9532,
53
- "eval_f1": 0.9531880161321298,
54
- "eval_loss": 0.11742018163204193,
55
- "eval_matthews_correlation": 0.90653053839975,
56
- "eval_precision": 0.9533958724202627,
57
- "eval_recall": 0.9531347036004765,
58
- "eval_runtime": 2.4626,
59
- "eval_samples_per_second": 4060.731,
60
- "eval_steps_per_second": 32.08,
61
- "step": 800
62
- },
63
- {
64
- "epoch": 1.6,
65
- "eval_accuracy": 0.9556,
66
- "eval_f1": 0.955599886335709,
67
- "eval_loss": 0.12430283427238464,
68
- "eval_matthews_correlation": 0.9112555051606778,
69
- "eval_precision": 0.955619879151055,
70
- "eval_recall": 0.9556356261456811,
71
- "eval_runtime": 2.4583,
72
- "eval_samples_per_second": 4067.886,
73
- "eval_steps_per_second": 32.136,
74
- "step": 1000
75
- },
76
- {
77
- "epoch": 1.92,
78
- "eval_accuracy": 0.9528,
79
- "eval_f1": 0.952781112444978,
80
- "eval_loss": 0.12060332298278809,
81
- "eval_matthews_correlation": 0.9058845708661534,
82
- "eval_precision": 0.9531780108538198,
83
- "eval_recall": 0.9527066826274487,
84
- "eval_runtime": 2.4638,
85
- "eval_samples_per_second": 4058.82,
86
- "eval_steps_per_second": 32.065,
87
- "step": 1200
88
- },
89
- {
90
- "epoch": 2.24,
91
- "eval_accuracy": 0.9543,
92
- "eval_f1": 0.9542984091276216,
93
- "eval_loss": 0.12013883143663406,
94
- "eval_matthews_correlation": 0.9085991278171274,
95
- "eval_precision": 0.954293666926086,
96
- "eval_recall": 0.9543054609675874,
97
- "eval_runtime": 2.4607,
98
- "eval_samples_per_second": 4063.951,
99
- "eval_steps_per_second": 32.105,
100
- "step": 1400
101
- },
102
- {
103
- "epoch": 2.56,
104
- "eval_accuracy": 0.9567,
105
- "eval_f1": 0.9566988737377059,
106
- "eval_loss": 0.11154153943061829,
107
- "eval_matthews_correlation": 0.9136779433109837,
108
- "eval_precision": 0.9568953679256471,
109
- "eval_recall": 0.956782582346535,
110
- "eval_runtime": 2.4593,
111
- "eval_samples_per_second": 4066.129,
112
- "eval_steps_per_second": 32.122,
113
- "step": 1600
114
- },
115
- {
116
- "epoch": 2.88,
117
- "eval_accuracy": 0.9561,
118
- "eval_f1": 0.9560997677677715,
119
- "eval_loss": 0.12107250094413757,
120
- "eval_matthews_correlation": 0.9122417773930729,
121
- "eval_precision": 0.9561110271995167,
122
- "eval_recall": 0.95613075040677,
123
- "eval_runtime": 2.4634,
124
- "eval_samples_per_second": 4059.49,
125
- "eval_steps_per_second": 32.07,
126
- "step": 1800
127
- },
128
- {
129
- "epoch": 3.2,
130
- "eval_accuracy": 0.9536,
131
- "eval_f1": 0.9535646273905611,
132
- "eval_loss": 0.1393684446811676,
133
- "eval_matthews_correlation": 0.9079401136271502,
134
- "eval_precision": 0.9544871828025743,
135
- "eval_recall": 0.9534535192224418,
136
- "eval_runtime": 2.4638,
137
- "eval_samples_per_second": 4058.767,
138
- "eval_steps_per_second": 32.064,
139
- "step": 2000
140
- },
141
- {
142
- "epoch": 3.52,
143
- "eval_accuracy": 0.9568,
144
- "eval_f1": 0.9567999982719999,
145
- "eval_loss": 0.13930311799049377,
146
- "eval_matthews_correlation": 0.9136884855064893,
147
- "eval_precision": 0.9568430000810835,
148
- "eval_recall": 0.956845485428786,
149
- "eval_runtime": 2.4619,
150
- "eval_samples_per_second": 4061.96,
151
- "eval_steps_per_second": 32.089,
152
- "step": 2200
153
- },
154
- {
155
- "epoch": 3.84,
156
- "eval_accuracy": 0.9571,
157
- "eval_f1": 0.9570994744685623,
158
- "eval_loss": 0.13920287787914276,
159
- "eval_matthews_correlation": 0.9142223987231565,
160
- "eval_precision": 0.9571000000000001,
161
- "eval_recall": 0.9571223989975509,
162
- "eval_runtime": 2.4602,
163
- "eval_samples_per_second": 4064.697,
164
- "eval_steps_per_second": 32.111,
165
- "step": 2400
166
- },
167
- {
168
- "epoch": 4.16,
169
- "eval_accuracy": 0.9561,
170
- "eval_f1": 0.9560997256232852,
171
- "eval_loss": 0.16130109131336212,
172
- "eval_matthews_correlation": 0.9123720449317049,
173
- "eval_precision": 0.9562076939079227,
174
- "eval_recall": 0.9561643520532506,
175
- "eval_runtime": 2.4609,
176
- "eval_samples_per_second": 4063.579,
177
- "eval_steps_per_second": 32.102,
178
- "step": 2600
179
- }
180
- ],
181
- "logging_steps": 100000,
182
- "max_steps": 3125,
183
- "num_train_epochs": 5,
184
- "save_steps": 200,
185
- "total_flos": 3.420443852459213e+16,
186
- "trial_name": null,
187
- "trial_params": null
188
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
demo_human_or_worm/base_3072/checkpoint-2600/training_args.bin DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:79807bdb0d13ffc0e5ccd912b83df31c706a82a57650d14ad46a53ba7288db15
3
- size 5265
 
 
 
 
demo_human_or_worm/base_3072/checkpoint-2800/config.json DELETED
@@ -1,27 +0,0 @@
1
- {
2
- "_name_or_path": "/root/NaN/dna-tokenizer/pretrain/models/base_3072/checkpoint-100000",
3
- "architectures": [
4
- "BertForSequenceClassification"
5
- ],
6
- "attention_probs_dropout_prob": 0.1,
7
- "classifier_dropout": null,
8
- "hidden_act": "gelu",
9
- "hidden_dropout_prob": 0.1,
10
- "hidden_size": 768,
11
- "initializer_range": 0.02,
12
- "intermediate_size": 3072,
13
- "layer_norm_eps": 1e-12,
14
- "max_length": 512,
15
- "max_position_embeddings": 512,
16
- "model_type": "bert",
17
- "num_attention_heads": 12,
18
- "num_hidden_layers": 12,
19
- "pad_token_id": 0,
20
- "position_embedding_type": "absolute",
21
- "problem_type": "single_label_classification",
22
- "torch_dtype": "float32",
23
- "transformers_version": "4.35.2",
24
- "type_vocab_size": 2,
25
- "use_cache": true,
26
- "vocab_size": 3072
27
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
demo_human_or_worm/base_3072/checkpoint-2800/model.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:b8bae6d8963fdaacefc24a0703837321cdaf07816acbad867461784e04e2646f
3
- size 353632152
 
