anchen1011 commited on
Commit
f0f7baf
·
verified ·
1 Parent(s): 29e7a76

Upload folder using huggingface_hub

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. all_results.json +11 -0
  2. checkpoint-1000/README.md +20 -0
  3. checkpoint-1000/adapter_config.json +26 -0
  4. checkpoint-1000/adapter_model.bin +3 -0
  5. checkpoint-1000/adapter_model/README.md +20 -0
  6. checkpoint-1000/adapter_model/adapter_config.json +26 -0
  7. checkpoint-1000/adapter_model/adapter_model.bin +3 -0
  8. checkpoint-1000/optimizer.pt +3 -0
  9. checkpoint-1000/rng_state.pth +3 -0
  10. checkpoint-1000/scheduler.pt +3 -0
  11. checkpoint-1000/special_tokens_map.json +12 -0
  12. checkpoint-1000/tokenizer.model +3 -0
  13. checkpoint-1000/tokenizer_config.json +71 -0
  14. checkpoint-1000/trainer_state.json +971 -0
  15. checkpoint-1000/training_args.bin +3 -0
  16. checkpoint-1200/README.md +20 -0
  17. checkpoint-1200/adapter_config.json +26 -0
  18. checkpoint-1200/adapter_model.bin +3 -0
  19. checkpoint-1200/adapter_model/README.md +20 -0
  20. checkpoint-1200/adapter_model/adapter_config.json +26 -0
  21. checkpoint-1200/adapter_model/adapter_model.bin +3 -0
  22. checkpoint-1200/optimizer.pt +3 -0
  23. checkpoint-1200/rng_state.pth +3 -0
  24. checkpoint-1200/scheduler.pt +3 -0
  25. checkpoint-1200/special_tokens_map.json +12 -0
  26. checkpoint-1200/tokenizer.model +3 -0
  27. checkpoint-1200/tokenizer_config.json +71 -0
  28. checkpoint-1200/trainer_state.json +1162 -0
  29. checkpoint-1200/training_args.bin +3 -0
  30. checkpoint-1400/README.md +20 -0
  31. checkpoint-1400/adapter_config.json +26 -0
  32. checkpoint-1400/adapter_model.bin +3 -0
  33. checkpoint-1400/adapter_model/README.md +20 -0
  34. checkpoint-1400/adapter_model/adapter_config.json +26 -0
  35. checkpoint-1400/adapter_model/adapter_model.bin +3 -0
  36. checkpoint-1400/optimizer.pt +3 -0
  37. checkpoint-1400/rng_state.pth +3 -0
  38. checkpoint-1400/scheduler.pt +3 -0
  39. checkpoint-1400/special_tokens_map.json +12 -0
  40. checkpoint-1400/tokenizer.model +3 -0
  41. checkpoint-1400/tokenizer_config.json +71 -0
  42. checkpoint-1400/trainer_state.json +1353 -0
  43. checkpoint-1400/training_args.bin +3 -0
  44. checkpoint-1600/README.md +20 -0
  45. checkpoint-1600/adapter_config.json +26 -0
  46. checkpoint-1600/adapter_model.bin +3 -0
  47. checkpoint-1600/adapter_model/README.md +20 -0
  48. checkpoint-1600/adapter_model/adapter_config.json +26 -0
  49. checkpoint-1600/adapter_model/adapter_model.bin +3 -0
  50. checkpoint-1600/optimizer.pt +3 -0
all_results.json ADDED
@@ -0,0 +1,11 @@
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 3.4,
3
+ "eval_loss": 1.151158332824707,
4
+ "eval_runtime": 572.7509,
5
+ "eval_samples_per_second": 1.746,
6
+ "eval_steps_per_second": 1.746,
7
+ "train_loss": 0.8699908837636312,
8
+ "train_runtime": 112632.6276,
9
+ "train_samples_per_second": 0.266,
10
+ "train_steps_per_second": 0.017
11
+ }
checkpoint-1000/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-1000/adapter_config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": "152334H/miqu-1-70b-sf",
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": 0.05,
12
+ "modules_to_save": null,
13
+ "peft_type": "LORA",
14
+ "r": 64,
15
+ "revision": null,
16
+ "target_modules": [
17
+ "gate_proj",
18
+ "down_proj",
19
+ "v_proj",
20
+ "up_proj",
21
+ "q_proj",
22
+ "k_proj",
23
+ "o_proj"
24
+ ],
25
+ "task_type": "CAUSAL_LM"
26
+ }
checkpoint-1000/adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8bc29d1b7b8d9fbb7bdf2819f6d0628cea3d5ab845cc689cb80acece39912a3b
3
+ size 1657155522
checkpoint-1000/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-1000/adapter_model/adapter_config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": "152334H/miqu-1-70b-sf",
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": 0.05,
12
+ "modules_to_save": null,
13
+ "peft_type": "LORA",
14
+ "r": 64,
15
+ "revision": null,
16
+ "target_modules": [
17
+ "gate_proj",
18
+ "down_proj",
19
+ "v_proj",
20
+ "up_proj",
21
+ "q_proj",
22
+ "k_proj",
23
+ "o_proj"
24
+ ],
25
+ "task_type": "CAUSAL_LM"
26
+ }
checkpoint-1000/adapter_model/adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8bc29d1b7b8d9fbb7bdf2819f6d0628cea3d5ab845cc689cb80acece39912a3b
3
+ size 1657155522
checkpoint-1000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7fff8ab4eb57b8e9147ca09f977042d1e861ca602fa63f83f85702beb67365ec
3
+ size 6627702922
checkpoint-1000/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e113e075e7ce260c7f7e75bb24de1ff504604347dbc91f90709c86d5a09023f2
3
+ size 14180
checkpoint-1000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:016164c1868d1353a972df97439a4a6f6ad10c19164c770b2c7d8301f524b82a
3
+ size 1064
checkpoint-1000/special_tokens_map.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": "<s>",
3
+ "eos_token": "</s>",
4
+ "pad_token": {
5
+ "content": "<unk>",
6
+ "lstrip": false,
7
+ "normalized": true,
8
+ "rstrip": false,
9
+ "single_word": false
10
+ },
11
+ "unk_token": "<unk>"
12
+ }
checkpoint-1000/tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
3
+ size 499723
checkpoint-1000/tokenizer_config.json ADDED
@@ -0,0 +1,71 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": true,
3
+ "add_eos_token": false,
4
+ "added_tokens_decoder": {
5
+ "0": {
6
+ "content": "<unk>",
7
+ "lstrip": false,
8
+ "normalized": true,
9
+ "rstrip": false,
10
+ "single_word": false,
11
+ "special": true
12
+ },
13
+ "1": {
14
+ "content": "<s>",
15
+ "lstrip": false,
16
+ "normalized": true,
17
+ "rstrip": false,
18
+ "single_word": false,
19
+ "special": true
20
+ },
21
+ "2": {
22
+ "content": "</s>",
23
+ "lstrip": false,
24
+ "normalized": true,
25
+ "rstrip": false,
26
+ "single_word": false,
27
+ "special": true
28
+ }
29
+ },
30
+ "bos_token": {
31
+ "__type": "AddedToken",
32
+ "content": "<s>",
33
+ "lstrip": false,
34
+ "normalized": true,
35
+ "rstrip": false,
36
+ "single_word": false
37
+ },
38
+ "chat_template": "{{ bos_token }}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if message['role'] == 'user' %}{{ '[INST] ' + message['content'] + ' [/INST]' }}{% elif message['role'] == 'assistant' %}{{ message['content'] + eos_token}}{% else %}{{ raise_exception('Only user and assistant roles are supported!') }}{% endif %}{% endfor %}",
39
+ "clean_up_tokenization_spaces": false,
40
+ "eos_token": {
41
+ "__type": "AddedToken",
42
+ "content": "</s>",
43
+ "lstrip": false,
44
+ "normalized": true,
45
+ "rstrip": false,
46
+ "single_word": false
47
+ },
48
+ "legacy": false,
49
+ "model_max_length": 1000000000000000019884624838656,
50
+ "pad_token": {
51
+ "__type": "AddedToken",
52
+ "content": "<unk>",
53
+ "lstrip": false,
54
+ "normalized": true,
55
+ "rstrip": false,
56
+ "single_word": false
57
+ },
58
+ "padding_side": "right",
59
+ "sp_model_kwargs": {},
60
+ "spaces_between_special_tokens": false,
61
+ "tokenizer_class": "LlamaTokenizer",
62
+ "unk_token": {
63
+ "__type": "AddedToken",
64
+ "content": "<unk>",
65
+ "lstrip": false,
66
+ "normalized": true,
67
+ "rstrip": false,
68
+ "single_word": false
69
+ },
70
+ "use_default_system_prompt": false
71
+ }
checkpoint-1000/trainer_state.json ADDED
@@ -0,0 +1,971 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 1.8136476989344819,
5
+ "global_step": 1000,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.02,
12
+ "learning_rate": 0.0001,
13
+ "loss": 1.1655,
14
+ "step": 10
15
+ },
16
+ {
17
+ "epoch": 0.04,
18
+ "learning_rate": 0.0001,
19
+ "loss": 1.001,
20
+ "step": 20
21
+ },
22
+ {
23
+ "epoch": 0.05,
24
+ "learning_rate": 0.0001,
25
+ "loss": 1.0287,
26
+ "step": 30
27
+ },
28
+ {
29
+ "epoch": 0.07,
30
+ "learning_rate": 0.0001,
31
+ "loss": 1.1578,
32
+ "step": 40
33
+ },
34
+ {
35
+ "epoch": 0.09,
36
+ "learning_rate": 0.0001,
37
+ "loss": 1.2146,
38
+ "step": 50
39
+ },
40
+ {
41
+ "epoch": 0.11,
42
+ "learning_rate": 0.0001,
43
+ "loss": 0.997,
44
+ "step": 60
45
+ },
46
+ {
47
+ "epoch": 0.13,
48
+ "learning_rate": 0.0001,
49
+ "loss": 0.9024,
50
+ "step": 70
51
+ },
52
+ {
53
+ "epoch": 0.15,
54
+ "learning_rate": 0.0001,
55
+ "loss": 0.9901,
56
+ "step": 80
57
+ },
58
+ {
59
+ "epoch": 0.16,
60
+ "learning_rate": 0.0001,
61
+ "loss": 1.1264,
62
+ "step": 90
63
+ },
64
+ {
65
+ "epoch": 0.18,
66
+ "learning_rate": 0.0001,
67
+ "loss": 1.2038,
68
+ "step": 100
69
+ },
70
+ {
71
+ "epoch": 0.2,
72
+ "learning_rate": 0.0001,
73
+ "loss": 0.8935,
74
+ "step": 110
75
+ },
76
+ {
77
+ "epoch": 0.22,
78
+ "learning_rate": 0.0001,
79
+ "loss": 0.9178,
80
+ "step": 120
81
+ },
82
+ {
83
+ "epoch": 0.24,
84
+ "learning_rate": 0.0001,
85
+ "loss": 0.9746,
86
+ "step": 130
87
+ },
88
+ {
89
+ "epoch": 0.25,
90
+ "learning_rate": 0.0001,
91
+ "loss": 1.1566,
92
+ "step": 140
93
+ },
94
+ {
95
+ "epoch": 0.27,
96
+ "learning_rate": 0.0001,
97
+ "loss": 1.2877,
98
+ "step": 150
99
+ },
100
+ {
101
+ "epoch": 0.29,
102
+ "learning_rate": 0.0001,
103
+ "loss": 0.9146,
104
+ "step": 160
105
+ },
106
+ {
107
+ "epoch": 0.31,
108
+ "learning_rate": 0.0001,
109
+ "loss": 0.8895,
110
+ "step": 170
111
+ },
112
+ {
113
+ "epoch": 0.33,
114
+ "learning_rate": 0.0001,
115
+ "loss": 1.0121,
116
+ "step": 180
117
+ },
118
+ {
119
+ "epoch": 0.34,
120
+ "eval_loss": 1.0215636491775513,
121
+ "eval_runtime": 950.138,
122
+ "eval_samples_per_second": 1.052,
123
+ "eval_steps_per_second": 1.052,
124
+ "step": 187
125
+ },
126
+ {
127
+ "epoch": 0.34,
128
+ "mmlu_eval_accuracy": 0.731892294851104,
129
+ "mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
130
+ "mmlu_eval_accuracy_anatomy": 0.7857142857142857,
131
+ "mmlu_eval_accuracy_astronomy": 0.6875,
132
+ "mmlu_eval_accuracy_business_ethics": 0.7272727272727273,
133
+ "mmlu_eval_accuracy_clinical_knowledge": 0.896551724137931,
134
+ "mmlu_eval_accuracy_college_biology": 0.875,
135
+ "mmlu_eval_accuracy_college_chemistry": 0.5,
136
+ "mmlu_eval_accuracy_college_computer_science": 0.7272727272727273,
137
+ "mmlu_eval_accuracy_college_mathematics": 0.45454545454545453,
138
+ "mmlu_eval_accuracy_college_medicine": 0.9090909090909091,
139
+ "mmlu_eval_accuracy_college_physics": 0.6363636363636364,
140
+ "mmlu_eval_accuracy_computer_security": 0.6363636363636364,
141
+ "mmlu_eval_accuracy_conceptual_physics": 0.6538461538461539,
142
+ "mmlu_eval_accuracy_econometrics": 0.8333333333333334,
143
+ "mmlu_eval_accuracy_electrical_engineering": 0.8125,
144
+ "mmlu_eval_accuracy_elementary_mathematics": 0.7073170731707317,
145
+ "mmlu_eval_accuracy_formal_logic": 0.5714285714285714,
146
+ "mmlu_eval_accuracy_global_facts": 0.5,
147
+ "mmlu_eval_accuracy_high_school_biology": 0.8125,
148
+ "mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
149
+ "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
150
+ "mmlu_eval_accuracy_high_school_european_history": 0.7777777777777778,
151
+ "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
152
+ "mmlu_eval_accuracy_high_school_government_and_politics": 0.9523809523809523,
153
+ "mmlu_eval_accuracy_high_school_macroeconomics": 0.7441860465116279,
154
+ "mmlu_eval_accuracy_high_school_mathematics": 0.3793103448275862,
155
+ "mmlu_eval_accuracy_high_school_microeconomics": 0.9615384615384616,
156
+ "mmlu_eval_accuracy_high_school_physics": 0.23529411764705882,
157
+ "mmlu_eval_accuracy_high_school_psychology": 0.9333333333333333,
158
+ "mmlu_eval_accuracy_high_school_statistics": 0.6956521739130435,
159
+ "mmlu_eval_accuracy_high_school_us_history": 0.9090909090909091,
160
+ "mmlu_eval_accuracy_high_school_world_history": 0.8076923076923077,
161
+ "mmlu_eval_accuracy_human_aging": 0.8260869565217391,
162
+ "mmlu_eval_accuracy_human_sexuality": 0.6666666666666666,
163
+ "mmlu_eval_accuracy_international_law": 1.0,
164
+ "mmlu_eval_accuracy_jurisprudence": 0.6363636363636364,
165
+ "mmlu_eval_accuracy_logical_fallacies": 0.7777777777777778,
166
+ "mmlu_eval_accuracy_machine_learning": 0.5454545454545454,
167
+ "mmlu_eval_accuracy_management": 0.9090909090909091,
168
+ "mmlu_eval_accuracy_marketing": 0.88,
169
+ "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
170
+ "mmlu_eval_accuracy_miscellaneous": 0.7906976744186046,
171
+ "mmlu_eval_accuracy_moral_disputes": 0.8157894736842105,
172
+ "mmlu_eval_accuracy_moral_scenarios": 0.59,
173
+ "mmlu_eval_accuracy_nutrition": 0.7878787878787878,
174
+ "mmlu_eval_accuracy_philosophy": 0.7941176470588235,
175
+ "mmlu_eval_accuracy_prehistory": 0.8285714285714286,
176
+ "mmlu_eval_accuracy_professional_accounting": 0.6451612903225806,
177
+ "mmlu_eval_accuracy_professional_law": 0.6294117647058823,
178
+ "mmlu_eval_accuracy_professional_medicine": 0.8064516129032258,
179
+ "mmlu_eval_accuracy_professional_psychology": 0.8115942028985508,
180
+ "mmlu_eval_accuracy_public_relations": 0.6666666666666666,
181
+ "mmlu_eval_accuracy_security_studies": 0.8148148148148148,
182
+ "mmlu_eval_accuracy_sociology": 0.9545454545454546,
183
+ "mmlu_eval_accuracy_us_foreign_policy": 1.0,
184
+ "mmlu_eval_accuracy_virology": 0.5,
185
+ "mmlu_eval_accuracy_world_religions": 0.8421052631578947,
186
+ "mmlu_loss": 1.326305795171384,
187
+ "step": 187
188
+ },
189
+ {
190
+ "epoch": 0.34,
191
+ "learning_rate": 0.0001,
192
+ "loss": 1.1133,
193
+ "step": 190
194
+ },
195
+ {
196
+ "epoch": 0.36,
197
+ "learning_rate": 0.0001,
198
+ "loss": 1.2485,
199
+ "step": 200
200
+ },
201
+ {
202
+ "epoch": 0.38,
203
+ "learning_rate": 0.0001,
204
+ "loss": 0.9653,
205
+ "step": 210
206
+ },
207
+ {
208
+ "epoch": 0.4,
209
+ "learning_rate": 0.0001,
210
+ "loss": 0.9455,
211
+ "step": 220
212
+ },
213
+ {
214
+ "epoch": 0.42,
215
+ "learning_rate": 0.0001,
216
+ "loss": 1.0373,
217
+ "step": 230
218
+ },
219
+ {
220
+ "epoch": 0.44,
221
+ "learning_rate": 0.0001,
222
+ "loss": 1.1425,
223
+ "step": 240
224
+ },
225
+ {
226
+ "epoch": 0.45,
227
+ "learning_rate": 0.0001,
228
+ "loss": 1.3136,
229
+ "step": 250
230
+ },
231
+ {
232
+ "epoch": 0.47,
233
+ "learning_rate": 0.0001,
234
+ "loss": 0.8695,
235
+ "step": 260
236
+ },
237
+ {
238
+ "epoch": 0.49,
239
+ "learning_rate": 0.0001,
240
+ "loss": 0.872,
241
+ "step": 270
242
+ },
243
+ {
244
+ "epoch": 0.51,
245
+ "learning_rate": 0.0001,
246
+ "loss": 1.0152,
247
+ "step": 280
248
+ },
249
+ {
250
+ "epoch": 0.53,
251
+ "learning_rate": 0.0001,
252
+ "loss": 1.1309,
253
+ "step": 290
254
+ },
255
+ {
256
+ "epoch": 0.54,
257
+ "learning_rate": 0.0001,
258
+ "loss": 1.267,
259
+ "step": 300
260
+ },
261
+ {
262
+ "epoch": 0.56,
263
+ "learning_rate": 0.0001,
264
+ "loss": 0.9249,
265
+ "step": 310
266
+ },
267
+ {
268
+ "epoch": 0.58,
269
+ "learning_rate": 0.0001,
270
+ "loss": 0.9148,
271
+ "step": 320
272
+ },
273
+ {
274
+ "epoch": 0.6,
275
+ "learning_rate": 0.0001,
276
+ "loss": 0.9864,
277
+ "step": 330
278
+ },
279
+ {
280
+ "epoch": 0.62,
281
+ "learning_rate": 0.0001,
282
+ "loss": 1.2312,
283
+ "step": 340
284
+ },
285
+ {
286
+ "epoch": 0.63,
287
+ "learning_rate": 0.0001,
288
+ "loss": 1.2354,
289
+ "step": 350
290
+ },
291
+ {
292
+ "epoch": 0.65,
293
+ "learning_rate": 0.0001,
294
+ "loss": 0.9126,
295
+ "step": 360
296
+ },
297
+ {
298
+ "epoch": 0.67,
299
+ "learning_rate": 0.0001,
300
+ "loss": 0.9213,
301
+ "step": 370
302
+ },
303
+ {
304
+ "epoch": 0.68,
305
+ "eval_loss": 1.0163359642028809,
306
+ "eval_runtime": 948.1151,
307
+ "eval_samples_per_second": 1.055,
308
+ "eval_steps_per_second": 1.055,
309
+ "step": 374
310
+ },
311
+ {
312
+ "epoch": 0.68,
313
+ "mmlu_eval_accuracy": 0.7395476061435284,
314
+ "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
315
+ "mmlu_eval_accuracy_anatomy": 0.7857142857142857,
316
+ "mmlu_eval_accuracy_astronomy": 0.75,
317
+ "mmlu_eval_accuracy_business_ethics": 0.7272727272727273,
318
+ "mmlu_eval_accuracy_clinical_knowledge": 0.896551724137931,
319
+ "mmlu_eval_accuracy_college_biology": 0.875,
320
+ "mmlu_eval_accuracy_college_chemistry": 0.5,
321
+ "mmlu_eval_accuracy_college_computer_science": 0.7272727272727273,
322
+ "mmlu_eval_accuracy_college_mathematics": 0.36363636363636365,
323
+ "mmlu_eval_accuracy_college_medicine": 0.9090909090909091,
324
