Hugging Face's logo

HydraLM
/
Nous-Hermes-llama-2-7b_7b_cluster031_partitioned_v3_standardized_031

PEFT
Model card Files Community
1
prateeky2806 commited on
Commit
1e3b791
•
1 Parent(s): 22efa04

Training in progress, step 800

Browse files
Files changed (16) hide show
  1. adapter_model.bin +1 -1
  2. checkpoint-600/adapter_model/adapter_model/README.md +20 -0
  3. checkpoint-600/adapter_model/adapter_model/adapter_config.json +26 -0
  4. checkpoint-600/adapter_model/adapter_model/adapter_model.bin +3 -0
  5. checkpoint-800/README.md +20 -0
  6. checkpoint-800/adapter_config.json +26 -0
  7. checkpoint-800/adapter_model.bin +3 -0
  8. checkpoint-800/added_tokens.json +3 -0
  9. checkpoint-800/optimizer.pt +3 -0
  10. checkpoint-800/rng_state.pth +3 -0
  11. checkpoint-800/scheduler.pt +3 -0
  12. checkpoint-800/special_tokens_map.json +6 -0
  13. checkpoint-800/tokenizer.model +3 -0
  14. checkpoint-800/tokenizer_config.json +35 -0
  15. checkpoint-800/trainer_state.json +780 -0
  16. checkpoint-800/training_args.bin +3 -0
adapter_model.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:91b270d8c7cc302c16e52324a7d8b57ea4328c3134621312c923acd826aae1fb
3
  size 319977229
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1164b4a136c841647f2491fa9c96cc6fb19eaa70e94706cc258cf196717b14af
3
  size 319977229
checkpoint-600/adapter_model/adapter_model/README.md ADDED
@@ -0,0 +1,20 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: peft
3
+ ---
4
+ ## Training procedure
5
+
6
+
7
+ The following `bitsandbytes` quantization config was used during training:
8
+ - load_in_8bit: False
9
+ - load_in_4bit: True
10
+ - llm_int8_threshold: 6.0
11
+ - llm_int8_skip_modules: None
12
+ - llm_int8_enable_fp32_cpu_offload: False
13
+ - llm_int8_has_fp16_weight: False
14
+ - bnb_4bit_quant_type: nf4
15
+ - bnb_4bit_use_double_quant: True
16
+ - bnb_4bit_compute_dtype: bfloat16
17
+ ### Framework versions
18
+
19
+
20
+ - PEFT 0.4.0
checkpoint-600/adapter_model/adapter_model/adapter_config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": "NousResearch/Nous-Hermes-llama-2-7b",
4
+ "bias": "none",
5
+ "fan_in_fan_out": false,
6
+ "inference_mode": true,
7
+ "init_lora_weights": true,
8
+ "layers_pattern": null,
9
+ "layers_to_transform": null,
10
+ "lora_alpha": 16.0,
11
+ "lora_dropout": 0.1,
12
+ "modules_to_save": null,
13
+ "peft_type": "LORA",
14
+ "r": 64,
15
+ "revision": null,
16
+ "target_modules": [
17
+ "v_proj",
18
+ "q_proj",
19
+ "down_proj",
20
+ "k_proj",
21
+ "gate_proj",
22
+ "o_proj",
23
+ "up_proj"
24
+ ],
25
+ "task_type": "CAUSAL_LM"
26
+ }
checkpoint-600/adapter_model/adapter_model/adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:91b270d8c7cc302c16e52324a7d8b57ea4328c3134621312c923acd826aae1fb
3
+ size 319977229
checkpoint-800/README.md ADDED
@@ -0,0 +1,20 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: peft
3
+ ---
4
+ ## Training procedure
5
+
6
+
7
+ The following `bitsandbytes` quantization config was used during training:
8
+ - load_in_8bit: False
9
+ - load_in_4bit: True
10
+ - llm_int8_threshold: 6.0
11
+ - llm_int8_skip_modules: None
12
+ - llm_int8_enable_fp32_cpu_offload: False
13
+ - llm_int8_has_fp16_weight: False
14
+ - bnb_4bit_quant_type: nf4
15
+ - bnb_4bit_use_double_quant: True
16
+ - bnb_4bit_compute_dtype: bfloat16
17
+ ### Framework versions
18
+
19
+
20
+ - PEFT 0.4.0
checkpoint-800/adapter_config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": "NousResearch/Nous-Hermes-llama-2-7b",
4
+ "bias": "none",
5
+ "fan_in_fan_out": false,
6
+ "inference_mode": true,
7
+ "init_lora_weights": true,
8
+ "layers_pattern": null,
9
+ "layers_to_transform": null,
10
+ "lora_alpha": 16.0,
11
+ "lora_dropout": 0.1,
12
+ "modules_to_save": null,
13
+ "peft_type": "LORA",
14
+ "r": 64,
15
+ "revision": null,
16
+ "target_modules": [
