Hugging Face's logo

Datasets:
lm-harmony
/
results

Dataset card Files
xet
Community
1
ghzhang commited on
Commit
5795f4d
·
1 Parent(s): 572d8d4

first commit

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. .gitignore +1 -0
  2. 01-ai/Yi-1.5-6B-Chat/results_2026-01-05-04-27-51.json +513 -0
  3. 01-ai/Yi-1.5-6B/results_2026-01-05-04-27-50.json +513 -0
  4. 01-ai/Yi-1.5-9B-Chat/results_2026-01-05-04-27-51.json +513 -0
  5. 01-ai/Yi-1.5-9B/results_2026-01-05-04-27-50.json +513 -0
  6. 01-ai/Yi-6B-Chat/results_2026-01-05-04-27-51.json +513 -0
  7. 01-ai/Yi-6B/results_2026-01-05-04-27-50.json +513 -0
  8. 01-ai/Yi-9B/results_2026-01-05-04-27-50.json +513 -0
  9. EleutherAI/pythia-1.4b-deduped/results_2026-01-05-04-27-50.json +513 -0
  10. EleutherAI/pythia-12b/results_2026-01-05-04-27-50.json +513 -0
  11. EleutherAI/pythia-160m-deduped/results_2026-01-05-04-27-50.json +513 -0
  12. EleutherAI/pythia-1b-deduped/results_2026-01-05-04-27-50.json +513 -0
  13. EleutherAI/pythia-2.8b-deduped/results_2026-01-05-04-27-50.json +513 -0
  14. EleutherAI/pythia-410m-deduped/results_2026-01-05-04-27-50.json +513 -0
  15. EleutherAI/pythia-6.9b-deduped/results_2026-01-05-04-27-50.json +513 -0
  16. EleutherAI/pythia-70m-deduped/results_2026-01-05-04-27-50.json +513 -0
  17. HuggingFaceTB/SmolLM3-3B-Base/results_2026-01-05-04-27-50.json +513 -0
  18. HuggingFaceTB/SmolLM3-3B/results_2026-01-05-04-27-51.json +513 -0
  19. LumiOpen/Llama-Poro-2-8B-Instruct/results_2026-01-05-04-27-51.json +513 -0
  20. LumiOpen/Llama-Poro-2-8B-base/results_2026-01-05-04-27-51.json +513 -0
  21. Qwen/Qwen1.5-0.5B-Chat/results_2026-01-05-04-27-51.json +513 -0
  22. Qwen/Qwen1.5-0.5B/results_2026-01-05-04-27-50.json +513 -0
  23. Qwen/Qwen1.5-1.8B-Chat/results_2026-01-05-04-27-51.json +513 -0
  24. Qwen/Qwen1.5-1.8B/results_2026-01-05-04-27-50.json +513 -0
  25. Qwen/Qwen1.5-14B-Chat/results_2026-01-05-04-27-51.json +513 -0
  26. Qwen/Qwen1.5-14B/results_2026-01-05-04-27-50.json +513 -0
  27. Qwen/Qwen1.5-4B-Chat/results_2026-01-05-04-27-51.json +513 -0
  28. Qwen/Qwen1.5-4B/results_2026-01-05-04-27-50.json +513 -0
  29. Qwen/Qwen1.5-7B-Chat/results_2026-01-05-04-27-51.json +513 -0
  30. Qwen/Qwen1.5-7B/results_2026-01-05-04-27-50.json +513 -0
  31. Qwen/Qwen2-0.5B-Instruct/results_2026-01-05-04-27-51.json +513 -0
  32. Qwen/Qwen2-0.5B/results_2026-01-05-04-27-50.json +513 -0
  33. Qwen/Qwen2-1.5B-Instruct/results_2026-01-05-04-27-51.json +513 -0
  34. Qwen/Qwen2-1.5B/results_2026-01-05-04-27-50.json +513 -0
  35. Qwen/Qwen2-7B-Instruct/results_2026-01-05-04-27-51.json +513 -0
  36. Qwen/Qwen2-7B/results_2026-01-05-04-27-50.json +513 -0
  37. Qwen/Qwen2.5-0.5B-Instruct/results_2026-01-05-04-27-51.json +513 -0
  38. Qwen/Qwen2.5-0.5B/results_2026-01-05-04-27-50.json +513 -0
  39. Qwen/Qwen2.5-1.5B-Instruct/results_2026-01-05-04-27-51.json +513 -0
  40. Qwen/Qwen2.5-1.5B/results_2026-01-05-04-27-50.json +513 -0
  41. Qwen/Qwen2.5-14B-Instruct/results_2026-01-05-04-27-51.json +513 -0
  42. Qwen/Qwen2.5-14B/results_2026-01-05-04-27-50.json +513 -0
  43. Qwen/Qwen2.5-3B-Instruct/results_2026-01-05-04-27-51.json +513 -0
  44. Qwen/Qwen2.5-3B/results_2026-01-05-04-27-50.json +513 -0
  45. Qwen/Qwen2.5-7B-Instruct/results_2026-01-05-04-27-51.json +513 -0
  46. Qwen/Qwen2.5-7B/results_2026-01-05-04-27-50.json +513 -0
  47. Qwen/Qwen3-0.6B-Base/results_2026-01-05-04-27-50.json +513 -0
  48. Qwen/Qwen3-0.6B/results_2026-01-05-04-27-51.json +513 -0
  49. Qwen/Qwen3-1.7B-Base/results_2026-01-05-04-27-50.json +513 -0
  50. Qwen/Qwen3-1.7B/results_2026-01-05-04-27-51.json +513 -0
.gitignore ADDED
@@ -0,0 +1 @@
 
 
1
+ *DS_Store
01-ai/Yi-1.5-6B-Chat/results_2026-01-05-04-27-51.json ADDED
@@ -0,0 +1,513 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "model_name": "01-ai/Yi-1.5-6B-Chat",
4
+ "model_type": "instruction-tuned",
5
+ "params": 6.0
6
+ },
7
+ "results": {
8
+ "medmcqa": {
9
+ "acc_norm": 0.4864929476452307,
10
+ "acc_norm_stderr": 0.0077289316463684445,
11
+ "args": {
12
+ "exp_name": "results/v1/medmcqa/01-ai/Yi-1.5-6B-Chat",
13
+ "task_name": "medmcqa",
14
+ "model_name": "01-ai/Yi-1.5-6B-Chat",
15
+ "dataset_param": {
16
+ "max_num_train": 50000,
17
+ "completion_only": false,
18
+ "apply_chat_template": true
19
+ },
20
+ "resume_from_checkpoint": false,
21
+ "train_param": {
22
+ "max_length": 1024,
23
+ "per_device_train_batch_size": 8,
24
+ "auto_find_batch_size": true,
25
+ "num_train_epochs": 1,
26
+ "max_steps": -1,
27
+ "logging_strategy": "steps",
28
+ "save_strategy": "steps",
29
+ "logging_steps": 0.2,
30
+ "save_steps": 0.2,
31
+ "gradient_accumulation_steps": 1,
32
+ "gradient_checkpointing": false,
33
+ "optim": "adamw_torch",
34
