Hugging Face's logo

HydraLM
/
mistralic-expert-16

Model card Files Community
pharaouk commited on
Commit
51e4d7c
•
1 Parent(s): 0531cc4

Training in progress, step 600, checkpoint

Browse files
Files changed (15) hide show
  1. checkpoint-600/README.md +21 -0
  2. checkpoint-600/adapter_config.json +26 -0
  3. checkpoint-600/adapter_model.bin +3 -0
  4. checkpoint-600/adapter_model/adapter_model/README.md +21 -0
  5. checkpoint-600/adapter_model/adapter_model/adapter_config.json +26 -0
  6. checkpoint-600/adapter_model/adapter_model/adapter_model.bin +3 -0
  7. checkpoint-600/added_tokens.json +8 -0
  8. checkpoint-600/optimizer.pt +3 -0
  9. checkpoint-600/rng_state.pth +3 -0
  10. checkpoint-600/scheduler.pt +3 -0
  11. checkpoint-600/special_tokens_map.json +11 -0
  12. checkpoint-600/tokenizer.model +3 -0
  13. checkpoint-600/tokenizer_config.json +74 -0
  14. checkpoint-600/trainer_state.json +592 -0
  15. checkpoint-600/training_args.bin +3 -0
checkpoint-600/README.md ADDED
@@ -0,0 +1,21 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: peft
3
+ ---
4
+ ## Training procedure
5
+
6
+
7
+ The following `bitsandbytes` quantization config was used during training:
8
+ - quant_method: bitsandbytes
9
+ - load_in_8bit: False
10
+ - load_in_4bit: True
11
+ - llm_int8_threshold: 6.0
12
+ - llm_int8_skip_modules: None
13
+ - llm_int8_enable_fp32_cpu_offload: False
14
+ - llm_int8_has_fp16_weight: False
15
+ - bnb_4bit_quant_type: nf4
16
+ - bnb_4bit_use_double_quant: True
17
+ - bnb_4bit_compute_dtype: bfloat16
18
+ ### Framework versions
19
+
20
+
21
+ - PEFT 0.4.0
checkpoint-600/adapter_config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": "SkunkworksAI/Mistralic-7B-1",
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
+ "q_proj",
18
+ "down_proj",
19
+ "o_proj",
20
+ "gate_proj",
21
+ "up_proj",
22
+ "k_proj",
23
+ "v_proj"
24
+ ],
25
+ "task_type": "CAUSAL_LM"
26
+ }
checkpoint-600/adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:04e77d9a77eb2ce3faa0eb51b3aeba31b860f779f97b1f28177ef5aa9af30690
3
+ size 335706314
checkpoint-600/adapter_model/adapter_model/README.md ADDED
@@ -0,0 +1,21 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: peft
3
+ ---
4
+ ## Training procedure
5
+
6
+
7
+ The following `bitsandbytes` quantization config was used during training:
8
+ - quant_method: bitsandbytes
9
+ - load_in_8bit: False
10
+ - load_in_4bit: True
11
+ - llm_int8_threshold: 6.0
12
+ - llm_int8_skip_modules: None
13
+ - llm_int8_enable_fp32_cpu_offload: False
14
+ - llm_int8_has_fp16_weight: False
15
+ - bnb_4bit_quant_type: nf4
16
+ - bnb_4bit_use_double_quant: True
17
+ - bnb_4bit_compute_dtype: bfloat16
18
+ ### Framework versions
19
+
20
+
21
+ - 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": "SkunkworksAI/Mistralic-7B-1",
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
+ "q_proj",
18
+ "down_proj",
19
+ "o_proj",
20
+ "gate_proj",
21
+ "up_proj",
22
+ "k_proj",
23
+ "v_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 sha256:04e77d9a77eb2ce3faa0eb51b3aeba31b860f779f97b1f28177ef5aa9af30690