+ "mmlu_eval_accuracy_college_physics": 0.6363636363636364,
325
+ "mmlu_eval_accuracy_computer_security": 0.6363636363636364,
326
+ "mmlu_eval_accuracy_conceptual_physics": 0.5769230769230769,
327
+ "mmlu_eval_accuracy_econometrics": 0.8333333333333334,
328
+ "mmlu_eval_accuracy_electrical_engineering": 0.875,
329
+ "mmlu_eval_accuracy_elementary_mathematics": 0.6829268292682927,
330
+ "mmlu_eval_accuracy_formal_logic": 0.6428571428571429,
331
+ "mmlu_eval_accuracy_global_facts": 0.5,
332
+ "mmlu_eval_accuracy_high_school_biology": 0.8125,
333
+ "mmlu_eval_accuracy_high_school_chemistry": 0.45454545454545453,
334
+ "mmlu_eval_accuracy_high_school_computer_science": 0.7777777777777778,
335
+ "mmlu_eval_accuracy_high_school_european_history": 0.7777777777777778,
336
+ "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
337
+ "mmlu_eval_accuracy_high_school_government_and_politics": 0.9523809523809523,
338
+ "mmlu_eval_accuracy_high_school_macroeconomics": 0.7674418604651163,
339
+ "mmlu_eval_accuracy_high_school_mathematics": 0.41379310344827586,
340
+ "mmlu_eval_accuracy_high_school_microeconomics": 0.9615384615384616,
341
+ "mmlu_eval_accuracy_high_school_physics": 0.23529411764705882,
342
+ "mmlu_eval_accuracy_high_school_psychology": 0.95,
343
+ "mmlu_eval_accuracy_high_school_statistics": 0.7391304347826086,
344
+ "mmlu_eval_accuracy_high_school_us_history": 0.9090909090909091,
345
+ "mmlu_eval_accuracy_high_school_world_history": 0.7692307692307693,
346
+ "mmlu_eval_accuracy_human_aging": 0.8260869565217391,
347
+ "mmlu_eval_accuracy_human_sexuality": 0.6666666666666666,
348
+ "mmlu_eval_accuracy_international_law": 1.0,
349
+ "mmlu_eval_accuracy_jurisprudence": 0.6363636363636364,
350
+ "mmlu_eval_accuracy_logical_fallacies": 0.7777777777777778,
351
+ "mmlu_eval_accuracy_machine_learning": 0.5454545454545454,
352
+ "mmlu_eval_accuracy_management": 0.9090909090909091,
353
+ "mmlu_eval_accuracy_marketing": 0.92,
354
+ "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
355
+ "mmlu_eval_accuracy_miscellaneous": 0.7790697674418605,
356
+ "mmlu_eval_accuracy_moral_disputes": 0.8157894736842105,
357
+ "mmlu_eval_accuracy_moral_scenarios": 0.57,
358
+ "mmlu_eval_accuracy_nutrition": 0.7272727272727273,
359
+ "mmlu_eval_accuracy_philosophy": 0.7941176470588235,
360
+ "mmlu_eval_accuracy_prehistory": 0.8571428571428571,
361
+ "mmlu_eval_accuracy_professional_accounting": 0.6774193548387096,
362
+ "mmlu_eval_accuracy_professional_law": 0.6411764705882353,
363
+ "mmlu_eval_accuracy_professional_medicine": 0.8387096774193549,
364
+ "mmlu_eval_accuracy_professional_psychology": 0.8115942028985508,
365
+ "mmlu_eval_accuracy_public_relations": 0.6666666666666666,
366
+ "mmlu_eval_accuracy_security_studies": 0.8148148148148148,
367
+ "mmlu_eval_accuracy_sociology": 0.9545454545454546,
368
+ "mmlu_eval_accuracy_us_foreign_policy": 1.0,
369
+ "mmlu_eval_accuracy_virology": 0.5,
370
+ "mmlu_eval_accuracy_world_religions": 0.8947368421052632,
371
+ "mmlu_loss": 1.2796503596061355,
372
+ "step": 374
373
+ },
374
+ {
375
+ "epoch": 0.69,
376
+ "learning_rate": 0.0001,
377
+ "loss": 0.9737,
378
+ "step": 380
379
+ },
380
+ {
381
+ "epoch": 0.71,
382
+ "learning_rate": 0.0001,
383
+ "loss": 1.157,
384
+ "step": 390
385
+ },
386
+ {
387
+ "epoch": 0.73,
388
+ "learning_rate": 0.0001,
389
+ "loss": 1.2106,
390
+ "step": 400
391
+ },
392
+ {
393
+ "epoch": 0.74,
394
+ "learning_rate": 0.0001,
395
+ "loss": 0.8687,
396
+ "step": 410
397
+ },
398
+ {
399
+ "epoch": 0.76,
400
+ "learning_rate": 0.0001,
401
+ "loss": 0.8742,
402
+ "step": 420
403
+ },
404
+ {
405
+ "epoch": 0.78,
406
+ "learning_rate": 0.0001,
407
+ "loss": 0.9901,
408
+ "step": 430
409
+ },
410
+ {
411
+ "epoch": 0.8,
412
+ "learning_rate": 0.0001,
413
+ "loss": 1.2238,
414
+ "step": 440
415
+ },
416
+ {
417
+ "epoch": 0.82,
418
+ "learning_rate": 0.0001,
419
+ "loss": 1.2604,
420
+ "step": 450
421
+ },
422
+ {
423
+ "epoch": 0.83,
424
+ "learning_rate": 0.0001,
425
+ "loss": 0.8756,
426
+ "step": 460
427
+ },
428
+ {
429
+ "epoch": 0.85,
430
+ "learning_rate": 0.0001,
431
+ "loss": 0.8683,
432
+ "step": 470
433
+ },
434
+ {
435
+ "epoch": 0.87,
436
+ "learning_rate": 0.0001,
437
+ "loss": 0.9824,
438
+ "step": 480
439
+ },
440
+ {
441
+ "epoch": 0.89,
442
+ "learning_rate": 0.0001,
443
+ "loss": 1.1574,
444
+ "step": 490
445
+ },
446
+ {
447
+ "epoch": 0.91,
448
+ "learning_rate": 0.0001,
449
+ "loss": 1.2687,
450
+ "step": 500
451
+ },
452
+ {
453
+ "epoch": 0.92,
454
+ "learning_rate": 0.0001,
455
+ "loss": 0.8657,
456
+ "step": 510
457
+ },
458
+ {
459
+ "epoch": 0.94,
460
+ "learning_rate": 0.0001,
461
+ "loss": 0.9207,
462
+ "step": 520
463
+ },
464
+ {
465
+ "epoch": 0.96,
466
+ "learning_rate": 0.0001,
467
+ "loss": 1.012,
468
+ "step": 530
469
+ },
470
+ {
471
+ "epoch": 0.98,
472
+ "learning_rate": 0.0001,
473
+ "loss": 1.1517,
474
+ "step": 540
475
+ },
476
+ {
477
+ "epoch": 1.0,
478
+ "learning_rate": 0.0001,
479
+ "loss": 1.1654,
480
+ "step": 550
481
+ },
482
+ {
483
+ "epoch": 1.02,
484
+ "learning_rate": 0.0001,
485
+ "loss": 0.8931,
486
+ "step": 560
487
+ },
488
+ {
489
+ "epoch": 1.02,
490
+ "eval_loss": 1.0150845050811768,
491
+ "eval_runtime": 949.8392,
492
+ "eval_samples_per_second": 1.053,
493
+ "eval_steps_per_second": 1.053,
494
+ "step": 561
495
+ },
496
+ {
497
+ "epoch": 1.02,
498
+ "mmlu_eval_accuracy": 0.7346397374699287,
499
+ "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
500
+ "mmlu_eval_accuracy_anatomy": 0.7142857142857143,
501
+ "mmlu_eval_accuracy_astronomy": 0.6875,
502
+ "mmlu_eval_accuracy_business_ethics": 0.8181818181818182,
503
+ "mmlu_eval_accuracy_clinical_knowledge": 0.8620689655172413,
504
+ "mmlu_eval_accuracy_college_biology": 0.875,
505
+ "mmlu_eval_accuracy_college_chemistry": 0.5,
506
+ "mmlu_eval_accuracy_college_computer_science": 0.7272727272727273,
507
+ "mmlu_eval_accuracy_college_mathematics": 0.36363636363636365,
508
+ "mmlu_eval_accuracy_college_medicine": 0.9090909090909091,
509
+ "mmlu_eval_accuracy_college_physics": 0.5454545454545454,
510
+ "mmlu_eval_accuracy_computer_security": 0.6363636363636364,
511
+ "mmlu_eval_accuracy_conceptual_physics": 0.6153846153846154,
512
+ "mmlu_eval_accuracy_econometrics": 0.8333333333333334,
513
+ "mmlu_eval_accuracy_electrical_engineering": 0.8125,
514
+ "mmlu_eval_accuracy_elementary_mathematics": 0.6829268292682927,
515
+ "mmlu_eval_accuracy_formal_logic": 0.6428571428571429,
516
+ "mmlu_eval_accuracy_global_facts": 0.5,
517
+ "mmlu_eval_accuracy_high_school_biology": 0.84375,
518
+ "mmlu_eval_accuracy_high_school_chemistry": 0.45454545454545453,
519
+ "mmlu_eval_accuracy_high_school_computer_science": 0.7777777777777778,
520
+ "mmlu_eval_accuracy_high_school_european_history": 0.7777777777777778,
521
+ "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
522
+ "mmlu_eval_accuracy_high_school_government_and_politics": 0.9523809523809523,
523
+ "mmlu_eval_accuracy_high_school_macroeconomics": 0.7674418604651163,
524
+ "mmlu_eval_accuracy_high_school_mathematics": 0.41379310344827586,
525
+ "mmlu_eval_accuracy_high_school_microeconomics": 0.9230769230769231,
526
+ "mmlu_eval_accuracy_high_school_physics": 0.23529411764705882,
527
+ "mmlu_eval_accuracy_high_school_psychology": 0.9666666666666667,
528
+ "mmlu_eval_accuracy_high_school_statistics": 0.7391304347826086,
529
+ "mmlu_eval_accuracy_high_school_us_history": 0.9090909090909091,
530
+ "mmlu_eval_accuracy_high_school_world_history": 0.8076923076923077,
531
+ "mmlu_eval_accuracy_human_aging": 0.8260869565217391,
532
+ "mmlu_eval_accuracy_human_sexuality": 0.6666666666666666,
533
+ "mmlu_eval_accuracy_international_law": 1.0,
534
+ "mmlu_eval_accuracy_jurisprudence": 0.6363636363636364,
535
+ "mmlu_eval_accuracy_logical_fallacies": 0.7222222222222222,
536
+ "mmlu_eval_accuracy_machine_learning": 0.6363636363636364,
537
+ "mmlu_eval_accuracy_management": 0.9090909090909091,
538
+ "mmlu_eval_accuracy_marketing": 0.88,
539
+ "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
540
+ "mmlu_eval_accuracy_miscellaneous": 0.7558139534883721,
541
+ "mmlu_eval_accuracy_moral_disputes": 0.8157894736842105,
542
+ "mmlu_eval_accuracy_moral_scenarios": 0.56,
543
+ "mmlu_eval_accuracy_nutrition": 0.7575757575757576,
544
+ "mmlu_eval_accuracy_philosophy": 0.7941176470588235,
545
+ "mmlu_eval_accuracy_prehistory": 0.8571428571428571,
546
+ "mmlu_eval_accuracy_professional_accounting": 0.6774193548387096,
547
+ "mmlu_eval_accuracy_professional_law": 0.6294117647058823,
548
+ "mmlu_eval_accuracy_professional_medicine": 0.8064516129032258,
549
+ "mmlu_eval_accuracy_professional_psychology": 0.8260869565217391,
550
+ "mmlu_eval_accuracy_public_relations": 0.6666666666666666,
551
+ "mmlu_eval_accuracy_security_studies": 0.8148148148148148,
552
+ "mmlu_eval_accuracy_sociology": 0.9090909090909091,
553
+ "mmlu_eval_accuracy_us_foreign_policy": 1.0,
554
+ "mmlu_eval_accuracy_virology": 0.5,
555
+ "mmlu_eval_accuracy_world_religions": 0.8421052631578947,
556
+ "mmlu_loss": 1.186674291658526,
557
+ "step": 561
558
+ },
559
+ {
560
+ "epoch": 1.03,
561
+ "learning_rate": 0.0001,
562
+ "loss": 0.8507,
563
+ "step": 570
564
+ },
565
+ {
566
+ "epoch": 1.05,
567
+ "learning_rate": 0.0001,
568
+ "loss": 0.9164,
569
+ "step": 580
570
+ },
571
+ {
572
+ "epoch": 1.07,
573
+ "learning_rate": 0.0001,
574
+ "loss": 1.0908,
575
+ "step": 590
576
+ },
577
+ {
578
+ "epoch": 1.09,
579
+ "learning_rate": 0.0001,
580
+ "loss": 1.0431,
581
+ "step": 600
582
+ },
583
+ {
584
+ "epoch": 1.11,
585
+ "learning_rate": 0.0001,
586
+ "loss": 0.8567,
587
+ "step": 610
588
+ },
589
+ {
590
+ "epoch": 1.12,
591
+ "learning_rate": 0.0001,
592
+ "loss": 0.8818,
593
+ "step": 620
594
+ },
595
+ {
596
+ "epoch": 1.14,
597
+ "learning_rate": 0.0001,
598
+ "loss": 0.9499,
599
+ "step": 630
600
+ },
601
+ {
602
+ "epoch": 1.16,
603
+ "learning_rate": 0.0001,
604
+ "loss": 1.0437,
605
+ "step": 640
606
+ },
607
+ {
608
+ "epoch": 1.18,
609
+ "learning_rate": 0.0001,
610
+ "loss": 1.0487,
611
+ "step": 650
612
+ },
613
+ {
614
+ "epoch": 1.2,
615
+ "learning_rate": 0.0001,
616
+ "loss": 0.8405,
617
+ "step": 660
618
+ },
619
+ {
620
+ "epoch": 1.22,
621
+ "learning_rate": 0.0001,
622
+ "loss": 0.8818,
623
+ "step": 670
624
+ },
625
+ {
626
+ "epoch": 1.23,
627
+ "learning_rate": 0.0001,
628
+ "loss": 0.9619,
629
+ "step": 680
630
+ },
631
+ {
632
+ "epoch": 1.25,
633
+ "learning_rate": 0.0001,
634
+ "loss": 1.0753,
635
+ "step": 690
636
+ },
637
+ {
638
+ "epoch": 1.27,
639
+ "learning_rate": 0.0001,
640
+ "loss": 1.0218,
641
+ "step": 700
642
+ },
643
+ {
644
+ "epoch": 1.29,
645
+ "learning_rate": 0.0001,
646
+ "loss": 0.8763,
647
+ "step": 710
648
+ },
649
+ {
650
+ "epoch": 1.31,
651
+ "learning_rate": 0.0001,
652
+ "loss": 0.8789,
653
+ "step": 720
654
+ },
655
+ {
656
+ "epoch": 1.32,
657
+ "learning_rate": 0.0001,
658
+ "loss": 0.8631,
659
+ "step": 730
660
+ },
661
+ {
662
+ "epoch": 1.34,
663
+ "learning_rate": 0.0001,
664
+ "loss": 0.9846,
665
+ "step": 740
666
+ },
667
+ {
668
+ "epoch": 1.36,
669
+ "eval_loss": 1.0305067300796509,
670
+ "eval_runtime": 948.7106,
671
+ "eval_samples_per_second": 1.054,
672
+ "eval_steps_per_second": 1.054,
673
+ "step": 748
674
+ },
675
+ {
676
+ "epoch": 1.36,
677
+ "mmlu_eval_accuracy": 0.7324229372189777,
678
+ "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
679
+ "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
680
+ "mmlu_eval_accuracy_astronomy": 0.6875,
681
+ "mmlu_eval_accuracy_business_ethics": 1.0,
682
+ "mmlu_eval_accuracy_clinical_knowledge": 0.896551724137931,
683
+ "mmlu_eval_accuracy_college_biology": 0.875,
684
+ "mmlu_eval_accuracy_college_chemistry": 0.5,
685
+ "mmlu_eval_accuracy_college_computer_science": 0.6363636363636364,
686
+ "mmlu_eval_accuracy_college_mathematics": 0.36363636363636365,
687
+ "mmlu_eval_accuracy_college_medicine": 0.8636363636363636,
688
+ "mmlu_eval_accuracy_college_physics": 0.6363636363636364,
689
+ "mmlu_eval_accuracy_computer_security": 0.6363636363636364,
690
+ "mmlu_eval_accuracy_conceptual_physics": 0.6153846153846154,
691
+ "mmlu_eval_accuracy_econometrics": 0.8333333333333334,
692
+ "mmlu_eval_accuracy_electrical_engineering": 0.8125,
693
+ "mmlu_eval_accuracy_elementary_mathematics": 0.7073170731707317,
694
+ "mmlu_eval_accuracy_formal_logic": 0.6428571428571429,
695
+ "mmlu_eval_accuracy_global_facts": 0.4,
696
+ "mmlu_eval_accuracy_high_school_biology": 0.8125,
697
+ "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
698
+ "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
699
+ "mmlu_eval_accuracy_high_school_european_history": 0.7777777777777778,
700
+ "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
701
+ "mmlu_eval_accuracy_high_school_government_and_politics": 0.9523809523809523,
702
+ "mmlu_eval_accuracy_high_school_macroeconomics": 0.7674418604651163,
703
+ "mmlu_eval_accuracy_high_school_mathematics": 0.3448275862068966,
704
+ "mmlu_eval_accuracy_high_school_microeconomics": 0.9615384615384616,
705
+ "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
706
+ "mmlu_eval_accuracy_high_school_psychology": 0.9333333333333333,
707
+ "mmlu_eval_accuracy_high_school_statistics": 0.782608695652174,
708
+ "mmlu_eval_accuracy_high_school_us_history": 0.9545454545454546,
709
+ "mmlu_eval_accuracy_high_school_world_history": 0.7692307692307693,
710
+ "mmlu_eval_accuracy_human_aging": 0.782608695652174,
711
+ "mmlu_eval_accuracy_human_sexuality": 0.6666666666666666,
712
+ "mmlu_eval_accuracy_international_law": 1.0,
713
+ "mmlu_eval_accuracy_jurisprudence": 0.6363636363636364,
714
+ "mmlu_eval_accuracy_logical_fallacies": 0.7777777777777778,
715
+ "mmlu_eval_accuracy_machine_learning": 0.6363636363636364,
716
+ "mmlu_eval_accuracy_management": 0.9090909090909091,
717
+ "mmlu_eval_accuracy_marketing": 0.96,
718
+ "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
719
+ "mmlu_eval_accuracy_miscellaneous": 0.7441860465116279,
720
+ "mmlu_eval_accuracy_moral_disputes": 0.8157894736842105,
721
+ "mmlu_eval_accuracy_moral_scenarios": 0.57,
722
+ "mmlu_eval_accuracy_nutrition": 0.7272727272727273,
723
+ "mmlu_eval_accuracy_philosophy": 0.7941176470588235,
724
+ "mmlu_eval_accuracy_prehistory": 0.8285714285714286,
725
+ "mmlu_eval_accuracy_professional_accounting": 0.6129032258064516,
726
+ "mmlu_eval_accuracy_professional_law": 0.6411764705882353,
727
+ "mmlu_eval_accuracy_professional_medicine": 0.8064516129032258,
728
+ "mmlu_eval_accuracy_professional_psychology": 0.8260869565217391,
729
+ "mmlu_eval_accuracy_public_relations": 0.6666666666666666,
730
+ "mmlu_eval_accuracy_security_studies": 0.8148148148148148,
731
+ "mmlu_eval_accuracy_sociology": 0.9090909090909091,
732
+ "mmlu_eval_accuracy_us_foreign_policy": 1.0,
733
+ "mmlu_eval_accuracy_virology": 0.5,
734
+ "mmlu_eval_accuracy_world_religions": 0.8421052631578947,
735
+ "mmlu_loss": 1.2988067958029479,
736
+ "step": 748
737
+ },
738
+ {
739
+ "epoch": 1.36,
740
+ "learning_rate": 0.0001,
741
+ "loss": 1.0735,
742
+ "step": 750
743
+ },
744
+ {
745
+ "epoch": 1.38,
746
+ "learning_rate": 0.0001,
747
+ "loss": 0.9066,
748
+ "step": 760
749
+ },
750
+ {
751
+ "epoch": 1.4,
752
+ "learning_rate": 0.0001,
753
+ "loss": 0.8716,
754
+ "step": 770
755
+ },
756
+ {
757
+ "epoch": 1.41,
758
+ "learning_rate": 0.0001,
759
+ "loss": 0.9144,
760
+ "step": 780
761
+ },
762
+ {
763
+ "epoch": 1.43,
764
+ "learning_rate": 0.0001,
765
+ "loss": 1.0338,
766
+ "step": 790
767
+ },
768
+ {
769
+ "epoch": 1.45,
770
+ "learning_rate": 0.0001,
771
+ "loss": 1.0275,
772
+ "step": 800
773
+ },
774
+ {
775
+ "epoch": 1.47,
776
+ "learning_rate": 0.0001,
777
+ "loss": 0.8382,
778
+ "step": 810
779
+ },
780
+ {
781
+ "epoch": 1.49,
782
+ "learning_rate": 0.0001,
783
+ "loss": 0.8489,
784
+ "step": 820
785
+ },
786
+ {
787
+ "epoch": 1.51,
788
+ "learning_rate": 0.0001,
789
+ "loss": 0.8931,
790
+ "step": 830
791
+ },
792
+ {
793
+ "epoch": 1.52,
794
+ "learning_rate": 0.0001,
795
+ "loss": 1.0515,
796
+ "step": 840
797
+ },
798
+ {
799
+ "epoch": 1.54,
800
+ "learning_rate": 0.0001,
801
+ "loss": 1.0965,
802