 
 
 
demo_human_or_worm/base_3072/checkpoint-2800/optimizer.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:97daacd5d9fdb5d7c2f982725d69a5d508139a93ea98f7f19970a5cb75ae92d6
3
- size 707385995
 
 
 
 
demo_human_or_worm/base_3072/checkpoint-2800/rng_state.pth DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:0c1b96607ba87c66bf218309c4bfe076bbba8cef6885e64c8bb7d2e994aa1db9
3
- size 14709
 
 
 
 
demo_human_or_worm/base_3072/checkpoint-2800/scheduler.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:a6cefee8a712b1b0d79adcf718e8524a4ab8ae4d33c34f47a5798f79151d383b
3
- size 1465
 
 
 
 
demo_human_or_worm/base_3072/checkpoint-2800/special_tokens_map.json DELETED
@@ -1,7 +0,0 @@
1
- {
2
- "cls_token": "[CLS]",
3
- "mask_token": "[MASK]",
4
- "pad_token": "[PAD]",
5
- "sep_token": "[SEP]",
6
- "unk_token": "[UNK]"
7
- }
 
 
 
 
 
 
 
 
demo_human_or_worm/base_3072/checkpoint-2800/tokenizer.json DELETED
The diff for this file is too large to render. See raw diff
 
demo_human_or_worm/base_3072/checkpoint-2800/tokenizer_config.json DELETED
@@ -1,56 +0,0 @@
1
- {
2
- "added_tokens_decoder": {
3
- "0": {
4
- "content": "[PAD]",
5
- "lstrip": false,
6
- "normalized": false,
7
- "rstrip": false,
8
- "single_word": false,
9
- "special": true
10
- },
11
- "1": {
12
- "content": "[UNK]",
13
- "lstrip": false,
14
- "normalized": false,
15
- "rstrip": false,
16
- "single_word": false,
17
- "special": true
18
- },
19
- "2": {
20
- "content": "[CLS]",
21
- "lstrip": false,
22
- "normalized": false,
23
- "rstrip": false,
24
- "single_word": false,
25
- "special": true
26
- },
27
- "3": {
28
- "content": "[SEP]",
29
- "lstrip": false,
30
- "normalized": false,
31
- "rstrip": false,
32
- "single_word": false,
33
- "special": true
34
- },
35
- "4": {
36
- "content": "[MASK]",
37
- "lstrip": false,
38
- "normalized": false,
39
- "rstrip": false,
40
- "single_word": false,
41
- "special": true
42
- }
43
- },
44
- "cache_dir": null,
45
- "clean_up_tokenization_spaces": true,
46
- "cls_token": "[CLS]",
47
- "mask_token": "[MASK]",
48
- "model_max_length": 512,
49
- "pad_token": "[PAD]",
50
- "padding_side": "right",
51
- "sep_token": "[SEP]",
52
- "tokenizer_class": "PreTrainedTokenizerFast",
53
- "trust_remote_code": true,
54
- "unk_token": "[UNK]",
55
- "use_fast": true
56
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
demo_human_or_worm/base_3072/checkpoint-2800/trainer_state.json DELETED
@@ -1,201 +0,0 @@
1
- {
2
- "best_metric": null,
3
- "best_model_checkpoint": null,
4
- "epoch": 4.48,
5
- "eval_steps": 200,
6
- "global_step": 2800,
7
- "is_hyper_param_search": false,
8
- "is_local_process_zero": true,
9
- "is_world_process_zero": true,
10
- "log_history": [
11
- {
12
- "epoch": 0.32,
13
- "eval_accuracy": 0.9465,
14
- "eval_f1": 0.9464975982771866,
15
- "eval_loss": 0.1371922791004181,
16
- "eval_matthews_correlation": 0.8929953669388209,
17
- "eval_precision": 0.9464958884715917,
18
- "eval_recall": 0.9464994784744453,
19
- "eval_runtime": 2.4801,
20
- "eval_samples_per_second": 4032.121,
21
- "eval_steps_per_second": 31.854,
22
- "step": 200
23
- },
24
- {
25
- "epoch": 0.64,
26
- "eval_accuracy": 0.9463,
27
- "eval_f1": 0.9462799898334131,
28
- "eval_loss": 0.1387450248003006,
29
- "eval_matthews_correlation": 0.8938709006332453,
30
- "eval_precision": 0.9473898846305053,
31
- "eval_recall": 0.946481477592402,
32
- "eval_runtime": 2.462,
33
- "eval_samples_per_second": 4061.723,
34
- "eval_steps_per_second": 32.088,
35
- "step": 400
36
- },
37
- {
38
- "epoch": 0.96,
39
- "eval_accuracy": 0.9521,
40
- "eval_f1": 0.9520968570747926,
41
- "eval_loss": 0.1217288225889206,
42
- "eval_matthews_correlation": 0.904628043960211,
43
- "eval_precision": 0.9524247130775669,
44
- "eval_recall": 0.9522033579645404,
45
- "eval_runtime": 2.4642,
46
- "eval_samples_per_second": 4058.034,
47
- "eval_steps_per_second": 32.058,
48
- "step": 600
49
- },
50
- {
51
- "epoch": 1.28,
52
- "eval_accuracy": 0.9532,
53
- "eval_f1": 0.9531880161321298,
54
- "eval_loss": 0.11742018163204193,
55
- "eval_matthews_correlation": 0.90653053839975,
56
- "eval_precision": 0.9533958724202627,
57
- "eval_recall": 0.9531347036004765,
58
- "eval_runtime": 2.4626,
59
- "eval_samples_per_second": 4060.731,
60
- "eval_steps_per_second": 32.08,
61
- "step": 800
62
- },
63
- {
64
- "epoch": 1.6,
65
- "eval_accuracy": 0.9556,
66
- "eval_f1": 0.955599886335709,
67
- "eval_loss": 0.12430283427238464,
68
- "eval_matthews_correlation": 0.9112555051606778,
69
- "eval_precision": 0.955619879151055,
70
- "eval_recall": 0.9556356261456811,
71
- "eval_runtime": 2.4583,
72
- "eval_samples_per_second": 4067.886,
73
- "eval_steps_per_second": 32.136,
74
- "step": 1000
75
- },
76
- {
77
- "epoch": 1.92,
78
- "eval_accuracy": 0.9528,
79
- "eval_f1": 0.952781112444978,
80
- "eval_loss": 0.12060332298278809,
81
- "eval_matthews_correlation": 0.9058845708661534,
82
- "eval_precision": 0.9531780108538198,
83
- "eval_recall": 0.9527066826274487,
84
- "eval_runtime": 2.4638,
85
- "eval_samples_per_second": 4058.82,
86
- "eval_steps_per_second": 32.065,
87
- "step": 1200
88
- },
89
- {
90
- "epoch": 2.24,
91
- "eval_accuracy": 0.9543,
92
- "eval_f1": 0.9542984091276216,
93
- "eval_loss": 0.12013883143663406,
94
- "eval_matthews_correlation": 0.9085991278171274,
95
- "eval_precision": 0.954293666926086,
96
- "eval_recall": 0.9543054609675874,
97
- "eval_runtime": 2.4607,
98
- "eval_samples_per_second": 4063.951,
99
- "eval_steps_per_second": 32.105,
100
- "step": 1400
101
- },
102
- {
103
- "epoch": 2.56,
104
- "eval_accuracy": 0.9567,
105
- "eval_f1": 0.9566988737377059,
106
- "eval_loss": 0.11154153943061829,
107
- "eval_matthews_correlation": 0.9136779433109837,
108
- "eval_precision": 0.9568953679256471,
109
- "eval_recall": 0.956782582346535,
110
- "eval_runtime": 2.4593,
111
- "eval_samples_per_second": 4066.129,
112
- "eval_steps_per_second": 32.122,
113
- "step": 1600
114
- },
115
- {
116
- "epoch": 2.88,
117
- "eval_accuracy": 0.9561,
118
- "eval_f1": 0.9560997677677715,
119
- "eval_loss": 0.12107250094413757,
120
- "eval_matthews_correlation": 0.9122417773930729,
121
- "eval_precision": 0.9561110271995167,
122
- "eval_recall": 0.95613075040677,
123
- "eval_runtime": 2.4634,
124
- "eval_samples_per_second": 4059.49,
125
- "eval_steps_per_second": 32.07,
126
- "step": 1800
127
- },
128
- {
129
- "epoch": 3.2,
130
- "eval_accuracy": 0.9536,
131
- "eval_f1": 0.9535646273905611,
132
- "eval_loss": 0.1393684446811676,
133
- "eval_matthews_correlation": 0.9079401136271502,
134
- "eval_precision": 0.9544871828025743,
135
- "eval_recall": 0.9534535192224418,
136
- "eval_runtime": 2.4638,
137
- "eval_samples_per_second": 4058.767,
138
- "eval_steps_per_second": 32.064,
139
- "step": 2000
140
- },
141
- {
142
- "epoch": 3.52,
143
- "eval_accuracy": 0.9568,
144
- "eval_f1": 0.9567999982719999,
145
- "eval_loss": 0.13930311799049377,
146
- "eval_matthews_correlation": 0.9136884855064893,
147
- "eval_precision": 0.9568430000810835,
148
- "eval_recall": 0.956845485428786,
149
- "eval_runtime": 2.4619,
150
- "eval_samples_per_second": 4061.96,
151
- "eval_steps_per_second": 32.089,
152
- "step": 2200
153
- },
154
- {
155
- "epoch": 3.84,
156
- "eval_accuracy": 0.9571,
157
- "eval_f1": 0.9570994744685623,
158
- "eval_loss": 0.13920287787914276,
159
- "eval_matthews_correlation": 0.9142223987231565,
160
- "eval_precision": 0.9571000000000001,
161
- "eval_recall": 0.9571223989975509,
162
- "eval_runtime": 2.4602,
163
- "eval_samples_per_second": 4064.697,
164
- "eval_steps_per_second": 32.111,
165
- "step": 2400
166
- },
167
- {
168
- "epoch": 4.16,
169
- "eval_accuracy": 0.9561,
170
- "eval_f1": 0.9560997256232852,
171
- "eval_loss": 0.16130109131336212,
172
- "eval_matthews_correlation": 0.9123720449317049,
173
- "eval_precision": 0.9562076939079227,
174
- "eval_recall": 0.9561643520532506,
175
- "eval_runtime": 2.4609,
176
- "eval_samples_per_second": 4063.579,
177
- "eval_steps_per_second": 32.102,
178
- "step": 2600
179
- },
180
- {
181
- "epoch": 4.48,
182
- "eval_accuracy": 0.9555,
183
- "eval_f1": 0.9554999781949893,
184
- "eval_loss": 0.1571253091096878,
185
- "eval_matthews_correlation": 0.9110758070613665,
186
- "eval_precision": 0.9555338855426506,
187
- "eval_recall": 0.9555419215541562,
188
- "eval_runtime": 2.4603,
189
- "eval_samples_per_second": 4064.486,
190
- "eval_steps_per_second": 32.109,
191
- "step": 2800
192
- }
193
- ],
194
- "logging_steps": 100000,
195
- "max_steps": 3125,
196
- "num_train_epochs": 5,
197
- "save_steps": 200,
198
- "total_flos": 3.683554918032998e+16,
199
- "trial_name": null,
200
- "trial_params": null
201
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
demo_human_or_worm/base_3072/checkpoint-2800/training_args.bin DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:79807bdb0d13ffc0e5ccd912b83df31c706a82a57650d14ad46a53ba7288db15
3
- size 5265
 
