joshuaclymer commited on
Commit
232c380
1 Parent(s): 28c2ecb

Upload folder using huggingface_hub

Browse files
README.md ADDED
@@ -0,0 +1,21 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: peft
3
+ ---
4
+ ## Training procedure
5
+
6
+
7
+ The following `bitsandbytes` quantization config was used during training:
8
+ - quant_method: bitsandbytes
9
+ - load_in_8bit: False
10
+ - load_in_4bit: True
11
+ - llm_int8_threshold: 6.0
12
+ - llm_int8_skip_modules: None
13
+ - llm_int8_enable_fp32_cpu_offload: False
14
+ - llm_int8_has_fp16_weight: False
15
+ - bnb_4bit_quant_type: nf4
16
+ - bnb_4bit_use_double_quant: True
17
+ - bnb_4bit_compute_dtype: float16
18
+ ### Framework versions
19
+
20
+
21
+ - PEFT 0.5.0
adapter_config.json ADDED
@@ -0,0 +1,21 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "auto_mapping": null,
3
+ "base_model_name_or_path": "models/llama-30b",
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,
11
+ "lora_dropout": 0.0,
12
+ "modules_to_save": null,
13
+ "peft_type": "LORA",
14
+ "r": 64,
15
+ "revision": null,
16
+ "target_modules": [
17
+ "q_proj",
18
+ "v_proj"
19
+ ],
20
+ "task_type": "SEQ_CLS"
21
+ }
adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7f71d6c6951c6c274f59f10f4eee211154b309bedb38e021f4bedeb11601df07
3
+ size 409081617
special_tokens_map.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<s>",
4
+ "lstrip": false,
5
+ "normalized": true,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "</s>",
11
+ "lstrip": false,
12
+ "normalized": true,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": "</s>",
17
+ "unk_token": {
18
+ "content": "<unk>",
19
+ "lstrip": false,
20
+ "normalized": true,
21
+ "rstrip": false,
22
+ "single_word": false
23
+ }
24
+ }
tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
3
+ size 499723
tokenizer_config.json ADDED
@@ -0,0 +1,32 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "__type": "AddedToken",
4
+ "content": "<s>",
5
+ "lstrip": false,
6
+ "normalized": true,
7
+ "rstrip": false,
8
+ "single_word": false
9
+ },
10
+ "clean_up_tokenization_spaces": false,
11
+ "eos_token": {
12
+ "__type": "AddedToken",
13
+ "content": "</s>",
14
+ "lstrip": false,
15
+ "normalized": true,
16
+ "rstrip": false,
17
+ "single_word": false
18
+ },
19
+ "model_max_length": 2048,
20
+ "pad_token": null,
21
+ "sp_model_kwargs": {},
22
+ "tokenizer_class": "LlamaTokenizer",
23
+ "unk_token": {
24
+ "__type": "AddedToken",
25
+ "content": "<unk>",
26
+ "lstrip": false,
27
+ "normalized": true,
28
+ "rstrip": false,
29
+ "single_word": false
30
+ },
31
+ "use_default_system_prompt": true
32
+ }
train_args.json ADDED
@@ -0,0 +1,123 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "initial_model_dir": "models/llama-30b",
3
+ "distribution_id": "math_hard",
4
+ "date_trained": "11/10/2023 08:18:27",
5
+ "output_dir": "models/classify_lora/llama-30b-math_hard",
6
+ "overwrite_output_dir": false,
7
+ "do_train": false,
8
+ "do_eval": true,
9
+ "do_predict": false,
10
+ "evaluation_strategy": "steps",
11
+ "prediction_loss_only": false,
12
+ "per_device_train_batch_size": 8,
13
+ "per_device_eval_batch_size": 8,
14
+ "per_gpu_train_batch_size": null,
15
+ "per_gpu_eval_batch_size": null,
16
+ "gradient_accumulation_steps": 1,
17
+ "eval_accumulation_steps": null,
18
+ "eval_delay": 0,
19
+ "learning_rate": 0.0002,
20
+ "weight_decay": 0.0,
21
+ "adam_beta1": 0.9,
22
+ "adam_beta2": 0.999,
23
+ "adam_epsilon": 1e-08,
24
+ "max_grad_norm": 0.3,
25
+ "num_train_epochs": 2,
26
+ "max_steps": 100,
27
+ "lr_scheduler_type": "constant",
28
+ "warmup_ratio": 0.03,