+ "learning_rate": 5e-05,
35
+ "weight_decay": 0.01,
36
+ "warmup_steps": 0,
37
+ "report_to": "tensorboard",
38
+ "save_safetensors": true,
39
+ "packing": false
40
+ },
41
+ "use_peft": true,
42
+ "peft_task_type": "CAUSAL_LM",
43
+ "peft_inference_mode": false,
44
+ "peft_r": 8,
45
+ "peft_lora_alpha": 32,
46
+ "peft_lora_dropout": 0.1,
47
+ "fsdp": "none",
48
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
49
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
50
+ "fsdp_fsdp_min_num_params": 100000000.0,
51
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
52
+ "fsdp_fsdp_forward_prefetch": false,
53
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
54
+ "fsdp_fsdp_offload_params": false,
55
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
56
+ "fsdp_fsdp_use_orig_params": true,
57
+ "fsdp_fsdp_activation_checkpointing": false,
58
+ "fsdp_fsdp_limit_all_gathers": true,
59
+ "fsdp_fsdp_sync_module_states": true,
60
+ "use_git": true,
61
+ "random_seed": 0
62
+ }
63
+ },
64
+ "nq_open": {
65
+ "exact_match,remove_whitespace": 0.22077562326869807,
66
+ "exact_match,remove_whitespace_stderr": 0.00690420032379699,
67
+ "args": {
68
+ "exp_name": "results/v1/nq_open/01-ai/Yi-1.5-6B-Chat",
69
+ "task_name": "nq_open",
70
+ "model_name": "01-ai/Yi-1.5-6B-Chat",
71
+ "dataset_param": {
72
+ "max_num_train": 50000,
73
+ "completion_only": false,
74
+ "apply_chat_template": true
75
+ },
76
+ "resume_from_checkpoint": false,
77
+ "train_param": {
78
+ "max_length": 1024,
79
+ "per_device_train_batch_size": 8,
80
+ "auto_find_batch_size": true,
81
+ "num_train_epochs": 1,
82
+ "max_steps": -1,
83
+ "logging_strategy": "steps",
84
+ "save_strategy": "steps",
85
+ "logging_steps": 0.2,
86
+ "save_steps": 0.2,
87
+ "gradient_accumulation_steps": 1,
88
+ "gradient_checkpointing": false,
89
+ "optim": "adamw_torch",
90
+ "learning_rate": 5e-05,
91
+ "weight_decay": 0.01,
92
+ "warmup_steps": 0,
93
+ "report_to": "tensorboard",
94
+ "save_safetensors": true,
95
+ "packing": false
96
+ },
97
+ "use_peft": true,
98
+ "peft_task_type": "CAUSAL_LM",
99
+ "peft_inference_mode": false,
100
+ "peft_r": 8,
101
+ "peft_lora_alpha": 32,
102
+ "peft_lora_dropout": 0.1,
103
+ "fsdp": "none",
104
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
105
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
106
+ "fsdp_fsdp_min_num_params": 100000000.0,
107
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
108
+ "fsdp_fsdp_forward_prefetch": false,
109
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
110
+ "fsdp_fsdp_offload_params": false,
111
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
112
+ "fsdp_fsdp_use_orig_params": true,
113
+ "fsdp_fsdp_activation_checkpointing": false,
114
+ "fsdp_fsdp_limit_all_gathers": true,
115
+ "fsdp_fsdp_sync_module_states": true,
116
+ "use_git": true,
117
+ "random_seed": 0
118
+ }
119
+ },
120
+ "winogrande": {
121
+ "acc": 0.7853196527229677,
122
+ "acc_stderr": 0.011539912734345388,
123
+ "args": {
124
+ "exp_name": "results/v1/winogrande/01-ai/Yi-1.5-6B-Chat",
125
+ "task_name": "winogrande",
126
+ "model_name": "01-ai/Yi-1.5-6B-Chat",
127
+ "dataset_param": {
128
+ "max_num_train": 50000,
129
+ "completion_only": false,
130
+ "apply_chat_template": true
131
+ },
132
+ "resume_from_checkpoint": false,
133
+ "train_param": {
134
+ "max_length": 1024,
135
+ "per_device_train_batch_size": 8,
136
+ "auto_find_batch_size": true,
137
+ "num_train_epochs": 1,
138
+ "max_steps": -1,
139
+ "logging_strategy": "steps",
140
+ "save_strategy": "steps",
141
+ "logging_steps": 0.2,
142
+ "save_steps": 0.2,
143
+ "gradient_accumulation_steps": 1,
144
+ "gradient_checkpointing": false,
145
+ "optim": "adamw_torch",
146
+ "learning_rate": 5e-05,
147
+ "weight_decay": 0.01,
148
+ "warmup_steps": 0,
149
+ "report_to": "tensorboard",
150
+ "save_safetensors": true,
151
+ "packing": false
152
+ },
153
+ "use_peft": true,
154
+ "peft_task_type": "CAUSAL_LM",
155
+ "peft_inference_mode": false,
156
+ "peft_r": 8,
157
+ "peft_lora_alpha": 32,
158
+ "peft_lora_dropout": 0.1,
159
+ "fsdp": "none",
160
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
161
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
162
+ "fsdp_fsdp_min_num_params": 100000000.0,
163
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
164
+ "fsdp_fsdp_forward_prefetch": false,
165
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
166
+ "fsdp_fsdp_offload_params": false,
167
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
168
+ "fsdp_fsdp_use_orig_params": true,
169
+ "fsdp_fsdp_activation_checkpointing": false,
170
+ "fsdp_fsdp_limit_all_gathers": true,
171
+ "fsdp_fsdp_sync_module_states": true,
172