 
 
 
demo_human_or_worm/base_3072/checkpoint-3000/config.json DELETED
@@ -1,27 +0,0 @@
1
- {
2
- "_name_or_path": "/root/NaN/dna-tokenizer/pretrain/models/base_3072/checkpoint-100000",
3
- "architectures": [
4
- "BertForSequenceClassification"
5
- ],
6
- "attention_probs_dropout_prob": 0.1,
7
- "classifier_dropout": null,
8
- "hidden_act": "gelu",
9
- "hidden_dropout_prob": 0.1,
10
- "hidden_size": 768,
11
- "initializer_range": 0.02,
12
- "intermediate_size": 3072,
13
- "layer_norm_eps": 1e-12,
14
- "max_length": 512,
15
- "max_position_embeddings": 512,
16
- "model_type": "bert",
17
- "num_attention_heads": 12,
18
- "num_hidden_layers": 12,
19
- "pad_token_id": 0,
20
- "position_embedding_type": "absolute",
21
- "problem_type": "single_label_classification",
22
- "torch_dtype": "float32",
23
- "transformers_version": "4.35.2",
24
- "type_vocab_size": 2,
25
- "use_cache": true,
26
- "vocab_size": 3072
27
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
demo_human_or_worm/base_3072/checkpoint-3000/model.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:e82a0e3d69594539db93a0f544d145c6ebe6bc2f58734cfeb753d850d2415656
3
- size 353632152
 
 
 
 
demo_human_or_worm/base_3072/checkpoint-3000/optimizer.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:ffffa07d617d2780c84670d2f667b363644d1df106ae144b3619da0210783413
3
- size 707385995
 
 
 
 
demo_human_or_worm/base_3072/checkpoint-3000/rng_state.pth DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:7fea070377f77fa1edf265afc140e46bc7a3392a48eeb057fbc8b549fd782262
3
- size 14709
 
 
 
 
demo_human_or_worm/base_3072/checkpoint-3000/scheduler.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:70c6df1acdfc8b00054d42e4c5bd530706da56c304e7106806e19d5f59082b67
3
- size 1465
 
 
 
 
demo_human_or_worm/base_3072/checkpoint-3000/special_tokens_map.json DELETED
@@ -1,7 +0,0 @@
1
- {
2
- "cls_token": "[CLS]",
3
- "mask_token": "[MASK]",
4
- "pad_token": "[PAD]",
5
- "sep_token": "[SEP]",
6
- "unk_token": "[UNK]"
7
- }
 
 
 
 
 