+ "step": 850
803
+ },
804
+ {
805
+ "epoch": 1.56,
806
+ "learning_rate": 0.0001,
807
+ "loss": 0.8928,
808
+ "step": 860
809
+ },
810
+ {
811
+ "epoch": 1.58,
812
+ "learning_rate": 0.0001,
813
+ "loss": 0.8608,
814
+ "step": 870
815
+ },
816
+ {
817
+ "epoch": 1.6,
818
+ "learning_rate": 0.0001,
819
+ "loss": 0.8831,
820
+ "step": 880
821
+ },
822
+ {
823
+ "epoch": 1.61,
824
+ "learning_rate": 0.0001,
825
+ "loss": 1.0253,
826
+ "step": 890
827
+ },
828
+ {
829
+ "epoch": 1.63,
830
+ "learning_rate": 0.0001,
831
+ "loss": 0.9905,
832
+ "step": 900
833
+ },
834
+ {
835
+ "epoch": 1.65,
836
+ "learning_rate": 0.0001,
837
+ "loss": 0.8487,
838
+ "step": 910
839
+ },
840
+ {
841
+ "epoch": 1.67,
842
+ "learning_rate": 0.0001,
843
+ "loss": 0.8568,
844
+ "step": 920
845
+ },
846
+ {
847
+ "epoch": 1.69,
848
+ "learning_rate": 0.0001,
849
+ "loss": 0.9047,
850
+ "step": 930
851
+ },
852
+ {
853
+ "epoch": 1.7,
854
+ "eval_loss": 1.0250624418258667,
855
+ "eval_runtime": 946.4035,
856
+ "eval_samples_per_second": 1.057,
857
+ "eval_steps_per_second": 1.057,
858
+ "step": 935
859
+ },
860
+ {
861
+ "epoch": 1.7,
862
+ "mmlu_eval_accuracy": 0.7288948695878031,
863
+ "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
864
+ "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
865
+ "mmlu_eval_accuracy_astronomy": 0.6875,
866
+ "mmlu_eval_accuracy_business_ethics": 0.9090909090909091,
867
+ "mmlu_eval_accuracy_clinical_knowledge": 0.8620689655172413,
868
+ "mmlu_eval_accuracy_college_biology": 0.875,
869
+ "mmlu_eval_accuracy_college_chemistry": 0.5,
870
+ "mmlu_eval_accuracy_college_computer_science": 0.7272727272727273,
871
+ "mmlu_eval_accuracy_college_mathematics": 0.36363636363636365,
872
+ "mmlu_eval_accuracy_college_medicine": 0.9090909090909091,
873
+ "mmlu_eval_accuracy_college_physics": 0.6363636363636364,
874
+ "mmlu_eval_accuracy_computer_security": 0.6363636363636364,
875
+ "mmlu_eval_accuracy_conceptual_physics": 0.6153846153846154,
876
+ "mmlu_eval_accuracy_econometrics": 0.8333333333333334,
877
+ "mmlu_eval_accuracy_electrical_engineering": 0.8125,
878
+ "mmlu_eval_accuracy_elementary_mathematics": 0.6097560975609756,
879
+ "mmlu_eval_accuracy_formal_logic": 0.6428571428571429,
880
+ "mmlu_eval_accuracy_global_facts": 0.5,
881
+ "mmlu_eval_accuracy_high_school_biology": 0.84375,
882
+ "mmlu_eval_accuracy_high_school_chemistry": 0.45454545454545453,
883
+ "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
884
+ "mmlu_eval_accuracy_high_school_european_history": 0.7777777777777778,
885
+ "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
886
+ "mmlu_eval_accuracy_high_school_government_and_politics": 0.9523809523809523,
887
+ "mmlu_eval_accuracy_high_school_macroeconomics": 0.7906976744186046,
888
+ "mmlu_eval_accuracy_high_school_mathematics": 0.3793103448275862,
889
+ "mmlu_eval_accuracy_high_school_microeconomics": 0.9230769230769231,
890
+ "mmlu_eval_accuracy_high_school_physics": 0.23529411764705882,
891
+ "mmlu_eval_accuracy_high_school_psychology": 0.9333333333333333,
892
+ "mmlu_eval_accuracy_high_school_statistics": 0.7391304347826086,
893
+ "mmlu_eval_accuracy_high_school_us_history": 0.9090909090909091,
894
+ "mmlu_eval_accuracy_high_school_world_history": 0.7692307692307693,
895
+ "mmlu_eval_accuracy_human_aging": 0.782608695652174,
896
+ "mmlu_eval_accuracy_human_sexuality": 0.6666666666666666,
897
+ "mmlu_eval_accuracy_international_law": 1.0,
898
+ "mmlu_eval_accuracy_jurisprudence": 0.5454545454545454,
899
+ "mmlu_eval_accuracy_logical_fallacies": 0.7222222222222222,
900
+ "mmlu_eval_accuracy_machine_learning": 0.6363636363636364,
901
+ "mmlu_eval_accuracy_management": 0.9090909090909091,
902
+ "mmlu_eval_accuracy_marketing": 0.96,
903
+ "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
904
+ "mmlu_eval_accuracy_miscellaneous": 0.7790697674418605,
905
+ "mmlu_eval_accuracy_moral_disputes": 0.8157894736842105,
906
+ "mmlu_eval_accuracy_moral_scenarios": 0.57,
907
+ "mmlu_eval_accuracy_nutrition": 0.7575757575757576,
908
+ "mmlu_eval_accuracy_philosophy": 0.8235294117647058,
909
+ "mmlu_eval_accuracy_prehistory": 0.8857142857142857,
910
+ "mmlu_eval_accuracy_professional_accounting": 0.6129032258064516,
911
+ "mmlu_eval_accuracy_professional_law": 0.6235294117647059,
912
+ "mmlu_eval_accuracy_professional_medicine": 0.8387096774193549,
913
+ "mmlu_eval_accuracy_professional_psychology": 0.8115942028985508,
914
+ "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
915
+ "mmlu_eval_accuracy_security_studies": 0.8148148148148148,
916
+ "mmlu_eval_accuracy_sociology": 0.9090909090909091,
917
+ "mmlu_eval_accuracy_us_foreign_policy": 1.0,
918
+ "mmlu_eval_accuracy_virology": 0.5,
919
+ "mmlu_eval_accuracy_world_religions": 0.8421052631578947,
920
+ "mmlu_loss": 1.243813282909306,
921
+ "step": 935
922
+ },
923
+ {
924
+ "epoch": 1.7,
925
+ "learning_rate": 0.0001,
926
+ "loss": 1.0174,
927
+ "step": 940
928
+ },
929
+ {
930
+ "epoch": 1.72,
931
+ "learning_rate": 0.0001,
932
+ "loss": 1.0302,
933
+ "step": 950
934
+ },
935
+ {
936
+ "epoch": 1.74,
937
+ "learning_rate": 0.0001,
938
+ "loss": 0.8799,
939
+ "step": 960
940
+ },
941
+ {
942
+ "epoch": 1.76,
943
+ "learning_rate": 0.0001,
944
+ "loss": 0.8447,
945
+ "step": 970
946
+ },
947
+ {
948
+ "epoch": 1.78,
949
+ "learning_rate": 0.0001,
950
+ "loss": 0.9053,
951
+ "step": 980
952
+ },
953
+ {
954
+ "epoch": 1.8,
955
+ "learning_rate": 0.0001,
956
+ "loss": 1.0331,
957
+ "step": 990
958
+ },
959
+ {
960
+ "epoch": 1.81,
961
+ "learning_rate": 0.0001,
962
+ "loss": 1.0412,
963
+ "step": 1000
964
+ }
965
+ ],
966
+ "max_steps": 1875,
967
+ "num_train_epochs": 4,
968
+ "total_flos": 1.158485717946876e+18,
969
+ "trial_name": null,
970
+ "trial_params": null
971
+ }
checkpoint-1000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3739b542a2914edbd2d7eb3b727d6fa2a7752c75b0d7a856f5e87dd807fa1ef9
3
+ size 6200
checkpoint-1200/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-1200/adapter_config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": "152334H/miqu-1-70b-sf",
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": 0.05,
12
+ "modules_to_save": null,
13
+ "peft_type": "LORA",
14
+ "r": 64,
15
+ "revision": null,
16
+ "target_modules": [
17
+ "gate_proj",
18
+ "down_proj",
19
+ "v_proj",
20
+ "up_proj",
21
+ "q_proj",
22
+ "k_proj",
23
+ "o_proj"
24
+ ],
25
+ "task_type": "CAUSAL_LM"
26
+ }
checkpoint-1200/adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ad32043ca97c48601084cb3f502e591bccca0879804c4979972d332fc79a801f
3
+ size 1657155522
checkpoint-1200/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-1200/adapter_model/adapter_config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": "152334H/miqu-1-70b-sf",
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": 0.05,
12
+ "modules_to_save": null,
13
+ "peft_type": "LORA",
14
+ "r": 64,
15
+ "revision": null,
16
+ "target_modules": [
17
+ "gate_proj",
18
+ "down_proj",
19
+ "v_proj",
20
+ "up_proj",
21
+ "q_proj",
22
+ "k_proj",
23
+ "o_proj"
24
+ ],
25
+ "task_type": "CAUSAL_LM"
26
+ }
checkpoint-1200/adapter_model/adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ad32043ca97c48601084cb3f502e591bccca0879804c4979972d332fc79a801f
3
+ size 1657155522
checkpoint-1200/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:de2f756f0c8b03dbc310fe2201c53f94e44de0a72ecbe1a58087f0f6916b3c1b
3
+ size 6627702922
checkpoint-1200/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:36539f193dde7f6dd2cc8b72d99a411c97b376a7260c2930cb324081a6c6ee3c
3
+ size 14180
checkpoint-1200/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2c5671c0de422189dff152cae166eb49cff39ac2aa88bdb353ba6e07d93451bf
3
+ size 1064
checkpoint-1200/special_tokens_map.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": "<s>",
3
+ "eos_token": "</s>",
4
+ "pad_token": {
5
+ "content": "<unk>",
6
+ "lstrip": false,
7
+ "normalized": true,
8
+ "rstrip": false,
9
+ "single_word": false
10
+ },
11
+ "unk_token": "<unk>"
12
+ }
checkpoint-1200/tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
3
+ size 499723
checkpoint-1200/tokenizer_config.json ADDED
@@ -0,0 +1,71 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": true,
3
+ "add_eos_token": false,
4
+ "added_tokens_decoder": {
5
+ "0": {
6
+ "content": "<unk>",
7
+ "lstrip": false,
8
+ "normalized": true,
9
+ "rstrip": false,
10
+ "single_word": false,
11
+ "special": true
12
+ },
13
+ "1": {
14
+ "content": "<s>",
15
+ "lstrip": false,
16
+ "normalized": true,
17
+ "rstrip": false,
18
+ "single_word": false,
19
+ "special": true
20
+ },
21
+ "2": {
22
+ "content": "</s>",
23
+ "lstrip": false,
24
+ "normalized": true,
25
+ "rstrip": false,
26
+ "single_word": false,
27
+ "special": true
28
+ }
29
+ },
30
+ "bos_token": {
31
+ "__type": "AddedToken",
32
+ "content": "<s>",
33
+ "lstrip": false,
34
+ "normalized": true,
35
+ "rstrip": false,
36
+ "single_word": false
37
+ },
38
+ "chat_template": "{{ bos_token }}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if message['role'] == 'user' %}{{ '[INST] ' + message['content'] + ' [/INST]' }}{% elif message['role'] == 'assistant' %}{{ message['content'] + eos_token}}{% else %}{{ raise_exception('Only user and assistant roles are supported!') }}{% endif %}{% endfor %}",
39
+ "clean_up_tokenization_spaces": false,
40
+ "eos_token": {
41
+ "__type": "AddedToken",
42
+ "content": "</s>",
43
+ "lstrip": false,
44
+ "normalized": true,
45
+ "rstrip": false,
46
+ "single_word": false
47
+ },
48
+ "legacy": false,
49
+ "model_max_length": 1000000000000000019884624838656,
50
+ "pad_token": {
51
+ "__type": "AddedToken",
52
+ "content": "<unk>",
53
+ "lstrip": false,
54
+ "normalized": true,
55
+ "rstrip": false,
56
+ "single_word": false
57
+ },
58
+ "padding_side": "right",
59
+ "sp_model_kwargs": {},
60
+ "spaces_between_special_tokens": false,
61
+ "tokenizer_class": "LlamaTokenizer",
62
+ "unk_token": {
63
+ "__type": "AddedToken",
64
+ "content": "<unk>",
65
+ "lstrip": false,
66
+ "normalized": true,
67
+ "rstrip": false,
68
+ "single_word": false
69
+ },
70
+ "use_default_system_prompt": false
71
+ }
checkpoint-1200/trainer_state.json ADDED
@@ -0,0 +1,1162 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 2.1763772387213782,
5
+ "global_step": 1200,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.02,
12
+ "learning_rate": 0.0001,
13
+ "loss": 1.1655,
14
+ "step": 10
15
+ },
16
+ {
17
+ "epoch": 0.04,
18
+ "learning_rate": 0.0001,
19
+ "loss": 1.001,
20
+ "step": 20
21
+ },
22
+ {
23
+ "epoch": 0.05,
24
+ "learning_rate": 0.0001,
25
+ "loss": 1.0287,
26
+ "step": 30
27
+ },
28
+ {
29
+ "epoch": 0.07,
30
+ "learning_rate": 0.0001,
31
+ "loss": 1.1578,
32
+ "step": 40
33
+ },
34
+ {
35
+ "epoch": 0.09,
36
+ "learning_rate": 0.0001,
37
+ "loss": 1.2146,
38
+ "step": 50
39
+ },
40
+ {
41
+ "epoch": 0.11,
42
+ "learning_rate": 0.0001,
43
+ "loss": 0.997,
44
+ "step": 60
45
+ },
46
+ {
47
+ "epoch": 0.13,
48
+ "learning_rate": 0.0001,
49
+ "loss": 0.9024,
50
+ "step": 70
51
+ },
52
+ {
53
+ "epoch": 0.15,
54
+ "learning_rate": 0.0001,
55
+ "loss": 0.9901,
56
+ "step": 80
57
+ },
58
+ {
59
+ "epoch": 0.16,
60
+ "learning_rate": 0.0001,
61
+ "loss": 1.1264,
62
+ "step": 90
63
+ },
64
+ {
65
+ "epoch": 0.18,
66
+ "learning_rate": 0.0001,
67
+ "loss": 1.2038,
68
+ "step": 100
69
+ },
70
+ {
71
+ "epoch": 0.2,
72
+ "learning_rate": 0.0001,
73
+ "loss": 0.8935,
74
+ "step": 110
75
+ },
76
+ {
77
+ "epoch": 0.22,
78
+ "learning_rate": 0.0001,
79
+ "loss": 0.9178,
80
+ "step": 120
81
+ },
82
+ {
83
+ "epoch": 0.24,
84
+ "learning_rate": 0.0001,
85
+ "loss": 0.9746,
86
+ "step": 130
87
+ },
88
+ {
89
+ "epoch": 0.25,
90
+ "learning_rate": 0.0001,
91
+ "loss": 1.1566,
92
+ "step": 140
93
+ },
94
+ {
95
+ "epoch": 0.27,
96
+ "learning_rate": 0.0001,
97
+ "loss": 1.2877,
98
+ "step": 150
99
+ },
100
+ {
101
+ "epoch": 0.29,
102
+ "learning_rate": 0.0001,
103
+ "loss": 0.9146,
104
+ "step": 160
105
+ },
106
+ {
107
+ "epoch": 0.31,
108
+ "learning_rate": 0.0001,
109
+ "loss": 0.8895,
110
+ "step": 170
111
+ },
112
+ {
113
+ "epoch": 0.33,
114
+ "learning_rate": 0.0001,
115
+ "loss": 1.0121,
116
+ "step": 180
117
+ },
118
+ {
119
+ "epoch": 0.34,
120
+ "eval_loss": 1.0215636491775513,
121
+ "eval_runtime": 950.138,
122
+ "eval_samples_per_second": 1.052,
123
+ "eval_steps_per_second": 1.052,
124
+ "step": 187
125
+ },
126
+ {
127
+ "epoch": 0.34,
128
+ "mmlu_eval_accuracy": 0.731892294851104,
129
+ "mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
130
+ "mmlu_eval_accuracy_anatomy": 0.7857142857142857,
131
+ "mmlu_eval_accuracy_astronomy": 0.6875,
132
+ "mmlu_eval_accuracy_business_ethics": 0.7272727272727273,
133
+ "mmlu_eval_accuracy_clinical_knowledge": 0.896551724137931,
134
+ "mmlu_eval_accuracy_college_biology": 0.875,
135
+ "mmlu_eval_accuracy_college_chemistry": 0.5,
136
+ "mmlu_eval_accuracy_college_computer_science": 0.7272727272727273,
137
+ "mmlu_eval_accuracy_college_mathematics": 0.45454545454545453,
138
+ "mmlu_eval_accuracy_college_medicine": 0.9090909090909091,
139
+ "mmlu_eval_accuracy_college_physics": 0.6363636363636364,
140
+ "mmlu_eval_accuracy_computer_security": 0.6363636363636364,
141
+ "mmlu_eval_accuracy_conceptual_physics": 0.6538461538461539,
142
+ "mmlu_eval_accuracy_econometrics": 0.8333333333333334,
143
+ "mmlu_eval_accuracy_electrical_engineering": 0.8125,
144
+ "mmlu_eval_accuracy_elementary_mathematics": 0.7073170731707317,
145
+ "mmlu_eval_accuracy_formal_logic": 0.5714285714285714,
146
+ "mmlu_eval_accuracy_global_facts": 0.5,
147
+ "mmlu_eval_accuracy_high_school_biology": 0.8125,
148
+ "mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
149
+ "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
150
+ "mmlu_eval_accuracy_high_school_european_history": 0.7777777777777778,
151
+ "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
152
+ "mmlu_eval_accuracy_high_school_government_and_politics": 0.9523809523809523,
153
+ "mmlu_eval_accuracy_high_school_macroeconomics": 0.7441860465116279,
154
+ "mmlu_eval_accuracy_high_school_mathematics": 0.3793103448275862,
155
+ "mmlu_eval_accuracy_high_school_microeconomics": 0.9615384615384616,
156
+ "mmlu_eval_accuracy_high_school_physics": 0.23529411764705882,
157
+ "mmlu_eval_accuracy_high_school_psychology": 0.9333333333333333,
158
+ "mmlu_eval_accuracy_high_school_statistics": 0.6956521739130435,
159
+ "mmlu_eval_accuracy_high_school_us_history": 0.9090909090909091,
160
+ "mmlu_eval_accuracy_high_school_world_history": 0.8076923076923077,
161
+ "mmlu_eval_accuracy_human_aging": 0.8260869565217391,
162
+ "mmlu_eval_accuracy_human_sexuality": 0.6666666666666666,
163
+ "mmlu_eval_accuracy_international_law": 1.0,
164
+ "mmlu_eval_accuracy_jurisprudence": 0.6363636363636364,
165
+ "mmlu_eval_accuracy_logical_fallacies": 0.7777777777777778,
166
+ "mmlu_eval_accuracy_machine_learning": 0.5454545454545454,
167
+ "mmlu_eval_accuracy_management": 0.9090909090909091,
168
+ "mmlu_eval_accuracy_marketing": 0.88,
169
+ "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
170
+ "mmlu_eval_accuracy_miscellaneous": 0.7906976744186046,
171
+ "mmlu_eval_accuracy_moral_disputes": 0.8157894736842105,
172
+ "mmlu_eval_accuracy_moral_scenarios": 0.59,
173
+ "mmlu_eval_accuracy_nutrition": 0.7878787878787878,
174
+ "mmlu_eval_accuracy_philosophy": 0.7941176470588235,
175
+ "mmlu_eval_accuracy_prehistory": 0.8285714285714286,
176
+ "mmlu_eval_accuracy_professional_accounting": 0.6451612903225806,
177
+ "mmlu_eval_accuracy_professional_law": 0.6294117647058823,
178
+ "mmlu_eval_accuracy_professional_medicine": 0.8064516129032258,
179
+ "mmlu_eval_accuracy_professional_psychology": 0.8115942028985508,
180
+ "mmlu_eval_accuracy_public_relations": 0.6666666666666666,
181
+ "mmlu_eval_accuracy_security_studies": 0.8148148148148148,
182
+ "mmlu_eval_accuracy_sociology": 0.9545454545454546,
183
+ "mmlu_eval_accuracy_us_foreign_policy": 1.0,
184
+ "mmlu_eval_accuracy_virology": 0.5,
185
+ "mmlu_eval_accuracy_world_religions": 0.8421052631578947,
186
+ "mmlu_loss": 1.326305795171384,
187
+ "step": 187
188
+ },
189
+ {