29
+ "warmup_steps": 0,
30
+ "log_level": "passive",
31
+ "log_level_replica": "warning",
32
+ "log_on_each_node": true,
33
+ "logging_dir": "models/classify_lora/llama-30b-math_hard/runs/Oct11_05-49-08_compute-permanent-node-975",
34
+ "logging_strategy": "steps",
35
+ "logging_first_step": false,
36
+ "logging_steps": 1,
37
+ "logging_nan_inf_filter": true,
38
+ "save_strategy": "steps",
39
+ "save_steps": 25,
40
+ "save_total_limit": 0,
41
+ "save_safetensors": false,
42
+ "save_on_each_node": false,
43
+ "no_cuda": false,
44
+ "use_cpu": false,
45
+ "use_mps_device": false,
46
+ "seed": 42,
47
+ "data_seed": null,
48
+ "jit_mode_eval": false,
49
+ "use_ipex": false,
50
+ "bf16": false,
51
+ "fp16": false,
52
+ "fp16_opt_level": "O1",
53
+ "half_precision_backend": "auto",
54
+ "bf16_full_eval": false,
55
+ "fp16_full_eval": false,
56
+ "tf32": null,
57
+ "local_rank": 0,
58
+ "ddp_backend": null,
59
+ "tpu_num_cores": null,
60
+ "tpu_metrics_debug": false,
61
+ "debug": [],
62
+ "dataloader_drop_last": false,
63
+ "eval_steps": 25,
64
+ "dataloader_num_workers": 0,
65
+ "past_index": -1,
66
+ "run_name": "train|models-classify_lora-llama-30b-math_hard",
67
+ "disable_tqdm": false,
68
+ "remove_unused_columns": false,
69
+ "label_names": null,
70
+ "load_best_model_at_end": false,
71
+ "metric_for_best_model": "eval_math_hard_score",
72
+ "greater_is_better": true,
73
+ "ignore_data_skip": false,
74
+ "sharded_ddp": [],
75
+ "fsdp": [],
76
+ "fsdp_min_num_params": 0,
77
+ "fsdp_config": {
78
+ "min_num_params": 0,
79
+ "xla": false,
80
+ "xla_fsdp_grad_ckpt": false
81
+ },
82
+ "fsdp_transformer_layer_cls_to_wrap": null,
83
+ "deepspeed": "configs/ds_zero_1.json",
84
+ "label_smoothing_factor": 0.0,
85
+ "optim": "paged_adamw_32bit",
86
+ "optim_args": null,
87
+ "adafactor": false,
88
+ "group_by_length": false,
89
+ "length_column_name": "length",
90
+ "report_to": [
91
+ "wandb"
92
+ ],
93
+ "ddp_find_unused_parameters": false,
94
+ "ddp_bucket_cap_mb": null,
95
+ "ddp_broadcast_buffers": null,
96
+ "dataloader_pin_memory": true,
97
+ "skip_memory_metrics": true,
98
+ "use_legacy_prediction_loop": false,
99
+ "push_to_hub": false,
100
+ "resume_from_checkpoint": null,
101
+ "hub_model_id": null,
102
+ "hub_strategy": "every_save",
103
+ "hub_token": null,
104
+ "hub_private_repo": false,
105
+ "hub_always_push": false,
106
+ "gradient_checkpointing": false,
107
+ "include_inputs_for_metrics": false,
108
+ "fp16_backend": "auto",
109
+ "push_to_hub_model_id": null,
110
+ "push_to_hub_organization": null,
111
+ "push_to_hub_token": null,
112
+ "_n_gpu": 1,
113
+ "mp_parameters": "",
114
+ "auto_find_batch_size": false,
115
+ "full_determinism": false,
116
+ "torchdynamo": null,
117
+ "ray_scope": "last",
118
+ "ddp_timeout": 1800,
119
+ "torch_compile": false,
120
+ "torch_compile_backend": null,
121
+ "torch_compile_mode": null,
122
+ "dispatch_batches": null
123
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:93a764fdaf689372fc9812c299850e445c3bc42ecfab1096e9ca2c21e2fbcbd3
3
+ size 5115
training_logs.json ADDED
@@ -0,0 +1,1067 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ [
2
+ {
3
+ "loss": 0.7313,
4
+ "learning_rate": 0.0002,
5
+ "epoch": 0.02,
6
+ "step": 1
7
+ },
8
+ {
9
+ "loss": 0.6794,
10
+ "learning_rate": 0.0002,
11
+ "epoch": 0.04,
12
+ "step": 2
13
+ },
14
+ {
15
+ "loss": 0.6862,
16
+ "learning_rate": 0.0002,
17
+ "epoch": 0.05,
18
+ "step": 3
19
+ },
20
+ {
21
+ "loss": 0.6799,
22
+ "learning_rate": 0.0002,
23
+ "epoch": 0.07,
24
+ "step": 4
25
+ },
26
+ {
27
+ "loss": 0.6825,
28
+ "learning_rate": 0.0002,
29