 
 
 
demo_human_or_worm/base_3072/checkpoint-3000/tokenizer.json DELETED
The diff for this file is too large to render. See raw diff
 
demo_human_or_worm/base_3072/checkpoint-3000/tokenizer_config.json DELETED
@@ -1,56 +0,0 @@
1
- {
2
- "added_tokens_decoder": {
3
- "0": {
4
- "content": "[PAD]",
5
- "lstrip": false,
6
- "normalized": false,
7
- "rstrip": false,
8
- "single_word": false,
9
- "special": true
10
- },
11
- "1": {
12
- "content": "[UNK]",
13
- "lstrip": false,
14
- "normalized": false,
15
- "rstrip": false,
16
- "single_word": false,
17
- "special": true
18
- },
19
- "2": {
20
- "content": "[CLS]",
21
- "lstrip": false,
22
- "normalized": false,
23
- "rstrip": false,
24
- "single_word": false,
25
- "special": true
26
- },
27
- "3": {
28
- "content": "[SEP]",
29
- "lstrip": false,
30
- "normalized": false,
31
- "rstrip": false,
32
- "single_word": false,
33
- "special": true
34
- },
35
- "4": {
36
- "content": "[MASK]",
37
- "lstrip": false,
38
- "normalized": false,
39
- "rstrip": false,
40
- "single_word": false,
41
- "special": true
42
- }
43
- },
44
- "cache_dir": null,
45
- "clean_up_tokenization_spaces": true,
46
- "cls_token": "[CLS]",
47
- "mask_token": "[MASK]",
48
- "model_max_length": 512,
49
- "pad_token": "[PAD]",
50
- "padding_side": "right",
51
- "sep_token": "[SEP]",
52
- "tokenizer_class": "PreTrainedTokenizerFast",
53
- "trust_remote_code": true,
54
- "unk_token": "[UNK]",
55
- "use_fast": true
56
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
demo_human_or_worm/base_3072/checkpoint-3000/trainer_state.json DELETED
@@ -1,214 +0,0 @@
1
- {
2
- "best_metric": null,
3
- "best_model_checkpoint": null,
4
- "epoch": 4.8,
5
- "eval_steps": 200,
6
- "global_step": 3000,
7
- "is_hyper_param_search": false,
8
- "is_local_process_zero": true,
9
- "is_world_process_zero": true,
10
- "log_history": [
11
- {
12
- "epoch": 0.32,
13
- "eval_accuracy": 0.9465,
14
- "eval_f1": 0.9464975982771866,
15
- "eval_loss": 0.1371922791004181,
16
- "eval_matthews_correlation": 0.8929953669388209,
17
- "eval_precision": 0.9464958884715917,
18
- "eval_recall": 0.9464994784744453,
19
- "eval_runtime": 2.4801,
20
- "eval_samples_per_second": 4032.121,
21
- "eval_steps_per_second": 31.854,
22
- "step": 200
23
- },
24
- {
25
- "epoch": 0.64,
26
- "eval_accuracy": 0.9463,
27
- "eval_f1": 0.9462799898334131,
28
- "eval_loss": 0.1387450248003006,
29
- "eval_matthews_correlation": 0.8938709006332453,
30
- "eval_precision": 0.9473898846305053,
31
- "eval_recall": 0.946481477592402,
32
- "eval_runtime": 2.462,
33
- "eval_samples_per_second": 4061.723,
34
- "eval_steps_per_second": 32.088,
35
- "step": 400
36
- },
37
- {
38
- "epoch": 0.96,
39
- "eval_accuracy": 0.9521,
40
- "eval_f1": 0.9520968570747926,
41
- "eval_loss": 0.1217288225889206,
42
- "eval_matthews_correlation": 0.904628043960211,
43
- "eval_precision": 0.9524247130775669,
44
- "eval_recall": 0.9522033579645404,
45
- "eval_runtime": 2.4642,
46
- "eval_samples_per_second": 4058.034,
47
- "eval_steps_per_second": 32.058,
48
- "step": 600
49
- },
50
- {
51
- "epoch": 1.28,
52
- "eval_accuracy": 0.9532,
53
- "eval_f1": 0.9531880161321298,
54
- "eval_loss": 0.11742018163204193,
55
- "eval_matthews_correlation": 0.90653053839975,
56
- "eval_precision": 0.9533958724202627,
57
- "eval_recall": 0.9531347036004765,
58
- "eval_runtime": 2.4626,
59
- "eval_samples_per_second": 4060.731,
60
- "eval_steps_per_second": 32.08,
61
- "step": 800
62
- },
63
- {
64
- "epoch": 1.6,
65
- "eval_accuracy": 0.9556,
66
- "eval_f1": 0.955599886335709,
67
- "eval_loss": 0.12430283427238464,
68
- "eval_matthews_correlation": 0.9112555051606778,
69
- "eval_precision": 0.955619879151055,
70
- "eval_recall": 0.9556356261456811,
71
- "eval_runtime": 2.4583,
72
- "eval_samples_per_second": 4067.886,
73
- "eval_steps_per_second": 32.136,
74
- "step": 1000
75
- },
76
- {
77
- "epoch": 1.92,
78
- "eval_accuracy": 0.9528,
79
- "eval_f1": 0.952781112444978,
80
- "eval_loss": 0.12060332298278809,
81
- "eval_matthews_correlation": 0.9058845708661534,
82
- "eval_precision": 0.9531780108538198,
83
- "eval_recall": 0.9527066826274487,
84
- "eval_runtime": 2.4638,
85
- "eval_samples_per_second": 4058.82,
86
- "eval_steps_per_second": 32.065,
87
- "step": 1200
88
- },
89
- {
90
- "epoch": 2.24,
91
- "eval_accuracy": 0.9543,
92
- "eval_f1": 0.9542984091276216,
93
- "eval_loss": 0.12013883143663406,
94
- "eval_matthews_correlation": 0.9085991278171274,
95
- "eval_precision": 0.954293666926086,
96
- "eval_recall": 0.9543054609675874,
97
- "eval_runtime": 2.4607,
98
- "eval_samples_per_second": 4063.951,
99
- "eval_steps_per_second": 32.105,
100
- "step": 1400
101
- },
102
- {
103
- "epoch": 2.56,
104
- "eval_accuracy": 0.9567,
105
- "eval_f1": 0.9566988737377059,
106
- "eval_loss": 0.11154153943061829,
107
- "eval_matthews_correlation": 0.9136779433109837,
108
- "eval_precision": 0.9568953679256471,
109
- "eval_recall": 0.956782582346535,
110
- "eval_runtime": 2.4593,
111
- "eval_samples_per_second": 4066.129,
112
- "eval_steps_per_second": 32.122,
113
- "step": 1600
114
- },
115
- {
116
- "epoch": 2.88,
117
- "eval_accuracy": 0.9561,
118
- "eval_f1": 0.9560997677677715,
119
- "eval_loss": 0.12107250094413757,
120
- "eval_matthews_correlation": 0.9122417773930729,
121
- "eval_precision": 0.9561110271995167,
122
- "eval_recall": 0.95613075040677,
123
- "eval_runtime": 2.4634,
124
- "eval_samples_per_second": 4059.49,
125
- "eval_steps_per_second": 32.07,
126
- "step": 1800
127
- },
128
- {
129
- "epoch": 3.2,
130
- "eval_accuracy": 0.9536,
131
- "eval_f1": 0.9535646273905611,
132
- "eval_loss": 0.1393684446811676,
133
- "eval_matthews_correlation": 0.9079401136271502,
134
- "eval_precision": 0.9544871828025743,
135
- "eval_recall": 0.9534535192224418,
136
- "eval_runtime": 2.4638,
137
- "eval_samples_per_second": 4058.767,
138
- "eval_steps_per_second": 32.064,
139
- "step": 2000
140
- },
141
- {
142
- "epoch": 3.52,
143
- "eval_accuracy": 0.9568,
144
- "eval_f1": 0.9567999982719999,
145
- "eval_loss": 0.13930311799049377,
146
- "eval_matthews_correlation": 0.9136884855064893,
147
- "eval_precision": 0.9568430000810835,
148
- "eval_recall": 0.956845485428786,
149
- "eval_runtime": 2.4619,
150
- "eval_samples_per_second": 4061.96,
151
- "eval_steps_per_second": 32.089,
152
- "step": 2200
153
- },
154
- {
155
- "epoch": 3.84,
156
- "eval_accuracy": 0.9571,
157
- "eval_f1": 0.9570994744685623,
158
- "eval_loss": 0.13920287787914276,
159
- "eval_matthews_correlation": 0.9142223987231565,
160
- "eval_precision": 0.9571000000000001,
161
- "eval_recall": 0.9571223989975509,
162
- "eval_runtime": 2.4602,
163
- "eval_samples_per_second": 4064.697,
164
- "eval_steps_per_second": 32.111,
165
- "step": 2400
166
- },
167
- {
168
- "epoch": 4.16,
169
- "eval_accuracy": 0.9561,
170
- "eval_f1": 0.9560997256232852,
171
- "eval_loss": 0.16130109131336212,
172
- "eval_matthews_correlation": 0.9123720449317049,
173
- "eval_precision": 0.9562076939079227,
174
- "eval_recall": 0.9561643520532506,
175
- "eval_runtime": 2.4609,
176
- "eval_samples_per_second": 4063.579,
177
- "eval_steps_per_second": 32.102,
178
- "step": 2600
179
- },
180
- {
181
- "epoch": 4.48,
182
- "eval_accuracy": 0.9555,
183
- "eval_f1": 0.9554999781949893,
184
- "eval_loss": 0.1571253091096878,
185
- "eval_matthews_correlation": 0.9110758070613665,
186
- "eval_precision": 0.9555338855426506,
187
- "eval_recall": 0.9555419215541562,
188
- "eval_runtime": 2.4603,
189
- "eval_samples_per_second": 4064.486,
190
- "eval_steps_per_second": 32.109,
191
- "step": 2800
192
- },
193
- {
194
- "epoch": 4.8,
195
- "eval_accuracy": 0.9557,
196
- "eval_f1": 0.95569987197263,
197
- "eval_loss": 0.16012302041053772,
198
- "eval_matthews_correlation": 0.9114534369754659,
199
- "eval_precision": 0.9557185061118392,
200
- "eval_recall": 0.9557349310116195,
201
- "eval_runtime": 2.4618,
202
- "eval_samples_per_second": 4062.019,
203
- "eval_steps_per_second": 32.09,
204
- "step": 3000
205
- }
206
- ],
207
- "logging_steps": 100000,
208
- "max_steps": 3125,
209
- "num_train_epochs": 5,
210
- "save_steps": 200,
211
- "total_flos": 3.946665983606784e+16,
212
- "trial_name": null,
213
- "trial_params": null
214
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
demo_human_or_worm/base_3072/checkpoint-3000/training_args.bin DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:79807bdb0d13ffc0e5ccd912b83df31c706a82a57650d14ad46a53ba7288db15
3
- size 5265
 