190
+ "epoch": 0.34,
191
+ "learning_rate": 0.0001,
192
+ "loss": 1.1133,
193
+ "step": 190
194
+ },
195
+ {
196
+ "epoch": 0.36,
197
+ "learning_rate": 0.0001,
198
+ "loss": 1.2485,
199
+ "step": 200
200
+ },
201
+ {
202
+ "epoch": 0.38,
203
+ "learning_rate": 0.0001,
204
+ "loss": 0.9653,
205
+ "step": 210
206
+ },
207
+ {
208
+ "epoch": 0.4,
209
+ "learning_rate": 0.0001,
210
+ "loss": 0.9455,
211
+ "step": 220
212
+ },
213
+ {
214
+ "epoch": 0.42,
215
+ "learning_rate": 0.0001,
216
+ "loss": 1.0373,
217
+ "step": 230
218
+ },
219
+ {
220
+ "epoch": 0.44,
221
+ "learning_rate": 0.0001,
222
+ "loss": 1.1425,
223
+ "step": 240
224
+ },
225
+ {
226
+ "epoch": 0.45,
227
+ "learning_rate": 0.0001,
228
+ "loss": 1.3136,
229
+ "step": 250
230
+ },
231
+ {
232
+ "epoch": 0.47,
233
+ "learning_rate": 0.0001,
234
+ "loss": 0.8695,
235
+ "step": 260
236
+ },
237
+ {
238
+ "epoch": 0.49,
239
+ "learning_rate": 0.0001,
240
+ "loss": 0.872,
241
+ "step": 270
242
+ },
243
+ {
244
+ "epoch": 0.51,
245
+ "learning_rate": 0.0001,
246
+ "loss": 1.0152,
247
+ "step": 280
248
+ },
249
+ {
250
+ "epoch": 0.53,
251
+ "learning_rate": 0.0001,
252
+ "loss": 1.1309,
253
+ "step": 290
254
+ },
255
+ {
256
+ "epoch": 0.54,
257
+ "learning_rate": 0.0001,
258
+ "loss": 1.267,
259
+ "step": 300
260
+ },
261
+ {
262
+ "epoch": 0.56,
263
+ "learning_rate": 0.0001,
264
+ "loss": 0.9249,
265
+ "step": 310
266
+ },
267
+ {
268
+ "epoch": 0.58,
269
+ "learning_rate": 0.0001,
270
+ "loss": 0.9148,
271
+ "step": 320
272
+ },
273
+ {
274
+ "epoch": 0.6,
275
+ "learning_rate": 0.0001,
276
+ "loss": 0.9864,
277
+ "step": 330
278
+ },
279
+ {
280
+ "epoch": 0.62,
281
+ "learning_rate": 0.0001,
282
+ "loss": 1.2312,
283
+ "step": 340
284
+ },
285
+ {
286
+ "epoch": 0.63,
287
+ "learning_rate": 0.0001,
288
+ "loss": 1.2354,
289
+ "step": 350
290
+ },
291
+ {
292
+ "epoch": 0.65,
293
+ "learning_rate": 0.0001,
294
+ "loss": 0.9126,
295
+ "step": 360
296
+ },
297
+ {
298
+ "epoch": 0.67,
299
+ "learning_rate": 0.0001,
300
+ "loss": 0.9213,
301
+ "step": 370
302
+ },
303
+ {
304
+ "epoch": 0.68,
305
+ "eval_loss": 1.0163359642028809,
306
+ "eval_runtime": 948.1151,
307
+ "eval_samples_per_second": 1.055,
308
+ "eval_steps_per_second": 1.055,
309
+ "step": 374
310
+ },
311
+ {
312
+ "epoch": 0.68,
313
+ "mmlu_eval_accuracy": 0.7395476061435284,
314
+ "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
315
+ "mmlu_eval_accuracy_anatomy": 0.7857142857142857,
316
+ "mmlu_eval_accuracy_astronomy": 0.75,
317
+ "mmlu_eval_accuracy_business_ethics": 0.7272727272727273,
318
+ "mmlu_eval_accuracy_clinical_knowledge": 0.896551724137931,
319
+ "mmlu_eval_accuracy_college_biology": 0.875,
320
+ "mmlu_eval_accuracy_college_chemistry": 0.5,
321
+ "mmlu_eval_accuracy_college_computer_science": 0.7272727272727273,
322
+ "mmlu_eval_accuracy_college_mathematics": 0.36363636363636365,
323
+ "mmlu_eval_accuracy_college_medicine": 0.9090909090909091,
324
+ "mmlu_eval_accuracy_college_physics": 0.6363636363636364,
325
+ "mmlu_eval_accuracy_computer_security": 0.6363636363636364,
326
+ "mmlu_eval_accuracy_conceptual_physics": 0.5769230769230769,
327
+ "mmlu_eval_accuracy_econometrics": 0.8333333333333334,
328
+ "mmlu_eval_accuracy_electrical_engineering": 0.875,
329
+ "mmlu_eval_accuracy_elementary_mathematics": 0.6829268292682927,
330
+ "mmlu_eval_accuracy_formal_logic": 0.6428571428571429,
331
+ "mmlu_eval_accuracy_global_facts": 0.5,
332
+ "mmlu_eval_accuracy_high_school_biology": 0.8125,
333
+ "mmlu_eval_accuracy_high_school_chemistry": 0.45454545454545453,
334
+ "mmlu_eval_accuracy_high_school_computer_science": 0.7777777777777778,
335
+ "mmlu_eval_accuracy_high_school_european_history": 0.7777777777777778,
336
+ "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
337
+ "mmlu_eval_accuracy_high_school_government_and_politics": 0.9523809523809523,
338
+ "mmlu_eval_accuracy_high_school_macroeconomics": 0.7674418604651163,
339
+ "mmlu_eval_accuracy_high_school_mathematics": 0.41379310344827586,
340
+ "mmlu_eval_accuracy_high_school_microeconomics": 0.9615384615384616,
341
+ "mmlu_eval_accuracy_high_school_physics": 0.23529411764705882,
342
+ "mmlu_eval_accuracy_high_school_psychology": 0.95,
343
+ "mmlu_eval_accuracy_high_school_statistics": 0.7391304347826086,
344
+ "mmlu_eval_accuracy_high_school_us_history": 0.9090909090909091,
345
+ "mmlu_eval_accuracy_high_school_world_history": 0.7692307692307693,
346
+ "mmlu_eval_accuracy_human_aging": 0.8260869565217391,
347
+ "mmlu_eval_accuracy_human_sexuality": 0.6666666666666666,
348
+ "mmlu_eval_accuracy_international_law": 1.0,
349
+ "mmlu_eval_accuracy_jurisprudence": 0.6363636363636364,
350
+ "mmlu_eval_accuracy_logical_fallacies": 0.7777777777777778,
351
+ "mmlu_eval_accuracy_machine_learning": 0.5454545454545454,
352
+ "mmlu_eval_accuracy_management": 0.9090909090909091,
353
+ "mmlu_eval_accuracy_marketing": 0.92,
354
+ "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
355
+ "mmlu_eval_accuracy_miscellaneous": 0.7790697674418605,
356
+ "mmlu_eval_accuracy_moral_disputes": 0.8157894736842105,
357
+ "mmlu_eval_accuracy_moral_scenarios": 0.57,
358
+ "mmlu_eval_accuracy_nutrition": 0.7272727272727273,
359
+ "mmlu_eval_accuracy_philosophy": 0.7941176470588235,
360
+ "mmlu_eval_accuracy_prehistory": 0.8571428571428571,
361
+ "mmlu_eval_accuracy_professional_accounting": 0.6774193548387096,
362
+ "mmlu_eval_accuracy_professional_law": 0.6411764705882353,
363
+ "mmlu_eval_accuracy_professional_medicine": 0.8387096774193549,
364
+ "mmlu_eval_accuracy_professional_psychology": 0.8115942028985508,
365
+ "mmlu_eval_accuracy_public_relations": 0.6666666666666666,
366
+ "mmlu_eval_accuracy_security_studies": 0.8148148148148148,
367
+ "mmlu_eval_accuracy_sociology": 0.9545454545454546,
368
+ "mmlu_eval_accuracy_us_foreign_policy": 1.0,
369
+ "mmlu_eval_accuracy_virology": 0.5,
370
+ "mmlu_eval_accuracy_world_religions": 0.8947368421052632,
371
+ "mmlu_loss": 1.2796503596061355,
372
+ "step": 374
373
+ },
374
+ {
375
+ "epoch": 0.69,
376
+ "learning_rate": 0.0001,
377
+ "loss": 0.9737,
378
+ "step": 380
379
+ },
380
+ {
381
+ "epoch": 0.71,
382
+ "learning_rate": 0.0001,
383
+ "loss": 1.157,
384
+ "step": 390
385
+ },
386
+ {
387
+ "epoch": 0.73,
388
+ "learning_rate": 0.0001,
389
+ "loss": 1.2106,
390
+ "step": 400
391
+ },
392
+ {
393
+ "epoch": 0.74,
394
+ "learning_rate": 0.0001,
395
+ "loss": 0.8687,
396
+ "step": 410
397
+ },
398
+ {
399
+ "epoch": 0.76,
400
+ "learning_rate": 0.0001,
401
+ "loss": 0.8742,
402
+ "step": 420
403
+ },
404
+ {
405
+ "epoch": 0.78,
406
+ "learning_rate": 0.0001,
407
+ "loss": 0.9901,
408
+ "step": 430
409
+ },
410
+ {
411
+ "epoch": 0.8,
412
+ "learning_rate": 0.0001,
413
+ "loss": 1.2238,
414
+ "step": 440
415
+ },
416
+ {
417
+ "epoch": 0.82,
418
+ "learning_rate": 0.0001,
419
+ "loss": 1.2604,
420
+ "step": 450
421
+ },
422
+ {
423
+ "epoch": 0.83,
424
+ "learning_rate": 0.0001,
425
+ "loss": 0.8756,
426
+ "step": 460
427
+ },
428
+ {
429
+ "epoch": 0.85,
430
+ "learning_rate": 0.0001,
431
+ "loss": 0.8683,
432
+ "step": 470
433
+ },
434
+ {
435
+ "epoch": 0.87,
436
+ "learning_rate": 0.0001,
437
+ "loss": 0.9824,
438
+ "step": 480
439
+ },
440
+ {
441
+ "epoch": 0.89,
442
+ "learning_rate": 0.0001,
443
+ "loss": 1.1574,
444
+ "step": 490
445
+ },
446
+ {
447
+ "epoch": 0.91,
448
+ "learning_rate": 0.0001,
449
+ "loss": 1.2687,
450
+ "step": 500
451
+ },
452
+ {
453
+ "epoch": 0.92,
454
+ "learning_rate": 0.0001,
455
+ "loss": 0.8657,
456
+ "step": 510
457
+ },
458
+ {
459
+ "epoch": 0.94,
460
+ "learning_rate": 0.0001,
461
+ "loss": 0.9207,
462
+ "step": 520
463
+ },
464
+ {
465
+ "epoch": 0.96,
466
+ "learning_rate": 0.0001,
467
+ "loss": 1.012,
468
+ "step": 530
469
+ },
470
+ {
471
+ "epoch": 0.98,
472
+ "learning_rate": 0.0001,
473
+ "loss": 1.1517,
474
+ "step": 540
475
+ },
476
+ {
477
+ "epoch": 1.0,
478
+ "learning_rate": 0.0001,
479
+ "loss": 1.1654,
480
+ "step": 550
481
+ },
482
+ {
483
+ "epoch": 1.02,
484
+ "learning_rate": 0.0001,
485
+ "loss": 0.8931,
486
+ "step": 560
487
+ },
488
+ {
489
+ "epoch": 1.02,
490
+ "eval_loss": 1.0150845050811768,
491
+ "eval_runtime": 949.8392,
492
+ "eval_samples_per_second": 1.053,
493
+ "eval_steps_per_second": 1.053,
494
+ "step": 561
495
+ },
496
+ {
497
+ "epoch": 1.02,
498
+ "mmlu_eval_accuracy": 0.7346397374699287,
499
+ "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
500
+ "mmlu_eval_accuracy_anatomy": 0.7142857142857143,
501
+ "mmlu_eval_accuracy_astronomy": 0.6875,
502
+ "mmlu_eval_accuracy_business_ethics": 0.8181818181818182,
503
+ "mmlu_eval_accuracy_clinical_knowledge": 0.8620689655172413,
504
+ "mmlu_eval_accuracy_college_biology": 0.875,
505
+ "mmlu_eval_accuracy_college_chemistry": 0.5,
506
+ "mmlu_eval_accuracy_college_computer_science": 0.7272727272727273,
507
+ "mmlu_eval_accuracy_college_mathematics": 0.36363636363636365,
508
+ "mmlu_eval_accuracy_college_medicine": 0.9090909090909091,
509
+ "mmlu_eval_accuracy_college_physics": 0.5454545454545454,
510
+ "mmlu_eval_accuracy_computer_security": 0.6363636363636364,
511
+ "mmlu_eval_accuracy_conceptual_physics": 0.6153846153846154,
512
+ "mmlu_eval_accuracy_econometrics": 0.8333333333333334,
513
+ "mmlu_eval_accuracy_electrical_engineering": 0.8125,
514
+ "mmlu_eval_accuracy_elementary_mathematics": 0.6829268292682927,
515
+ "mmlu_eval_accuracy_formal_logic": 0.6428571428571429,
516
+ "mmlu_eval_accuracy_global_facts": 0.5,
517
+ "mmlu_eval_accuracy_high_school_biology": 0.84375,
518
+ "mmlu_eval_accuracy_high_school_chemistry": 0.45454545454545453,
519
+ "mmlu_eval_accuracy_high_school_computer_science": 0.7777777777777778,
520
+ "mmlu_eval_accuracy_high_school_european_history": 0.7777777777777778,
521
+ "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
522
+ "mmlu_eval_accuracy_high_school_government_and_politics": 0.9523809523809523,
523
+ "mmlu_eval_accuracy_high_school_macroeconomics": 0.7674418604651163,
524
+ "mmlu_eval_accuracy_high_school_mathematics": 0.41379310344827586,
525
+ "mmlu_eval_accuracy_high_school_microeconomics": 0.9230769230769231,
526
+ "mmlu_eval_accuracy_high_school_physics": 0.23529411764705882,
527
+ "mmlu_eval_accuracy_high_school_psychology": 0.9666666666666667,
528
+ "mmlu_eval_accuracy_high_school_statistics": 0.7391304347826086,
529
+ "mmlu_eval_accuracy_high_school_us_history": 0.9090909090909091,
530
+ "mmlu_eval_accuracy_high_school_world_history": 0.8076923076923077,
531
+ "mmlu_eval_accuracy_human_aging": 0.8260869565217391,
532
+ "mmlu_eval_accuracy_human_sexuality": 0.6666666666666666,
533
+ "mmlu_eval_accuracy_international_law": 1.0,
534
+ "mmlu_eval_accuracy_jurisprudence": 0.6363636363636364,
535
+ "mmlu_eval_accuracy_logical_fallacies": 0.7222222222222222,
536
+ "mmlu_eval_accuracy_machine_learning": 0.6363636363636364,
537
+ "mmlu_eval_accuracy_management": 0.9090909090909091,
538
+ "mmlu_eval_accuracy_marketing": 0.88,
539
+ "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
540
+ "mmlu_eval_accuracy_miscellaneous": 0.7558139534883721,
541
+ "mmlu_eval_accuracy_moral_disputes": 0.8157894736842105,
542
+ "mmlu_eval_accuracy_moral_scenarios": 0.56,
543
+ "mmlu_eval_accuracy_nutrition": 0.7575757575757576,
544
+ "mmlu_eval_accuracy_philosophy": 0.7941176470588235,
545
+ "mmlu_eval_accuracy_prehistory": 0.8571428571428571,
546
+ "mmlu_eval_accuracy_professional_accounting": 0.6774193548387096,
547
+ "mmlu_eval_accuracy_professional_law": 0.6294117647058823,
548
+ "mmlu_eval_accuracy_professional_medicine": 0.8064516129032258,
549
+ "mmlu_eval_accuracy_professional_psychology": 0.8260869565217391,
550
+ "mmlu_eval_accuracy_public_relations": 0.6666666666666666,
551
+ "mmlu_eval_accuracy_security_studies": 0.8148148148148148,
552
+ "mmlu_eval_accuracy_sociology": 0.9090909090909091,
553
+ "mmlu_eval_accuracy_us_foreign_policy": 1.0,
554
+ "mmlu_eval_accuracy_virology": 0.5,
555
+ "mmlu_eval_accuracy_world_religions": 0.8421052631578947,
556
+ "mmlu_loss": 1.186674291658526,
557
+ "step": 561
558
+ },
559
+ {
560
+ "epoch": 1.03,
561
+ "learning_rate": 0.0001,
562
+ "loss": 0.8507,
563
+ "step": 570
564
+ },
565
+ {
566
+ "epoch": 1.05,
567
+ "learning_rate": 0.0001,
568
+ "loss": 0.9164,
569
+ "step": 580
570
+ },
571
+ {
572
+ "epoch": 1.07,
573
+ "learning_rate": 0.0001,
574
+ "loss": 1.0908,
575
+ "step": 590
576
+ },
577
+ {
578
+ "epoch": 1.09,
579
+ "learning_rate": 0.0001,
580
+ "loss": 1.0431,
581
+ "step": 600
582
+ },
583
+ {
584
+ "epoch": 1.11,
585
+ "learning_rate": 0.0001,
586
+ "loss": 0.8567,
587
+ "step": 610
588
+ },
589
+ {
590
+ "epoch": 1.12,
591
+ "learning_rate": 0.0001,
592
+ "loss": 0.8818,
593
+ "step": 620
594
+ },
595
+ {
596
+ "epoch": 1.14,
597
+ "learning_rate": 0.0001,
598
+ "loss": 0.9499,
599
+ "step": 630
600
+ },
601
+ {
602
+ "epoch": 1.16,
603
+ "learning_rate": 0.0001,
604
+ "loss": 1.0437,
605
+ "step": 640
606
+ },
607
+ {
608
+ "epoch": 1.18,
609
+ "learning_rate": 0.0001,
610
+ "loss": 1.0487,
611
+ "step": 650
612
+ },
613
+ {
614
+ "epoch": 1.2,
615
+ "learning_rate": 0.0001,
616
+ "loss": 0.8405,
617
+ "step": 660
618
+ },
619
+ {
620
+ "epoch": 1.22,
621
+ "learning_rate": 0.0001,
622
+ "loss": 0.8818,
623
+ "step": 670
624
+ },
625
+ {
626
+ "epoch": 1.23,
627
+ "learning_rate": 0.0001,
628
+ "loss": 0.9619,
629
+ "step": 680
630
+ },
631
+ {
632
+ "epoch": 1.25,
633
+ "learning_rate": 0.0001,
634
+ "loss": 1.0753,
635
+ "step": 690
636
+ },
637
+ {
638
+ "epoch": 1.27,
639
+ "learning_rate": 0.0001,
640
+ "loss": 1.0218,
641
+ "step": 700
642
+ },
643
+ {
644
+ "epoch": 1.29,
645
+ "learning_rate": 0.0001,
646
+ "loss": 0.8763,
647
+ "step": 710
648
+ },
649
+ {
650
+ "epoch": 1.31,
651
+ "learning_rate": 0.0001,
652
+ "loss": 0.8789,
653
+ "step": 720
654
+ },
655
+ {
656
+ "epoch": 1.32,
657
+ "learning_rate": 0.0001,
658
+ "loss": 0.8631,
659
+ "step": 730
660
+ },
661
+ {
662
+ "epoch": 1.34,
663
+ "learning_rate": 0.0001,
664
+ "loss": 0.9846,
665
+ "step": 740
666
+ },
667
+ {
668
+ "epoch": 1.36,
669
+ "eval_loss": 1.0305067300796509,
670
+ "eval_runtime": 948.7106,
671
+ "eval_samples_per_second": 1.054,
672
+ "eval_steps_per_second": 1.054,
673
+ "step": 748
674
+ },
675
+ {
676
+ "epoch": 1.36,
677
+ "mmlu_eval_accuracy": 0.7324229372189777,
678
+ "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
679
+ "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
680
+ "mmlu_eval_accuracy_astronomy": 0.6875,
681
+ "mmlu_eval_accuracy_business_ethics": 1.0,
682
+ "mmlu_eval_accuracy_clinical_knowledge": 0.896551724137931,
683
+ "mmlu_eval_accuracy_college_biology": 0.875,
684
+ "mmlu_eval_accuracy_college_chemistry": 0.5,
685
+ "mmlu_eval_accuracy_college_computer_science": 0.6363636363636364,
686
+ "mmlu_eval_accuracy_college_mathematics": 0.36363636363636365,
687
+ "mmlu_eval_accuracy_college_medicine": 0.8636363636363636,
688
+ "mmlu_eval_accuracy_college_physics": 0.6363636363636364,
689
+ "mmlu_eval_accuracy_computer_security": 0.6363636363636364,
690
+ "mmlu_eval_accuracy_conceptual_physics": 0.6153846153846154,
691
+ "mmlu_eval_accuracy_econometrics": 0.8333333333333334,
692
+ "mmlu_eval_accuracy_electrical_engineering": 0.8125,
693
+ "mmlu_eval_accuracy_elementary_mathematics": 0.7073170731707317,
694
+ "mmlu_eval_accuracy_formal_logic": 0.6428571428571429,
695
+ "mmlu_eval_accuracy_global_facts": 0.4,
696
+ "mmlu_eval_accuracy_high_school_biology": 0.8125,
697
+ "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
698
+ "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
699
+ "mmlu_eval_accuracy_high_school_european_history": 0.7777777777777778,
700
+ "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
701
+ "mmlu_eval_accuracy_high_school_government_and_politics": 0.9523809523809523,
702
+ "mmlu_eval_accuracy_high_school_macroeconomics": 0.7674418604651163,
703
+ "mmlu_eval_accuracy_high_school_mathematics": 0.3448275862068966,
704
+ "mmlu_eval_accuracy_high_school_microeconomics": 0.9615384615384616,
705
+ "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
706
+ "mmlu_eval_accuracy_high_school_psychology": 0.9333333333333333,
707