+ "use_git": true,
173
+ "random_seed": 0
174
+ }
175
+ },
176
+ "hellaswag": {
177
+ "acc_norm": 0.7854,
178
+ "acc_norm_stderr": 0.004105650939493204,
179
+ "args": {
180
+ "exp_name": "results/v1/hellaswag/01-ai/Yi-1.5-6B-Chat",
181
+ "task_name": "hellaswag",
182
+ "model_name": "01-ai/Yi-1.5-6B-Chat",
183
+ "dataset_param": {
184
+ "max_num_train": 50000,
185
+ "completion_only": false,
186
+ "apply_chat_template": true
187
+ },
188
+ "resume_from_checkpoint": false,
189
+ "train_param": {
190
+ "max_length": 1024,
191
+ "per_device_train_batch_size": 8,
192
+ "auto_find_batch_size": true,
193
+ "num_train_epochs": 1,
194
+ "max_steps": -1,
195
+ "logging_strategy": "steps",
196
+ "save_strategy": "steps",
197
+ "logging_steps": 0.2,
198
+ "save_steps": 0.2,
199
+ "gradient_accumulation_steps": 1,
200
+ "gradient_checkpointing": false,
201
+ "optim": "adamw_torch",
202
+ "learning_rate": 5e-05,
203
+ "weight_decay": 0.01,
204
+ "warmup_steps": 0,
205
+ "report_to": "tensorboard",
206
+ "save_safetensors": true,
207
+ "packing": false
208
+ },
209
+ "use_peft": true,
210
+ "peft_task_type": "CAUSAL_LM",
211
+ "peft_inference_mode": false,
212
+ "peft_r": 8,
213
+ "peft_lora_alpha": 32,
214
+ "peft_lora_dropout": 0.1,
215
+ "fsdp": "none",
216
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
217
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
218
+ "fsdp_fsdp_min_num_params": 100000000.0,
219
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
220
+ "fsdp_fsdp_forward_prefetch": false,
221
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
222
+ "fsdp_fsdp_offload_params": false,
223
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
224
+ "fsdp_fsdp_use_orig_params": true,
225
+ "fsdp_fsdp_activation_checkpointing": false,
226
+ "fsdp_fsdp_limit_all_gathers": true,
227
+ "fsdp_fsdp_sync_module_states": true,
228
+ "use_git": true,
229
+ "random_seed": 0
230
+ }
231
+ },
232
+ "social_iqa": {
233
+ "acc": 0.5634595701125895,
234
+ "acc_stderr": 0.011222574420844783,
235
+ "args": {
236
+ "exp_name": "results/v1/social_iqa/01-ai/Yi-1.5-6B-Chat",
237
+ "task_name": "social_iqa",
238
+ "model_name": "01-ai/Yi-1.5-6B-Chat",
239
+ "dataset_param": {
240
+ "max_num_train": 50000,
241
+ "completion_only": false,
242
+ "apply_chat_template": true
243
+ },
244
+ "resume_from_checkpoint": false,
245
+ "train_param": {
246
+ "max_length": 1024,
247
+ "per_device_train_batch_size": 8,
248
+ "auto_find_batch_size": true,
249
+ "num_train_epochs": 1,
250
+ "max_steps": -1,
251
+ "logging_strategy": "steps",
252
+ "save_strategy": "steps",
253
+ "logging_steps": 0.2,
254
+ "save_steps": 0.2,
255
+ "gradient_accumulation_steps": 1,
256
+ "gradient_checkpointing": false,
257
+ "optim": "adamw_torch",
258
+ "learning_rate": 5e-05,
259
+ "weight_decay": 0.01,
260
+ "warmup_steps": 0,
261
+ "report_to": "tensorboard",
262
+ "save_safetensors": true,
263
+ "packing": false
264
+ },
265
+ "use_peft": true,
266
+ "peft_task_type": "CAUSAL_LM",
267
+ "peft_inference_mode": false,
268
+ "peft_r": 8,
269
+ "peft_lora_alpha": 32,
270
+ "peft_lora_dropout": 0.1,
271
+ "fsdp": "none",
272
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
273
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
274
+ "fsdp_fsdp_min_num_params": 100000000.0,
275
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
276
+ "fsdp_fsdp_forward_prefetch": false,
277
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
278
+ "fsdp_fsdp_offload_params": false,
279
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
280
+ "fsdp_fsdp_use_orig_params": true,
281
+ "fsdp_fsdp_activation_checkpointing": false,
282
+ "fsdp_fsdp_limit_all_gathers": true,
283
+ "fsdp_fsdp_sync_module_states": true,
284
+ "use_git": true,
285
+ "random_seed": 0
286
+ }
287
+ },
288
+ "piqa": {
289
+ "acc_norm": 0.8193688792165397,
290
+ "acc_norm_stderr": 0.008975971037338047,
291
+ "args": {
292
+ "exp_name": "results/v2/piqa/01-ai/Yi-1.5-6B-Chat",
293
+ "task_name": "piqa",
294
+ "model_name": "01-ai/Yi-1.5-6B-Chat",
295
+ "dataset_param": {
296
+ "max_num_train": 50000,
297
+ "completion_only": false,
298
+ "apply_chat_template": true
299
+ },
300
+ "resume_from_checkpoint": false,
301
+ "train_param": {
302
+ "max_length": 1024,
303
+ "per_device_train_batch_size": 8,
304
+ "auto_find_batch_size": true,
305
+ "num_train_epochs": 5,
306
+ "max_steps": -1,
307
+ "logging_strategy": "epoch",
308
+ "save_strategy": "epoch",
309
+ "logging_steps": 0.2,
310
+ "save_steps": 0.2,
311
+ "gradient_accumulation_steps": 1,
312
+ "gradient_checkpointing": false,
313
+ "optim": "adamw_torch",
314
+ "learning_rate": 5e-05,
315
+ "weight_decay": 0.01,
316
+ "warmup_steps": 0,