+ "epoch": 0.09,
30
+ "step": 5
31
+ },
32
+ {
33
+ "loss": 0.6914,
34
+ "learning_rate": 0.0002,
35
+ "epoch": 0.11,
36
+ "step": 6
37
+ },
38
+ {
39
+ "loss": 0.6325,
40
+ "learning_rate": 0.0002,
41
+ "epoch": 0.12,
42
+ "step": 7
43
+ },
44
+ {
45
+ "loss": 0.6633,
46
+ "learning_rate": 0.0002,
47
+ "epoch": 0.14,
48
+ "step": 8
49
+ },
50
+ {
51
+ "loss": 0.6252,
52
+ "learning_rate": 0.0002,
53
+ "epoch": 0.16,
54
+ "step": 9
55
+ },
56
+ {
57
+ "loss": 0.5798,
58
+ "learning_rate": 0.0002,
59
+ "epoch": 0.18,
60
+ "step": 10
61
+ },
62
+ {
63
+ "loss": 0.5536,
64
+ "learning_rate": 0.0002,
65
+ "epoch": 0.19,
66
+ "step": 11
67
+ },
68
+ {
69
+ "loss": 0.5727,
70
+ "learning_rate": 0.0002,
71
+ "epoch": 0.21,
72
+ "step": 12
73
+ },
74
+ {
75
+ "loss": 0.4765,
76
+ "learning_rate": 0.0002,
77
+ "epoch": 0.23,
78
+ "step": 13
79
+ },
80
+ {
81
+ "loss": 0.3769,
82
+ "learning_rate": 0.0002,
83
+ "epoch": 0.25,
84
+ "step": 14
85
+ },
86
+ {
87
+ "loss": 0.2708,
88
+ "learning_rate": 0.0002,
89
+ "epoch": 0.26,
90
+ "step": 15
91
+ },
92
+ {
93
+ "loss": 0.2132,
94
+ "learning_rate": 0.0002,
95
+ "epoch": 0.28,
96
+ "step": 16
97
+ },
98
+ {
99
+ "loss": 0.2204,
100
+ "learning_rate": 0.0002,
101
+ "epoch": 0.3,
102
+ "step": 17
103
+ },
104
+ {
105
+ "loss": 0.3512,
106
+ "learning_rate": 0.0002,
107
+ "epoch": 0.32,
108
+ "step": 18
109
+ },
110
+ {
111
+ "loss": 0.1736,
112
+ "learning_rate": 0.0002,
113
+ "epoch": 0.33,
114
+ "step": 19
115
+ },
116
+ {
117
+ "loss": 0.3119,
118
+ "learning_rate": 0.0002,
119
+ "epoch": 0.35,
120
+ "step": 20
121
+ },
122
+ {
123
+ "loss": 0.1878,
124
+ "learning_rate": 0.0002,
125
+ "epoch": 0.37,
126
+ "step": 21
127
+ },
128
+ {
129
+ "loss": 0.2817,
130
+ "learning_rate": 0.0002,
131
+ "epoch": 0.39,
132
+ "step": 22
133
+ },
134
+ {
135
+ "loss": 0.4016,
136
+ "learning_rate": 0.0002,
137
+ "epoch": 0.4,
138
+ "step": 23
139
+ },
140
+ {
141
+ "loss": 0.156,
142
+ "learning_rate": 0.0002,
143
+ "epoch": 0.42,
144
+ "step": 24
145
+ },
146
+ {
147
+ "loss": 0.1408,
148
+ "learning_rate": 0.0002,
149
+ "epoch": 0.44,
150
+ "step": 25
151
+ },
152
+ {
153
+ "eval_math_hard_loss": 0.2689576745033264,
154
+ "eval_math_hard_score": -0.07232501357793808,
155
+ "eval_math_hard_brier_score": 0.07232501357793808,
156
+ "eval_math_hard_average_probability": 0.8480575084686279,
157
+ "eval_math_hard_accuracy": 0.91,
158
+ "eval_math_hard_probabilities": [
159
+ 0.12067427486181259,
160
+ 0.1707480251789093,
161
+ 0.22421833872795105,
162
+ 0.9965168237686157,
163
+ 0.9968488812446594,
164
+ 0.8897665739059448,
165
+ 0.8757681846618652,
166
+ 0.9160223007202148,
167
+ 0.4495812952518463,
168
+ 0.9733548760414124,
169
+ 0.8542622327804565,
170
+ 0.8829213976860046,
171
+ 0.7457072734832764,
172
+ 0.7373321056365967,
173
+ 0.7001582980155945,
174
+ 0.9495631456375122,
175
+ 0.9757802486419678,
176
+ 0.9814878702163696,
177
+ 0.8223016262054443,
178
+ 0.8257706761360168,
179
+ 0.7114197611808777,
180
+ 0.9997543692588806,
181
+ 0.9997970461845398,
182
+ 0.9997699856758118,
183
+ 0.9983037710189819,
184
+ 0.9911226034164429,
185
+ 0.9890031218528748,
186
+ 0.9952235817909241,
187
+ 0.9955497980117798,
188
+ 0.9985560774803162,
189
+ 0.9976275563240051,
190
+ 0.9978576302528381,
191
+ 0.9976431727409363,
192
+ 0.9857828617095947,
193
+ 0.9856427311897278,
194
+ 0.9890527129173279,
195
+ 0.9993459582328796,
196
+ 0.99762362241745,
197
+ 0.9985696077346802,
198
+ 0.9795147180557251,
199
+ 0.9767017364501953,
200
+ 0.9040936827659607,
201
+ 0.989809513092041,
202
+ 0.9877877831459045,
203
+ 0.9814877510070801,
204