+ "mmlu_eval_accuracy_high_school_statistics": 0.782608695652174,
708
+ "mmlu_eval_accuracy_high_school_us_history": 0.9545454545454546,
709
+ "mmlu_eval_accuracy_high_school_world_history": 0.7692307692307693,
710
+ "mmlu_eval_accuracy_human_aging": 0.782608695652174,
711
+ "mmlu_eval_accuracy_human_sexuality": 0.6666666666666666,
712
+ "mmlu_eval_accuracy_international_law": 1.0,
713
+ "mmlu_eval_accuracy_jurisprudence": 0.6363636363636364,
714
+ "mmlu_eval_accuracy_logical_fallacies": 0.7777777777777778,
715
+ "mmlu_eval_accuracy_machine_learning": 0.6363636363636364,
716
+ "mmlu_eval_accuracy_management": 0.9090909090909091,
717
+ "mmlu_eval_accuracy_marketing": 0.96,
718
+ "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
719
+ "mmlu_eval_accuracy_miscellaneous": 0.7441860465116279,
720
+ "mmlu_eval_accuracy_moral_disputes": 0.8157894736842105,
721
+ "mmlu_eval_accuracy_moral_scenarios": 0.57,
722
+ "mmlu_eval_accuracy_nutrition": 0.7272727272727273,
723
+ "mmlu_eval_accuracy_philosophy": 0.7941176470588235,
724
+ "mmlu_eval_accuracy_prehistory": 0.8285714285714286,
725
+ "mmlu_eval_accuracy_professional_accounting": 0.6129032258064516,
726
+ "mmlu_eval_accuracy_professional_law": 0.6411764705882353,
727
+ "mmlu_eval_accuracy_professional_medicine": 0.8064516129032258,
728
+ "mmlu_eval_accuracy_professional_psychology": 0.8260869565217391,
729
+ "mmlu_eval_accuracy_public_relations": 0.6666666666666666,
730
+ "mmlu_eval_accuracy_security_studies": 0.8148148148148148,
731
+ "mmlu_eval_accuracy_sociology": 0.9090909090909091,
732
+ "mmlu_eval_accuracy_us_foreign_policy": 1.0,
733
+ "mmlu_eval_accuracy_virology": 0.5,
734
+ "mmlu_eval_accuracy_world_religions": 0.8421052631578947,
735
+ "mmlu_loss": 1.2988067958029479,
736
+ "step": 748
737
+ },
738
+ {
739
+ "epoch": 1.36,
740
+ "learning_rate": 0.0001,
741
+ "loss": 1.0735,
742
+ "step": 750
743
+ },
744
+ {
745
+ "epoch": 1.38,
746
+ "learning_rate": 0.0001,
747
+ "loss": 0.9066,
748
+ "step": 760
749
+ },
750
+ {
751
+ "epoch": 1.4,
752
+ "learning_rate": 0.0001,
753
+ "loss": 0.8716,
754
+ "step": 770
755
+ },
756
+ {
757
+ "epoch": 1.41,
758
+ "learning_rate": 0.0001,
759
+ "loss": 0.9144,
760
+ "step": 780
761
+ },
762
+ {
763
+ "epoch": 1.43,
764
+ "learning_rate": 0.0001,
765
+ "loss": 1.0338,
766
+ "step": 790
767
+ },
768
+ {
769
+ "epoch": 1.45,
770
+ "learning_rate": 0.0001,
771
+ "loss": 1.0275,
772
+ "step": 800
773
+ },
774
+ {
775
+ "epoch": 1.47,
776
+ "learning_rate": 0.0001,
777
+ "loss": 0.8382,
778
+ "step": 810
779
+ },
780
+ {
781
+ "epoch": 1.49,
782
+ "learning_rate": 0.0001,
783
+ "loss": 0.8489,
784
+ "step": 820
785
+ },
786
+ {
787
+ "epoch": 1.51,
788
+ "learning_rate": 0.0001,
789
+ "loss": 0.8931,
790
+ "step": 830
791
+ },
792
+ {
793
+ "epoch": 1.52,
794
+ "learning_rate": 0.0001,
795
+ "loss": 1.0515,
796
+ "step": 840
797
+ },
798
+ {
799
+ "epoch": 1.54,
800
+ "learning_rate": 0.0001,
801
+ "loss": 1.0965,
802
+ "step": 850
803
+ },
804
+ {
805
+ "epoch": 1.56,
806
+ "learning_rate": 0.0001,
807
+ "loss": 0.8928,
808
+ "step": 860
809
+ },
810
+ {
811
+ "epoch": 1.58,
812
+ "learning_rate": 0.0001,
813
+ "loss": 0.8608,
814
+ "step": 870
815
+ },
816
+ {
817
+ "epoch": 1.6,
818
+ "learning_rate": 0.0001,
819
+ "loss": 0.8831,
820
+ "step": 880
821
+ },
822
+ {
823
+ "epoch": 1.61,
824
+ "learning_rate": 0.0001,
825
+ "loss": 1.0253,
826
+ "step": 890
827
+ },
828
+ {
829
+ "epoch": 1.63,
830
+ "learning_rate": 0.0001,
831
+ "loss": 0.9905,
832
+ "step": 900
833
+ },
834
+ {
835
+ "epoch": 1.65,
836
+ "learning_rate": 0.0001,
837
+ "loss": 0.8487,
838
+ "step": 910
839
+ },
840
+ {
841
+ "epoch": 1.67,
842
+ "learning_rate": 0.0001,
843
+ "loss": 0.8568,
844
+ "step": 920
845
+ },
846
+ {
847
+ "epoch": 1.69,
848
+ "learning_rate": 0.0001,
849
+ "loss": 0.9047,
850
+ "step": 930
851
+ },
852
+ {
853
+ "epoch": 1.7,
854
+ "eval_loss": 1.0250624418258667,
855
+ "eval_runtime": 946.4035,
856
+ "eval_samples_per_second": 1.057,
857
+ "eval_steps_per_second": 1.057,
858
+ "step": 935
859
+ },
860
+ {
861
+ "epoch": 1.7,
862
+ "mmlu_eval_accuracy": 0.7288948695878031,
863
+ "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
864
+ "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
865
+ "mmlu_eval_accuracy_astronomy": 0.6875,
866
+ "mmlu_eval_accuracy_business_ethics": 0.9090909090909091,
867
+ "mmlu_eval_accuracy_clinical_knowledge": 0.8620689655172413,
868
+ "mmlu_eval_accuracy_college_biology": 0.875,
869
+ "mmlu_eval_accuracy_college_chemistry": 0.5,
870
+ "mmlu_eval_accuracy_college_computer_science": 0.7272727272727273,
871
+ "mmlu_eval_accuracy_college_mathematics": 0.36363636363636365,
872
+ "mmlu_eval_accuracy_college_medicine": 0.9090909090909091,
873
+ "mmlu_eval_accuracy_college_physics": 0.6363636363636364,
874
+ "mmlu_eval_accuracy_computer_security": 0.6363636363636364,
875
+ "mmlu_eval_accuracy_conceptual_physics": 0.6153846153846154,
876
+ "mmlu_eval_accuracy_econometrics": 0.8333333333333334,
877
+ "mmlu_eval_accuracy_electrical_engineering": 0.8125,
878
+ "mmlu_eval_accuracy_elementary_mathematics": 0.6097560975609756,
879
+ "mmlu_eval_accuracy_formal_logic": 0.6428571428571429,
880
+ "mmlu_eval_accuracy_global_facts": 0.5,
881
+ "mmlu_eval_accuracy_high_school_biology": 0.84375,
882
+ "mmlu_eval_accuracy_high_school_chemistry": 0.45454545454545453,
883
+ "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
884
+ "mmlu_eval_accuracy_high_school_european_history": 0.7777777777777778,
885
+ "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
886
+ "mmlu_eval_accuracy_high_school_government_and_politics": 0.9523809523809523,
887
+ "mmlu_eval_accuracy_high_school_macroeconomics": 0.7906976744186046,
888
+ "mmlu_eval_accuracy_high_school_mathematics": 0.3793103448275862,
889
+ "mmlu_eval_accuracy_high_school_microeconomics": 0.9230769230769231,
890
+ "mmlu_eval_accuracy_high_school_physics": 0.23529411764705882,
891
+ "mmlu_eval_accuracy_high_school_psychology": 0.9333333333333333,
892
+ "mmlu_eval_accuracy_high_school_statistics": 0.7391304347826086,
893
+ "mmlu_eval_accuracy_high_school_us_history": 0.9090909090909091,
894
+ "mmlu_eval_accuracy_high_school_world_history": 0.7692307692307693,
895
+ "mmlu_eval_accuracy_human_aging": 0.782608695652174,
896
+ "mmlu_eval_accuracy_human_sexuality": 0.6666666666666666,
897
+ "mmlu_eval_accuracy_international_law": 1.0,
898
+ "mmlu_eval_accuracy_jurisprudence": 0.5454545454545454,
899
+ "mmlu_eval_accuracy_logical_fallacies": 0.7222222222222222,
900
+ "mmlu_eval_accuracy_machine_learning": 0.6363636363636364,
901
+ "mmlu_eval_accuracy_management": 0.9090909090909091,
902
+ "mmlu_eval_accuracy_marketing": 0.96,
903
+ "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
904
+ "mmlu_eval_accuracy_miscellaneous": 0.7790697674418605,
905
+ "mmlu_eval_accuracy_moral_disputes": 0.8157894736842105,
906
+ "mmlu_eval_accuracy_moral_scenarios": 0.57,
907
+ "mmlu_eval_accuracy_nutrition": 0.7575757575757576,
908
+ "mmlu_eval_accuracy_philosophy": 0.8235294117647058,
909
+ "mmlu_eval_accuracy_prehistory": 0.8857142857142857,
910
+ "mmlu_eval_accuracy_professional_accounting": 0.6129032258064516,
911
+ "mmlu_eval_accuracy_professional_law": 0.6235294117647059,
912
+ "mmlu_eval_accuracy_professional_medicine": 0.8387096774193549,
913
+ "mmlu_eval_accuracy_professional_psychology": 0.8115942028985508,
914
+ "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
915
+ "mmlu_eval_accuracy_security_studies": 0.8148148148148148,
916
+ "mmlu_eval_accuracy_sociology": 0.9090909090909091,
917
+ "mmlu_eval_accuracy_us_foreign_policy": 1.0,
918
+ "mmlu_eval_accuracy_virology": 0.5,
919
+ "mmlu_eval_accuracy_world_religions": 0.8421052631578947,
920
+ "mmlu_loss": 1.243813282909306,
921
+ "step": 935
922
+ },
923
+ {
924
+ "epoch": 1.7,
925
+ "learning_rate": 0.0001,
926
+ "loss": 1.0174,
927
+ "step": 940
928
+ },
929
+ {
930
+ "epoch": 1.72,
931
+ "learning_rate": 0.0001,
932
+ "loss": 1.0302,
933
+ "step": 950
934
+ },
935
+ {
936
+ "epoch": 1.74,
937
+ "learning_rate": 0.0001,
938
+ "loss": 0.8799,
939
+ "step": 960
940
+ },
941
+ {
942
+ "epoch": 1.76,
943
+ "learning_rate": 0.0001,
944
+ "loss": 0.8447,
945
+ "step": 970
946
+ },
947
+ {
948
+ "epoch": 1.78,
949
+ "learning_rate": 0.0001,
950
+ "loss": 0.9053,
951
+ "step": 980
952
+ },
953
+ {
954
+ "epoch": 1.8,
955
+ "learning_rate": 0.0001,
956
+ "loss": 1.0331,
957
+ "step": 990
958
+ },
959
+ {
960
+ "epoch": 1.81,
961
+ "learning_rate": 0.0001,
962
+ "loss": 1.0412,
963
+ "step": 1000
964
+ },
965
+ {
966
+ "epoch": 1.83,
967
+ "learning_rate": 0.0001,
968
+ "loss": 0.8753,
969
+ "step": 1010
970
+ },
971
+ {
972
+ "epoch": 1.85,
973
+ "learning_rate": 0.0001,
974
+ "loss": 0.8744,
975
+ "step": 1020
976
+ },
977
+ {
978
+ "epoch": 1.87,
979
+ "learning_rate": 0.0001,
980
+ "loss": 0.8899,
981
+ "step": 1030
982
+ },
983
+ {
984
+ "epoch": 1.89,
985
+ "learning_rate": 0.0001,
986
+ "loss": 1.0053,
987
+ "step": 1040
988
+ },
989
+ {
990
+ "epoch": 1.9,
991
+ "learning_rate": 0.0001,
992
+ "loss": 1.0127,
993
+ "step": 1050
994
+ },
995
+ {
996
+ "epoch": 1.92,
997
+ "learning_rate": 0.0001,
998
+ "loss": 0.8023,
999
+ "step": 1060
1000
+ },
1001
+ {
1002
+ "epoch": 1.94,
1003
+ "learning_rate": 0.0001,
1004
+ "loss": 0.8349,
1005
+ "step": 1070
1006
+ },
1007
+ {
1008
+ "epoch": 1.96,
1009
+ "learning_rate": 0.0001,
1010
+ "loss": 0.9742,
1011
+ "step": 1080
1012
+ },
1013
+ {
1014
+ "epoch": 1.98,
1015
+ "learning_rate": 0.0001,
1016
+ "loss": 1.0971,
1017
+ "step": 1090
1018
+ },
1019
+ {
1020
+ "epoch": 2.0,
1021
+ "learning_rate": 0.0001,
1022
+ "loss": 1.0728,
1023
+ "step": 1100
1024
+ },
1025
+ {
1026
+ "epoch": 2.01,
1027
+ "learning_rate": 0.0001,
1028
+ "loss": 0.7724,
1029
+ "step": 1110
1030
+ },
1031
+ {
1032
+ "epoch": 2.03,
1033
+ "learning_rate": 0.0001,
1034
+ "loss": 0.7675,
1035
+ "step": 1120
1036
+ },
1037
+ {
1038
+ "epoch": 2.03,
1039
+ "eval_loss": 1.052681565284729,
1040
+ "eval_runtime": 942.0722,
1041
+ "eval_samples_per_second": 1.061,
1042
+ "eval_steps_per_second": 1.061,
1043
+ "step": 1122
1044
+ },
1045
+ {
1046
+ "epoch": 2.03,
1047
+ "mmlu_eval_accuracy": 0.7373981967098951,
1048
+ "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
1049
+ "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
1050
+ "mmlu_eval_accuracy_astronomy": 0.6875,
1051
+ "mmlu_eval_accuracy_business_ethics": 0.9090909090909091,
1052
+ "mmlu_eval_accuracy_clinical_knowledge": 0.896551724137931,
1053
+ "mmlu_eval_accuracy_college_biology": 0.875,
1054
+ "mmlu_eval_accuracy_college_chemistry": 0.5,
1055
+ "mmlu_eval_accuracy_college_computer_science": 0.5454545454545454,
1056
+ "mmlu_eval_accuracy_college_mathematics": 0.36363636363636365,
1057
+ "mmlu_eval_accuracy_college_medicine": 0.9090909090909091,
1058
+ "mmlu_eval_accuracy_college_physics": 0.5454545454545454,
1059
+ "mmlu_eval_accuracy_computer_security": 0.6363636363636364,
1060
+ "mmlu_eval_accuracy_conceptual_physics": 0.6153846153846154,
1061
+ "mmlu_eval_accuracy_econometrics": 0.8333333333333334,
1062
+ "mmlu_eval_accuracy_electrical_engineering": 0.875,
1063
+ "mmlu_eval_accuracy_elementary_mathematics": 0.5853658536585366,
1064
+ "mmlu_eval_accuracy_formal_logic": 0.7142857142857143,
1065
+ "mmlu_eval_accuracy_global_facts": 0.4,
1066
+ "mmlu_eval_accuracy_high_school_biology": 0.84375,
1067
+ "mmlu_eval_accuracy_high_school_chemistry": 0.45454545454545453,
1068
+ "mmlu_eval_accuracy_high_school_computer_science": 0.7777777777777778,
1069
+ "mmlu_eval_accuracy_high_school_european_history": 0.7777777777777778,
1070
+ "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
1071
+ "mmlu_eval_accuracy_high_school_government_and_politics": 0.9523809523809523,
1072
+ "mmlu_eval_accuracy_high_school_macroeconomics": 0.7674418604651163,
1073
+ "mmlu_eval_accuracy_high_school_mathematics": 0.41379310344827586,
1074
+ "mmlu_eval_accuracy_high_school_microeconomics": 0.9230769230769231,
1075
+ "mmlu_eval_accuracy_high_school_physics": 0.23529411764705882,
1076
+ "mmlu_eval_accuracy_high_school_psychology": 0.95,
1077
+ "mmlu_eval_accuracy_high_school_statistics": 0.782608695652174,
1078
+ "mmlu_eval_accuracy_high_school_us_history": 0.9090909090909091,
1079
+ "mmlu_eval_accuracy_high_school_world_history": 0.7692307692307693,
1080
+ "mmlu_eval_accuracy_human_aging": 0.8260869565217391,
1081
+ "mmlu_eval_accuracy_human_sexuality": 0.6666666666666666,
1082
+ "mmlu_eval_accuracy_international_law": 1.0,
1083
+ "mmlu_eval_accuracy_jurisprudence": 0.6363636363636364,
1084
+ "mmlu_eval_accuracy_logical_fallacies": 0.7777777777777778,
1085
+ "mmlu_eval_accuracy_machine_learning": 0.6363636363636364,
1086
+ "mmlu_eval_accuracy_management": 0.9090909090909091,
1087
+ "mmlu_eval_accuracy_marketing": 0.92,
1088
+ "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
1089
+ "mmlu_eval_accuracy_miscellaneous": 0.7906976744186046,
1090
+ "mmlu_eval_accuracy_moral_disputes": 0.8157894736842105,
1091
+ "mmlu_eval_accuracy_moral_scenarios": 0.62,
1092
+ "mmlu_eval_accuracy_nutrition": 0.7575757575757576,
1093
+ "mmlu_eval_accuracy_philosophy": 0.8235294117647058,
1094
+ "mmlu_eval_accuracy_prehistory": 0.8857142857142857,
1095
+ "mmlu_eval_accuracy_professional_accounting": 0.6774193548387096,
1096
+ "mmlu_eval_accuracy_professional_law": 0.6294117647058823,
1097
+ "mmlu_eval_accuracy_professional_medicine": 0.8709677419354839,
1098
+ "mmlu_eval_accuracy_professional_psychology": 0.782608695652174,
1099
+ "mmlu_eval_accuracy_public_relations": 0.6666666666666666,
1100
+ "mmlu_eval_accuracy_security_studies": 0.8148148148148148,
1101
+ "mmlu_eval_accuracy_sociology": 0.9090909090909091,
1102
+ "mmlu_eval_accuracy_us_foreign_policy": 1.0,
1103
+ "mmlu_eval_accuracy_virology": 0.5,
1104
+ "mmlu_eval_accuracy_world_religions": 0.8421052631578947,
1105
+ "mmlu_loss": 1.2340081441760609,
1106
+ "step": 1122
1107
+ },
1108
+ {
1109
+ "epoch": 2.05,
1110
+ "learning_rate": 0.0001,
1111
+ "loss": 0.7194,
1112
+ "step": 1130
1113
+ },
1114
+ {
1115
+ "epoch": 2.07,
1116
+ "learning_rate": 0.0001,
1117
+ "loss": 0.8236,
1118
+ "step": 1140
1119
+ },
1120
+ {
1121
+ "epoch": 2.09,
1122
+ "learning_rate": 0.0001,
1123
+ "loss": 0.6652,
1124
+ "step": 1150
1125
+ },
1126
+ {
1127
+ "epoch": 2.1,
1128
+ "learning_rate": 0.0001,
1129
+ "loss": 0.7177,
1130
+ "step": 1160
1131
+ },
1132
+ {
1133
+ "epoch": 2.12,
1134
+ "learning_rate": 0.0001,
1135
+ "loss": 0.7788,
1136
+ "step": 1170
1137
+ },
1138
+ {
1139
+ "epoch": 2.14,
1140
+ "learning_rate": 0.0001,
1141
+ "loss": 0.8117,
1142
+ "step": 1180
1143
+ },
1144
+ {
1145
+ "epoch": 2.16,
1146
+ "learning_rate": 0.0001,
1147
+ "loss": 0.8145,
1148
+ "step": 1190
1149
+ },
1150
+ {
1151
+ "epoch": 2.18,
1152
+ "learning_rate": 0.0001,
1153
+ "loss": 0.6984,
1154
+ "step": 1200
1155
+ }
1156
+ ],
1157
+ "max_steps": 1875,
1158
+ "num_train_epochs": 4,
1159
+ "total_flos": 1.3906525682785812e+18,
1160
+ "trial_name": null,
1161
+ "trial_params": null
1162
+ }
checkpoint-1200/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3739b542a2914edbd2d7eb3b727d6fa2a7752c75b0d7a856f5e87dd807fa1ef9
3
+ size 6200
checkpoint-1400/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-1400/adapter_config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": "152334H/miqu-1-70b-sf",
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": 0.05,
12
+ "modules_to_save": null,
13
+ "peft_type": "LORA",
14
+ "r": 64,
15
+ "revision": null,
16
+ "target_modules": [
17
+ "gate_proj",
18
+ "down_proj",
19
+ "v_proj",
20
+ "up_proj",
21
+ "q_proj",
22
+ "k_proj",
23
+ "o_proj"
24
+ ],
25
+ "task_type": "CAUSAL_LM"
26
+ }
checkpoint-1400/adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d1533cea48affa188e9822cffe2f4176c550f93d32a8939add43058d977fbd29
3
+ size 1657155522
checkpoint-1400/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-1400/adapter_model/adapter_config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": "152334H/miqu-1-70b-sf",
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": 0.05,
12
+ "modules_to_save": null,
13