 
 
 
demo_human_or_worm/base_3072/results/base_3072_hg38_BPE_3e-5_demo_human_or_worm_seed42/eval_results.json DELETED
@@ -1 +0,0 @@
1
- {"eval_loss": 0.16287651658058167, "eval_accuracy": 0.957, "eval_f1": 0.9569996129965169, "eval_matthews_correlation": 0.9140930182013325, "eval_precision": 0.957065706570657, "eval_recall": 0.957027312436999, "eval_runtime": 2.4537, "eval_samples_per_second": 4075.463, "eval_steps_per_second": 32.196, "epoch": 5.0}
 
 
demo_human_or_worm/base_4096/checkpoint-2600/config.json DELETED
@@ -1,27 +0,0 @@
1
- {
2
- "_name_or_path": "/root/NaN/dna-tokenizer/pretrain/models/base_4096/checkpoint-100000",
3
- "architectures": [
4
- "BertForSequenceClassification"
5
- ],
6
- "attention_probs_dropout_prob": 0.1,
7
- "classifier_dropout": null,
8
- "hidden_act": "gelu",
9
- "hidden_dropout_prob": 0.1,
10
- "hidden_size": 768,
11
- "initializer_range": 0.02,
12
- "intermediate_size": 3072,
13
- "layer_norm_eps": 1e-12,
14
- "max_length": 512,
15
- "max_position_embeddings": 512,
16
- "model_type": "bert",
17
- "num_attention_heads": 12,
18
- "num_hidden_layers": 12,
19
- "pad_token_id": 0,
20
- "position_embedding_type": "absolute",
21
- "problem_type": "single_label_classification",
22
- "torch_dtype": "float32",
23
- "transformers_version": "4.35.2",
24
- "type_vocab_size": 2,
25
- "use_cache": true,
26
- "vocab_size": 4096
27
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
demo_human_or_worm/base_4096/checkpoint-2600/model.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:4c4199d9b467e89072d8ec84285704525dd2f3ca707a7bc11ed22a0a375c6f95
3
- size 356777880
 
 
 
 
demo_human_or_worm/base_4096/checkpoint-2600/optimizer.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:4c17908ab733851aa67e7382a69f96f75b4ecdea8f975275618cc241b708dc0a
3
- size 713677451
 
 
 
 
demo_human_or_worm/base_4096/checkpoint-2600/rng_state.pth DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:2378bcb27aed37ebdd3529313a295e8022b1fd7da7cb9b7ce5737ea8840f6e9b
3
- size 14709
 
 
 
 
demo_human_or_worm/base_4096/checkpoint-2600/scheduler.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:bddc3849c97826c2754f7b8e56907ee68f9747753bacae1ed49cffe8ff782a2b
3
- size 1465
 
 
 
 
demo_human_or_worm/base_4096/checkpoint-2600/special_tokens_map.json DELETED
@@ -1,7 +0,0 @@
1
- {
2
- "cls_token": "[CLS]",
3
- "mask_token": "[MASK]",
4
- "pad_token": "[PAD]",
5
- "sep_token": "[SEP]",
6
- "unk_token": "[UNK]"
7
- }
 
 
 
 
 