+ 0.9997876286506653,
205
+ 0.9965195655822754,
206
+ 0.9839633107185364,
207
+ 0.44732266664505005,
208
+ 0.5352095365524292,
209
+ 0.5794621109962463,
210
+ 0.9967998266220093,
211
+ 0.9975603818893433,
212
+ 0.9964653253555298,
213
+ 0.9998288154602051,
214
+ 0.9998891353607178,
215
+ 0.9999051094055176,
216
+ 0.2663218379020691,
217
+ 0.4094083905220032,
218
+ 0.7260100245475769,
219
+ 0.9965078234672546,
220
+ 0.9969045519828796,
221
+ 0.998040497303009,
222
+ 0.9707159399986267,
223
+ 0.9864614605903625,
224
+ 0.9651244878768921,
225
+ 0.9996368885040283,
226
+ 0.9995032548904419,
227
+ 0.9982764720916748,
228
+ 0.6076152920722961,
229
+ 0.585483729839325,
230
+ 0.5877071619033813,
231
+ 0.9988491535186768,
232
+ 0.9986512064933777,
233
+ 0.9980589747428894,
234
+ 0.9256494045257568,
235
+ 0.9805311560630798,
236
+ 0.9392337799072266,
237
+ 0.7655142545700073,
238
+ 0.6185799241065979,
239
+ 0.541168212890625,
240
+ 0.782398521900177,
241
+ 0.8448206782341003,
242
+ 0.7403358817100525,
243
+ 0.8463240265846252,
244
+ 0.8631186485290527,
245
+ 0.8114821314811707,
246
+ 0.9599024057388306,
247
+ 0.7725688815116882,
248
+ 0.3576231896877289,
249
+ 0.027788693085312843,
250
+ 0.8065083622932434,
251
+ 0.8443471789360046,
252
+ 0.8439624905586243,
253
+ 0.727486789226532,
254
+ 0.8130761384963989,
255
+ 0.9100518822669983,
256
+ 0.9845901727676392,
257
+ 0.9753082990646362,
258
+ 0.9655715823173523
259
+ ],
260
+ "eval_math_hard_runtime": 86.0301,
261
+ "eval_math_hard_samples_per_second": 1.162,
262
+ "eval_math_hard_steps_per_second": 0.046,
263
+ "epoch": 0.44,
264
+ "step": 25
265
+ },
266
+ {
267
+ "loss": 0.1166,
268
+ "learning_rate": 0.0002,
269
+ "epoch": 0.46,
270
+ "step": 26
271
+ },
272
+ {
273
+ "loss": 0.1934,
274
+ "learning_rate": 0.0002,
275
+ "epoch": 0.47,
276
+ "step": 27
277
+ },
278
+ {
279
+ "loss": 0.1695,
280
+ "learning_rate": 0.0002,
281
+ "epoch": 0.49,
282
+ "step": 28
283
+ },
284
+ {
285
+ "loss": 0.1415,
286
+ "learning_rate": 0.0002,
287
+ "epoch": 0.51,
288
+ "step": 29
289
+ },
290
+ {
291
+ "loss": 0.2946,
292
+ "learning_rate": 0.0002,
293
+ "epoch": 0.53,
294
+ "step": 30
295
+ },
296
+ {
297
+ "loss": 0.1945,
298
+ "learning_rate": 0.0002,
299
+ "epoch": 0.54,
300
+ "step": 31
301
+ },
302
+ {
303
+ "loss": 0.1672,
304
+ "learning_rate": 0.0002,
305
+ "epoch": 0.56,
306
+ "step": 32
307
+ },
308
+ {
309
+ "loss": 0.1961,
310
+ "learning_rate": 0.0002,
311
+ "epoch": 0.58,
312
+ "step": 33
313
+ },
314
+ {
315
+ "loss": 0.3305,
316
+ "learning_rate": 0.0002,
317
+ "epoch": 0.6,
318
+ "step": 34
319
+ },
320
+ {
321
+ "loss": 0.0823,
322
+ "learning_rate": 0.0002,
323
+ "epoch": 0.61,
324
+ "step": 35
325
+ },
326
+ {
327
+ "loss": 0.1063,
328
+ "learning_rate": 0.0002,
329
+ "epoch": 0.63,
330
+ "step": 36
331
+ },
332
+ {
333
+ "loss": 0.1566,
334
+ "learning_rate": 0.0002,
335
+ "epoch": 0.65,
336
+ "step": 37
337
+ },
338
+ {
339
+ "loss": 0.1957,
340
+ "learning_rate": 0.0002,
341
+ "epoch": 0.67,
342
+ "step": 38
343
+ },
344
+ {
345
+ "loss": 0.0823,
346
+ "learning_rate": 0.0002,
347
+ "epoch": 0.68,
348
+ "step": 39
349
+ },
350
+ {
351
+ "loss": 0.0602,
352
+ "learning_rate": 0.0002,
353
+ "epoch": 0.7,
354
+ "step": 40
355
+ },
356
+ {
357
+ "loss": 0.1042,
358
+ "learning_rate": 0.0002,
359
+ "epoch": 0.72,
360
+ "step": 41
361
+ },
362
+ {
363
+ "loss": 0.0696,
364
+ "learning_rate": 0.0002,
365
+ "epoch": 0.74,
366
+ "step": 42
367
+ },
368
+ {
369
+ "loss": 0.0618,
370
+ "learning_rate": 0.0002,
371
+ "epoch": 0.75,
372
+ "step": 43
373
+ },
374
+ {
375
+ "loss": 0.1516,
376