17
+ "v_proj",
18
+ "q_proj",
19
+ "down_proj",
20
+ "k_proj",
21
+ "gate_proj",
22
+ "o_proj",
23
+ "up_proj"
24
+ ],
25
+ "task_type": "CAUSAL_LM"
26
+ }
checkpoint-800/adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1164b4a136c841647f2491fa9c96cc6fb19eaa70e94706cc258cf196717b14af
3
+ size 319977229
checkpoint-800/added_tokens.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ {
2
+ "<pad>": 32000
3
+ }
checkpoint-800/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:656eb4f61799a80bead0d6ce487d997816b681665ed0d5be13d1d5ecc8b18adc
3
+ size 1279539973
checkpoint-800/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:92ac00e4e09154d1d68be5402bf049fe6bac7d270e4feeb9f9c273eacf3842b6
3
+ size 14511
checkpoint-800/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:de7840bcb72f2f480fd301578d289cdfa174589e831b0d33e5772f3956b6beae
3
+ size 627
checkpoint-800/special_tokens_map.json ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": "<s>",
3
+ "eos_token": "</s>",
4
+ "pad_token": "<unk>",
5
+ "unk_token": "<unk>"
6
+ }
checkpoint-800/tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
3
+ size 499723
checkpoint-800/tokenizer_config.json ADDED
@@ -0,0 +1,35 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": true,
3
+ "add_eos_token": false,
4
+ "bos_token": {
5
+ "__type": "AddedToken",
6
+ "content": "<s>",
7
+ "lstrip": false,
8
+ "normalized": true,
9
+ "rstrip": false,
10
+ "single_word": false
11
+ },
12
+ "clean_up_tokenization_spaces": false,
13
+ "eos_token": {
14
+ "__type": "AddedToken",
15
+ "content": "</s>",
16
+ "lstrip": false,
17
+ "normalized": true,
18
+ "rstrip": false,
19
+ "single_word": false
20
+ },
21
+ "legacy": false,
22
+ "model_max_length": 1000000000000000019884624838656,
23
+ "pad_token": null,
24
+ "padding_side": "right",
25
+ "sp_model_kwargs": {},
26
+ "tokenizer_class": "LlamaTokenizer",
27
+ "unk_token": {
28
+ "__type": "AddedToken",
29
+ "content": "<unk>",
30
+ "lstrip": false,
31
+ "normalized": true,
32
+ "rstrip": false,
33
+ "single_word": false
34
+ }
35
+ }
checkpoint-800/trainer_state.json ADDED
@@ -0,0 +1,780 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 0.8391241431236267,
3
+ "best_model_checkpoint": "./output_v2/7b_cluster031_Nous-Hermes-llama-2-7b_partitioned_v3_standardized_031/checkpoint-800",
4
+ "epoch": 0.5503955968352253,
5
+ "global_step": 800,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.01,
12
+ "learning_rate": 0.0002,
13
+ "loss": 0.9502,
14
+ "step": 10
15
+ },
16
+ {
17
+ "epoch": 0.01,
18
+ "learning_rate": 0.0002,
19
+ "loss": 0.9426,
20
+ "step": 20
21
+ },
22
+ {
23
+ "epoch": 0.02,
24
+ "learning_rate": 0.0002,
25
+ "loss": 0.9682,
26
+ "step": 30
27
+ },
28
+ {
29
+ "epoch": 0.03,
30
+ "learning_rate": 0.0002,
31
+ "loss": 0.8517,
32
+ "step": 40
33
+ },
34
+ {
35
+ "epoch": 0.03,
36
+ "learning_rate": 0.0002,
37
+ "loss": 0.8944,
38
+ "step": 50
39
+ },
40
+ {
41
+ "epoch": 0.04,
42
+ "learning_rate": 0.0002,
43
+ "loss": 0.8608,
44
+ "step": 60
45
+ },
46
+ {
47
+ "epoch": 0.05,
48
+ "learning_rate": 0.0002,
49
+ "loss": 0.915,
50
+ "step": 70
51
+ },
52
+ {
53
+ "epoch": 0.06,
54
+ "learning_rate": 0.0002,
55
+ "loss": 0.8349,
56
+ "step": 80
57
+ },
58
+ {
59
+ "epoch": 0.06,
60
+ "learning_rate": 0.0002,
61
+ "loss": 0.8912,
62
+ "step": 90
63
+ },
64
+ {
65
+ "epoch": 0.07,
66
+ "learning_rate": 0.0002,
67
+ "loss": 0.876,
68
+ "step": 100
69
+ },
70
+ {
71
+ "epoch": 0.08,
72
+ "learning_rate": 0.0002,
73
+ "loss": 0.8697,
74
+ "step": 110
75
+ },
76
+ {
77
+ "epoch": 0.08,
78
+ "learning_rate": 0.0002,
79
+ "loss": 0.8604,
80
+ "step": 120
81
+ },
82
+ {
83
+ "epoch": 0.09,
84
+ "learning_rate": 0.0002,
85
+ "loss": 0.8654,
86
+ "step": 130
87
+ },
88
+ {
89
+ "epoch": 0.1,
90
+ "learning_rate": 0.0002,
91
+ "loss": 0.8804,
92
+ "step": 140
93
+ },
94
+ {