317
+ "report_to": "tensorboard",
318
+ "save_safetensors": true,
319
+ "packing": false
320
+ },
321
+ "use_peft": true,
322
+ "peft_task_type": "CAUSAL_LM",
323
+ "peft_inference_mode": false,
324
+ "peft_r": 8,
325
+ "peft_lora_alpha": 32,
326
+ "peft_lora_dropout": 0.1,
327
+ "fsdp": "none",
328
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
329
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
330
+ "fsdp_fsdp_min_num_params": 100000000.0,
331
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
332
+ "fsdp_fsdp_forward_prefetch": false,
333
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
334
+ "fsdp_fsdp_offload_params": false,
335
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
336
+ "fsdp_fsdp_use_orig_params": true,
337
+ "fsdp_fsdp_activation_checkpointing": false,
338
+ "fsdp_fsdp_limit_all_gathers": true,
339
+ "fsdp_fsdp_sync_module_states": true,
340
+ "use_git": true,
341
+ "random_seed": 0
342
+ }
343
+ },
344
+ "commonsense_qa": {
345
+ "acc": 0.800982800982801,
346
+ "acc_stderr": 0.011430809442838401,
347
+ "args": {
348
+ "exp_name": "results/v2/commonsense_qa/01-ai/Yi-1.5-6B-Chat",
349
+ "task_name": "commonsense_qa",
350
+ "model_name": "01-ai/Yi-1.5-6B-Chat",
351
+ "dataset_param": {
352
+ "max_num_train": 50000,
353
+ "completion_only": false,
354
+ "apply_chat_template": true
355
+ },
356
+ "resume_from_checkpoint": false,
357
+ "train_param": {
358
+ "max_length": 1024,
359
+ "per_device_train_batch_size": 8,
360
+ "auto_find_batch_size": true,
361
+ "num_train_epochs": 5,
362
+ "max_steps": -1,
363
+ "logging_strategy": "epoch",
364
+ "save_strategy": "epoch",
365
+ "logging_steps": 0.2,
366
+ "save_steps": 0.2,
367
+ "gradient_accumulation_steps": 1,
368
+ "gradient_checkpointing": false,
369
+ "optim": "adamw_torch",
370
+ "learning_rate": 5e-05,
371
+ "weight_decay": 0.01,
372
+ "warmup_steps": 0,
373
+ "report_to": "tensorboard",
374
+ "save_safetensors": true,
375
+ "packing": false
376
+ },
377
+ "use_peft": true,
378
+ "peft_task_type": "CAUSAL_LM",
379
+ "peft_inference_mode": false,
380
+ "peft_r": 8,
381
+ "peft_lora_alpha": 32,
382
+ "peft_lora_dropout": 0.1,
383
+ "fsdp": "none",
384
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
385
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
386
+ "fsdp_fsdp_min_num_params": 100000000.0,
387
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
388
+ "fsdp_fsdp_forward_prefetch": false,
389
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
390
+ "fsdp_fsdp_offload_params": false,
391
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
392
+ "fsdp_fsdp_use_orig_params": true,
393
+ "fsdp_fsdp_activation_checkpointing": false,
394
+ "fsdp_fsdp_limit_all_gathers": true,
395
+ "fsdp_fsdp_sync_module_states": true,
396
+ "use_git": true,
397
+ "random_seed": 0
398
+ }
399
+ },
400
+ "gsm8k": {
401
+ "exact_match,flexible-extract": 0.5572403335860501,
402
+ "exact_match,flexible-extract_stderr": 0.01368193719176463,
403
+ "args": {
404
+ "exp_name": "results/v2/gsm8k/01-ai/Yi-1.5-6B-Chat",
405
+ "task_name": "gsm8k",
406
+ "model_name": "01-ai/Yi-1.5-6B-Chat",
407
+ "dataset_param": {
408
+ "max_num_train": 50000,
409
+ "completion_only": false,
410
+ "apply_chat_template": true
411
+ },
412
+ "resume_from_checkpoint": false,
413
+ "train_param": {
414
+ "max_length": 1024,
415
+ "per_device_train_batch_size": 8,
416
+ "auto_find_batch_size": true,
417
+ "num_train_epochs": 5,
418
+ "max_steps": -1,
419
+ "logging_strategy": "epoch",
420
+ "save_strategy": "epoch",
421
+ "logging_steps": 0.2,
422
+ "save_steps": 0.2,
423
+ "gradient_accumulation_steps": 1,
424
+ "gradient_checkpointing": false,
425
+ "optim": "adamw_torch",
426
+ "learning_rate": 5e-05,
427
+ "weight_decay": 0.01,
428
+ "warmup_steps": 0,
429
+ "report_to": "tensorboard",
430
+ "save_safetensors": true,
431
+ "packing": false
432
+ },
433
+ "use_peft": true,
434
+ "peft_task_type": "CAUSAL_LM",
435
+ "peft_inference_mode": false,
436
+ "peft_r": 8,
437
+ "peft_lora_alpha": 32,
438
+ "peft_lora_dropout": 0.1,
439
+ "fsdp": "none",
440
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
441
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
442
+ "fsdp_fsdp_min_num_params": 100000000.0,
443
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
444
+ "fsdp_fsdp_forward_prefetch": false,
445
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
446
+ "fsdp_fsdp_offload_params": false,
447
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
448
+ "fsdp_fsdp_use_orig_params": true,
449
+ "fsdp_fsdp_activation_checkpointing": false,
450
+ "fsdp_fsdp_limit_all_gathers": true,
451
+ "fsdp_fsdp_sync_module_states": true,
452
+ "use_git": true,