+ "learning_rate": 0.0002,
377
+ "epoch": 0.77,
378
+ "step": 44
379
+ },
380
+ {
381
+ "loss": 0.1877,
382
+ "learning_rate": 0.0002,
383
+ "epoch": 0.79,
384
+ "step": 45
385
+ },
386
+ {
387
+ "loss": 0.2814,
388
+ "learning_rate": 0.0002,
389
+ "epoch": 0.81,
390
+ "step": 46
391
+ },
392
+ {
393
+ "loss": 0.3869,
394
+ "learning_rate": 0.0002,
395
+ "epoch": 0.82,
396
+ "step": 47
397
+ },
398
+ {
399
+ "loss": 0.0865,
400
+ "learning_rate": 0.0002,
401
+ "epoch": 0.84,
402
+ "step": 48
403
+ },
404
+ {
405
+ "loss": 0.5889,
406
+ "learning_rate": 0.0002,
407
+ "epoch": 0.86,
408
+ "step": 49
409
+ },
410
+ {
411
+ "loss": 0.108,
412
+ "learning_rate": 0.0002,
413
+ "epoch": 0.88,
414
+ "step": 50
415
+ },
416
+ {
417
+ "eval_math_hard_loss": 0.4195532202720642,
418
+ "eval_math_hard_score": -0.08378518372774124,
419
+ "eval_math_hard_brier_score": 0.08378518372774124,
420
+ "eval_math_hard_average_probability": 0.8607721924781799,
421
+ "eval_math_hard_accuracy": 0.89,
422
+ "eval_math_hard_probabilities": [
423
+ 0.001347638200968504,
424
+ 0.0014310380211099982,
425
+ 0.0292666032910347,
426
+ 0.9999281167984009,
427
+ 0.9999291896820068,
428
+ 0.9515820741653442,
429
+ 0.9991993308067322,
430
+ 0.9985817670822144,
431
+ 0.4427696764469147,
432
+ 0.9999922513961792,
433
+ 0.5870345830917358,
434
+ 0.9338187575340271,
435
+ 0.9995662569999695,
436
+ 0.999679446220398,
437
+ 0.9990984201431274,
438
+ 0.8128757476806641,
439
+ 0.8317935466766357,
440
+ 0.8578028082847595,
441
+ 0.767571747303009,
442
+ 0.7276169061660767,
443
+ 0.16569289565086365,
444
+ 0.9999608993530273,
445
+ 0.9999557733535767,
446
+ 0.9999984502792358,
447
+ 0.9999668598175049,
448
+ 0.9999934434890747,
449
+ 0.9996861219406128,
450
+ 0.999996542930603,
451
+ 0.9999967813491821,
452
+ 0.9999997615814209,
453
+ 0.999997615814209,
454
+ 0.9999953508377075,
455
+ 0.9999957084655762,
456
+ 0.9973000884056091,
457
+ 0.9996415376663208,
458
+ 0.9997121691703796,
459
+ 0.9999990463256836,
460
+ 0.9999995231628418,
461
+ 0.9999998807907104,
462
+ 0.9920561909675598,
463
+ 0.9743956327438354,
464
+ 0.518930196762085,
465
+ 0.9998225569725037,
466
+ 0.9935445785522461,
467
+ 0.9992994070053101,
468
+ 0.9982439279556274,
469
+ 0.9694525599479675,
470
+ 0.8038814067840576,
471
+ 0.5726821422576904,
472
+ 0.4458751678466797,
473
+ 0.741835355758667,
474
+ 0.9999998807907104,
475
+ 1.0,
476
+ 0.9999977350234985,
477
+ 0.9999512434005737,
478
+ 1.0,
479
+ 1.0,
480
+ 0.1753261685371399,
481
+ 0.2425578236579895,
482
+ 0.8445780873298645,
483
+ 0.9961126446723938,
484
+ 0.9984958171844482,
485
+ 0.9912378191947937,
486
+ 0.9592325687408447,
487
+ 0.9976963400840759,
488
+ 0.866991400718689,
489
+ 0.9999998807907104,
490
+ 0.9999804496765137,
491
+ 0.9998711347579956,
492
+ 0.26644328236579895,
493
+ 0.829546332359314,
494
+ 0.8019995093345642,
495
+ 0.9999985694885254,
496
+ 0.9999955892562866,
497
+ 0.9999771118164062,
498
+ 0.9985451698303223,
499
+ 0.9999727010726929,
500
+ 0.9993960857391357,
501
+ 0.9808396100997925,
502
+ 0.8794113993644714,
503
+ 0.8073936700820923,
504
+ 0.963763952255249,
505
+ 0.9844830632209778,
506
+ 0.958784818649292,
507
+ 0.939505934715271,
508
+ 0.9660537838935852,
509
+ 0.817690372467041,
510
+ 0.9990904331207275,
511
+ 0.7585221529006958,
512
+ 0.4643564522266388,
513
+ 0.11916442960500717,
514
+ 0.9887877106666565,
515
+ 0.9978222846984863,
516
+ 0.8583523631095886,
517
+ 0.8125313520431519,
518
+ 0.8561255931854248,
519
+ 0.8449611067771912,
520
+ 0.9997431635856628,
521
+ 0.9992627501487732,
522
+ 0.9998737573623657
523
+ ],
524
+ "eval_math_hard_runtime": 86.0735,