95
+ "epoch": 0.1,
96
+ "learning_rate": 0.0002,
97
+ "loss": 0.8355,
98
+ "step": 150
99
+ },
100
+ {
101
+ "epoch": 0.11,
102
+ "learning_rate": 0.0002,
103
+ "loss": 0.8704,
104
+ "step": 160
105
+ },
106
+ {
107
+ "epoch": 0.12,
108
+ "learning_rate": 0.0002,
109
+ "loss": 0.8241,
110
+ "step": 170
111
+ },
112
+ {
113
+ "epoch": 0.12,
114
+ "learning_rate": 0.0002,
115
+ "loss": 0.8523,
116
+ "step": 180
117
+ },
118
+ {
119
+ "epoch": 0.13,
120
+ "learning_rate": 0.0002,
121
+ "loss": 0.845,
122
+ "step": 190
123
+ },
124
+ {
125
+ "epoch": 0.14,
126
+ "learning_rate": 0.0002,
127
+ "loss": 0.8681,
128
+ "step": 200
129
+ },
130
+ {
131
+ "epoch": 0.14,
132
+ "eval_loss": 0.8539559245109558,
133
+ "eval_runtime": 189.7934,
134
+ "eval_samples_per_second": 5.269,
135
+ "eval_steps_per_second": 2.634,
136
+ "step": 200
137
+ },
138
+ {
139
+ "epoch": 0.14,
140
+ "mmlu_eval_accuracy": 0.47295530665862845,
141
+ "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
142
+ "mmlu_eval_accuracy_anatomy": 0.5,
143
+ "mmlu_eval_accuracy_astronomy": 0.375,
144
+ "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
145
+ "mmlu_eval_accuracy_clinical_knowledge": 0.4482758620689655,
146
+ "mmlu_eval_accuracy_college_biology": 0.4375,
147
+ "mmlu_eval_accuracy_college_chemistry": 0.125,
148
+ "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
149
+ "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
150
+ "mmlu_eval_accuracy_college_medicine": 0.36363636363636365,
151
+ "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
152
+ "mmlu_eval_accuracy_computer_security": 0.45454545454545453,
153
+ "mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231,
154
+ "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
155
+ "mmlu_eval_accuracy_electrical_engineering": 0.5,
156
+ "mmlu_eval_accuracy_elementary_mathematics": 0.34146341463414637,
157
+ "mmlu_eval_accuracy_formal_logic": 0.35714285714285715,
158
+ "mmlu_eval_accuracy_global_facts": 0.5,
159
+ "mmlu_eval_accuracy_high_school_biology": 0.34375,
160
+ "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
161
+ "mmlu_eval_accuracy_high_school_computer_science": 0.7777777777777778,
162
+ "mmlu_eval_accuracy_high_school_european_history": 0.7222222222222222,
163
+ "mmlu_eval_accuracy_high_school_geography": 0.7272727272727273,
164
+ "mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666,
165
+ "mmlu_eval_accuracy_high_school_macroeconomics": 0.3023255813953488,
166
+ "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
167
+ "mmlu_eval_accuracy_high_school_microeconomics": 0.34615384615384615,
168
+ "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
169
+ "mmlu_eval_accuracy_high_school_psychology": 0.7166666666666667,
170
+ "mmlu_eval_accuracy_high_school_statistics": 0.30434782608695654,
171
+ "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
172
+ "mmlu_eval_accuracy_high_school_world_history": 0.5769230769230769,
173
+ "mmlu_eval_accuracy_human_aging": 0.6956521739130435,
174
+ "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
175
+ "mmlu_eval_accuracy_international_law": 0.6923076923076923,
176
+ "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
177
+ "mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556,
178
+ "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
179
+ "mmlu_eval_accuracy_management": 0.7272727272727273,
180
+ "mmlu_eval_accuracy_marketing": 0.76,
181
+ "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
182
+ "mmlu_eval_accuracy_miscellaneous": 0.6744186046511628,
183
+ "mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
184
+ "mmlu_eval_accuracy_moral_scenarios": 0.24,
185
+ "mmlu_eval_accuracy_nutrition": 0.5757575757575758,
186
+ "mmlu_eval_accuracy_philosophy": 0.4117647058823529,
187
+ "mmlu_eval_accuracy_prehistory": 0.42857142857142855,
188