3
+ size 335706314
checkpoint-600/added_tokens.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "</s>": 2,
3
+ "<s>": 1,
4
+ "<unk>": 0,
5
+ "<|im_end|>": 32000,
6
+ "<|im_start|>": 32001,
7
+ "[PAD]": 32002
8
+ }
checkpoint-600/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c04085789b60226cb2bb0d78408f4b0620abf63143359ddd1cf148058f01b24a
3
+ size 1342453434
checkpoint-600/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7caf8ca6f9fc07e869e1a2b2a0a8594bb9576a3a95e960aca35ef5c79a3dd868
3
+ size 14180
checkpoint-600/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6d85fd0dcc980904bdfd3763281a80094639a2b2b63af9676380dd57c838a883
3
+ size 1064
checkpoint-600/special_tokens_map.json ADDED
@@ -0,0 +1,11 @@
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "additional_special_tokens": [
3
+ "<unk>",
4
+ "<s>",
5
+ "</s>"
6
+ ],
7
+ "bos_token": "<s>",
8
+ "eos_token": "</s>",
9
+ "pad_token": "[PAD]",
10
+ "unk_token": "<unk>"
11
+ }
checkpoint-600/tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:dadfd56d766715c61d2ef780a525ab43b8e6da4de6865bda3d95fdef5e134055
3
+ size 493443
checkpoint-600/tokenizer_config.json ADDED
@@ -0,0 +1,74 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": true,
3
+ "add_eos_token": false,
4
+ "added_tokens_decoder": {
5
+ "0": {
6
+ "content": "<unk>",
7
+ "lstrip": true,
8
+ "normalized": false,
9
+ "rstrip": true,
10
+ "single_word": false,
11
+ "special": true
12
+ },
13
+ "1": {
14
+ "content": "<s>",
15
+ "lstrip": true,
16
+ "normalized": false,
17
+ "rstrip": true,
18
+ "single_word": false,
19
+ "special": true
20
+ },
21
+ "2": {
22
+ "content": "</s>",
23
+ "lstrip": true,
24
+ "normalized": false,
25
+ "rstrip": true,
26
+ "single_word": false,
27
+ "special": true
28
+ },
29
+ "32000": {
30
+ "content": "<|im_end|>",
31
+ "lstrip": true,
32
+ "normalized": true,
33
+ "rstrip": true,
34
+ "single_word": false,
35
+ "special": false
36
+ },
37
+ "32001": {
38
+ "content": "<|im_start|>",
39
+ "lstrip": true,
40
+ "normalized": true,
41
+ "rstrip": true,
42
+ "single_word": false,
43
+ "special": false
44
+ },
45
+ "32002": {
46
+ "content": "[PAD]",
47
+ "lstrip": true,
48
+ "normalized": false,
49
+ "rstrip": true,
50
+ "single_word": false,
51
+ "special": true
52
+ }
53
+ },
54
+ "additional_special_tokens": [
55
+ "<unk>",
56
+ "<s>",
57
+ "</s>"
58
+ ],
59
+ "bos_token": "<s>",
60
+ "clean_up_tokenization_spaces": false,
61
+ "eos_token": "</s>",
62
+ "legacy": true,
63
+ "model_max_length": 1000000000000000019884624838656,
64
+ "pad_token": "[PAD]",
65
+ "padding_side": "right",
66
+ "sp_model_kwargs": {},
67
+ "spaces_between_special_tokens": false,
68
+ "tokenizer_class": "LlamaTokenizer",
69
+ "tokenizer_file": null,
70
+ "trust_remote_code": false,
71
+ "unk_token": "<unk>",
72
+ "use_default_system_prompt": true,
73
+ "use_fast": true
74
+ }
checkpoint-600/trainer_state.json ADDED
@@ -0,0 +1,592 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 0.7431616187095642,
3
+ "best_model_checkpoint": "experts/mistralic-expert-16/checkpoint-600",
4