525
+ "eval_math_hard_samples_per_second": 1.162,
526
+ "eval_math_hard_steps_per_second": 0.046,
527
+ "epoch": 0.88,
528
+ "step": 50
529
+ },
530
+ {
531
+ "loss": 0.0879,
532
+ "learning_rate": 0.0002,
533
+ "epoch": 0.89,
534
+ "step": 51
535
+ },
536
+ {
537
+ "loss": 0.2006,
538
+ "learning_rate": 0.0002,
539
+ "epoch": 0.91,
540
+ "step": 52
541
+ },
542
+ {
543
+ "loss": 0.1101,
544
+ "learning_rate": 0.0002,
545
+ "epoch": 0.93,
546
+ "step": 53
547
+ },
548
+ {
549
+ "loss": 0.1182,
550
+ "learning_rate": 0.0002,
551
+ "epoch": 0.95,
552
+ "step": 54
553
+ },
554
+ {
555
+ "loss": 0.1047,
556
+ "learning_rate": 0.0002,
557
+ "epoch": 0.96,
558
+ "step": 55
559
+ },
560
+ {
561
+ "loss": 0.1168,
562
+ "learning_rate": 0.0002,
563
+ "epoch": 0.98,
564
+ "step": 56
565
+ },
566
+ {
567
+ "loss": 0.1458,
568
+ "learning_rate": 0.0002,
569
+ "epoch": 1.0,
570
+ "step": 57
571
+ },
572
+ {
573
+ "loss": 0.0369,
574
+ "learning_rate": 0.0002,
575
+ "epoch": 1.02,
576
+ "step": 58
577
+ },
578
+ {
579
+ "loss": 0.1703,
580
+ "learning_rate": 0.0002,
581
+ "epoch": 1.04,
582
+ "step": 59
583
+ },
584
+ {
585
+ "loss": 0.0705,
586
+ "learning_rate": 0.0002,
587
+ "epoch": 1.05,
588
+ "step": 60
589
+ },
590
+ {
591
+ "loss": 0.0306,
592
+ "learning_rate": 0.0002,
593
+ "epoch": 1.07,
594
+ "step": 61
595
+ },
596
+ {
597
+ "loss": 0.0199,
598
+ "learning_rate": 0.0002,
599
+ "epoch": 1.09,
600
+ "step": 62
601
+ },
602
+ {
603
+ "loss": 0.0059,
604
+ "learning_rate": 0.0002,
605
+ "epoch": 1.11,
606
+ "step": 63
607
+ },
608
+ {
609
+ "loss": 0.0019,
610
+ "learning_rate": 0.0002,
611
+ "epoch": 1.12,
612
+ "step": 64
613
+ },
614
+ {
615
+ "loss": 0.05,
616
+ "learning_rate": 0.0002,
617
+ "epoch": 1.14,
618
+ "step": 65
619
+ },
620
+ {
621
+ "loss": 0.0192,
622
+ "learning_rate": 0.0002,
623
+ "epoch": 1.16,
624
+ "step": 66
625
+ },
626
+ {
627
+ "loss": 0.2126,
628
+ "learning_rate": 0.0002,
629
+ "epoch": 1.18,
630
+ "step": 67
631
+ },
632
+ {
633
+ "loss": 0.0073,
634
+ "learning_rate": 0.0002,
635
+ "epoch": 1.19,
636
+ "step": 68
637
+ },
638
+ {
639
+ "loss": 0.0883,
640
+ "learning_rate": 0.0002,
641
+ "epoch": 1.21,
642
+ "step": 69
643
+ },
644
+ {
645
+ "loss": 0.0251,
646
+ "learning_rate": 0.0002,
647
+ "epoch": 1.23,
648
+ "step": 70
649
+ },
650
+ {
651
+ "loss": 0.0673,
652
+ "learning_rate": 0.0002,
653
+ "epoch": 1.25,
654
+ "step": 71
655
+ },
656
+ {
657
+ "loss": 0.0042,
658
+ "learning_rate": 0.0002,
659
+ "epoch": 1.26,
660
+ "step": 72
661
+ },
662
+ {
663
+ "loss": 0.0769,
664
+ "learning_rate": 0.0002,
665
+ "epoch": 1.28,
666
+ "step": 73
667
+ },
668
+ {
669
+ "loss": 0.1407,
670
+ "learning_rate": 0.0002,
671
+ "epoch": 1.3,
672
+ "step": 74
673
+ },
674
+ {
675
+ "loss": 0.0613,
676
+ "learning_rate": 0.0002,
677
+ "epoch": 1.32,
678
+ "step": 75
679
+ },
680
+ {
681
+ "eval_math_hard_loss": 0.40808790922164917,
682
+ "eval_math_hard_score": -0.07436995208263397,
683
+ "eval_math_hard_brier_score": 0.07436995208263397,
684
+ "eval_math_hard_average_probability": 0.9035561084747314,
685
+ "eval_math_hard_accuracy": 0.92,
686
+ "eval_math_hard_probabilities": [
687
+ 0.01734217070043087,
688
+ 0.025694051757454872,
689
+ 0.1909039467573166,
690
+ 1.0,
691
+ 1.0,
692
+ 0.9912930130958557,
693
+ 0.9999858140945435,
694
+ 0.9999847412109375,
695
+ 0.9637160897254944,
696
+ 1.0,
697
+ 0.9999996423721313,
698
+ 0.9999971389770508,
699
+ 0.9771614074707031,
700
+ 0.9833316802978516,
701
+ 0.966616153717041,
702
+ 0.9999768733978271,
703
+ 0.9999898672103882,
704
+ 0.9999890327453613,
705
+ 0.9881917834281921,