+ "mmlu_eval_accuracy_professional_accounting": 0.2903225806451613,
189
+ "mmlu_eval_accuracy_professional_law": 0.32941176470588235,
190
+ "mmlu_eval_accuracy_professional_medicine": 0.41935483870967744,
191
+ "mmlu_eval_accuracy_professional_psychology": 0.391304347826087,
192
+ "mmlu_eval_accuracy_public_relations": 0.6666666666666666,
193
+ "mmlu_eval_accuracy_security_studies": 0.5185185185185185,
194
+ "mmlu_eval_accuracy_sociology": 0.6818181818181818,
195
+ "mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
196
+ "mmlu_eval_accuracy_virology": 0.4444444444444444,
197
+ "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
198
+ "mmlu_loss": 1.2187759077891358,
199
+ "step": 200
200
+ },
201
+ {
202
+ "epoch": 0.14,
203
+ "learning_rate": 0.0002,
204
+ "loss": 0.8902,
205
+ "step": 210
206
+ },
207
+ {
208
+ "epoch": 0.15,
209
+ "learning_rate": 0.0002,
210
+ "loss": 0.9119,
211
+ "step": 220
212
+ },
213
+ {
214
+ "epoch": 0.16,
215
+ "learning_rate": 0.0002,
216
+ "loss": 0.8849,
217
+ "step": 230
218
+ },
219
+ {
220
+ "epoch": 0.17,
221
+ "learning_rate": 0.0002,
222
+ "loss": 0.8345,
223
+ "step": 240
224
+ },
225
+ {
226
+ "epoch": 0.17,
227
+ "learning_rate": 0.0002,
228
+ "loss": 0.8418,
229
+ "step": 250
230
+ },
231
+ {
232
+ "epoch": 0.18,
233
+ "learning_rate": 0.0002,
234
+ "loss": 0.8305,
235
+ "step": 260
236
+ },
237
+ {
238
+ "epoch": 0.19,
239
+ "learning_rate": 0.0002,
240
+ "loss": 0.8525,
241
+ "step": 270
242
+ },
243
+ {
244
+ "epoch": 0.19,
245
+ "learning_rate": 0.0002,
246
+ "loss": 0.8523,
247
+ "step": 280
248
+ },
249
+ {
250
+ "epoch": 0.2,
251
+ "learning_rate": 0.0002,
252
+ "loss": 0.8511,
253
+ "step": 290
254
+ },
255
+ {
256
+ "epoch": 0.21,
257
+ "learning_rate": 0.0002,
258
+ "loss": 0.8297,
259
+ "step": 300
260
+ },
261
+ {
262
+ "epoch": 0.21,
263
+ "learning_rate": 0.0002,
264
+ "loss": 0.8213,
265
+ "step": 310
266
+ },
267
+ {
268
+ "epoch": 0.22,
269
+ "learning_rate": 0.0002,
270
+ "loss": 0.8225,
271
+ "step": 320
272
+ },
273
+ {
274
+ "epoch": 0.23,
275
+ "learning_rate": 0.0002,
276
+ "loss": 0.8359,
277
+ "step": 330
278
+ },
279
+ {
280
+ "epoch": 0.23,
281
+ "learning_rate": 0.0002,
282
+ "loss": 0.8326,
283
+ "step": 340
284
+ },
285
+ {
286
+ "epoch": 0.24,
287
+ "learning_rate": 0.0002,
288
+ "loss": 0.832,
289
+ "step": 350
290
+ },
291
+ {
292
+ "epoch": 0.25,
293
+ "learning_rate": 0.0002,
294
+ "loss": 0.8208,
295
+ "step": 360
296
+ },
297
+ {
298
+ "epoch": 0.25,
299
+ "learning_rate": 0.0002,
300
+ "loss": 0.8427,
301
+ "step": 370
302
+ },
303
+ {
304
+ "epoch": 0.26,
305
+ "learning_rate": 0.0002,
306
+ "loss": 0.8812,
307
+ "step": 380
308
+ },
309
+ {
310
+ "epoch": 0.27,
311
+ "learning_rate": 0.0002,
312
+ "loss": 0.9016,
313
+ "step": 390
314
+ },
315
+ {
316
+ "epoch": 0.28,
317
+ "learning_rate": 0.0002,
318
+ "loss": 0.8694,
319
+ "step": 400
320
+ },
321
+ {
322
+ "epoch": 0.28,
323
+ "eval_loss": 0.8468822240829468,
324
+ "eval_runtime": 189.8705,
325
+ "eval_samples_per_second": 5.267,
326
+ "eval_steps_per_second": 2.633,
327
+ "step": 400
328
+ },
329
+ {
330
+ "epoch": 0.28,
331
+ "mmlu_eval_accuracy": 0.45864815481629223,
332
+ "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
333
+ "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
334
+ "mmlu_eval_accuracy_astronomy": 0.375,
335
+ "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
336
+ "mmlu_eval_accuracy_clinical_knowledge": 0.4482758620689655,
337
+ "mmlu_eval_accuracy_college_biology": 0.4375,
338
+ "mmlu_eval_accuracy_college_chemistry": 0.125,
339
+ "mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
340
+ "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
341
+ "mmlu_eval_accuracy_college_medicine": 0.3181818181818182,
342
+ "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
343
+ "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