+ "peft_type": "LORA",
14
+ "r": 64,
15
+ "revision": null,
16
+ "target_modules": [
17
+ "gate_proj",
18
+ "down_proj",
19
+ "v_proj",
20
+ "up_proj",
21
+ "q_proj",
22
+ "k_proj",
23
+ "o_proj"
24
+ ],
25
+ "task_type": "CAUSAL_LM"
26
+ }
checkpoint-1400/adapter_model/adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d1533cea48affa188e9822cffe2f4176c550f93d32a8939add43058d977fbd29
3
+ size 1657155522
checkpoint-1400/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:55ab8de049e35c0670e2b97c2dcffcc99fcb8e1c34ca4e4de2f3d6c3717664bf
3
+ size 6627702922
checkpoint-1400/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6bb2503442a3c1d2dd808417344726e9ee1fe213f212edf8a440ebcb1863ef6f
3
+ size 14180
checkpoint-1400/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:90d628735beb51fe7df523288b184f458e4b2c3ef9d3bc30ab77bfb1a76e0ba0
3
+ size 1064
checkpoint-1400/special_tokens_map.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": "<s>",
3
+ "eos_token": "</s>",
4
+ "pad_token": {
5
+ "content": "<unk>",
6
+ "lstrip": false,
7
+ "normalized": true,
8
+ "rstrip": false,
9
+ "single_word": false
10
+ },
11
+ "unk_token": "<unk>"
12
+ }
checkpoint-1400/tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
3
+ size 499723
checkpoint-1400/tokenizer_config.json ADDED
@@ -0,0 +1,71 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": true,
3
+ "add_eos_token": false,
4
+ "added_tokens_decoder": {
5
+ "0": {
6
+ "content": "<unk>",
7
+ "lstrip": false,
8
+ "normalized": true,
9
+ "rstrip": false,
10
+ "single_word": false,
11
+ "special": true
12
+ },
13
+ "1": {
14
+ "content": "<s>",
15
+ "lstrip": false,
16
+ "normalized": true,
17
+ "rstrip": false,
18
+ "single_word": false,
19
+ "special": true
20
+ },
21
+ "2": {
22
+ "content": "</s>",
23
+ "lstrip": false,
24
+ "normalized": true,
25
+ "rstrip": false,
26
+ "single_word": false,
27
+ "special": true
28
+ }
29
+ },
30
+ "bos_token": {
31
+ "__type": "AddedToken",
32
+ "content": "<s>",
33
+ "lstrip": false,
34
+ "normalized": true,
35
+ "rstrip": false,
36
+ "single_word": false
37
+ },
38
+ "chat_template": "{{ bos_token }}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if message['role'] == 'user' %}{{ '[INST] ' + message['content'] + ' [/INST]' }}{% elif message['role'] == 'assistant' %}{{ message['content'] + eos_token}}{% else %}{{ raise_exception('Only user and assistant roles are supported!') }}{% endif %}{% endfor %}",
39
+ "clean_up_tokenization_spaces": false,
40
+ "eos_token": {
41
+ "__type": "AddedToken",
42
+ "content": "</s>",
43
+ "lstrip": false,
44
+ "normalized": true,
45
+ "rstrip": false,
46
+ "single_word": false
47
+ },
48
+ "legacy": false,
49
+ "model_max_length": 1000000000000000019884624838656,
50
+ "pad_token": {
51
+ "__type": "AddedToken",
52
+ "content": "<unk>",
53
+ "lstrip": false,
54
+ "normalized": true,
55
+ "rstrip": false,
56
+ "single_word": false
57
+ },
58
+ "padding_side": "right",
59
+ "sp_model_kwargs": {},
60
+ "spaces_between_special_tokens": false,
61
+ "tokenizer_class": "LlamaTokenizer",
62
+ "unk_token": {
63
+ "__type": "AddedToken",
64
+ "content": "<unk>",
65
+ "lstrip": false,
66
+ "normalized": true,
67
+ "rstrip": false,
68
+ "single_word": false
69
+ },
70
+ "use_default_system_prompt": false
71
+ }
checkpoint-1400/trainer_state.json ADDED
@@ -0,0 +1,1353 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 2.539106778508275,
5
+ "global_step": 1400,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.02,
12
+ "learning_rate": 0.0001,
13
+ "loss": 1.1655,
14
+ "step": 10
15
+ },
16
+ {
17
+ "epoch": 0.04,
18
+ "learning_rate": 0.0001,
19
+ "loss": 1.001,
20
+ "step": 20
21
+ },
22
+ {
23
+ "epoch": 0.05,
24
+ "learning_rate": 0.0001,
25
+ "loss": 1.0287,
26
+ "step": 30
27
+ },
28
+ {
29
+ "epoch": 0.07,
30
+ "learning_rate": 0.0001,
31
+ "loss": 1.1578,
32
+ "step": 40
33
+ },
34
+ {
35
+ "epoch": 0.09,
36
+ "learning_rate": 0.0001,
37
+ "loss": 1.2146,
38
+ "step": 50
39
+ },
40
+ {
41
+ "epoch": 0.11,
42
+ "learning_rate": 0.0001,
43
+ "loss": 0.997,
44
+ "step": 60
45
+ },
46
+ {
47
+ "epoch": 0.13,
48
+ "learning_rate": 0.0001,
49
+ "loss": 0.9024,
50
+ "step": 70
51
+ },
52
+ {
53
+ "epoch": 0.15,
54
+ "learning_rate": 0.0001,
55
+ "loss": 0.9901,
56
+ "step": 80
57
+ },
58
+ {
59
+ "epoch": 0.16,
60
+ "learning_rate": 0.0001,
61
+ "loss": 1.1264,
62
+ "step": 90
63
+ },
64
+ {
65
+ "epoch": 0.18,
66
+ "learning_rate": 0.0001,
67
+ "loss": 1.2038,
68
+ "step": 100
69
+ },
70
+ {
71
+ "epoch": 0.2,
72
+ "learning_rate": 0.0001,
73
+ "loss": 0.8935,
74
+ "step": 110
75
+ },
76
+ {
77
+ "epoch": 0.22,
78
+ "learning_rate": 0.0001,
79
+ "loss": 0.9178,
80
+ "step": 120
81
+ },
82
+ {
83
+ "epoch": 0.24,
84
+ "learning_rate": 0.0001,
85
+ "loss": 0.9746,
86
+ "step": 130
87
+ },
88
+ {
89
+ "epoch": 0.25,
90
+ "learning_rate": 0.0001,
91
+ "loss": 1.1566,
92
+ "step": 140
93
+ },
94
+ {
95
+ "epoch": 0.27,
96
+ "learning_rate": 0.0001,
97
+ "loss": 1.2877,
98
+ "step": 150
99
+ },
100
+ {
101
+ "epoch": 0.29,
102
+ "learning_rate": 0.0001,
103
+ "loss": 0.9146,
104
+ "step": 160
105
+ },
106
+ {
107
+ "epoch": 0.31,
108
+ "learning_rate": 0.0001,
109
+ "loss": 0.8895,
110
+ "step": 170
111
+ },
112
+ {
113
+ "epoch": 0.33,
114
+ "learning_rate": 0.0001,
115
+ "loss": 1.0121,
116
+ "step": 180
117
+ },
118
+ {
119
+ "epoch": 0.34,
120
+ "eval_loss": 1.0215636491775513,
121
+ "eval_runtime": 950.138,
122
+ "eval_samples_per_second": 1.052,
123
+ "eval_steps_per_second": 1.052,
124
+ "step": 187
125
+ },
126
+ {
127
+ "epoch": 0.34,
128
+ "mmlu_eval_accuracy": 0.731892294851104,
129
+ "mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
130
+ "mmlu_eval_accuracy_anatomy": 0.7857142857142857,
131
+ "mmlu_eval_accuracy_astronomy": 0.6875,
132
+ "mmlu_eval_accuracy_business_ethics": 0.7272727272727273,
133
+ "mmlu_eval_accuracy_clinical_knowledge": 0.896551724137931,
134
+ "mmlu_eval_accuracy_college_biology": 0.875,
135
+ "mmlu_eval_accuracy_college_chemistry": 0.5,
136
+ "mmlu_eval_accuracy_college_computer_science": 0.7272727272727273,
137
+ "mmlu_eval_accuracy_college_mathematics": 0.45454545454545453,
138
+ "mmlu_eval_accuracy_college_medicine": 0.9090909090909091,
139
+ "mmlu_eval_accuracy_college_physics": 0.6363636363636364,
140
+ "mmlu_eval_accuracy_computer_security": 0.6363636363636364,
141
+ "mmlu_eval_accuracy_conceptual_physics": 0.6538461538461539,
142
+ "mmlu_eval_accuracy_econometrics": 0.8333333333333334,
143
+ "mmlu_eval_accuracy_electrical_engineering": 0.8125,
144
+ "mmlu_eval_accuracy_elementary_mathematics": 0.7073170731707317,
145
+ "mmlu_eval_accuracy_formal_logic": 0.5714285714285714,
146
+ "mmlu_eval_accuracy_global_facts": 0.5,
147
+ "mmlu_eval_accuracy_high_school_biology": 0.8125,
148
+ "mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
149
+ "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
150
+ "mmlu_eval_accuracy_high_school_european_history": 0.7777777777777778,
151
+ "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
152
+ "mmlu_eval_accuracy_high_school_government_and_politics": 0.9523809523809523,
153
+ "mmlu_eval_accuracy_high_school_macroeconomics": 0.7441860465116279,
154
+ "mmlu_eval_accuracy_high_school_mathematics": 0.3793103448275862,
155
+ "mmlu_eval_accuracy_high_school_microeconomics": 0.9615384615384616,
156
+ "mmlu_eval_accuracy_high_school_physics": 0.23529411764705882,
157
+ "mmlu_eval_accuracy_high_school_psychology": 0.9333333333333333,
158
+ "mmlu_eval_accuracy_high_school_statistics": 0.6956521739130435,
159
+ "mmlu_eval_accuracy_high_school_us_history": 0.9090909090909091,
160
+ "mmlu_eval_accuracy_high_school_world_history": 0.8076923076923077,
161
+ "mmlu_eval_accuracy_human_aging": 0.8260869565217391,
162
+ "mmlu_eval_accuracy_human_sexuality": 0.6666666666666666,
163
+ "mmlu_eval_accuracy_international_law": 1.0,
164
+ "mmlu_eval_accuracy_jurisprudence": 0.6363636363636364,
165
+ "mmlu_eval_accuracy_logical_fallacies": 0.7777777777777778,
166
+ "mmlu_eval_accuracy_machine_learning": 0.5454545454545454,
167
+ "mmlu_eval_accuracy_management": 0.9090909090909091,
168
+ "mmlu_eval_accuracy_marketing": 0.88,
169
+ "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
170
+ "mmlu_eval_accuracy_miscellaneous": 0.7906976744186046,
171
+ "mmlu_eval_accuracy_moral_disputes": 0.8157894736842105,
172
+ "mmlu_eval_accuracy_moral_scenarios": 0.59,
173
+ "mmlu_eval_accuracy_nutrition": 0.7878787878787878,
174
+ "mmlu_eval_accuracy_philosophy": 0.7941176470588235,
175
+ "mmlu_eval_accuracy_prehistory": 0.8285714285714286,
176
+ "mmlu_eval_accuracy_professional_accounting": 0.6451612903225806,
177
+ "mmlu_eval_accuracy_professional_law": 0.6294117647058823,
178
+ "mmlu_eval_accuracy_professional_medicine": 0.8064516129032258,
179
+ "mmlu_eval_accuracy_professional_psychology": 0.8115942028985508,
180
+ "mmlu_eval_accuracy_public_relations": 0.6666666666666666,
181
+ "mmlu_eval_accuracy_security_studies": 0.8148148148148148,
182
+ "mmlu_eval_accuracy_sociology": 0.9545454545454546,
183
+ "mmlu_eval_accuracy_us_foreign_policy": 1.0,
184
+ "mmlu_eval_accuracy_virology": 0.5,
185
+ "mmlu_eval_accuracy_world_religions": 0.8421052631578947,
186
+ "mmlu_loss": 1.326305795171384,
187
+ "step": 187
188
+ },
189
+ {
190
+ "epoch": 0.34,
191
+ "learning_rate": 0.0001,
192
+ "loss": 1.1133,
193
+ "step": 190
194
+ },
195
+ {
196
+ "epoch": 0.36,
197
+ "learning_rate": 0.0001,
198
+ "loss": 1.2485,
199
+ "step": 200
200
+ },
201
+ {
202
+ "epoch": 0.38,
203
+ "learning_rate": 0.0001,
204
+ "loss": 0.9653,
205
+ "step": 210
206
+ },
207
+ {
208
+ "epoch": 0.4,
209
+ "learning_rate": 0.0001,
210
+ "loss": 0.9455,
211
+ "step": 220
212
+ },
213
+ {
214
+ "epoch": 0.42,
215
+ "learning_rate": 0.0001,
216
+ "loss": 1.0373,
217
+ "step": 230
218
+ },
219
+ {
220
+ "epoch": 0.44,
221
+ "learning_rate": 0.0001,
222
+ "loss": 1.1425,
223
+ "step": 240
224
+ },
225
+ {
226
+ "epoch": 0.45,
227
+ "learning_rate": 0.0001,
228
+ "loss": 1.3136,
229
+ "step": 250
230
+ },
231
+ {
232
+ "epoch": 0.47,
233
+ "learning_rate": 0.0001,
234
+ "loss": 0.8695,
235
+ "step": 260
236
+ },
237
+ {
238
+ "epoch": 0.49,
239
+ "learning_rate": 0.0001,
240
+ "loss": 0.872,
241
+ "step": 270
242
+ },
243
+ {
244
+ "epoch": 0.51,
245
+ "learning_rate": 0.0001,
246
+ "loss": 1.0152,
247
+ "step": 280
248
+ },
249
+ {
250
+ "epoch": 0.53,
251
+ "learning_rate": 0.0001,
252
+ "loss": 1.1309,
253
+ "step": 290
254
+ },
255
+ {
256
+ "epoch": 0.54,
257
+ "learning_rate": 0.0001,
258
+ "loss": 1.267,
259
+ "step": 300
260
+ },
261
+ {
262
+ "epoch": 0.56,
263
+ "learning_rate": 0.0001,
264
+ "loss": 0.9249,
265
+ "step": 310
266
+ },
267
+ {
268
+ "epoch": 0.58,
269
+ "learning_rate": 0.0001,
270
+ "loss": 0.9148,
271
+ "step": 320
272
+ },
273
+ {
274
+ "epoch": 0.6,
275
+ "learning_rate": 0.0001,
276
+ "loss": 0.9864,
277
+ "step": 330
278
+ },
279
+ {
280
+ "epoch": 0.62,
281
+ "learning_rate": 0.0001,
282
+ "loss": 1.2312,
283
+ "step": 340
284
+ },
285
+ {
286
+ "epoch": 0.63,
287
+ "learning_rate": 0.0001,
288
+ "loss": 1.2354,
289
+ "step": 350
290
+ },
291
+ {
292
+ "epoch": 0.65,
293
+ "learning_rate": 0.0001,
294
+ "loss": 0.9126,
295
+ "step": 360
296
+ },
297
+ {
298
+ "epoch": 0.67,
299
+ "learning_rate": 0.0001,
300
+ "loss": 0.9213,
301
+ "step": 370
302
+ },
303
+ {
304
+ "epoch": 0.68,
305
+ "eval_loss": 1.0163359642028809,
306
+ "eval_runtime": 948.1151,
307
+ "eval_samples_per_second": 1.055,
308
+ "eval_steps_per_second": 1.055,
309
+ "step": 374
310
+ },
311
+ {
312
+ "epoch": 0.68,
313
+ "mmlu_eval_accuracy": 0.7395476061435284,
314
+ "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
315
+ "mmlu_eval_accuracy_anatomy": 0.7857142857142857,
316
+ "mmlu_eval_accuracy_astronomy": 0.75,
317
+ "mmlu_eval_accuracy_business_ethics": 0.7272727272727273,
318
+ "mmlu_eval_accuracy_clinical_knowledge": 0.896551724137931,
319
+ "mmlu_eval_accuracy_college_biology": 0.875,
320
+ "mmlu_eval_accuracy_college_chemistry": 0.5,
321
+ "mmlu_eval_accuracy_college_computer_science": 0.7272727272727273,
322
+ "mmlu_eval_accuracy_college_mathematics": 0.36363636363636365,
323
+ "mmlu_eval_accuracy_college_medicine": 0.9090909090909091,
324
+ "mmlu_eval_accuracy_college_physics": 0.6363636363636364,
325
+ "mmlu_eval_accuracy_computer_security": 0.6363636363636364,
326
+ "mmlu_eval_accuracy_conceptual_physics": 0.5769230769230769,
327
+ "mmlu_eval_accuracy_econometrics": 0.8333333333333334,
328
+ "mmlu_eval_accuracy_electrical_engineering": 0.875,
329
+ "mmlu_eval_accuracy_elementary_mathematics": 0.6829268292682927,
330
+ "mmlu_eval_accuracy_formal_logic": 0.6428571428571429,
331
+ "mmlu_eval_accuracy_global_facts": 0.5,
332
+ "mmlu_eval_accuracy_high_school_biology": 0.8125,
333
+ "mmlu_eval_accuracy_high_school_chemistry": 0.45454545454545453,
334
+ "mmlu_eval_accuracy_high_school_computer_science": 0.7777777777777778,
335
+ "mmlu_eval_accuracy_high_school_european_history": 0.7777777777777778,
336
+ "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
337
+ "mmlu_eval_accuracy_high_school_government_and_politics": 0.9523809523809523,
338
+ "mmlu_eval_accuracy_high_school_macroeconomics": 0.7674418604651163,
339
+ "mmlu_eval_accuracy_high_school_mathematics": 0.41379310344827586,
340
+ "mmlu_eval_accuracy_high_school_microeconomics": 0.9615384615384616,
341
+ "mmlu_eval_accuracy_high_school_physics": 0.23529411764705882,
342
+ "mmlu_eval_accuracy_high_school_psychology": 0.95,
343
+ "mmlu_eval_accuracy_high_school_statistics": 0.7391304347826086,
344
+ "mmlu_eval_accuracy_high_school_us_history": 0.9090909090909091,
345
+ "mmlu_eval_accuracy_high_school_world_history": 0.7692307692307693,
346
+ "mmlu_eval_accuracy_human_aging": 0.8260869565217391,
347
+ "mmlu_eval_accuracy_human_sexuality": 0.6666666666666666,
348
+ "mmlu_eval_accuracy_international_law": 1.0,
349
+ "mmlu_eval_accuracy_jurisprudence": 0.6363636363636364,
350
+ "mmlu_eval_accuracy_logical_fallacies": 0.7777777777777778,
351
+ "mmlu_eval_accuracy_machine_learning": 0.5454545454545454,
352
+ "mmlu_eval_accuracy_management": 0.9090909090909091,
353
+ "mmlu_eval_accuracy_marketing": 0.92,
354
+ "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
355
+ "mmlu_eval_accuracy_miscellaneous": 0.7790697674418605,
356
+ "mmlu_eval_accuracy_moral_disputes": 0.8157894736842105,
357
+ "mmlu_eval_accuracy_moral_scenarios": 0.57,
358
+ "mmlu_eval_accuracy_nutrition": 0.7272727272727273,
359
+ "mmlu_eval_accuracy_philosophy": 0.7941176470588235,
360
+ "mmlu_eval_accuracy_prehistory": 0.8571428571428571,
361
+ "mmlu_eval_accuracy_professional_accounting": 0.6774193548387096,
362
+ "mmlu_eval_accuracy_professional_law": 0.6411764705882353,
363
+ "mmlu_eval_accuracy_professional_medicine": 0.8387096774193549,
364
+ "mmlu_eval_accuracy_professional_psychology": 0.8115942028985508,
365
+ "mmlu_eval_accuracy_public_relations": 0.6666666666666666,
366
+ "mmlu_eval_accuracy_security_studies": 0.8148148148148148,
367
+ "mmlu_eval_accuracy_sociology": 0.9545454545454546,
368
+ "mmlu_eval_accuracy_us_foreign_policy": 1.0,
369
+ "mmlu_eval_accuracy_virology": 0.5,
370
+ "mmlu_eval_accuracy_world_religions": 0.8947368421052632,
371
+ "mmlu_loss": 1.2796503596061355,
372
+ "step": 374
373
+ },
374
+ {
375
+ "epoch": 0.69,
376
+ "learning_rate": 0.0001,
377
+ "loss": 0.9737,
378
+ "step": 380
379
+ },
380
+ {
381
+ "epoch": 0.71,
382
+ "learning_rate": 0.0001,
383
+ "loss": 1.157,
384
+ "step": 390
385
+ },
386
+ {
387
+ "epoch": 0.73,
388
+ "learning_rate": 0.0001,
389
+ "loss": 1.2106,
390
+ "step": 400
391
+ },
392
+ {
393
+ "epoch": 0.74,
394
+ "learning_rate": 0.0001,
395
+ "loss": 0.8687,
396
+ "step": 410
397
+ },
398
+ {
399
+ "epoch": 0.76,
400
+ "learning_rate": 0.0001,
401
+ "loss": 0.8742,
402
+ "step": 420
403
+ },
404
+ {
405
+ "epoch": 0.78,
406
+ "learning_rate": 0.0001,
407
+ "loss": 0.9901,
408
+ "step": 430
409
+ },
410
+ {
411
+ "epoch": 0.8,
412
+ "learning_rate": 0.0001,
413
+ "loss": 1.2238,
414
+ "step": 440
415
+ },
416
+ {
417
+ "epoch": 0.82,
418
+ "learning_rate": 0.0001,
419
+ "loss": 1.2604,
420
+ "step": 450