706
+ 0.9601098895072937,
707
+ 0.08695478737354279,
708
+ 1.0,
709
+ 1.0,
710
+ 1.0,
711
+ 1.0,
712
+ 1.0,
713
+ 0.9999998807907104,
714
+ 1.0,
715
+ 1.0,
716
+ 1.0,
717
+ 1.0,
718
+ 1.0,
719
+ 1.0,
720
+ 1.0,
721
+ 1.0,
722
+ 1.0,
723
+ 1.0,
724
+ 1.0,
725
+ 1.0,
726
+ 0.9999986886978149,
727
+ 0.9999992847442627,
728
+ 0.6397180557250977,
729
+ 0.9994465708732605,
730
+ 0.9973899722099304,
731
+ 0.9993732571601868,
732
+ 1.0,
733
+ 0.9999997615814209,
734
+ 0.99986732006073,
735
+ 0.9971725940704346,
736
+ 0.8597153425216675,
737
+ 0.9993921518325806,
738
+ 1.0,
739
+ 1.0,
740
+ 1.0,
741
+ 1.0,
742
+ 1.0,
743
+ 1.0,
744
+ 0.0010394210694357753,
745
+ 0.04478495568037033,
746
+ 0.9842627644538879,
747
+ 0.9999969005584717,
748
+ 0.9999995231628418,
749
+ 0.709723174571991,
750
+ 0.9994927644729614,
751
+ 0.9999569654464722,
752
+ 0.9997255206108093,
753
+ 1.0,
754
+ 1.0,
755
+ 1.0,
756
+ 0.98405921459198,
757
+ 0.9982985854148865,
758
+ 0.9936298727989197,
759
+ 1.0,
760
+ 0.9999997615814209,
761
+ 0.9999988079071045,
762
+ 0.9999998807907104,
763
+ 1.0,
764
+ 0.9999998807907104,
765
+ 0.9991475343704224,
766
+ 0.8117725253105164,
767
+ 0.20917673408985138,
768
+ 0.9854962229728699,
769
+ 0.998820960521698,
770
+ 0.9700117111206055,
771
+ 0.7556695938110352,
772
+ 0.8359659314155579,
773
+ 0.9189376831054688,
774
+ 1.0,
775
+ 0.8505204916000366,
776
+ 0.6733924150466919,
777
+ 3.9892434870125726e-05,
778
+ 0.9999310970306396,
779
+ 0.9999845027923584,
780
+ 0.9993377327919006,
781
+ 0.9925682544708252,
782
+ 0.9994389414787292,
783
+ 0.9971237778663635,
784
+ 1.0,
785
+ 1.0,
786
+ 0.9999998807907104
787
+ ],
788
+ "eval_math_hard_runtime": 86.0452,
789
+ "eval_math_hard_samples_per_second": 1.162,
790
+ "eval_math_hard_steps_per_second": 0.046,
791
+ "epoch": 1.32,
792
+ "step": 75
793
+ },
794
+ {
795
+ "loss": 0.0338,
796
+ "learning_rate": 0.0002,
797
+ "epoch": 1.33,
798
+ "step": 76
799
+ },
800
+ {
801
+ "loss": 0.0115,
802
+ "learning_rate": 0.0002,
803
+ "epoch": 1.35,
804
+ "step": 77
805
+ },
806
+ {
807
+ "loss": 0.0876,
808
+ "learning_rate": 0.0002,
809
+ "epoch": 1.37,
810
+ "step": 78
811
+ },
812
+ {
813
+ "loss": 0.0071,
814
+ "learning_rate": 0.0002,
815
+ "epoch": 1.39,
816
+ "step": 79
817
+ },
818
+ {
819
+ "loss": 0.0739,
820
+ "learning_rate": 0.0002,
821
+ "epoch": 1.4,
822
+ "step": 80
823
+ },
824
+ {
825
+ "loss": 0.0016,
826
+ "learning_rate": 0.0002,
827
+ "epoch": 1.42,
828
+ "step": 81
829
+ },
830
+ {
831
+ "loss": 0.1257,
832
+ "learning_rate": 0.0002,
833
+ "epoch": 1.44,
834
+ "step": 82
835
+ },
836
+ {
837
+ "loss": 0.0048,
838
+ "learning_rate": 0.0002,
839
+ "epoch": 1.46,
840
+ "step": 83
841
+ },
842
+ {
843
+ "loss": 0.007,
844
+ "learning_rate": 0.0002,
845
+ "epoch": 1.47,
846
+ "step": 84
847
+ },
848
+ {
849
+ "loss": 0.0162,
850
+ "learning_rate": 0.0002,
851
+ "epoch": 1.49,
852
+ "step": 85
853
+ },
854
+ {
855
+ "loss": 0.042,
856
+ "learning_rate": 0.0002,
857
+ "epoch": 1.51,
858
+ "step": 86
859
+ },
860
+ {
861
+ "loss": 0.0231,
862
+ "learning_rate": 0.0002,
863
+ "epoch": 1.53,
864
+ "step": 87
865
+ },
866
+ {
867
+ "loss": 0.0103,
868
+ "learning_rate": 0.0002,
869
+ "epoch": 1.54,
870
+ "step": 88
871
+ },
872
+ {
873
+ "loss": 0.0077,
874
+ "learning_rate": 0.0002,
875
+ "epoch": 1.56,
876
+ "step": 89
877
+ },
878
+ {
879
+ "loss": 0.0014,
880
+ "learning_rate": 0.0002,
881
+ "epoch": 1.58,
882
+ "step": 90
883
+ },
884
+ {
885
+ "loss": 0.014,
886
+ "learning_rate": 0.0002,
887
+ "epoch": 1.6,
888
+ "step": 91
889
+ },
890
+ {
891
+ "loss": 0.0054,
892
+ "learning_rate": 0.0002,