344
+ "mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
345
+ "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
346
+ "mmlu_eval_accuracy_electrical_engineering": 0.375,
347
+ "mmlu_eval_accuracy_elementary_mathematics": 0.34146341463414637,
348
+ "mmlu_eval_accuracy_formal_logic": 0.35714285714285715,
349
+ "mmlu_eval_accuracy_global_facts": 0.5,
350
+ "mmlu_eval_accuracy_high_school_biology": 0.34375,
351
+ "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
352
+ "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
353
+ "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
354
+ "mmlu_eval_accuracy_high_school_geography": 0.6818181818181818,
355
+ "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
356
+ "mmlu_eval_accuracy_high_school_macroeconomics": 0.32558139534883723,
357
+ "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
358
+ "mmlu_eval_accuracy_high_school_microeconomics": 0.34615384615384615,
359
+ "mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
360
+ "mmlu_eval_accuracy_high_school_psychology": 0.7166666666666667,
361
+ "mmlu_eval_accuracy_high_school_statistics": 0.34782608695652173,
362
+ "mmlu_eval_accuracy_high_school_us_history": 0.6818181818181818,
363
+ "mmlu_eval_accuracy_high_school_world_history": 0.5384615384615384,
364
+ "mmlu_eval_accuracy_human_aging": 0.6956521739130435,
365
+ "mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
366
+ "mmlu_eval_accuracy_international_law": 0.7692307692307693,
367
+ "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
368
+ "mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556,
369
+ "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
370
+ "mmlu_eval_accuracy_management": 0.7272727272727273,
371
+ "mmlu_eval_accuracy_marketing": 0.68,
372
+ "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
373
+ "mmlu_eval_accuracy_miscellaneous": 0.6627906976744186,
374
+ "mmlu_eval_accuracy_moral_disputes": 0.5,
375
+ "mmlu_eval_accuracy_moral_scenarios": 0.24,
376
+ "mmlu_eval_accuracy_nutrition": 0.5757575757575758,
377
+ "mmlu_eval_accuracy_philosophy": 0.4411764705882353,
378
+ "mmlu_eval_accuracy_prehistory": 0.42857142857142855,
379
+ "mmlu_eval_accuracy_professional_accounting": 0.3870967741935484,
380
+ "mmlu_eval_accuracy_professional_law": 0.3588235294117647,
381
+ "mmlu_eval_accuracy_professional_medicine": 0.45161290322580644,
382
+ "mmlu_eval_accuracy_professional_psychology": 0.4057971014492754,
383
+ "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
384
+ "mmlu_eval_accuracy_security_studies": 0.5185185185185185,
385
+ "mmlu_eval_accuracy_sociology": 0.5454545454545454,
386
+ "mmlu_eval_accuracy_us_foreign_policy": 0.5454545454545454,
387
+ "mmlu_eval_accuracy_virology": 0.4444444444444444,
388
+ "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
389
+ "mmlu_loss": 0.9554785659084433,
390
+ "step": 400
391
+ },
392
+ {
393
+ "epoch": 0.28,
394
+ "learning_rate": 0.0002,
395
+ "loss": 0.8383,
396
+ "step": 410
397
+ },
398
+ {
399
+ "epoch": 0.29,
400
+ "learning_rate": 0.0002,
401
+ "loss": 0.8315,
402
+ "step": 420
403
+ },
404
+ {
405
+ "epoch": 0.3,
406
+ "learning_rate": 0.0002,
407
+ "loss": 0.8613,
408
+ "step": 430
409
+ },
410
+ {
411
+ "epoch": 0.3,
412
+ "learning_rate": 0.0002,
413
+ "loss": 0.865,
414
+ "step": 440
415
+ },
416
+ {
417
+ "epoch": 0.31,
418
+ "learning_rate": 0.0002,
419
+ "loss": 0.8028,
420
+ "step": 450
421
+ },
422
+ {
423
+ "epoch": 0.32,
424
+ "learning_rate": 0.0002,
425
+ "loss": 0.8557,
426
+ "step": 460
427
+ },
428
+ {
429
+ "epoch": 0.32,
430
+ "learning_rate": 0.0002,
431
+ "loss": 0.8834,
432
+ "step": 470
433
+ },
434
+ {
435
+ "epoch": 0.33,
436
+ "learning_rate": 0.0002,
437
+ "loss": 0.8468,
438
+ "step": 480
439
+ },
440
+ {
441
+ "epoch": 0.34,
442
+ "learning_rate": 0.0002,