453
+ "random_seed": 0
454
+ }
455
+ },
456
+ "arc_challenge": {
457
+ "acc_norm": 0.5580204778156996,
458
+ "acc_norm_stderr": 0.014512682523128345,
459
+ "args": {
460
+ "exp_name": "results/v2/arc_challenge/01-ai/Yi-1.5-6B-Chat",
461
+ "task_name": "arc_challenge",
462
+ "model_name": "01-ai/Yi-1.5-6B-Chat",
463
+ "dataset_param": {
464
+ "max_num_train": 50000,
465
+ "completion_only": false,
466
+ "apply_chat_template": true
467
+ },
468
+ "resume_from_checkpoint": false,
469
+ "train_param": {
470
+ "max_length": 1024,
471
+ "per_device_train_batch_size": 8,
472
+ "auto_find_batch_size": true,
473
+ "num_train_epochs": 5,
474
+ "max_steps": -1,
475
+ "logging_strategy": "epoch",
476
+ "save_strategy": "epoch",
477
+ "logging_steps": 0.2,
478
+ "save_steps": 0.2,
479
+ "gradient_accumulation_steps": 1,
480
+ "gradient_checkpointing": false,
481
+ "optim": "adamw_torch",
482
+ "learning_rate": 5e-05,
483
+ "weight_decay": 0.01,
484
+ "warmup_steps": 0,
485
+ "report_to": "tensorboard",
486
+ "save_safetensors": true,
487
+ "packing": false
488
+ },
489
+ "use_peft": true,
490
+ "peft_task_type": "CAUSAL_LM",
491
+ "peft_inference_mode": false,
492
+ "peft_r": 8,
493
+ "peft_lora_alpha": 32,
494
+ "peft_lora_dropout": 0.1,
495
+ "fsdp": "none",
496
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
497
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
498
+ "fsdp_fsdp_min_num_params": 100000000.0,
499
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
500
+ "fsdp_fsdp_forward_prefetch": false,
501
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
502
+ "fsdp_fsdp_offload_params": false,
503
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
504
+ "fsdp_fsdp_use_orig_params": true,
505
+ "fsdp_fsdp_activation_checkpointing": false,
506
+ "fsdp_fsdp_limit_all_gathers": true,
507
+ "fsdp_fsdp_sync_module_states": true,
508
+ "use_git": true,
509
+ "random_seed": 0
510
+ }
511
+ }
512
+ }
513
+ }
01-ai/Yi-1.5-6B/results_2026-01-05-04-27-50.json ADDED
@@ -0,0 +1,513 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "model_name": "01-ai/Yi-1.5-6B",
4
+ "model_type": "pretrained",
5
+ "params": 6.0
6
+ },
7
+ "results": {
8
+ "medmcqa": {
9
+ "acc_norm": 0.5058570404016256,
10
+ "acc_norm_stderr": 0.007731222837978417,
11
+ "args": {
12
+ "exp_name": "results/v1/medmcqa/01-ai/Yi-1.5-6B",
13
+ "task_name": "medmcqa",
14
+ "model_name": "01-ai/Yi-1.5-6B",
15
+ "dataset_param": {
16
+ "max_num_train": 50000,
17
+ "completion_only": false,
18
+ "apply_chat_template": false
19
+ },
20
+ "resume_from_checkpoint": false,
21
+ "train_param": {
22
+ "max_length": 1024,
23
+ "per_device_train_batch_size": 8,
24
+ "auto_find_batch_size": true,
25
+ "num_train_epochs": 1,
26
+ "max_steps": -1,
27
+ "logging_strategy": "steps",
28
+ "save_strategy": "steps",
29
+ "logging_steps": 0.2,
30
+ "save_steps": 0.2,
31
+ "gradient_accumulation_steps": 1,
32
+ "gradient_checkpointing": false,
33
+ "optim": "adamw_torch",
34
+ "learning_rate": 5e-05,
35
+ "weight_decay": 0.01,
36
+ "warmup_steps": 0,
37
+ "report_to": "tensorboard",
38
+ "save_safetensors": true,
39
+ "packing": false
40
+ },
41
+ "use_peft": true,
42
+ "peft_task_type": "CAUSAL_LM",
43
+ "peft_inference_mode": false,
44
+ "peft_r": 8,
45
+ "peft_lora_alpha": 32,
46
+ "peft_lora_dropout": 0.1,
47
+ "fsdp": "none",
48
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
49
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
50
+ "fsdp_fsdp_min_num_params": 100000000.0,
51
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
52
+ "fsdp_fsdp_forward_prefetch": false,
53
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
54
+ "fsdp_fsdp_offload_params": false,
55
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
56
+ "fsdp_fsdp_use_orig_params": true,
57
+ "fsdp_fsdp_activation_checkpointing": false,
58
+ "fsdp_fsdp_limit_all_gathers": true,
59
+ "fsdp_fsdp_sync_module_states": true,
60
+ "use_git": true,
61
+ "random_seed": 0
62
+ }
63
+ },
64
+ "nq_open": {
65
+ "exact_match,remove_whitespace": 0.26869806094182824,
66
+ "exact_match,remove_whitespace_stderr": 0.007378824611197684,
67
+ "args": {
68
+ "exp_name": "results/v1/nq_open/01-ai/Yi-1.5-6B",
69
+ "task_name": "nq_open",
70
+ "model_name": "01-ai/Yi-1.5-6B",
71
+ "dataset_param": {
72
+ "max_num_train": 50000,
73
+ "completion_only": false,
74
+ "apply_chat_template": false
75
+ },
76
+ "resume_from_checkpoint": false,
77
+ "train_param": {
78
+ "max_length": 1024,
79
+ "per_device_train_batch_size": 8,
80
+ "auto_find_batch_size": true,
81
+ "num_train_epochs": 1,
82
+ "max_steps": -1,
83
+ "logging_strategy": "steps",
84
+ "save_strategy": "steps",
85
+ "logging_steps": 0.2,
86