 
 
 
demo_human_or_worm/base_4096/checkpoint-2600/tokenizer.json DELETED
The diff for this file is too large to render. See raw diff
 
demo_human_or_worm/base_4096/checkpoint-2600/tokenizer_config.json DELETED
@@ -1,56 +0,0 @@
1
- {
2
- "added_tokens_decoder": {
3
- "0": {
4
- "content": "[PAD]",
5
- "lstrip": false,
6
- "normalized": false,
7
- "rstrip": false,
8
- "single_word": false,
9
- "special": true
10
- },
11
- "1": {
12
- "content": "[UNK]",
13
- "lstrip": false,
14
- "normalized": false,
15
- "rstrip": false,
16
- "single_word": false,
17
- "special": true
18
- },
19
- "2": {
20
- "content": "[CLS]",
21
- "lstrip": false,
22
- "normalized": false,
23
- "rstrip": false,
24
- "single_word": false,
25
- "special": true
26
- },
27
- "3": {
28
- "content": "[SEP]",
29
- "lstrip": false,
30
- "normalized": false,
31
- "rstrip": false,
32
- "single_word": false,
33
- "special": true
34
- },
35
- "4": {
36
- "content": "[MASK]",
37
- "lstrip": false,
38
- "normalized": false,
39
- "rstrip": false,
40
- "single_word": false,
41
- "special": true
42
- }
43
- },
44
- "cache_dir": null,
45
- "clean_up_tokenization_spaces": true,
46
- "cls_token": "[CLS]",
47
- "mask_token": "[MASK]",
48
- "model_max_length": 512,
49
- "pad_token": "[PAD]",
50
- "padding_side": "right",
51
- "sep_token": "[SEP]",
52
- "tokenizer_class": "PreTrainedTokenizerFast",
53
- "trust_remote_code": true,
54
- "unk_token": "[UNK]",
55
- "use_fast": true
56
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
demo_human_or_worm/base_4096/checkpoint-2600/trainer_state.json DELETED
@@ -1,188 +0,0 @@
1
- {
2
- "best_metric": null,
3
- "best_model_checkpoint": null,
4
- "epoch": 4.16,
5
- "eval_steps": 200,
6
- "global_step": 2600,
7
- "is_hyper_param_search": false,
8
- "is_local_process_zero": true,
9
- "is_world_process_zero": true,
10
- "log_history": [
11
- {
12
- "epoch": 0.32,
13
- "eval_accuracy": 0.9487,
14
- "eval_f1": 0.9486749323236963,
15
- "eval_loss": 0.13615772128105164,
16
- "eval_matthews_correlation": 0.897782989791398,
17
- "eval_precision": 0.9491914090394853,
18
- "eval_recall": 0.9485917809972688,
19
- "eval_runtime": 2.4907,
20
- "eval_samples_per_second": 4014.887,
21
- "eval_steps_per_second": 31.718,
22
- "step": 200
23
- },
24
- {
25
- "epoch": 0.64,
26
- "eval_accuracy": 0.9501,
27
- "eval_f1": 0.9500897938619468,
28
- "eval_loss": 0.1309155374765396,
29
- "eval_matthews_correlation": 0.9002808940351357,
30
- "eval_precision": 0.9502344613903062,
31
- "eval_recall": 0.9500464522761616,
32
- "eval_runtime": 2.4635,
33
- "eval_samples_per_second": 4059.185,
34
- "eval_steps_per_second": 32.068,
35
- "step": 400
36
- },
37
- {
38
- "epoch": 0.96,
39
- "eval_accuracy": 0.9511,
40
- "eval_f1": 0.9510999173588603,
41
- "eval_loss": 0.1206379160284996,
42
- "eval_matthews_correlation": 0.9023308867241059,
43
- "eval_precision": 0.9511751803046169,
44
- "eval_recall": 0.9511557066296248,
45
- "eval_runtime": 2.4733,
46
- "eval_samples_per_second": 4043.177,
47
- "eval_steps_per_second": 31.941,
48
- "step": 600
49
- },
50
- {
51
- "epoch": 1.28,
52
- "eval_accuracy": 0.9541,
53
- "eval_f1": 0.9540868867557264,
54
- "eval_loss": 0.12018074095249176,
55
- "eval_matthews_correlation": 0.9083609162890578,
56
- "eval_precision": 0.9543325197890132,
57
- "eval_recall": 0.9540284473939223,
58
- "eval_runtime": 2.4741,
59
- "eval_samples_per_second": 4041.799,
60
- "eval_steps_per_second": 31.93,
61
- "step": 800
62
- },
63
- {
64
- "epoch": 1.6,
65
- "eval_accuracy": 0.9558,
66
- "eval_f1": 0.9557952188108665,
67
- "eval_loss": 0.11954936385154724,
68
- "eval_matthews_correlation": 0.9116125405332579,
69
- "eval_precision": 0.9558385098858226,
70
- "eval_recall": 0.9557740329276134,
71
- "eval_runtime": 2.4726,
72
- "eval_samples_per_second": 4044.315,
73
- "eval_steps_per_second": 31.95,
74
- "step": 1000
75
- },
76
- {
77
- "epoch": 1.92,
78
- "eval_accuracy": 0.9548,
79
- "eval_f1": 0.95479583398406,
80
- "eval_loss": 0.11968192458152771,
81
- "eval_matthews_correlation": 0.9096045792184586,
82
- "eval_precision": 0.9548249961524273,
83
- "eval_recall": 0.9547795841996258,
84
- "eval_runtime": 2.4707,
85
- "eval_samples_per_second": 4047.467,
86
- "eval_steps_per_second": 31.975,
87
- "step": 1200
88
- },
89
- {
90
- "epoch": 2.24,
91
- "eval_accuracy": 0.9533,
92
- "eval_f1": 0.9532840114529197,
93
- "eval_loss": 0.13409386575222015,
94
- "eval_matthews_correlation": 0.906820368393804,
95
- "eval_precision": 0.9536032429186267,
96
- "eval_recall": 0.9532172076431745,
97
- "eval_runtime": 2.4704,
98
- "eval_samples_per_second": 4047.94,
99
- "eval_steps_per_second": 31.979,
100
- "step": 1400
101
- },
102
- {
103
- "epoch": 2.56,
104
- "eval_accuracy": 0.9567,
105
- "eval_f1": 0.9566970168574913,
106
- "eval_loss": 0.11604487150907516,
107
- "eval_matthews_correlation": 0.913842105364093,
108
- "eval_precision": 0.9570371514118503,
109
- "eval_recall": 0.9568049834441887,
110
- "eval_runtime": 2.4694,
111
- "eval_samples_per_second": 4049.647,
112
- "eval_steps_per_second": 31.992,
113
- "step": 1600
114
- },
115
- {
116
- "epoch": 2.88,
117
- "eval_accuracy": 0.9541,
118
- "eval_f1": 0.9540846233995303,
119
- "eval_loss": 0.12985098361968994,
120
- "eval_matthews_correlation": 0.909392229490647,
121
- "eval_precision": 0.955117760658301,
122
- "eval_recall": 0.954274859468114,
123
- "eval_runtime": 2.4903,
124
- "eval_samples_per_second": 4015.533,
125
- "eval_steps_per_second": 31.723,
126
- "step": 1800
127
- },
128
- {
129
- "epoch": 3.2,
130
- "eval_accuracy": 0.9571,
131
- "eval_f1": 0.9570945213994375,
132
- "eval_loss": 0.14071306586265564,
133
- "eval_matthews_correlation": 0.914224445647042,
134
- "eval_precision": 0.9571566536432305,
135
- "eval_recall": 0.9570677963220198,
136
- "eval_runtime": 2.472,
137
- "eval_samples_per_second": 4045.239,
138
- "eval_steps_per_second": 31.957,
139
- "step": 2000
140
- },
141
- {
142
- "epoch": 3.52,
143
- "eval_accuracy": 0.9551,
144
- "eval_f1": 0.9550986417339125,
145
- "eval_loss": 0.14618055522441864,
146
- "eval_matthews_correlation": 0.910495816078335,
147
- "eval_precision": 0.9553105206086772,
148
- "eval_recall": 0.9551853040798999,
149
- "eval_runtime": 2.4737,
150
- "eval_samples_per_second": 4042.588,
151
- "eval_steps_per_second": 31.936,
152
- "step": 2200
153
- },
154
- {
155
- "epoch": 3.84,
156
- "eval_accuracy": 0.9562,
157
- "eval_f1": 0.9561977292902865,
158
- "eval_loss": 0.13146208226680756,
159
- "eval_matthews_correlation": 0.9123955350841443,
160
- "eval_precision": 0.9561990814617009,
161
- "eval_recall": 0.9561964536262277,
162
- "eval_runtime": 2.4746,
163
- "eval_samples_per_second": 4041.016,
164
- "eval_steps_per_second": 31.924,
165
- "step": 2400
166
- },
167
- {
168
- "epoch": 4.16,
169
- "eval_accuracy": 0.9552,
170
- "eval_f1": 0.9551999551999553,
171
- "eval_loss": 0.1614685356616974,
172
- "eval_matthews_correlation": 0.9104686874069903,
173
- "eval_precision": 0.9552288807220181,
174
- "eval_recall": 0.9552398067505308,
175
- "eval_runtime": 2.4733,
176
- "eval_samples_per_second": 4043.166,
177
- "eval_steps_per_second": 31.941,
178
- "step": 2600
179
- }
180
- ],
181
- "logging_steps": 100000,
182
- "max_steps": 3125,
183
- "num_train_epochs": 5,
184
- "save_steps": 200,
185
- "total_flos": 3.420443852459213e+16,
186
- "trial_name": null,
187
- "trial_params": null
188
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
demo_human_or_worm/base_4096/checkpoint-2600/training_args.bin DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:23dc7c47452d2c4c37a3e1948d7b222dd3f9ffc01b9fcb9497c9911eed901833
3
- size 5265
 