+ "epoch": 0.19011406844106463,
5
+ "eval_steps": 200,
6
+ "global_step": 600,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.0,
13
+ "learning_rate": 0.0002,
14
+ "loss": 0.7873,
15
+ "step": 10
16
+ },
17
+ {
18
+ "epoch": 0.01,
19
+ "learning_rate": 0.0002,
20
+ "loss": 0.8128,
21
+ "step": 20
22
+ },
23
+ {
24
+ "epoch": 0.01,
25
+ "learning_rate": 0.0002,
26
+ "loss": 0.8641,
27
+ "step": 30
28
+ },
29
+ {
30
+ "epoch": 0.01,
31
+ "learning_rate": 0.0002,
32
+ "loss": 0.8246,
33
+ "step": 40
34
+ },
35
+ {
36
+ "epoch": 0.02,
37
+ "learning_rate": 0.0002,
38
+ "loss": 0.7867,
39
+ "step": 50
40
+ },
41
+ {
42
+ "epoch": 0.02,
43
+ "learning_rate": 0.0002,
44
+ "loss": 0.7705,
45
+ "step": 60
46
+ },
47
+ {
48
+ "epoch": 0.02,
49
+ "learning_rate": 0.0002,
50
+ "loss": 0.7671,
51
+ "step": 70
52
+ },
53
+ {
54
+ "epoch": 0.03,
55
+ "learning_rate": 0.0002,
56
+ "loss": 0.8725,
57
+ "step": 80
58
+ },
59
+ {
60
+ "epoch": 0.03,
61
+ "learning_rate": 0.0002,
62
+ "loss": 0.8337,
63
+ "step": 90
64
+ },
65
+ {
66
+ "epoch": 0.03,
67
+ "learning_rate": 0.0002,
68
+ "loss": 0.7819,
69
+ "step": 100
70
+ },
71
+ {
72
+ "epoch": 0.03,
73
+ "learning_rate": 0.0002,
74
+ "loss": 0.7729,
75
+ "step": 110
76
+ },
77
+ {
78
+ "epoch": 0.04,
79
+ "learning_rate": 0.0002,
80
+ "loss": 0.8169,
81
+ "step": 120
82
+ },
83
+ {
84
+ "epoch": 0.04,
85
+ "learning_rate": 0.0002,
86
+ "loss": 0.7988,
87
+ "step": 130
88
+ },
89
+ {
90
+ "epoch": 0.04,
91
+ "learning_rate": 0.0002,
92
+ "loss": 0.8958,
93
+ "step": 140
94
+ },
95
+ {
96
+ "epoch": 0.05,
97
+ "learning_rate": 0.0002,
98
+ "loss": 0.7682,
99
+ "step": 150
100
+ },
101
+ {
102
+ "epoch": 0.05,
103
+ "learning_rate": 0.0002,
104
+ "loss": 0.7729,
105
+ "step": 160
106
+ },
107
+ {
108
+ "epoch": 0.05,
109
+ "learning_rate": 0.0002,
110
+ "loss": 0.7375,
111
+ "step": 170
112
+ },
113
+ {
114
+ "epoch": 0.06,
115
+ "learning_rate": 0.0002,
116
+ "loss": 0.7756,
117
+ "step": 180
118
+ },
119
+ {
120
+ "epoch": 0.06,
121
+ "learning_rate": 0.0002,
122
+ "loss": 0.8256,
123
+ "step": 190
124
+ },
125
+ {
126
+ "epoch": 0.06,
127
+ "learning_rate": 0.0002,
128
+ "loss": 0.8504,
129
+ "step": 200
130
+ },
131
+ {
132
+ "epoch": 0.06,
133
+ "eval_loss": 0.7519773244857788,
134
+ "eval_runtime": 153.135,
135
+ "eval_samples_per_second": 6.53,
136
+ "eval_steps_per_second": 3.265,
137
+ "step": 200
138
+ },
139
+ {
140
+ "epoch": 0.06,
141
+ "mmlu_eval_accuracy": 0.5986624199855438,
142
+ "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
143
+ "mmlu_eval_accuracy_anatomy": 0.42857142857142855,
144
+ "mmlu_eval_accuracy_astronomy": 0.6875,
145
+ "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
146
+ "mmlu_eval_accuracy_clinical_knowledge": 0.5862068965517241,
147
+ "mmlu_eval_accuracy_college_biology": 0.5625,
148
+ "mmlu_eval_accuracy_college_chemistry": 0.375,
149
+ "mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
150
+ "mmlu_eval_accuracy_college_mathematics": 0.6363636363636364,
151
+ "mmlu_eval_accuracy_college_medicine": 0.6363636363636364,
152
+ "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