893
+ "epoch": 1.61,
894
+ "step": 92
895
+ },
896
+ {
897
+ "loss": 0.0349,
898
+ "learning_rate": 0.0002,
899
+ "epoch": 1.63,
900
+ "step": 93
901
+ },
902
+ {
903
+ "loss": 0.0961,
904
+ "learning_rate": 0.0002,
905
+ "epoch": 1.65,
906
+ "step": 94
907
+ },
908
+ {
909
+ "loss": 0.0058,
910
+ "learning_rate": 0.0002,
911
+ "epoch": 1.67,
912
+ "step": 95
913
+ },
914
+ {
915
+ "loss": 0.0311,
916
+ "learning_rate": 0.0002,
917
+ "epoch": 1.68,
918
+ "step": 96
919
+ },
920
+ {
921
+ "loss": 0.0143,
922
+ "learning_rate": 0.0002,
923
+ "epoch": 1.7,
924
+ "step": 97
925
+ },
926
+ {
927
+ "loss": 0.0007,
928
+ "learning_rate": 0.0002,
929
+ "epoch": 1.72,
930
+ "step": 98
931
+ },
932
+ {
933
+ "loss": 0.1104,
934
+ "learning_rate": 0.0002,
935
+ "epoch": 1.74,
936
+ "step": 99
937
+ },
938
+ {
939
+ "loss": 0.0337,
940
+ "learning_rate": 0.0002,
941
+ "epoch": 1.75,
942
+ "step": 100
943
+ },
944
+ {
945
+ "eval_math_hard_loss": 0.7644428014755249,
946
+ "eval_math_hard_score": -0.1077132597565651,
947
+ "eval_math_hard_brier_score": 0.1077132597565651,
948
+ "eval_math_hard_average_probability": 0.8629195690155029,
949
+ "eval_math_hard_accuracy": 0.88,
950
+ "eval_math_hard_probabilities": [
951
+ 0.01672576367855072,
952
+ 0.0001243828737642616,
953
+ 0.09437736123800278,
954
+ 1.0,
955
+ 1.0,
956
+ 0.9999958276748657,
957
+ 1.0,
958
+ 1.0,
959
+ 0.9949105381965637,
960
+ 0.9999998807907104,
961
+ 0.9999997615814209,
962
+ 0.9999969005584717,
963
+ 0.9774806499481201,
964
+ 0.9809224009513855,
965
+ 0.06269947439432144,
966
+ 0.9999384880065918,
967
+ 1.0,
968
+ 1.0,
969
+ 0.7058337330818176,
970
+ 0.6948679089546204,
971
+ 0.33711233735084534,
972
+ 1.0,
973
+ 1.0,
974
+ 1.0,
975
+ 1.0,
976
+ 1.0,
977
+ 1.0,
978
+ 1.0,
979
+ 1.0,
980
+ 1.0,
981
+ 1.0,
982
+ 1.0,
983
+ 1.0,
984
+ 1.0,
985
+ 1.0,
986
+ 1.0,
987
+ 1.0,
988
+ 1.0,
989
+ 1.0,
990
+ 1.0,
991
+ 1.0,
992
+ 0.13868287205696106,
993
+ 0.9946752786636353,
994
+ 0.7222106456756592,
995
+ 0.864886999130249,
996
+ 1.0,
997
+ 1.0,
998
+ 0.9999991655349731,
999
+ 0.9912755489349365,
1000
+ 0.07769276201725006,
1001
+ 0.996044397354126,
1002
+ 1.0,
1003
+ 1.0,
1004
+ 1.0,
1005
+ 1.0,
1006
+ 1.0,
1007
+ 1.0,
1008
+ 5.921328556723893e-05,
1009
+ 0.054774124175310135,
1010
+ 0.8164430856704712,
1011
+ 1.0,
1012
+ 1.0,
1013
+ 0.538241446018219,
1014
+ 0.9993982315063477,
1015
+ 0.9931572079658508,
1016
+ 0.999657154083252,
1017
+ 1.0,
1018
+ 1.0,
1019
+ 1.0,
1020
+ 0.7599838376045227,
1021
+ 0.9545750617980957,
1022
+ 0.9728327989578247,
1023
+ 1.0,
1024
+ 1.0,
1025
+ 1.0,
1026
+ 1.0,
1027
+ 1.0,
1028
+ 1.0,
1029
+ 0.9995208978652954,
1030
+ 0.7313249111175537,
1031
+ 0.005399945192039013,
1032
+ 0.5453264713287354,
1033
+ 0.9951422214508057,
1034
+ 0.9464260935783386,
1035
+ 0.43800926208496094,
1036
+ 0.897794246673584,
1037
+ 0.9954219460487366,
1038
+ 1.0,
1039
+ 0.9996411800384521,
1040
+ 0.9987420439720154,
1041
+ 1.534920940349238e-10,
1042
+ 0.9999021291732788,
1043
+ 0.9999223947525024,
1044
+ 0.9999954700469971,
1045
+ 0.9998113512992859,
1046
+ 0.9999969005584717,
1047
+ 0.9999964237213135,
1048
+ 1.0,
1049
+ 1.0,
1050
+ 1.0
1051
+ ],
1052
+ "eval_math_hard_runtime": 86.022,
1053
+ "eval_math_hard_samples_per_second": 1.162,
1054
+ "eval_math_hard_steps_per_second": 0.046,
1055
+ "epoch": 1.75,
1056
+ "step": 100
1057
+ },
1058
+ {
1059
+ "train_runtime": 8937.8559,
1060
+ "train_samples_per_second": 0.358,
1061
+ "train_steps_per_second": 0.011,
1062
+ "total_flos": 0.0,
1063
+ "train_loss": 0.18627140538301318,
1064
+ "epoch": 1.75,
1065
+ "step": 100
1066
+ }
1067
+ ]