421
+ },
422
+ {
423
+ "epoch": 0.83,
424
+ "learning_rate": 0.0001,
425
+ "loss": 0.8756,
426
+ "step": 460
427
+ },
428
+ {
429
+ "epoch": 0.85,
430
+ "learning_rate": 0.0001,
431
+ "loss": 0.8683,
432
+ "step": 470
433
+ },
434
+ {
435
+ "epoch": 0.87,
436
+ "learning_rate": 0.0001,
437
+ "loss": 0.9824,
438
+ "step": 480
439
+ },
440
+ {
441
+ "epoch": 0.89,
442
+ "learning_rate": 0.0001,
443
+ "loss": 1.1574,
444
+ "step": 490
445
+ },
446
+ {
447
+ "epoch": 0.91,
448
+ "learning_rate": 0.0001,
449
+ "loss": 1.2687,
450
+ "step": 500
451
+ },
452
+ {
453
+ "epoch": 0.92,
454
+ "learning_rate": 0.0001,
455
+ "loss": 0.8657,
456
+ "step": 510
457
+ },
458
+ {
459
+ "epoch": 0.94,
460
+ "learning_rate": 0.0001,
461
+ "loss": 0.9207,
462
+ "step": 520
463
+ },
464
+ {
465
+ "epoch": 0.96,
466
+ "learning_rate": 0.0001,
467
+ "loss": 1.012,
468
+ "step": 530
469
+ },
470
+ {
471
+ "epoch": 0.98,
472
+ "learning_rate": 0.0001,
473
+ "loss": 1.1517,
474
+ "step": 540
475
+ },
476
+ {
477
+ "epoch": 1.0,
478
+ "learning_rate": 0.0001,
479
+ "loss": 1.1654,
480
+ "step": 550
481
+ },
482
+ {
483
+ "epoch": 1.02,
484
+ "learning_rate": 0.0001,
485
+ "loss": 0.8931,
486
+ "step": 560
487
+ },
488
+ {
489
+ "epoch": 1.02,
490
+ "eval_loss": 1.0150845050811768,
491
+ "eval_runtime": 949.8392,
492
+ "eval_samples_per_second": 1.053,
493
+ "eval_steps_per_second": 1.053,
494
+ "step": 561
495
+ },
496
+ {
497
+ "epoch": 1.02,
498
+ "mmlu_eval_accuracy": 0.7346397374699287,
499
+ "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
500
+ "mmlu_eval_accuracy_anatomy": 0.7142857142857143,
501
+ "mmlu_eval_accuracy_astronomy": 0.6875,
502
+ "mmlu_eval_accuracy_business_ethics": 0.8181818181818182,
503
+ "mmlu_eval_accuracy_clinical_knowledge": 0.8620689655172413,
504
+ "mmlu_eval_accuracy_college_biology": 0.875,
505
+ "mmlu_eval_accuracy_college_chemistry": 0.5,
506
+ "mmlu_eval_accuracy_college_computer_science": 0.7272727272727273,
507
+ "mmlu_eval_accuracy_college_mathematics": 0.36363636363636365,
508
+ "mmlu_eval_accuracy_college_medicine": 0.9090909090909091,
509
+ "mmlu_eval_accuracy_college_physics": 0.5454545454545454,
510
+ "mmlu_eval_accuracy_computer_security": 0.6363636363636364,
511
+ "mmlu_eval_accuracy_conceptual_physics": 0.6153846153846154,
512
+ "mmlu_eval_accuracy_econometrics": 0.8333333333333334,
513
+ "mmlu_eval_accuracy_electrical_engineering": 0.8125,
514
+ "mmlu_eval_accuracy_elementary_mathematics": 0.6829268292682927,
515
+ "mmlu_eval_accuracy_formal_logic": 0.6428571428571429,
516
+ "mmlu_eval_accuracy_global_facts": 0.5,
517
+ "mmlu_eval_accuracy_high_school_biology": 0.84375,
518
+ "mmlu_eval_accuracy_high_school_chemistry": 0.45454545454545453,
519
+ "mmlu_eval_accuracy_high_school_computer_science": 0.7777777777777778,
520
+ "mmlu_eval_accuracy_high_school_european_history": 0.7777777777777778,
521
+ "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
522
+ "mmlu_eval_accuracy_high_school_government_and_politics": 0.9523809523809523,
523
+ "mmlu_eval_accuracy_high_school_macroeconomics": 0.7674418604651163,
524
+ "mmlu_eval_accuracy_high_school_mathematics": 0.41379310344827586,
525
+ "mmlu_eval_accuracy_high_school_microeconomics": 0.9230769230769231,
526
+ "mmlu_eval_accuracy_high_school_physics": 0.23529411764705882,
527
+ "mmlu_eval_accuracy_high_school_psychology": 0.9666666666666667,
528
+ "mmlu_eval_accuracy_high_school_statistics": 0.7391304347826086,
529
+ "mmlu_eval_accuracy_high_school_us_history": 0.9090909090909091,
530
+ "mmlu_eval_accuracy_high_school_world_history": 0.8076923076923077,
531
+ "mmlu_eval_accuracy_human_aging": 0.8260869565217391,
532
+ "mmlu_eval_accuracy_human_sexuality": 0.6666666666666666,
533
+ "mmlu_eval_accuracy_international_law": 1.0,
534
+ "mmlu_eval_accuracy_jurisprudence": 0.6363636363636364,
535
+ "mmlu_eval_accuracy_logical_fallacies": 0.7222222222222222,
536
+ "mmlu_eval_accuracy_machine_learning": 0.6363636363636364,
537
+ "mmlu_eval_accuracy_management": 0.9090909090909091,
538
+ "mmlu_eval_accuracy_marketing": 0.88,
539
+ "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
540
+ "mmlu_eval_accuracy_miscellaneous": 0.7558139534883721,
541
+ "mmlu_eval_accuracy_moral_disputes": 0.8157894736842105,
542
+ "mmlu_eval_accuracy_moral_scenarios": 0.56,
543
+ "mmlu_eval_accuracy_nutrition": 0.7575757575757576,
544
+ "mmlu_eval_accuracy_philosophy": 0.7941176470588235,
545
+ "mmlu_eval_accuracy_prehistory": 0.8571428571428571,
546
+ "mmlu_eval_accuracy_professional_accounting": 0.6774193548387096,
547
+ "mmlu_eval_accuracy_professional_law": 0.6294117647058823,
548
+ "mmlu_eval_accuracy_professional_medicine": 0.8064516129032258,
549
+ "mmlu_eval_accuracy_professional_psychology": 0.8260869565217391,
550
+ "mmlu_eval_accuracy_public_relations": 0.6666666666666666,
551
+ "mmlu_eval_accuracy_security_studies": 0.8148148148148148,
552
+ "mmlu_eval_accuracy_sociology": 0.9090909090909091,
553
+ "mmlu_eval_accuracy_us_foreign_policy": 1.0,
554
+ "mmlu_eval_accuracy_virology": 0.5,
555
+ "mmlu_eval_accuracy_world_religions": 0.8421052631578947,
556
+ "mmlu_loss": 1.186674291658526,
557
+ "step": 561
558
+ },
559
+ {
560
+ "epoch": 1.03,
561
+ "learning_rate": 0.0001,
562
+ "loss": 0.8507,
563
+ "step": 570
564
+ },
565
+ {
566
+ "epoch": 1.05,
567
+ "learning_rate": 0.0001,
568
+ "loss": 0.9164,
569
+ "step": 580
570
+ },
571
+ {
572
+ "epoch": 1.07,
573
+ "learning_rate": 0.0001,
574
+ "loss": 1.0908,
575
+ "step": 590
576
+ },
577
+ {
578
+ "epoch": 1.09,
579
+ "learning_rate": 0.0001,
580
+ "loss": 1.0431,
581
+ "step": 600
582
+ },
583
+ {
584
+ "epoch": 1.11,
585
+ "learning_rate": 0.0001,
586
+ "loss": 0.8567,
587
+ "step": 610
588
+ },
589
+ {
590
+ "epoch": 1.12,
591
+ "learning_rate": 0.0001,
592
+ "loss": 0.8818,
593
+ "step": 620
594
+ },
595
+ {
596
+ "epoch": 1.14,
597
+ "learning_rate": 0.0001,
598
+ "loss": 0.9499,
599
+ "step": 630
600
+ },
601
+ {
602
+ "epoch": 1.16,
603
+ "learning_rate": 0.0001,
604
+ "loss": 1.0437,
605
+ "step": 640
606
+ },
607
+ {
608
+ "epoch": 1.18,
609
+ "learning_rate": 0.0001,
610
+ "loss": 1.0487,
611
+ "step": 650
612
+ },
613
+ {
614
+ "epoch": 1.2,
615
+ "learning_rate": 0.0001,
616
+ "loss": 0.8405,
617
+ "step": 660
618
+ },
619
+ {
620
+ "epoch": 1.22,
621
+ "learning_rate": 0.0001,
622
+ "loss": 0.8818,
623
+ "step": 670
624
+ },
625
+ {
626
+ "epoch": 1.23,
627
+ "learning_rate": 0.0001,
628
+ "loss": 0.9619,
629
+ "step": 680
630
+ },
631
+ {
632
+ "epoch": 1.25,
633
+ "learning_rate": 0.0001,
634
+ "loss": 1.0753,
635
+ "step": 690
636
+ },
637
+ {
638
+ "epoch": 1.27,
639
+ "learning_rate": 0.0001,
640
+ "loss": 1.0218,
641
+ "step": 700
642
+ },
643
+ {
644
+ "epoch": 1.29,
645
+ "learning_rate": 0.0001,
646
+ "loss": 0.8763,
647
+ "step": 710
648
+ },
649
+ {
650
+ "epoch": 1.31,
651
+ "learning_rate": 0.0001,
652
+ "loss": 0.8789,
653
+ "step": 720
654
+ },
655
+ {
656
+ "epoch": 1.32,
657
+ "learning_rate": 0.0001,
658
+ "loss": 0.8631,
659
+ "step": 730
660
+ },
661
+ {
662
+ "epoch": 1.34,
663
+ "learning_rate": 0.0001,
664
+ "loss": 0.9846,
665
+ "step": 740
666
+ },
667
+ {
668
+ "epoch": 1.36,
669
+ "eval_loss": 1.0305067300796509,
670
+ "eval_runtime": 948.7106,
671
+ "eval_samples_per_second": 1.054,
672
+ "eval_steps_per_second": 1.054,
673
+ "step": 748
674
+ },
675
+ {
676
+ "epoch": 1.36,
677
+ "mmlu_eval_accuracy": 0.7324229372189777,
678
+ "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
679
+ "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
680
+ "mmlu_eval_accuracy_astronomy": 0.6875,
681
+ "mmlu_eval_accuracy_business_ethics": 1.0,
682
+ "mmlu_eval_accuracy_clinical_knowledge": 0.896551724137931,
683
+ "mmlu_eval_accuracy_college_biology": 0.875,
684
+ "mmlu_eval_accuracy_college_chemistry": 0.5,
685
+ "mmlu_eval_accuracy_college_computer_science": 0.6363636363636364,
686
+ "mmlu_eval_accuracy_college_mathematics": 0.36363636363636365,
687
+ "mmlu_eval_accuracy_college_medicine": 0.8636363636363636,
688
+ "mmlu_eval_accuracy_college_physics": 0.6363636363636364,
689
+ "mmlu_eval_accuracy_computer_security": 0.6363636363636364,
690
+ "mmlu_eval_accuracy_conceptual_physics": 0.6153846153846154,
691
+ "mmlu_eval_accuracy_econometrics": 0.8333333333333334,
692
+ "mmlu_eval_accuracy_electrical_engineering": 0.8125,
693
+ "mmlu_eval_accuracy_elementary_mathematics": 0.7073170731707317,
694
+ "mmlu_eval_accuracy_formal_logic": 0.6428571428571429,
695
+ "mmlu_eval_accuracy_global_facts": 0.4,
696
+ "mmlu_eval_accuracy_high_school_biology": 0.8125,
697
+ "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
698
+ "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
699
+ "mmlu_eval_accuracy_high_school_european_history": 0.7777777777777778,
700
+ "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
701
+ "mmlu_eval_accuracy_high_school_government_and_politics": 0.9523809523809523,
702
+ "mmlu_eval_accuracy_high_school_macroeconomics": 0.7674418604651163,
703
+ "mmlu_eval_accuracy_high_school_mathematics": 0.3448275862068966,
704
+ "mmlu_eval_accuracy_high_school_microeconomics": 0.9615384615384616,
705
+ "mmlu_eval_accuracy_high_school_physics": 0.35294117647058826,
706
+ "mmlu_eval_accuracy_high_school_psychology": 0.9333333333333333,
707
+ "mmlu_eval_accuracy_high_school_statistics": 0.782608695652174,
708
+ "mmlu_eval_accuracy_high_school_us_history": 0.9545454545454546,
709
+ "mmlu_eval_accuracy_high_school_world_history": 0.7692307692307693,
710
+ "mmlu_eval_accuracy_human_aging": 0.782608695652174,
711
+ "mmlu_eval_accuracy_human_sexuality": 0.6666666666666666,
712
+ "mmlu_eval_accuracy_international_law": 1.0,
713
+ "mmlu_eval_accuracy_jurisprudence": 0.6363636363636364,
714
+ "mmlu_eval_accuracy_logical_fallacies": 0.7777777777777778,
715
+ "mmlu_eval_accuracy_machine_learning": 0.6363636363636364,
716
+ "mmlu_eval_accuracy_management": 0.9090909090909091,
717
+ "mmlu_eval_accuracy_marketing": 0.96,
718
+ "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
719
+ "mmlu_eval_accuracy_miscellaneous": 0.7441860465116279,
720
+ "mmlu_eval_accuracy_moral_disputes": 0.8157894736842105,
721
+ "mmlu_eval_accuracy_moral_scenarios": 0.57,
722
+ "mmlu_eval_accuracy_nutrition": 0.7272727272727273,
723
+ "mmlu_eval_accuracy_philosophy": 0.7941176470588235,
724
+ "mmlu_eval_accuracy_prehistory": 0.8285714285714286,
725
+ "mmlu_eval_accuracy_professional_accounting": 0.6129032258064516,
726
+ "mmlu_eval_accuracy_professional_law": 0.6411764705882353,
727
+ "mmlu_eval_accuracy_professional_medicine": 0.8064516129032258,
728
+ "mmlu_eval_accuracy_professional_psychology": 0.8260869565217391,
729
+ "mmlu_eval_accuracy_public_relations": 0.6666666666666666,
730
+ "mmlu_eval_accuracy_security_studies": 0.8148148148148148,
731
+ "mmlu_eval_accuracy_sociology": 0.9090909090909091,
732
+ "mmlu_eval_accuracy_us_foreign_policy": 1.0,
733
+ "mmlu_eval_accuracy_virology": 0.5,
734
+ "mmlu_eval_accuracy_world_religions": 0.8421052631578947,
735
+ "mmlu_loss": 1.2988067958029479,
736
+ "step": 748
737
+ },
738
+ {
739
+ "epoch": 1.36,
740
+ "learning_rate": 0.0001,
741
+ "loss": 1.0735,
742
+ "step": 750
743
+ },
744
+ {
745
+ "epoch": 1.38,
746
+ "learning_rate": 0.0001,
747
+ "loss": 0.9066,
748
+ "step": 760
749
+ },
750
+ {
751
+ "epoch": 1.4,
752
+ "learning_rate": 0.0001,
753
+ "loss": 0.8716,
754
+ "step": 770
755
+ },
756
+ {
757
+ "epoch": 1.41,
758
+ "learning_rate": 0.0001,
759
+ "loss": 0.9144,
760
+ "step": 780
761
+ },
762
+ {
763
+ "epoch": 1.43,
764
+ "learning_rate": 0.0001,
765
+ "loss": 1.0338,
766
+ "step": 790
767
+ },
768
+ {
769
+ "epoch": 1.45,
770
+ "learning_rate": 0.0001,
771
+ "loss": 1.0275,
772
+ "step": 800
773
+ },
774
+ {
775
+ "epoch": 1.47,
776
+ "learning_rate": 0.0001,
777
+ "loss": 0.8382,
778
+ "step": 810
779
+ },
780
+ {
781
+ "epoch": 1.49,
782
+ "learning_rate": 0.0001,
783
+ "loss": 0.8489,
784
+ "step": 820
785
+ },
786
+ {
787
+ "epoch": 1.51,
788
+ "learning_rate": 0.0001,
789
+ "loss": 0.8931,
790
+ "step": 830
791
+ },
792
+ {
793
+ "epoch": 1.52,
794
+ "learning_rate": 0.0001,
795
+ "loss": 1.0515,
796
+ "step": 840
797
+ },
798
+ {
799
+ "epoch": 1.54,
800
+ "learning_rate": 0.0001,
801
+ "loss": 1.0965,
802
+ "step": 850
803
+ },
804
+ {
805
+ "epoch": 1.56,
806
+ "learning_rate": 0.0001,
807
+ "loss": 0.8928,
808
+ "step": 860
809
+ },
810
+ {
811
+ "epoch": 1.58,
812
+ "learning_rate": 0.0001,
813
+ "loss": 0.8608,
814
+ "step": 870
815
+ },
816
+ {
817
+ "epoch": 1.6,
818
+ "learning_rate": 0.0001,
819
+ "loss": 0.8831,
820
+ "step": 880
821
+ },
822
+ {
823
+ "epoch": 1.61,
824
+ "learning_rate": 0.0001,
825
+ "loss": 1.0253,
826
+ "step": 890
827
+ },
828
+ {
829
+ "epoch": 1.63,
830
+ "learning_rate": 0.0001,
831
+ "loss": 0.9905,
832
+ "step": 900
833
+ },
834
+ {
835
+ "epoch": 1.65,
836
+ "learning_rate": 0.0001,
837
+ "loss": 0.8487,
838
+ "step": 910
839
+ },
840
+ {
841
+ "epoch": 1.67,
842
+ "learning_rate": 0.0001,
843
+ "loss": 0.8568,
844
+ "step": 920
845
+ },
846
+ {
847
+ "epoch": 1.69,
848
+ "learning_rate": 0.0001,
849
+ "loss": 0.9047,
850
+ "step": 930
851
+ },
852
+ {
853
+ "epoch": 1.7,
854
+ "eval_loss": 1.0250624418258667,
855
+ "eval_runtime": 946.4035,
856
+ "eval_samples_per_second": 1.057,
857
+ "eval_steps_per_second": 1.057,
858
+ "step": 935
859
+ },
860
+ {
861
+ "epoch": 1.7,
862
+ "mmlu_eval_accuracy": 0.7288948695878031,
863
+ "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
864
+ "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
865
+ "mmlu_eval_accuracy_astronomy": 0.6875,
866
+ "mmlu_eval_accuracy_business_ethics": 0.9090909090909091,
867
+ "mmlu_eval_accuracy_clinical_knowledge": 0.8620689655172413,
868
+ "mmlu_eval_accuracy_college_biology": 0.875,
869
+ "mmlu_eval_accuracy_college_chemistry": 0.5,
870
+ "mmlu_eval_accuracy_college_computer_science": 0.7272727272727273,
871
+ "mmlu_eval_accuracy_college_mathematics": 0.36363636363636365,
872
+ "mmlu_eval_accuracy_college_medicine": 0.9090909090909091,
873
+ "mmlu_eval_accuracy_college_physics": 0.6363636363636364,
874
+ "mmlu_eval_accuracy_computer_security": 0.6363636363636364,
875
+ "mmlu_eval_accuracy_conceptual_physics": 0.6153846153846154,
876
+ "mmlu_eval_accuracy_econometrics": 0.8333333333333334,
877
+ "mmlu_eval_accuracy_electrical_engineering": 0.8125,
878
+ "mmlu_eval_accuracy_elementary_mathematics": 0.6097560975609756,
879
+ "mmlu_eval_accuracy_formal_logic": 0.6428571428571429,
880
+ "mmlu_eval_accuracy_global_facts": 0.5,
881
+ "mmlu_eval_accuracy_high_school_biology": 0.84375,
882
+ "mmlu_eval_accuracy_high_school_chemistry": 0.45454545454545453,
883
+ "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
884
+ "mmlu_eval_accuracy_high_school_european_history": 0.7777777777777778,
885
+ "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
886
+ "mmlu_eval_accuracy_high_school_government_and_politics": 0.9523809523809523,
887
+ "mmlu_eval_accuracy_high_school_macroeconomics": 0.7906976744186046,
888
+ "mmlu_eval_accuracy_high_school_mathematics": 0.3793103448275862,
889
+ "mmlu_eval_accuracy_high_school_microeconomics": 0.9230769230769231,
890
+ "mmlu_eval_accuracy_high_school_physics": 0.23529411764705882,
891
+ "mmlu_eval_accuracy_high_school_psychology": 0.9333333333333333,
892
+ "mmlu_eval_accuracy_high_school_statistics": 0.7391304347826086,
893
+ "mmlu_eval_accuracy_high_school_us_history": 0.9090909090909091,
894
+ "mmlu_eval_accuracy_high_school_world_history": 0.7692307692307693,
895
+ "mmlu_eval_accuracy_human_aging": 0.782608695652174,
896
+ "mmlu_eval_accuracy_human_sexuality": 0.6666666666666666,
897
+ "mmlu_eval_accuracy_international_law": 1.0,
898
+ "mmlu_eval_accuracy_jurisprudence": 0.5454545454545454,
899
+ "mmlu_eval_accuracy_logical_fallacies": 0.7222222222222222,
900
+ "mmlu_eval_accuracy_machine_learning": 0.6363636363636364,
901
+ "mmlu_eval_accuracy_management": 0.9090909090909091,
902
+ "mmlu_eval_accuracy_marketing": 0.96,
903
+ "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
904
+ "mmlu_eval_accuracy_miscellaneous": 0.7790697674418605,
905
+ "mmlu_eval_accuracy_moral_disputes": 0.8157894736842105,