153
+ "mmlu_eval_accuracy_computer_security": 0.5454545454545454,
154
+ "mmlu_eval_accuracy_conceptual_physics": 0.5384615384615384,
155
+ "mmlu_eval_accuracy_econometrics": 0.5833333333333334,
156
+ "mmlu_eval_accuracy_electrical_engineering": 0.625,
157
+ "mmlu_eval_accuracy_elementary_mathematics": 0.43902439024390244,
158
+ "mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
159
+ "mmlu_eval_accuracy_global_facts": 0.4,
160
+ "mmlu_eval_accuracy_high_school_biology": 0.59375,
161
+ "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
162
+ "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
163
+ "mmlu_eval_accuracy_high_school_european_history": 0.8333333333333334,
164
+ "mmlu_eval_accuracy_high_school_geography": 0.8636363636363636,
165
+ "mmlu_eval_accuracy_high_school_government_and_politics": 0.7142857142857143,
166
+ "mmlu_eval_accuracy_high_school_macroeconomics": 0.6046511627906976,
167
+ "mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
168
+ "mmlu_eval_accuracy_high_school_microeconomics": 0.5769230769230769,
169
+ "mmlu_eval_accuracy_high_school_physics": 0.23529411764705882,
170
+ "mmlu_eval_accuracy_high_school_psychology": 0.85,
171
+ "mmlu_eval_accuracy_high_school_statistics": 0.34782608695652173,
172
+ "mmlu_eval_accuracy_high_school_us_history": 0.7727272727272727,
173
+ "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
174
+ "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
175
+ "mmlu_eval_accuracy_human_sexuality": 0.5,
176
+ "mmlu_eval_accuracy_international_law": 0.9230769230769231,
177
+ "mmlu_eval_accuracy_jurisprudence": 0.6363636363636364,
178
+ "mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
179
+ "mmlu_eval_accuracy_machine_learning": 0.45454545454545453,
180
+ "mmlu_eval_accuracy_management": 0.9090909090909091,
181
+ "mmlu_eval_accuracy_marketing": 0.88,
182
+ "mmlu_eval_accuracy_medical_genetics": 1.0,
183
+ "mmlu_eval_accuracy_miscellaneous": 0.7558139534883721,
184
+ "mmlu_eval_accuracy_moral_disputes": 0.5526315789473685,
185
+ "mmlu_eval_accuracy_moral_scenarios": 0.27,
186
+ "mmlu_eval_accuracy_nutrition": 0.696969696969697,
187
+ "mmlu_eval_accuracy_philosophy": 0.7647058823529411,
188
+ "mmlu_eval_accuracy_prehistory": 0.5142857142857142,
189
+ "mmlu_eval_accuracy_professional_accounting": 0.6451612903225806,
190
+ "mmlu_eval_accuracy_professional_law": 0.4,
191
+ "mmlu_eval_accuracy_professional_medicine": 0.6129032258064516,
192
+ "mmlu_eval_accuracy_professional_psychology": 0.6231884057971014,
193
+ "mmlu_eval_accuracy_public_relations": 0.5,
194
+ "mmlu_eval_accuracy_security_studies": 0.6296296296296297,
195
+ "mmlu_eval_accuracy_sociology": 0.8181818181818182,
196
+ "mmlu_eval_accuracy_us_foreign_policy": 0.9090909090909091,
197
+ "mmlu_eval_accuracy_virology": 0.5,
198
+ "mmlu_eval_accuracy_world_religions": 0.8421052631578947,
199
+ "mmlu_loss": 1.2932990876174781,
200
+ "step": 200
201
+ },
202
+ {
203
+ "epoch": 0.07,
204
+ "learning_rate": 0.0002,
205
+ "loss": 0.823,
206
+ "step": 210
207
+ },
208
+ {
209
+ "epoch": 0.07,