443
+ "loss": 0.8556,
444
+ "step": 490
445
+ },
446
+ {
447
+ "epoch": 0.34,
448
+ "learning_rate": 0.0002,
449
+ "loss": 0.8591,
450
+ "step": 500
451
+ },
452
+ {
453
+ "epoch": 0.35,
454
+ "learning_rate": 0.0002,
455
+ "loss": 0.8683,
456
+ "step": 510
457
+ },
458
+ {
459
+ "epoch": 0.36,
460
+ "learning_rate": 0.0002,
461
+ "loss": 0.8278,
462
+ "step": 520
463
+ },
464
+ {
465
+ "epoch": 0.36,
466
+ "learning_rate": 0.0002,
467
+ "loss": 0.8208,
468
+ "step": 530
469
+ },
470
+ {
471
+ "epoch": 0.37,
472
+ "learning_rate": 0.0002,
473
+ "loss": 0.8372,
474
+ "step": 540
475
+ },
476
+ {
477
+ "epoch": 0.38,
478
+ "learning_rate": 0.0002,
479
+ "loss": 0.8144,
480
+ "step": 550
481
+ },
482
+ {
483
+ "epoch": 0.39,
484
+ "learning_rate": 0.0002,
485
+ "loss": 0.8783,
486
+ "step": 560
487
+ },
488
+ {
489
+ "epoch": 0.39,
490
+ "learning_rate": 0.0002,
491
+ "loss": 0.8337,
492
+ "step": 570
493
+ },
494
+ {
495
+ "epoch": 0.4,
496
+ "learning_rate": 0.0002,
497
+ "loss": 0.8572,
498
+ "step": 580
499
+ },
500
+ {
501
+ "epoch": 0.41,
502
+ "learning_rate": 0.0002,
503
+ "loss": 0.8275,
504
+ "step": 590
505
+ },
506
+ {
507
+ "epoch": 0.41,
508
+ "learning_rate": 0.0002,
509
+ "loss": 0.8599,
510
+ "step": 600
511
+ },
512
+ {
513
+ "epoch": 0.41,
514
+ "eval_loss": 0.84308922290802,
515
+ "eval_runtime": 189.8586,
516
+ "eval_samples_per_second": 5.267,
517
+ "eval_steps_per_second": 2.634,
518
+ "step": 600
519
+ },
520
+ {
521
+ "epoch": 0.41,
522
+ "mmlu_eval_accuracy": 0.45447446268908714,
523
+ "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
524
+ "mmlu_eval_accuracy_anatomy": 0.5,
525
+ "mmlu_eval_accuracy_astronomy": 0.375,
526
+ "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
527
+ "mmlu_eval_accuracy_clinical_knowledge": 0.4482758620689655,
528
+ "mmlu_eval_accuracy_college_biology": 0.375,
529
+ "mmlu_eval_accuracy_college_chemistry": 0.125,
530
+ "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
531
+ "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
532
+ "mmlu_eval_accuracy_college_medicine": 0.3181818181818182,
533
+ "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
534
+ "mmlu_eval_accuracy_computer_security": 0.36363636363636365,
535
+ "mmlu_eval_accuracy_conceptual_physics": 0.34615384615384615,
536
+ "mmlu_eval_accuracy_econometrics": 0.16666666666666666,
537
+ "mmlu_eval_accuracy_electrical_engineering": 0.375,
538
+ "mmlu_eval_accuracy_elementary_mathematics": 0.3902439024390244,
539
+ "mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
540
+ "mmlu_eval_accuracy_global_facts": 0.4,
541
+ "mmlu_eval_accuracy_high_school_biology": 0.3125,
542
+ "mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
543
+ "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
544
+ "mmlu_eval_accuracy_high_school_european_history": 0.6111111111111112,
545
+ "mmlu_eval_accuracy_high_school_geography": 0.7272727272727273,
546
+ "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
547
+ "mmlu_eval_accuracy_high_school_macroeconomics": 0.3023255813953488,
548
+ "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
549
+ "mmlu_eval_accuracy_high_school_microeconomics": 0.23076923076923078,
550
+ "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
551
+ "mmlu_eval_accuracy_high_school_psychology": 0.6833333333333333,
552
+ "mmlu_eval_accuracy_high_school_statistics": 0.30434782608695654,
553
+ "mmlu_eval_accuracy_high_school_us_history": 0.7272727272727273,
554
+ "mmlu_eval_accuracy_high_school_world_history": 0.5769230769230769,
555
+ "mmlu_eval_accuracy_human_aging": 0.6521739130434783,
556
+ "mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
557
+ "mmlu_eval_accuracy_international_law": 0.6923076923076923,
558
+ "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
559
+ "mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556,