+ "save_steps": 0.2,
87
+ "gradient_accumulation_steps": 1,
88
+ "gradient_checkpointing": false,
89
+ "optim": "adamw_torch",
90
+ "learning_rate": 5e-05,
91
+ "weight_decay": 0.01,
92
+ "warmup_steps": 0,
93
+ "report_to": "tensorboard",
94
+ "save_safetensors": true,
95
+ "packing": false
96
+ },
97
+ "use_peft": true,
98
+ "peft_task_type": "CAUSAL_LM",
99
+ "peft_inference_mode": false,
100
+ "peft_r": 8,
101
+ "peft_lora_alpha": 32,
102
+ "peft_lora_dropout": 0.1,
103
+ "fsdp": "none",
104
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
105
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
106
+ "fsdp_fsdp_min_num_params": 100000000.0,
107
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
108
+ "fsdp_fsdp_forward_prefetch": false,
109
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
110
+ "fsdp_fsdp_offload_params": false,
111
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
112
+ "fsdp_fsdp_use_orig_params": true,
113
+ "fsdp_fsdp_activation_checkpointing": false,
114
+ "fsdp_fsdp_limit_all_gathers": true,
115
+ "fsdp_fsdp_sync_module_states": true,
116
+ "use_git": true,
117
+ "random_seed": 0
118
+ }
119
+ },
120
+ "winogrande": {
121
+ "acc": 0.7892659826361483,
122
+ "acc_stderr": 0.011462046419710676,
123
+ "args": {
124
+ "exp_name": "results/v1/winogrande/01-ai/Yi-1.5-6B",
125
+ "task_name": "winogrande",
126
+ "model_name": "01-ai/Yi-1.5-6B",
127
+ "dataset_param": {
128
+ "max_num_train": 50000,
129
+ "completion_only": false,
130
+ "apply_chat_template": false
131
+ },
132
+ "resume_from_checkpoint": false,
133
+ "train_param": {
134
+ "max_length": 1024,
135
+ "per_device_train_batch_size": 8,
136
+ "auto_find_batch_size": true,
137
+ "num_train_epochs": 1,
138
+ "max_steps": -1,
139
+ "logging_strategy": "steps",
140
+ "save_strategy": "steps",
141
+ "logging_steps": 0.2,
142
+ "save_steps": 0.2,
143
+ "gradient_accumulation_steps": 1,
144
+ "gradient_checkpointing": false,
145
+ "optim": "adamw_torch",
146
+ "learning_rate": 5e-05,
147
+ "weight_decay": 0.01,
148
+ "warmup_steps": 0,
149
+ "report_to": "tensorboard",
150
+ "save_safetensors": true,
151
+ "packing": false
152
+ },
153
+ "use_peft": true,
154
+ "peft_task_type": "CAUSAL_LM",
155
+ "peft_inference_mode": false,
156
+ "peft_r": 8,
157
+ "peft_lora_alpha": 32,
158
+ "peft_lora_dropout": 0.1,
159
+ "fsdp": "none",
160
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
161
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
162
+ "fsdp_fsdp_min_num_params": 100000000.0,
163
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
164
+ "fsdp_fsdp_forward_prefetch": false,
165
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
166
+ "fsdp_fsdp_offload_params": false,
167
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
168
+ "fsdp_fsdp_use_orig_params": true,
169
+ "fsdp_fsdp_activation_checkpointing": false,
170
+ "fsdp_fsdp_limit_all_gathers": true,
171
+ "fsdp_fsdp_sync_module_states": true,
172
+ "use_git": true,
173
+ "random_seed": 0
174
+ }
175
+ },
176
+ "hellaswag": {
177
+ "acc_norm": 0.7956,
178
+ "acc_norm_stderr": 0.004032826597937932,
179
+ "args": {
180
+ "exp_name": "results/v1/hellaswag/01-ai/Yi-1.5-6B",
181
+ "task_name": "hellaswag",
182
+ "model_name": "01-ai/Yi-1.5-6B",
183
+ "dataset_param": {
184
+ "max_num_train": 50000,
185
+ "completion_only": false,
186
+ "apply_chat_template": false
187
+ },
188
+ "resume_from_checkpoint": false,
189
+ "train_param": {
190
+ "max_length": 1024,
191
+ "per_device_train_batch_size": 8,
192
+ "auto_find_batch_size": true,
193
+ "num_train_epochs": 1,
194
+ "max_steps": -1,
195
+ "logging_strategy": "steps",
196
+ "save_strategy": "steps",
197
+ "logging_steps": 0.2,
198
+ "save_steps": 0.2,
199
+ "gradient_accumulation_steps": 1,
200
+ "gradient_checkpointing": false,
201
+ "optim": "adamw_torch",
202
+ "learning_rate": 5e-05,
203
+ "weight_decay": 0.01,
204
+ "warmup_steps": 0,
205
+ "report_to": "tensorboard",
206
+ "save_safetensors": true,
207
+ "packing": false
208
+ },
209
+ "use_peft": true,
210
+ "peft_task_type": "CAUSAL_LM",
211
+ "peft_inference_mode": false,
212
+ "peft_r": 8,
213
+ "peft_lora_alpha": 32,
214
+ "peft_lora_dropout": 0.1,
215
+ "fsdp": "none",
216
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
217
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
218
+ "fsdp_fsdp_min_num_params": 100000000.0,
219
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
220
+ "fsdp_fsdp_forward_prefetch": false,
221
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
222
+ "fsdp_fsdp_offload_params": false,
223
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
224
+ "fsdp_fsdp_use_orig_params": true,
225