210
+ "learning_rate": 0.0002,
211
+ "loss": 0.8241,
212
+ "step": 220
213
+ },
214
+ {
215
+ "epoch": 0.07,
216
+ "learning_rate": 0.0002,
217
+ "loss": 0.8277,
218
+ "step": 230
219
+ },
220
+ {
221
+ "epoch": 0.08,
222
+ "learning_rate": 0.0002,
223
+ "loss": 0.8068,
224
+ "step": 240
225
+ },
226
+ {
227
+ "epoch": 0.08,
228
+ "learning_rate": 0.0002,
229
+ "loss": 0.7698,
230
+ "step": 250
231
+ },
232
+ {
233
+ "epoch": 0.08,
234
+ "learning_rate": 0.0002,
235
+ "loss": 0.8068,
236
+ "step": 260
237
+ },
238
+ {
239
+ "epoch": 0.09,
240
+ "learning_rate": 0.0002,
241
+ "loss": 0.7913,
242
+ "step": 270
243
+ },
244
+ {
245
+ "epoch": 0.09,
246
+ "learning_rate": 0.0002,
247
+ "loss": 0.8086,
248
+ "step": 280
249
+ },
250
+ {
251
+ "epoch": 0.09,
252
+ "learning_rate": 0.0002,
253
+ "loss": 0.8127,
254
+ "step": 290
255
+ },
256
+ {
257
+ "epoch": 0.1,
258
+ "learning_rate": 0.0002,
259
+ "loss": 0.7804,
260
+ "step": 300
261
+ },
262
+ {
263
+ "epoch": 0.1,
264
+ "learning_rate": 0.0002,
265
+ "loss": 0.7667,
266
+ "step": 310
267
+ },
268
+ {
269
+ "epoch": 0.1,
270
+ "learning_rate": 0.0002,
271
+ "loss": 0.757,
272
+ "step": 320
273
+ },
274
+ {
275
+ "epoch": 0.1,
276
+ "learning_rate": 0.0002,
277
+ "loss": 0.7438,
278
+ "step": 330
279
+ },
280
+ {
281
+ "epoch": 0.11,
282
+ "learning_rate": 0.0002,
283
+ "loss": 0.768,
284
+ "step": 340
285
+ },
286
+ {
287
+ "epoch": 0.11,
288
+ "learning_rate": 0.0002,
289
+ "loss": 0.8151,
290
+ "step": 350
291
+ },
292
+ {
293
+ "epoch": 0.11,
294
+ "learning_rate": 0.0002,
295
+ "loss": 0.7718,
296
+ "step": 360
297
+ },
298
+ {
299
+ "epoch": 0.12,
300
+ "learning_rate": 0.0002,
301
+ "loss": 0.7903,
302
+ "step": 370
303
+ },
304
+ {
305
+ "epoch": 0.12,
306
+ "learning_rate": 0.0002,
307
+ "loss": 0.7447,
308
+ "step": 380
309
+ },
310
+ {
311
+ "epoch": 0.12,
312
+ "learning_rate": 0.0002,
313
+ "loss": 0.7712,
314
+ "step": 390
315
+ },
316
+ {
317
+ "epoch": 0.13,
318
+ "learning_rate": 0.0002,
319
+ "loss": 0.7808,
320
+ "step": 400
321
+ },
322
+ {
323
+ "epoch": 0.13,
324
+ "eval_loss": 0.7457320690155029,
325
+ "eval_runtime": 152.8304,
326
+ "eval_samples_per_second": 6.543,
327
+ "eval_steps_per_second": 3.272,
328
+ "step": 400
329
+ },
330
+ {
331
+ "epoch": 0.13,
332
+ "mmlu_eval_accuracy": 0.5941853117334526,
333
+ "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
334
+ "mmlu_eval_accuracy_anatomy": 0.5,
335
+ "mmlu_eval_accuracy_astronomy": 0.75,
336
+ "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
337
+ "mmlu_eval_accuracy_clinical_knowledge": 0.6206896551724138,
338
+ "mmlu_eval_accuracy_college_biology": 0.5625,
339
+ "mmlu_eval_accuracy_college_chemistry": 0.25,
340
+ "mmlu_eval_accuracy_college_computer_science": 0.2727272727272727,
341
+ "mmlu_eval_accuracy_college_mathematics": 0.5454545454545454,
342
+ "mmlu_eval_accuracy_college_medicine": 0.6363636363636364,
343
+ "mmlu_eval_accuracy_college_physics": 0.5454545454545454,
344
+ "mmlu_eval_accuracy_computer_security": 0.7272727272727273,
345
+ "mmlu_eval_accuracy_conceptual_physics": 0.5384615384615384,