560
+ "mmlu_eval_accuracy_machine_learning": 0.18181818181818182,
561
+ "mmlu_eval_accuracy_management": 0.45454545454545453,
562
+ "mmlu_eval_accuracy_marketing": 0.72,
563
+ "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
564
+ "mmlu_eval_accuracy_miscellaneous": 0.6511627906976745,
565
+ "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
566
+ "mmlu_eval_accuracy_moral_scenarios": 0.22,
567
+ "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
568
+ "mmlu_eval_accuracy_philosophy": 0.5294117647058824,
569
+ "mmlu_eval_accuracy_prehistory": 0.4,
570
+ "mmlu_eval_accuracy_professional_accounting": 0.3548387096774194,
571
+ "mmlu_eval_accuracy_professional_law": 0.3235294117647059,
572
+ "mmlu_eval_accuracy_professional_medicine": 0.41935483870967744,
573
+ "mmlu_eval_accuracy_professional_psychology": 0.43478260869565216,
574
+ "mmlu_eval_accuracy_public_relations": 0.6666666666666666,
575
+ "mmlu_eval_accuracy_security_studies": 0.5185185185185185,
576
+ "mmlu_eval_accuracy_sociology": 0.5454545454545454,
577
+ "mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
578
+ "mmlu_eval_accuracy_virology": 0.5,
579
+ "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
580
+ "mmlu_loss": 0.9984549878082449,
581
+ "step": 600
582
+ },
583
+ {
584
+ "epoch": 0.42,
585
+ "learning_rate": 0.0002,
586
+ "loss": 0.8279,
587
+ "step": 610
588
+ },
589
+ {
590
+ "epoch": 0.43,
591
+ "learning_rate": 0.0002,
592
+ "loss": 0.8363,
593
+ "step": 620
594
+ },
595
+ {
596
+ "epoch": 0.43,
597
+ "learning_rate": 0.0002,
598
+ "loss": 0.8698,
599
+ "step": 630
600
+ },
601
+ {
602
+ "epoch": 0.44,
603
+ "learning_rate": 0.0002,
604
+ "loss": 0.8407,
605
+ "step": 640
606
+ },
607
+ {
608
+ "epoch": 0.45,
609
+ "learning_rate": 0.0002,
610
+ "loss": 0.8555,
611
+ "step": 650
612
+ },
613
+ {
614
+ "epoch": 0.45,
615
+ "learning_rate": 0.0002,
616
+ "loss": 0.8393,
617
+ "step": 660
618
+ },
619
+ {
620
+ "epoch": 0.46,
621
+ "learning_rate": 0.0002,
622
+ "loss": 0.8527,
623
+ "step": 670
624
+ },
625
+ {
626
+ "epoch": 0.47,
627
+ "learning_rate": 0.0002,
628
+ "loss": 0.8666,
629
+ "step": 680
630
+ },
631
+ {
632
+ "epoch": 0.47,
633
+ "learning_rate": 0.0002,
634
+ "loss": 0.8584,
635
+ "step": 690
636
+ },
637
+ {
638
+ "epoch": 0.48,
639
+ "learning_rate": 0.0002,
640
+ "loss": 0.8583,
641
+ "step": 700
642
+ },
643
+ {
644
+ "epoch": 0.49,
645
+ "learning_rate": 0.0002,
646
+ "loss": 0.8524,
647
+ "step": 710
648
+ },
649
+ {
650
+ "epoch": 0.5,
651
+ "learning_rate": 0.0002,
652
+ "loss": 0.8543,
653
+ "step": 720
654
+ },
655
+ {
656
+ "epoch": 0.5,
657
+ "learning_rate": 0.0002,
658
+ "loss": 0.8687,
659
+ "step": 730
660
+ },
661
+ {
662
+ "epoch": 0.51,
663
+ "learning_rate": 0.0002,
664
+ "loss": 0.8593,
665
+ "step": 740
666
+ },
667
+ {
668
+ "epoch": 0.52,
669
+ "learning_rate": 0.0002,
670
+ "loss": 0.8365,
671
+ "step": 750
672
+ },
673
+ {
674
+ "epoch": 0.52,
675
+ "learning_rate": 0.0002,
676
+ "loss": 0.8718,
677
+ "step": 760
678
+ },
679
+ {
680
+ "epoch": 0.53,
681
+ "learning_rate": 0.0002,
682
+ "loss": 0.8362,
683
+ "step": 770
684
+ },
685
+ {
686
+ "epoch": 0.54,
687
+ "learning_rate": 0.0002,
688
+ "loss": 0.8114,
689
+ "step": 780
690
+ },
691
+ {
692
+ "epoch": 0.54,
693
+ "learning_rate": 0.0002,
694
+ "loss": 0.8611,
695
+ "step": 790
696
+ },
697
+ {
698
+ "epoch": 0.55,
699
+ "learning_rate": 0.0002,
700
+ "loss": 0.8521,
701
+ "step": 800
702
+ },
703
+ {
704
+ "epoch": 0.55,
705
+ "eval_loss": 0.8391241431236267,
706
+ "eval_runtime": 189.8022,
707
+ "eval_samples_per_second": 5.269,
708
+ "eval_steps_per_second": 2.634,
709
+ "step": 800
710
+ },
711
+ {
712
+ "epoch": 0.55,
713
+ "mmlu_eval_accuracy": 0.4694480562846655,
714
+ "mmlu_eval_accuracy_abstract_algebra": 0.45454545454545453,
715
+ "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