+ "fsdp_fsdp_activation_checkpointing": false,
226
+ "fsdp_fsdp_limit_all_gathers": true,
227
+ "fsdp_fsdp_sync_module_states": true,
228
+ "use_git": true,
229
+ "random_seed": 0
230
+ }
231
+ },
232
+ "social_iqa": {
233
+ "acc": 0.5726714431934493,
234
+ "acc_stderr": 0.011193930340551272,
235
+ "args": {
236
+ "exp_name": "results/v1/social_iqa/01-ai/Yi-1.5-6B",
237
+ "task_name": "social_iqa",
238
+ "model_name": "01-ai/Yi-1.5-6B",
239
+ "dataset_param": {
240
+ "max_num_train": 50000,
241
+ "completion_only": false,
242
+ "apply_chat_template": false
243
+ },
244
+ "resume_from_checkpoint": false,
245
+ "train_param": {
246
+ "max_length": 1024,
247
+ "per_device_train_batch_size": 8,
248
+ "auto_find_batch_size": true,
249
+ "num_train_epochs": 1,
250
+ "max_steps": -1,
251
+ "logging_strategy": "steps",
252
+ "save_strategy": "steps",
253
+ "logging_steps": 0.2,
254
+ "save_steps": 0.2,
255
+ "gradient_accumulation_steps": 1,
256
+ "gradient_checkpointing": false,
257
+ "optim": "adamw_torch",
258
+ "learning_rate": 5e-05,
259
+ "weight_decay": 0.01,
260
+ "warmup_steps": 0,
261
+ "report_to": "tensorboard",
262
+ "save_safetensors": true,
263
+ "packing": false
264
+ },
265
+ "use_peft": true,
266
+ "peft_task_type": "CAUSAL_LM",
267
+ "peft_inference_mode": false,
268
+ "peft_r": 8,
269
+ "peft_lora_alpha": 32,
270
+ "peft_lora_dropout": 0.1,
271
+ "fsdp": "none",
272
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
273
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
274
+ "fsdp_fsdp_min_num_params": 100000000.0,
275
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
276
+ "fsdp_fsdp_forward_prefetch": false,
277
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
278
+ "fsdp_fsdp_offload_params": false,
279
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
280
+ "fsdp_fsdp_use_orig_params": true,
281
+ "fsdp_fsdp_activation_checkpointing": false,
282
+ "fsdp_fsdp_limit_all_gathers": true,
283
+ "fsdp_fsdp_sync_module_states": true,
284
+ "use_git": true,
285
+ "random_seed": 0
286
+ }
287
+ },
288
+ "piqa": {
289
+ "acc_norm": 0.8269858541893362,
290
+ "acc_norm_stderr": 0.008825417687866427,
291
+ "args": {
292
+ "exp_name": "results/v2/piqa/01-ai/Yi-1.5-6B",
293
+ "task_name": "piqa",
294
+ "model_name": "01-ai/Yi-1.5-6B",
295
+ "dataset_param": {
296
+ "max_num_train": 50000,
297
+ "completion_only": false,
298
+ "apply_chat_template": false
299
+ },
300
+ "resume_from_checkpoint": false,
301
+ "train_param": {
302
+ "max_length": 1024,
303
+ "per_device_train_batch_size": 8,
304
+ "auto_find_batch_size": true,
305
+ "num_train_epochs": 5,
306
+ "max_steps": -1,
307
+ "logging_strategy": "epoch",
308
+ "save_strategy": "epoch",
309
+ "logging_steps": 0.2,
310
+ "save_steps": 0.2,
311
+ "gradient_accumulation_steps": 1,
312
+ "gradient_checkpointing": false,
313
+ "optim": "adamw_torch",
314
+ "learning_rate": 5e-05,
315
+ "weight_decay": 0.01,
316
+ "warmup_steps": 0,
317
+ "report_to": "tensorboard",
318
+ "save_safetensors": true,
319
+ "packing": false
320
+ },
321
+ "use_peft": true,
322
+ "peft_task_type": "CAUSAL_LM",
323
+ "peft_inference_mode": false,
324
+ "peft_r": 8,
325
+ "peft_lora_alpha": 32,
326
+ "peft_lora_dropout": 0.1,
327
+ "fsdp": "none",
328
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
329
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
330
+ "fsdp_fsdp_min_num_params": 100000000.0,
331
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
332
+ "fsdp_fsdp_forward_prefetch": false,
333
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
334
+ "fsdp_fsdp_offload_params": false,
335
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
336
+ "fsdp_fsdp_use_orig_params": true,
337
+ "fsdp_fsdp_activation_checkpointing": false,
338
+ "fsdp_fsdp_limit_all_gathers": true,
339
+ "fsdp_fsdp_sync_module_states": true,
340
+ "use_git": true,
341
+ "random_seed": 0
342
+ }
343
+ },
344
+ "commonsense_qa": {
345
+ "acc": 0.8149058149058149,
346
+ "acc_stderr": 0.011119113942559872,
347
+ "args": {
348
+ "exp_name": "results/v2/commonsense_qa/01-ai/Yi-1.5-6B",
349
+ "task_name": "commonsense_qa",
350
+ "model_name": "01-ai/Yi-1.5-6B",
351
+ "dataset_param": {
352
+ "max_num_train": 50000,
353
+ "completion_only": false,
354
+ "apply_chat_template": false
355
+ },
356
+ "resume_from_checkpoint": false,
357
+ "train_param": {
358
+ "max_length": 1024,
359
+ "per_device_train_batch_size": 8,
360
+ "auto_find_batch_size": true,
361
+ "num_train_epochs": 5,
362
+ "max_steps": -1,
363
+ "logging_strategy": "epoch",
364
+ "save_strategy": "epoch",
365
+ "logging_steps": 0.2,
366
+ "save_steps": 0.2,
367
+ "gradient_accumulation_steps": 1,
368
+ "gradient_checkpointing": false,
369