906
+ "mmlu_eval_accuracy_moral_scenarios": 0.57,
907
+ "mmlu_eval_accuracy_nutrition": 0.7575757575757576,
908
+ "mmlu_eval_accuracy_philosophy": 0.8235294117647058,
909
+ "mmlu_eval_accuracy_prehistory": 0.8857142857142857,
910
+ "mmlu_eval_accuracy_professional_accounting": 0.6129032258064516,
911
+ "mmlu_eval_accuracy_professional_law": 0.6235294117647059,
912
+ "mmlu_eval_accuracy_professional_medicine": 0.8387096774193549,
913
+ "mmlu_eval_accuracy_professional_psychology": 0.8115942028985508,
914
+ "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
915
+ "mmlu_eval_accuracy_security_studies": 0.8148148148148148,
916
+ "mmlu_eval_accuracy_sociology": 0.9090909090909091,
917
+ "mmlu_eval_accuracy_us_foreign_policy": 1.0,
918
+ "mmlu_eval_accuracy_virology": 0.5,
919
+ "mmlu_eval_accuracy_world_religions": 0.8421052631578947,
920
+ "mmlu_loss": 1.243813282909306,
921
+ "step": 935
922
+ },
923
+ {
924
+ "epoch": 1.7,
925
+ "learning_rate": 0.0001,
926
+ "loss": 1.0174,
927
+ "step": 940
928
+ },
929
+ {
930
+ "epoch": 1.72,
931
+ "learning_rate": 0.0001,
932
+ "loss": 1.0302,
933
+ "step": 950
934
+ },
935
+ {
936
+ "epoch": 1.74,
937
+ "learning_rate": 0.0001,
938
+ "loss": 0.8799,
939
+ "step": 960
940
+ },
941
+ {
942
+ "epoch": 1.76,
943
+ "learning_rate": 0.0001,
944
+ "loss": 0.8447,
945
+ "step": 970
946
+ },
947
+ {
948
+ "epoch": 1.78,
949
+ "learning_rate": 0.0001,
950
+ "loss": 0.9053,
951
+ "step": 980
952
+ },
953
+ {
954
+ "epoch": 1.8,
955
+ "learning_rate": 0.0001,
956
+ "loss": 1.0331,
957
+ "step": 990
958
+ },
959
+ {
960
+ "epoch": 1.81,
961
+ "learning_rate": 0.0001,
962
+ "loss": 1.0412,
963
+ "step": 1000
964
+ },
965
+ {
966
+ "epoch": 1.83,
967
+ "learning_rate": 0.0001,
968
+ "loss": 0.8753,
969
+ "step": 1010
970
+ },
971
+ {
972
+ "epoch": 1.85,
973
+ "learning_rate": 0.0001,
974
+ "loss": 0.8744,
975
+ "step": 1020
976
+ },
977
+ {
978
+ "epoch": 1.87,
979
+ "learning_rate": 0.0001,
980
+ "loss": 0.8899,
981
+ "step": 1030
982
+ },
983
+ {
984
+ "epoch": 1.89,
985
+ "learning_rate": 0.0001,
986
+ "loss": 1.0053,
987
+ "step": 1040
988
+ },
989
+ {
990
+ "epoch": 1.9,
991
+ "learning_rate": 0.0001,
992
+ "loss": 1.0127,
993
+ "step": 1050
994
+ },
995
+ {
996
+ "epoch": 1.92,
997
+ "learning_rate": 0.0001,
998
+ "loss": 0.8023,
999
+ "step": 1060
1000
+ },
1001
+ {
1002
+ "epoch": 1.94,
1003
+ "learning_rate": 0.0001,
1004
+ "loss": 0.8349,
1005
+ "step": 1070
1006
+ },
1007
+ {
1008
+ "epoch": 1.96,
1009
+ "learning_rate": 0.0001,
1010
+ "loss": 0.9742,
1011
+ "step": 1080
1012
+ },
1013
+ {
1014
+ "epoch": 1.98,
1015
+ "learning_rate": 0.0001,
1016
+ "loss": 1.0971,
1017
+ "step": 1090
1018
+ },
1019
+ {
1020
+ "epoch": 2.0,
1021
+ "learning_rate": 0.0001,
1022
+ "loss": 1.0728,
1023
+ "step": 1100
1024
+ },
1025
+ {
1026
+ "epoch": 2.01,
1027
+ "learning_rate": 0.0001,
1028
+ "loss": 0.7724,
1029
+ "step": 1110
1030
+ },
1031
+ {
1032
+ "epoch": 2.03,
1033
+ "learning_rate": 0.0001,
1034
+ "loss": 0.7675,
1035
+ "step": 1120
1036
+ },
1037
+ {
1038
+ "epoch": 2.03,
1039
+ "eval_loss": 1.052681565284729,
1040
+ "eval_runtime": 942.0722,
1041
+ "eval_samples_per_second": 1.061,
1042
+ "eval_steps_per_second": 1.061,
1043
+ "step": 1122
1044
+ },
1045
+ {
1046
+ "epoch": 2.03,
1047
+ "mmlu_eval_accuracy": 0.7373981967098951,
1048
+ "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
1049
+ "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
1050
+ "mmlu_eval_accuracy_astronomy": 0.6875,
1051
+ "mmlu_eval_accuracy_business_ethics": 0.9090909090909091,
1052
+ "mmlu_eval_accuracy_clinical_knowledge": 0.896551724137931,
1053
+ "mmlu_eval_accuracy_college_biology": 0.875,
1054
+ "mmlu_eval_accuracy_college_chemistry": 0.5,
1055
+ "mmlu_eval_accuracy_college_computer_science": 0.5454545454545454,
1056
+ "mmlu_eval_accuracy_college_mathematics": 0.36363636363636365,
1057
+ "mmlu_eval_accuracy_college_medicine": 0.9090909090909091,
1058
+ "mmlu_eval_accuracy_college_physics": 0.5454545454545454,
1059
+ "mmlu_eval_accuracy_computer_security": 0.6363636363636364,
1060
+ "mmlu_eval_accuracy_conceptual_physics": 0.6153846153846154,
1061
+ "mmlu_eval_accuracy_econometrics": 0.8333333333333334,
1062
+ "mmlu_eval_accuracy_electrical_engineering": 0.875,
1063
+ "mmlu_eval_accuracy_elementary_mathematics": 0.5853658536585366,
1064
+ "mmlu_eval_accuracy_formal_logic": 0.7142857142857143,
1065
+ "mmlu_eval_accuracy_global_facts": 0.4,
1066
+ "mmlu_eval_accuracy_high_school_biology": 0.84375,
1067
+ "mmlu_eval_accuracy_high_school_chemistry": 0.45454545454545453,
1068
+ "mmlu_eval_accuracy_high_school_computer_science": 0.7777777777777778,
1069
+ "mmlu_eval_accuracy_high_school_european_history": 0.7777777777777778,
1070
+ "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
1071
+ "mmlu_eval_accuracy_high_school_government_and_politics": 0.9523809523809523,
1072
+ "mmlu_eval_accuracy_high_school_macroeconomics": 0.7674418604651163,
1073
+ "mmlu_eval_accuracy_high_school_mathematics": 0.41379310344827586,
1074
+ "mmlu_eval_accuracy_high_school_microeconomics": 0.9230769230769231,
1075
+ "mmlu_eval_accuracy_high_school_physics": 0.23529411764705882,
1076
+ "mmlu_eval_accuracy_high_school_psychology": 0.95,
1077
+ "mmlu_eval_accuracy_high_school_statistics": 0.782608695652174,
1078
+ "mmlu_eval_accuracy_high_school_us_history": 0.9090909090909091,
1079
+ "mmlu_eval_accuracy_high_school_world_history": 0.7692307692307693,
1080
+ "mmlu_eval_accuracy_human_aging": 0.8260869565217391,
1081
+ "mmlu_eval_accuracy_human_sexuality": 0.6666666666666666,
1082
+ "mmlu_eval_accuracy_international_law": 1.0,
1083
+ "mmlu_eval_accuracy_jurisprudence": 0.6363636363636364,
1084
+ "mmlu_eval_accuracy_logical_fallacies": 0.7777777777777778,
1085
+ "mmlu_eval_accuracy_machine_learning": 0.6363636363636364,
1086
+ "mmlu_eval_accuracy_management": 0.9090909090909091,
1087
+ "mmlu_eval_accuracy_marketing": 0.92,
1088
+ "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
1089
+ "mmlu_eval_accuracy_miscellaneous": 0.7906976744186046,
1090
+ "mmlu_eval_accuracy_moral_disputes": 0.8157894736842105,
1091
+ "mmlu_eval_accuracy_moral_scenarios": 0.62,
1092
+ "mmlu_eval_accuracy_nutrition": 0.7575757575757576,
1093
+ "mmlu_eval_accuracy_philosophy": 0.8235294117647058,
1094
+ "mmlu_eval_accuracy_prehistory": 0.8857142857142857,
1095
+ "mmlu_eval_accuracy_professional_accounting": 0.6774193548387096,
1096
+ "mmlu_eval_accuracy_professional_law": 0.6294117647058823,
1097
+ "mmlu_eval_accuracy_professional_medicine": 0.8709677419354839,
1098
+ "mmlu_eval_accuracy_professional_psychology": 0.782608695652174,
1099
+ "mmlu_eval_accuracy_public_relations": 0.6666666666666666,
1100
+ "mmlu_eval_accuracy_security_studies": 0.8148148148148148,
1101
+ "mmlu_eval_accuracy_sociology": 0.9090909090909091,
1102
+ "mmlu_eval_accuracy_us_foreign_policy": 1.0,
1103
+ "mmlu_eval_accuracy_virology": 0.5,
1104
+ "mmlu_eval_accuracy_world_religions": 0.8421052631578947,
1105
+ "mmlu_loss": 1.2340081441760609,
1106
+ "step": 1122
1107
+ },
1108
+ {
1109
+ "epoch": 2.05,
1110
+ "learning_rate": 0.0001,
1111
+ "loss": 0.7194,
1112
+ "step": 1130
1113
+ },
1114
+ {
1115
+ "epoch": 2.07,
1116
+ "learning_rate": 0.0001,
1117
+ "loss": 0.8236,
1118
+ "step": 1140
1119
+ },
1120
+ {
1121
+ "epoch": 2.09,
1122
+ "learning_rate": 0.0001,
1123
+ "loss": 0.6652,
1124
+ "step": 1150
1125
+ },
1126
+ {
1127
+ "epoch": 2.1,
1128
+ "learning_rate": 0.0001,
1129
+ "loss": 0.7177,
1130
+ "step": 1160
1131
+ },
1132
+ {
1133
+ "epoch": 2.12,
1134
+ "learning_rate": 0.0001,
1135
+ "loss": 0.7788,
1136
+ "step": 1170
1137
+ },
1138
+ {
1139
+ "epoch": 2.14,
1140
+ "learning_rate": 0.0001,
1141
+ "loss": 0.8117,
1142
+ "step": 1180
1143
+ },
1144
+ {
1145
+ "epoch": 2.16,
1146
+ "learning_rate": 0.0001,
1147
+ "loss": 0.8145,
1148
+ "step": 1190
1149
+ },
1150
+ {
1151
+ "epoch": 2.18,
1152
+ "learning_rate": 0.0001,
1153
+ "loss": 0.6984,
1154
+ "step": 1200
1155
+ },
1156
+ {
1157
+ "epoch": 2.19,
1158
+ "learning_rate": 0.0001,
1159
+ "loss": 0.7011,
1160
+ "step": 1210
1161
+ },
1162
+ {
1163
+ "epoch": 2.21,
1164
+ "learning_rate": 0.0001,
1165
+ "loss": 0.769,
1166
+ "step": 1220
1167
+ },
1168
+ {
1169
+ "epoch": 2.23,
1170
+ "learning_rate": 0.0001,
1171
+ "loss": 0.7705,
1172
+ "step": 1230
1173
+ },
1174
+ {
1175
+ "epoch": 2.25,
1176
+ "learning_rate": 0.0001,
1177
+ "loss": 0.8066,
1178
+ "step": 1240
1179
+ },
1180
+ {
1181
+ "epoch": 2.27,
1182
+ "learning_rate": 0.0001,
1183
+ "loss": 0.6622,
1184
+ "step": 1250
1185
+ },
1186
+ {
1187
+ "epoch": 2.29,
1188
+ "learning_rate": 0.0001,
1189
+ "loss": 0.6641,
1190
+ "step": 1260
1191
+ },
1192
+ {
1193
+ "epoch": 2.3,
1194
+ "learning_rate": 0.0001,
1195
+ "loss": 0.7239,
1196
+ "step": 1270
1197
+ },
1198
+ {
1199
+ "epoch": 2.32,
1200
+ "learning_rate": 0.0001,
1201
+ "loss": 0.7618,
1202
+ "step": 1280
1203
+ },
1204
+ {
1205
+ "epoch": 2.34,
1206
+ "learning_rate": 0.0001,
1207
+ "loss": 0.7845,
1208
+ "step": 1290
1209
+ },
1210
+ {
1211
+ "epoch": 2.36,
1212
+ "learning_rate": 0.0001,
1213
+ "loss": 0.719,
1214
+ "step": 1300
1215
+ },
1216
+ {
1217
+ "epoch": 2.37,
1218
+ "eval_loss": 1.1104822158813477,
1219
+ "eval_runtime": 948.1299,
1220
+ "eval_samples_per_second": 1.055,
1221
+ "eval_steps_per_second": 1.055,
1222
+ "step": 1309
1223
+ },
1224
+ {
1225
+ "epoch": 2.37,
1226
+ "mmlu_eval_accuracy": 0.7369285730399766,
1227
+ "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
1228
+ "mmlu_eval_accuracy_anatomy": 0.6428571428571429,
1229
+ "mmlu_eval_accuracy_astronomy": 0.6875,
1230
+ "mmlu_eval_accuracy_business_ethics": 1.0,
1231
+ "mmlu_eval_accuracy_clinical_knowledge": 0.8275862068965517,
1232
+ "mmlu_eval_accuracy_college_biology": 0.875,
1233
+ "mmlu_eval_accuracy_college_chemistry": 0.5,
1234
+ "mmlu_eval_accuracy_college_computer_science": 0.6363636363636364,
1235
+ "mmlu_eval_accuracy_college_mathematics": 0.45454545454545453,
1236
+ "mmlu_eval_accuracy_college_medicine": 0.8636363636363636,
1237
+ "mmlu_eval_accuracy_college_physics": 0.6363636363636364,
1238
+ "mmlu_eval_accuracy_computer_security": 0.7272727272727273,
1239
+ "mmlu_eval_accuracy_conceptual_physics": 0.6923076923076923,
1240
+ "mmlu_eval_accuracy_econometrics": 0.75,
1241
+ "mmlu_eval_accuracy_electrical_engineering": 0.8125,
1242
+ "mmlu_eval_accuracy_elementary_mathematics": 0.6341463414634146,
1243
+ "mmlu_eval_accuracy_formal_logic": 0.7857142857142857,
1244
+ "mmlu_eval_accuracy_global_facts": 0.5,
1245
+ "mmlu_eval_accuracy_high_school_biology": 0.84375,
1246
+ "mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
1247
+ "mmlu_eval_accuracy_high_school_computer_science": 0.8888888888888888,
1248
+ "mmlu_eval_accuracy_high_school_european_history": 0.7777777777777778,
1249
+ "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
1250
+ "mmlu_eval_accuracy_high_school_government_and_politics": 0.9523809523809523,
1251
+ "mmlu_eval_accuracy_high_school_macroeconomics": 0.7906976744186046,
1252
+ "mmlu_eval_accuracy_high_school_mathematics": 0.3793103448275862,
1253
+ "mmlu_eval_accuracy_high_school_microeconomics": 0.9230769230769231,
1254
+ "mmlu_eval_accuracy_high_school_physics": 0.17647058823529413,
1255
+ "mmlu_eval_accuracy_high_school_psychology": 0.95,
1256
+ "mmlu_eval_accuracy_high_school_statistics": 0.6956521739130435,
1257
+ "mmlu_eval_accuracy_high_school_us_history": 0.9090909090909091,
1258
+ "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
1259
+ "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
1260
+ "mmlu_eval_accuracy_human_sexuality": 0.6666666666666666,
1261
+ "mmlu_eval_accuracy_international_law": 1.0,
1262
+ "mmlu_eval_accuracy_jurisprudence": 0.5454545454545454,
1263
+ "mmlu_eval_accuracy_logical_fallacies": 0.7222222222222222,
1264
+ "mmlu_eval_accuracy_machine_learning": 0.6363636363636364,
1265
+ "mmlu_eval_accuracy_management": 0.9090909090909091,
1266
+ "mmlu_eval_accuracy_marketing": 0.88,
1267
+ "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
1268
+ "mmlu_eval_accuracy_miscellaneous": 0.7906976744186046,
1269
+ "mmlu_eval_accuracy_moral_disputes": 0.8157894736842105,
1270
+ "mmlu_eval_accuracy_moral_scenarios": 0.57,
1271
+ "mmlu_eval_accuracy_nutrition": 0.7575757575757576,
1272
+ "mmlu_eval_accuracy_philosophy": 0.8529411764705882,
1273
+ "mmlu_eval_accuracy_prehistory": 0.8571428571428571,
1274
+ "mmlu_eval_accuracy_professional_accounting": 0.6774193548387096,
1275
+ "mmlu_eval_accuracy_professional_law": 0.6058823529411764,
1276
+ "mmlu_eval_accuracy_professional_medicine": 0.8709677419354839,
1277
+ "mmlu_eval_accuracy_professional_psychology": 0.7681159420289855,
1278
+ "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
1279
+ "mmlu_eval_accuracy_security_studies": 0.8148148148148148,
1280
+ "mmlu_eval_accuracy_sociology": 0.9090909090909091,
1281
+ "mmlu_eval_accuracy_us_foreign_policy": 1.0,
1282
+ "mmlu_eval_accuracy_virology": 0.5555555555555556,
1283
+ "mmlu_eval_accuracy_world_religions": 0.8421052631578947,
1284
+ "mmlu_loss": 1.0866562834095908,
1285
+ "step": 1309
1286
+ },
1287
+ {
1288
+ "epoch": 2.38,
1289
+ "learning_rate": 0.0001,
1290
+ "loss": 0.7093,
1291
+ "step": 1310
1292
+ },
1293
+ {
1294
+ "epoch": 2.39,
1295
+ "learning_rate": 0.0001,
1296
+ "loss": 0.7684,
1297
+ "step": 1320
1298
+ },
1299
+ {
1300
+ "epoch": 2.41,
1301
+ "learning_rate": 0.0001,
1302
+ "loss": 0.7501,
1303
+ "step": 1330
1304
+ },
1305
+ {
1306
+ "epoch": 2.43,
1307
+ "learning_rate": 0.0001,
1308
+ "loss": 0.8043,
1309
+ "step": 1340
1310
+ },
1311
+ {
1312
+ "epoch": 2.45,
1313
+ "learning_rate": 0.0001,
1314
+ "loss": 0.6927,
1315
+ "step": 1350
1316
+ },
1317
+ {
1318
+ "epoch": 2.47,
1319
+ "learning_rate": 0.0001,
1320
+ "loss": 0.7278,
1321
+ "step": 1360
1322
+ },
1323
+ {
1324
+ "epoch": 2.48,
1325
+ "learning_rate": 0.0001,
1326
+ "loss": 0.8095,
1327
+ "step": 1370
1328
+ },
1329
+ {
1330
+ "epoch": 2.5,
1331
+ "learning_rate": 0.0001,
1332
+ "loss": 0.7463,
1333
+ "step": 1380
1334
+ },
1335
+ {
1336
+ "epoch": 2.52,
1337
+ "learning_rate": 0.0001,
1338
+ "loss": 0.7707,
1339
+ "step": 1390
1340
+ },
1341
+ {
1342
+ "epoch": 2.54,
1343
+ "learning_rate": 0.0001,
1344
+ "loss": 0.7152,
1345
+ "step": 1400
1346
+ }
1347
+ ],
1348
+ "max_steps": 1875,
1349
+ "num_train_epochs": 4,
1350
+ "total_flos": 1.6231629152399524e+18,
1351
+ "trial_name": null,
1352
+ "trial_params": null
1353
+ }
checkpoint-1400/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3739b542a2914edbd2d7eb3b727d6fa2a7752c75b0d7a856f5e87dd807fa1ef9
3
+ size 6200
checkpoint-1600/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-1600/adapter_config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": "152334H/miqu-1-70b-sf",
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": 0.05,
12
+ "modules_to_save": null,
13
+ "peft_type": "LORA",
14
+ "r": 64,
15
+ "revision": null,
16
+ "target_modules": [
17
+ "gate_proj",
18
+ "down_proj",
19
+ "v_proj",
20
+ "up_proj",
21
+ "q_proj",
22
+ "k_proj",
23
+ "o_proj"
24
+ ],
25
+ "task_type": "CAUSAL_LM"
26
+ }
checkpoint-1600/adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e9a0f9e046389866de742b94122dd6dbb44196a645e8a8912ab9becd3b6e9ee2
3
+ size 1657155522
checkpoint-1600/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-1600/adapter_model/adapter_config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": "152334H/miqu-1-70b-sf",
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": 0.05,
12
+ "modules_to_save": null,
13
+ "peft_type": "LORA",
14
+ "r": 64,
15
+ "revision": null,
16
+ "target_modules": [
17
+ "gate_proj",
18
+ "down_proj",
19
+ "v_proj",
20
+ "up_proj",
21
+ "q_proj",
22
+ "k_proj",
23
+ "o_proj"
24
+ ],
25
+ "task_type": "CAUSAL_LM"
26
+ }
checkpoint-1600/adapter_model/adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e9a0f9e046389866de742b94122dd6dbb44196a645e8a8912ab9becd3b6e9ee2
3
+ size 1657155522
checkpoint-1600/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:991313c3544fd29570cdd6d6c35cee055932460db308158ecc93c1bf6e12e312
3
+ size 6627702922