716
+ "mmlu_eval_accuracy_astronomy": 0.3125,
717
+ "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
718
+ "mmlu_eval_accuracy_clinical_knowledge": 0.4482758620689655,
719
+ "mmlu_eval_accuracy_college_biology": 0.5,
720
+ "mmlu_eval_accuracy_college_chemistry": 0.25,
721
+ "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
722
+ "mmlu_eval_accuracy_college_mathematics": 0.36363636363636365,
723
+ "mmlu_eval_accuracy_college_medicine": 0.3181818181818182,
724
+ "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
725
+ "mmlu_eval_accuracy_computer_security": 0.45454545454545453,
726
+ "mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
727
+ "mmlu_eval_accuracy_econometrics": 0.25,
728
+ "mmlu_eval_accuracy_electrical_engineering": 0.3125,
729
+ "mmlu_eval_accuracy_elementary_mathematics": 0.34146341463414637,
730
+ "mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
731
+ "mmlu_eval_accuracy_global_facts": 0.5,
732
+ "mmlu_eval_accuracy_high_school_biology": 0.4375,
733
+ "mmlu_eval_accuracy_high_school_chemistry": 0.22727272727272727,
734
+ "mmlu_eval_accuracy_high_school_computer_science": 0.4444444444444444,
735
+ "mmlu_eval_accuracy_high_school_european_history": 0.4444444444444444,
736
+ "mmlu_eval_accuracy_high_school_geography": 0.8636363636363636,
737
+ "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
738
+ "mmlu_eval_accuracy_high_school_macroeconomics": 0.32558139534883723,
739
+ "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
740
+ "mmlu_eval_accuracy_high_school_microeconomics": 0.46153846153846156,
741
+ "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
742
+ "mmlu_eval_accuracy_high_school_psychology": 0.7166666666666667,
743
+ "mmlu_eval_accuracy_high_school_statistics": 0.2608695652173913,
744
+ "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
745
+ "mmlu_eval_accuracy_high_school_world_history": 0.5384615384615384,
746
+ "mmlu_eval_accuracy_human_aging": 0.6956521739130435,
747
+ "mmlu_eval_accuracy_human_sexuality": 0.3333333333333333,
748
+ "mmlu_eval_accuracy_international_law": 0.8461538461538461,
749
+ "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
750
+ "mmlu_eval_accuracy_logical_fallacies": 0.5555555555555556,
751
+ "mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
752
+ "mmlu_eval_accuracy_management": 0.7272727272727273,
753
+ "mmlu_eval_accuracy_marketing": 0.72,
754
+ "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
755
+ "mmlu_eval_accuracy_miscellaneous": 0.6627906976744186,
756
+ "mmlu_eval_accuracy_moral_disputes": 0.47368421052631576,
757
+ "mmlu_eval_accuracy_moral_scenarios": 0.23,
758
+ "mmlu_eval_accuracy_nutrition": 0.6666666666666666,
759
+ "mmlu_eval_accuracy_philosophy": 0.4411764705882353,
760
+ "mmlu_eval_accuracy_prehistory": 0.5428571428571428,
761
+ "mmlu_eval_accuracy_professional_accounting": 0.22580645161290322,
762
+ "mmlu_eval_accuracy_professional_law": 0.31176470588235294,
763
+ "mmlu_eval_accuracy_professional_medicine": 0.41935483870967744,
764
+ "mmlu_eval_accuracy_professional_psychology": 0.4057971014492754,
765
+ "mmlu_eval_accuracy_public_relations": 0.5,
766
+ "mmlu_eval_accuracy_security_studies": 0.48148148148148145,
767
+ "mmlu_eval_accuracy_sociology": 0.7272727272727273,
768
+ "mmlu_eval_accuracy_us_foreign_policy": 0.6363636363636364,
769
+ "mmlu_eval_accuracy_virology": 0.3888888888888889,
770
+ "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
771
+ "mmlu_loss": 1.0093606002337006,
772
+ "step": 800
773
+ }
774
+ ],
775
+ "max_steps": 5000,
776
+ "num_train_epochs": 4,
777
+ "total_flos": 1.4473563545228083e+17,
778
+ "trial_name": null,
779
+ "trial_params": null
780
+ }
checkpoint-800/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5c0932e1068e6c603217c581d0394ebfe8507d5a833b5c14738371df28a02c94
3
+ size 6011