346
+ "mmlu_eval_accuracy_econometrics": 0.5,
347
+ "mmlu_eval_accuracy_electrical_engineering": 0.5625,
348
+ "mmlu_eval_accuracy_elementary_mathematics": 0.4634146341463415,
349
+ "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
350
+ "mmlu_eval_accuracy_global_facts": 0.4,
351
+ "mmlu_eval_accuracy_high_school_biology": 0.625,
352
+ "mmlu_eval_accuracy_high_school_chemistry": 0.45454545454545453,
353
+ "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
354
+ "mmlu_eval_accuracy_high_school_european_history": 0.7777777777777778,
355
+ "mmlu_eval_accuracy_high_school_geography": 0.8181818181818182,
356
+ "mmlu_eval_accuracy_high_school_government_and_politics": 0.7619047619047619,
357
+ "mmlu_eval_accuracy_high_school_macroeconomics": 0.6046511627906976,
358
+ "mmlu_eval_accuracy_high_school_mathematics": 0.1724137931034483,
359
+ "mmlu_eval_accuracy_high_school_microeconomics": 0.6923076923076923,
360
+ "mmlu_eval_accuracy_high_school_physics": 0.17647058823529413,
361
+ "mmlu_eval_accuracy_high_school_psychology": 0.8666666666666667,
362
+ "mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
363
+ "mmlu_eval_accuracy_high_school_us_history": 0.7727272727272727,
364
+ "mmlu_eval_accuracy_high_school_world_history": 0.6538461538461539,
365
+ "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
366
+ "mmlu_eval_accuracy_human_sexuality": 0.5,
367
+ "mmlu_eval_accuracy_international_law": 1.0,
368
+ "mmlu_eval_accuracy_jurisprudence": 0.6363636363636364,
369
+ "mmlu_eval_accuracy_logical_fallacies": 0.7222222222222222,
370
+ "mmlu_eval_accuracy_machine_learning": 0.5454545454545454,
371
+ "mmlu_eval_accuracy_management": 0.9090909090909091,
372
+ "mmlu_eval_accuracy_marketing": 0.88,
373
+ "mmlu_eval_accuracy_medical_genetics": 1.0,
374
+ "mmlu_eval_accuracy_miscellaneous": 0.7441860465116279,
375
+ "mmlu_eval_accuracy_moral_disputes": 0.6052631578947368,
376
+ "mmlu_eval_accuracy_moral_scenarios": 0.32,
377
+ "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
378
+ "mmlu_eval_accuracy_philosophy": 0.7058823529411765,
379
+ "mmlu_eval_accuracy_prehistory": 0.6285714285714286,
380
+ "mmlu_eval_accuracy_professional_accounting": 0.5483870967741935,
381
+ "mmlu_eval_accuracy_professional_law": 0.4294117647058823,
382
+ "mmlu_eval_accuracy_professional_medicine": 0.6129032258064516,
383
+ "mmlu_eval_accuracy_professional_psychology": 0.5797101449275363,
384
+ "mmlu_eval_accuracy_public_relations": 0.5,
385
+ "mmlu_eval_accuracy_security_studies": 0.5555555555555556,
386
+ "mmlu_eval_accuracy_sociology": 0.8181818181818182,
387
+ "mmlu_eval_accuracy_us_foreign_policy": 0.8181818181818182,
388
+ "mmlu_eval_accuracy_virology": 0.3888888888888889,
389
+ "mmlu_eval_accuracy_world_religions": 0.8421052631578947,
390
+ "mmlu_loss": 1.0765447558606573,
391
+ "step": 400
392
+ },
393
+ {
394
+ "epoch": 0.13,
395
+ "learning_rate": 0.0002,
396
+ "loss": 0.7945,
397
+ "step": 410
398
+ },
399
+ {
400
+ "epoch": 0.13,
401
+ "learning_rate": 0.0002,
402
+ "loss": 0.8205,
403
+ "step": 420
404
+ },
405
+ {
406