 
 
 
demo_human_or_worm/base_4096/checkpoint-2800/config.json DELETED
@@ -1,27 +0,0 @@
1
- {
2
- "_name_or_path": "/root/NaN/dna-tokenizer/pretrain/models/base_4096/checkpoint-100000",
3
- "architectures": [
4
- "BertForSequenceClassification"
5
- ],
6
- "attention_probs_dropout_prob": 0.1,
7
- "classifier_dropout": null,
8
- "hidden_act": "gelu",
9
- "hidden_dropout_prob": 0.1,
10
- "hidden_size": 768,
11
- "initializer_range": 0.02,
12
- "intermediate_size": 3072,
13
- "layer_norm_eps": 1e-12,
14
- "max_length": 512,
15
- "max_position_embeddings": 512,
16
- "model_type": "bert",
17
- "num_attention_heads": 12,
18
- "num_hidden_layers": 12,
19
- "pad_token_id": 0,
20
- "position_embedding_type": "absolute",
21
- "problem_type": "single_label_classification",
22
- "torch_dtype": "float32",
23
- "transformers_version": "4.35.2",
24
- "type_vocab_size": 2,
25
- "use_cache": true,
26
- "vocab_size": 4096
27
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
demo_human_or_worm/base_4096/checkpoint-2800/model.safetensors DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:2ebebba497c07b2c1a8a5c099799bd35c47c2754138870f9dd292503cb875317
3
- size 356777880
 
 
 
 
demo_human_or_worm/base_4096/checkpoint-2800/optimizer.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:6fec096e67974898b467707cb29c0cdab2561d31bcce718789cbf476006632f3
3
- size 713677451
 
 
 
 
demo_human_or_worm/base_4096/checkpoint-2800/rng_state.pth DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:0c1b96607ba87c66bf218309c4bfe076bbba8cef6885e64c8bb7d2e994aa1db9
3
- size 14709
 
 
 
 
demo_human_or_worm/base_4096/checkpoint-2800/scheduler.pt DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:a6cefee8a712b1b0d79adcf718e8524a4ab8ae4d33c34f47a5798f79151d383b
3
- size 1465
 
 
 
 
demo_human_or_worm/base_4096/checkpoint-2800/special_tokens_map.json DELETED
@@ -1,7 +0,0 @@
1
- {
2
- "cls_token": "[CLS]",
3
- "mask_token": "[MASK]",
4
- "pad_token": "[PAD]",
5
- "sep_token": "[SEP]",
6
- "unk_token": "[UNK]"
7
- }
 
 
 
 
 
 
 
 
demo_human_or_worm/base_4096/checkpoint-2800/tokenizer.json DELETED
The diff for this file is too large to render. See raw diff
 
demo_human_or_worm/base_4096/checkpoint-2800/tokenizer_config.json DELETED
@@ -1,56 +0,0 @@
1
- {
2
- "added_tokens_decoder": {
3
- "0": {
4
- "content": "[PAD]",
5
- "lstrip": false,
6
- "normalized": false,
7
- "rstrip": false,
8
- "single_word": false,
9
- "special": true
10
- },
11
- "1": {
12
- "content": "[UNK]",
13
- "lstrip": false,
14
- "normalized": false,
15
- "rstrip": false,
16
- "single_word": false,
17
- "special": true
18
- },
19
- "2": {
20
- "content": "[CLS]",
21
- "lstrip": false,
22
- "normalized": false,
23
- "rstrip": false,
24
- "single_word": false,
25
- "special": true
26
- },
27
- "3": {
28
- "content": "[SEP]",
29
- "lstrip": false,
30
- "normalized": false,
31
- "rstrip": false,
32
- "single_word": false,
33
- "special": true
34
- },
35
- "4": {
36
- "content": "[MASK]",
37
- "lstrip": false,
38
- "normalized": false,
39
- "rstrip": false,
40
- "single_word": false,
41
- "special": true
42
- }
43
- },
44
- "cache_dir": null,
45
- "clean_up_tokenization_spaces": true,
46
- "cls_token": "[CLS]",
47
- "mask_token": "[MASK]",
48
- "model_max_length": 512,
49
- "pad_token": "[PAD]",
50
- "padding_side": "right",
51
- "sep_token": "[SEP]",
52
- "tokenizer_class": "PreTrainedTokenizerFast",
53
- "trust_remote_code": true,
54
- "unk_token": "[UNK]",
55
- "use_fast": true
56
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
demo_human_or_worm/base_4096/checkpoint-2800/trainer_state.json DELETED
@@ -1,201 +0,0 @@
1
- {
2
- "best_metric": null,
3
- "best_model_checkpoint": null,
4
- "epoch": 4.48,
5
- "eval_steps": 200,
6
- "global_step": 2800,
7
- "is_hyper_param_search": false,
8
- "is_local_process_zero": true,
9
- "is_world_process_zero": true,
10
- "log_history": [
11
- {
12
- "epoch": 0.32,
13
- "eval_accuracy": 0.9487,
14
- "eval_f1": 0.9486749323236963,
15
- "eval_loss": 0.13615772128105164,
16
- "eval_matthews_correlation": 0.897782989791398,
17
- "eval_precision": 0.9491914090394853,
18
- "eval_recall": 0.9485917809972688,
19
- "eval_runtime": 2.4907,
20
- "eval_samples_per_second": 4014.887,
21
- "eval_steps_per_second": 31.718,
22
- "step": 200
23
- },
24
- {
25
- "epoch": 0.64,
26
- "eval_accuracy": 0.9501,
27
- "eval_f1": 0.9500897938619468,
28
- "eval_loss": 0.1309155374765396,
29
- "eval_matthews_correlation": 0.9002808940351357,
30
- "eval_precision": 0.9502344613903062,
31
- "eval_recall": 0.9500464522761616,
32
- "eval_runtime": 2.4635,
33
- "eval_samples_per_second": 4059.185,
34
- "eval_steps_per_second": 32.068,
35
- "step": 400
36
- },
37
- {
38
- "epoch": 0.96,
39
- "eval_accuracy": 0.9511,
40
- "eval_f1": 0.9510999173588603,
41
- "eval_loss": 0.1206379160284996,
42
- "eval_matthews_correlation": 0.9023308867241059,
43
- "eval_precision": 0.9511751803046169,
44
- "eval_recall": 0.9511557066296248,
45
- "eval_runtime": 2.4733,
46
- "eval_samples_per_second": 4043.177,
47
- "eval_steps_per_second": 31.941,
48
- "step": 600
49
- },
50
- {
51
- "epoch": 1.28,
52
- "eval_accuracy": 0.9541,
53
- "eval_f1": 0.9540868867557264,
54
- "eval_loss": 0.12018074095249176,
55
- "eval_matthews_correlation": 0.9083609162890578,
56
- "eval_precision": 0.9543325197890132,
57
- "eval_recall": 0.9540284473939223,
58
- "eval_runtime": 2.4741,
59
- "eval_samples_per_second": 4041.799,
60
- "eval_steps_per_second": 31.93,
61
- "step": 800
62
- },
63
- {
64
- "epoch": 1.6,
65
- "eval_accuracy": 0.9558,
66
- "eval_f1": 0.9557952188108665,
67
- "eval_loss": 0.11954936385154724,
68
- "eval_matthews_correlation": 0.9116125405332579,
69
- "eval_precision": 0.9558385098858226,
70
- "eval_recall": 0.9557740329276134,
71
- "eval_runtime": 2.4726,
72
- "eval_samples_per_second": 4044.315,
73
- "eval_steps_per_second": 31.95,
74
- "step": 1000
75
- },
76
- {
77
- "epoch": 1.92,
78
- "eval_accuracy": 0.9548,
79
- "eval_f1": 0.95479583398406,
80
- "eval_loss": 0.11968192458152771,
81
- "eval_matthews_correlation": 0.9096045792184586,
82
- "eval_precision": 0.9548249961524273,
83
- "eval_recall": 0.9547795841996258,
84
- "eval_runtime": 2.4707,
85
- "eval_samples_per_second": 4047.467,
86
- "eval_steps_per_second": 31.975,
87
- "step": 1200
88
- },
89
- {
90
- "epoch": 2.24,
91
- "eval_accuracy": 0.9533,
92
- "eval_f1": 0.9532840114529197,
93
- "eval_loss": 0.13409386575222015,
94
- "eval_matthews_correlation": 0.906820368393804,
95
- "eval_precision": 0.9536032429186267,
96
- "eval_recall": 0.9532172076431745,
97
- "eval_runtime": 2.4704,
98
- "eval_samples_per_second": 4047.94,
99
- "eval_steps_per_second": 31.979,
100
- "step": 1400
101
- },
102
- {
103
- "epoch": 2.56,
104
- "eval_accuracy": 0.9567,
105
- "eval_f1": 0.9566970168574913,
106
- "eval_loss": 0.11604487150907516,
107
- "eval_matthews_correlation": 0.913842105364093,
108
- "eval_precision": 0.9570371514118503,
109
- "eval_recall": 0.9568049834441887,
110
- "eval_runtime": 2.4694,
111
- "eval_samples_per_second": 4049.647,
112
- "eval_steps_per_second": 31.992,
113
- "step": 1600
114
- },
115
- {
116
- "epoch": 2.88,
117
- "eval_accuracy": 0.9541,
118
- "eval_f1": 0.9540846233995303,
119
- "eval_loss": 0.12985098361968994,
120
- "eval_matthews_correlation": 0.909392229490647,
121
- "eval_precision": 0.955117760658301,
122
- "eval_recall": 0.954274859468114,
123
- "eval_runtime": 2.4903,
124
- "eval_samples_per_second": 4015.533,
125
- "eval_steps_per_second": 31.723,
126
- "step": 1800
127
- },
128
- {
129
- "epoch": 3.2,
130
- "eval_accuracy": 0.9571,
131
- "eval_f1": 0.9570945213994375,
132
- "eval_loss": 0.14071306586265564,
133
- "eval_matthews_correlation": 0.914224445647042,
134
- "eval_precision": 0.9571566536432305,
135
- "eval_recall": 0.9570677963220198,
136
- "eval_runtime": 2.472,
137
- "eval_samples_per_second": 4045.239,
138
- "eval_steps_per_second": 31.957,
139
- "step": 2000
140
- },
141
- {
142
- "epoch": 3.52,
143
- "eval_accuracy": 0.9551,
144
- "eval_f1": 0.9550986417339125,
145
- "eval_loss": 0.14618055522441864,
146
- "eval_matthews_correlation": 0.910495816078335,
147
- "eval_precision": 0.9553105206086772,
148
- "eval_recall": 0.9551853040798999,
149
- "eval_runtime": 2.4737,
150
- "eval_samples_per_second": 4042.588,
151
- "eval_steps_per_second": 31.936,
152
- "step": 2200
153
- },
154
- {
155
- "epoch": 3.84,
156
- "eval_accuracy": 0.9562,
157
- "eval_f1": 0.9561977292902865,
158
- "eval_loss": 0.13146208226680756,
159
- "eval_matthews_correlation": 0.9123955350841443,
160
- "eval_precision": 0.9561990814617009,
161
- "eval_recall": 0.9561964536262277,
162
- "eval_runtime": 2.4746,
163
- "eval_samples_per_second": 4041.016,
164
- "eval_steps_per_second": 31.924,
165
- "step": 2400
166
- },
167
- {
168
- "epoch": 4.16,
169
- "eval_accuracy": 0.9552,
170
- "eval_f1": 0.9551999551999553,
171
- "eval_loss": 0.1614685356616974,
172
- "eval_matthews_correlation": 0.9104686874069903,
173
- "eval_precision": 0.9552288807220181,
174
- "eval_recall": 0.9552398067505308,
175
- "eval_runtime": 2.4733,
176
- "eval_samples_per_second": 4043.166,
177
- "eval_steps_per_second": 31.941,
178
- "step": 2600
179
- },
180
- {
181
- "epoch": 4.48,
182
- "eval_accuracy": 0.9571,
183
- "eval_f1": 0.957097586739254,
184
- "eval_loss": 0.1590801477432251,
185
- "eval_matthews_correlation": 0.9141956520800557,
186
- "eval_precision": 0.9571012544802868,
187
- "eval_recall": 0.9570943976254837,
188
- "eval_runtime": 2.4731,
189
- "eval_samples_per_second": 4043.566,
190
- "eval_steps_per_second": 31.944,
191
- "step": 2800
192
- }
193
- ],
194
- "logging_steps": 100000,
195
- "max_steps": 3125,
196
- "num_train_epochs": 5,
197
- "save_steps": 200,
198
- "total_flos": 3.683554918032998e+16,
199
- "trial_name": null,
200
- "trial_params": null
201
- }