+ "epoch": 0.14,
407
+ "learning_rate": 0.0002,
408
+ "loss": 0.8022,
409
+ "step": 430
410
+ },
411
+ {
412
+ "epoch": 0.14,
413
+ "learning_rate": 0.0002,
414
+ "loss": 0.7947,
415
+ "step": 440
416
+ },
417
+ {
418
+ "epoch": 0.14,
419
+ "learning_rate": 0.0002,
420
+ "loss": 0.8091,
421
+ "step": 450
422
+ },
423
+ {
424
+ "epoch": 0.15,
425
+ "learning_rate": 0.0002,
426
+ "loss": 0.8249,
427
+ "step": 460
428
+ },
429
+ {
430
+ "epoch": 0.15,
431
+ "learning_rate": 0.0002,
432
+ "loss": 0.8089,
433
+ "step": 470
434
+ },
435
+ {
436
+ "epoch": 0.15,
437
+ "learning_rate": 0.0002,
438
+ "loss": 0.7972,
439
+ "step": 480
440
+ },
441
+ {
442
+ "epoch": 0.16,
443
+ "learning_rate": 0.0002,
444
+ "loss": 0.8133,
445
+ "step": 490
446
+ },
447
+ {
448
+ "epoch": 0.16,
449
+ "learning_rate": 0.0002,
450
+ "loss": 0.8118,
451
+ "step": 500
452
+ },
453
+ {
454
+ "epoch": 0.16,
455
+ "learning_rate": 0.0002,
456
+ "loss": 0.8342,
457
+ "step": 510
458
+ },
459
+ {
460
+ "epoch": 0.16,
461
+ "learning_rate": 0.0002,
462
+ "loss": 0.8183,
463
+ "step": 520
464
+ },
465
+ {
466
+ "epoch": 0.17,
467
+ "learning_rate": 0.0002,
468
+ "loss": 0.7988,
469
+ "step": 530
470
+ },
471
+ {
472
+ "epoch": 0.17,
473
+ "learning_rate": 0.0002,
474
+ "loss": 0.8296,
475
+ "step": 540
476
+ },
477
+ {
478
+ "epoch": 0.17,
479
+ "learning_rate": 0.0002,
480
+ "loss": 0.7557,
481
+ "step": 550
482
+ },
483
+ {
484
+ "epoch": 0.18,
485
+ "learning_rate": 0.0002,
486
+ "loss": 0.7607,
487
+ "step": 560
488
+ },
489
+ {
490
+ "epoch": 0.18,
491
+ "learning_rate": 0.0002,
492
+ "loss": 0.7358,
493
+ "step": 570
494
+ },
495
+ {
496
+ "epoch": 0.18,
497
+ "learning_rate": 0.0002,
498
+ "loss": 0.7788,
499
+ "step": 580
500
+ },
501
+ {
502
+ "epoch": 0.19,
503
+ "learning_rate": 0.0002,
504
+ "loss": 0.7277,
505
+ "step": 590
506
+ },
507
+ {
508
+ "epoch": 0.19,
509
+ "learning_rate": 0.0002,
510
+ "loss": 0.7814,
511
+ "step": 600
512
+ },
513
+ {
514
+ "epoch": 0.19,
515
+ "eval_loss": 0.7431616187095642,
516
+ "eval_runtime": 152.711,
517
+ "eval_samples_per_second": 6.548,
518
+ "eval_steps_per_second": 3.274,
519
+ "step": 600
520
+ },
521
+ {
522
+ "epoch": 0.19,
523
+ "mmlu_eval_accuracy": 0.5964258160221073,
524
+ "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
525
+ "mmlu_eval_accuracy_anatomy": 0.5,
526
+ "mmlu_eval_accuracy_astronomy": 0.6875,
527
+ "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
528
+ "mmlu_eval_accuracy_clinical_knowledge": 0.6551724137931034,
529
+ "mmlu_eval_accuracy_college_biology": 0.625,
530
+ "mmlu_eval_accuracy_college_chemistry": 0.25,
531
+ "mmlu_eval_accuracy_college_computer_science": 0.2727272727272727,
532
+ "mmlu_eval_accuracy_college_mathematics": 0.5454545454545454,
533
+ "mmlu_eval_accuracy_college_medicine": 0.6363636363636364,
534
+ "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
535
+ "mmlu_eval_accuracy_computer_security": 0.7272727272727273,
536
+ "mmlu_eval_accuracy_conceptual_physics": 0.46153846153846156,
537
+ "mmlu_eval_accuracy_econometrics": 0.5,
538
+ "mmlu_eval_accuracy_electrical_engineering": 0.5625,
539
+ "mmlu_eval_accuracy_elementary_mathematics": 0.4878048780487805,
540
+ "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
541
+ "mmlu_eval_accuracy_global_facts": 0.4,
542
+ "mmlu_eval_accuracy_high_school_biology": 0.59375,
543
+ "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
544
+ "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
545
+ "mmlu_eval_accuracy_high_school_european_history": 0.8333333333333334,
546
+ "mmlu_eval_accuracy_high_school_geography": 0.8636363636363636,
547
+ "mmlu_eval_accuracy_high_school_government_and_politics": 0.7619047619047619,
548
+ "mmlu_eval_accuracy_high_school_macroeconomics": 0.6046511627906976,
549
+ "mmlu_eval_accuracy_high_school_mathematics": 0.3103448275862069,
550
+ "mmlu_eval_accuracy_high_school_microeconomics": 0.6153846153846154,
551
+ "mmlu_eval_accuracy_high_school_physics": 0.11764705882352941,
552
+ "mmlu_eval_accuracy_high_school_psychology": 0.8666666666666667,
553
+ "mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
554
+ "mmlu_eval_accuracy_high_school_us_history": 0.8181818181818182,
555
+ "mmlu_eval_accuracy_high_school_world_history": 0.6923076923076923,
556
+ "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
557
+ "mmlu_eval_accuracy_human_sexuality": 0.5,
558
+ "mmlu_eval_accuracy_international_law": 1.0,
559
+ "mmlu_eval_accuracy_jurisprudence": 0.6363636363636364,
560
+ "mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
561
+ "mmlu_eval_accuracy_machine_learning": 0.6363636363636364,
562
+ "mmlu_eval_accuracy_management": 0.9090909090909091,
563
+ "mmlu_eval_accuracy_marketing": 0.84,
564
+ "mmlu_eval_accuracy_medical_genetics": 1.0,
565
+ "mmlu_eval_accuracy_miscellaneous": 0.7441860465116279,
566
+ "mmlu_eval_accuracy_moral_disputes": 0.5789473684210527,
567
+ "mmlu_eval_accuracy_moral_scenarios": 0.33,
568
+ "mmlu_eval_accuracy_nutrition": 0.7272727272727273,
569
+ "mmlu_eval_accuracy_philosophy": 0.7352941176470589,
570
+ "mmlu_eval_accuracy_prehistory": 0.5428571428571428,
571
+ "mmlu_eval_accuracy_professional_accounting": 0.5161290322580645,
572
+ "mmlu_eval_accuracy_professional_law": 0.4294117647058823,
573
+ "mmlu_eval_accuracy_professional_medicine": 0.5806451612903226,
574
+ "mmlu_eval_accuracy_professional_psychology": 0.6086956521739131,
575
+ "mmlu_eval_accuracy_public_relations": 0.5,
576
+ "mmlu_eval_accuracy_security_studies": 0.5925925925925926,
577
+ "mmlu_eval_accuracy_sociology": 0.9545454545454546,
578
+ "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
579
+ "mmlu_eval_accuracy_virology": 0.5,
580
+ "mmlu_eval_accuracy_world_religions": 0.8421052631578947,
581
+ "mmlu_loss": 1.2551592476374176,
582
+ "step": 600
583
+ }
584
+ ],
585
+ "logging_steps": 10,
586
+ "max_steps": 9468,
587
+ "num_train_epochs": 3,
588
+ "save_steps": 200,
589
+ "total_flos": 3.678923566747484e+17,
590
+ "trial_name": null,
591
+ "trial_params": null
592
+ }
checkpoint-600/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6d0e0baa1bb2aa5937233bd631ead3549b3c1482ab16d538a9907b26a462c171
3
+ size 6392