+ "optim": "adamw_torch",
370
+ "learning_rate": 5e-05,
371
+ "weight_decay": 0.01,
372
+ "warmup_steps": 0,
373
+ "report_to": "tensorboard",
374
+ "save_safetensors": true,
375
+ "packing": false
376
+ },
377
+ "use_peft": true,
378
+ "peft_task_type": "CAUSAL_LM",
379
+ "peft_inference_mode": false,
380
+ "peft_r": 8,
381
+ "peft_lora_alpha": 32,
382
+ "peft_lora_dropout": 0.1,
383
+ "fsdp": "none",
384
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
385
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
386
+ "fsdp_fsdp_min_num_params": 100000000.0,
387
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
388
+ "fsdp_fsdp_forward_prefetch": false,
389
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
390
+ "fsdp_fsdp_offload_params": false,
391
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
392
+ "fsdp_fsdp_use_orig_params": true,
393
+ "fsdp_fsdp_activation_checkpointing": false,
394
+ "fsdp_fsdp_limit_all_gathers": true,
395
+ "fsdp_fsdp_sync_module_states": true,
396
+ "use_git": true,
397
+ "random_seed": 0
398
+ }
399
+ },
400
+ "gsm8k": {
401
+ "exact_match,flexible-extract": 0.5731614859742229,
402
+ "exact_match,flexible-extract_stderr": 0.01362424969659522,
403
+ "args": {
404
+ "exp_name": "results/v2/gsm8k/01-ai/Yi-1.5-6B",
405
+ "task_name": "gsm8k",
406
+ "model_name": "01-ai/Yi-1.5-6B",
407
+ "dataset_param": {
408
+ "max_num_train": 50000,
409
+ "completion_only": false,
410
+ "apply_chat_template": false
411
+ },
412
+ "resume_from_checkpoint": false,
413
+ "train_param": {
414
+ "max_length": 1024,
415
+ "per_device_train_batch_size": 8,
416
+ "auto_find_batch_size": true,
417
+ "num_train_epochs": 5,
418
+ "max_steps": -1,
419
+ "logging_strategy": "epoch",
420
+ "save_strategy": "epoch",
421
+ "logging_steps": 0.2,
422
+ "save_steps": 0.2,
423
+ "gradient_accumulation_steps": 1,
424
+ "gradient_checkpointing": false,
425
+ "optim": "adamw_torch",
426
+ "learning_rate": 5e-05,
427
+ "weight_decay": 0.01,
428
+ "warmup_steps": 0,
429
+ "report_to": "tensorboard",
430
+ "save_safetensors": true,
431
+ "packing": false
432
+ },
433
+ "use_peft": true,
434
+ "peft_task_type": "CAUSAL_LM",
435
+ "peft_inference_mode": false,
436
+ "peft_r": 8,
437
+ "peft_lora_alpha": 32,
438
+ "peft_lora_dropout": 0.1,
439
+ "fsdp": "none",
440
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
441
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
442
+ "fsdp_fsdp_min_num_params": 100000000.0,
443
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
444
+ "fsdp_fsdp_forward_prefetch": false,
445
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
446
+ "fsdp_fsdp_offload_params": false,
447
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
448
+ "fsdp_fsdp_use_orig_params": true,
449
+ "fsdp_fsdp_activation_checkpointing": false,
450
+ "fsdp_fsdp_limit_all_gathers": true,
451
+ "fsdp_fsdp_sync_module_states": true,
452
+ "use_git": true,
453
+ "random_seed": 0
454
+ }
455
+ },
456
+ "arc_challenge": {
457
+ "acc_norm": 0.591296928327645,
458
+ "acc_norm_stderr": 0.014365750345427006,
459
+ "args": {
460
+ "exp_name": "results/v2/arc_challenge/01-ai/Yi-1.5-6B",
461
+ "task_name": "arc_challenge",
462
+ "model_name": "01-ai/Yi-1.5-6B",
463
+ "dataset_param": {
464
+ "max_num_train": 50000,
465
+ "completion_only": false,
466
+ "apply_chat_template": false
467
+ },
468
+ "resume_from_checkpoint": false,
469
+ "train_param": {
470
+ "max_length": 1024,
471
+ "per_device_train_batch_size": 8,
472
+ "auto_find_batch_size": true,
473
+ "num_train_epochs": 5,
474
+ "max_steps": -1,
475
+ "logging_strategy": "epoch",
476
+ "save_strategy": "epoch",
477
+ "logging_steps": 0.2,
478
+ "save_steps": 0.2,
479
+ "gradient_accumulation_steps": 1,
480
+ "gradient_checkpointing": false,
481
+ "optim": "adamw_torch",
482
+ "learning_rate": 5e-05,
483
+ "weight_decay": 0.01,
484
+ "warmup_steps": 0,
485
+ "report_to": "tensorboard",
486
+ "save_safetensors": true,
487
+ "packing": false
488
+ },
489
+ "use_peft": true,
490
+ "peft_task_type": "CAUSAL_LM",
491
+ "peft_inference_mode": false,
492
+ "peft_r": 8,
493
+ "peft_lora_alpha": 32,
494
+ "peft_lora_dropout": 0.1,
495
+ "fsdp": "none",
496
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
497
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
498
+ "fsdp_fsdp_min_num_params": 100000000.0,
499
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
500
+ "fsdp_fsdp_forward_prefetch": false,
501
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
502
+ "fsdp_fsdp_offload_params": false,
503
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
504
+ "fsdp_fsdp_use_orig_params": true,
505
+ "fsdp_fsdp_activation_checkpointing": false,
506
+ "fsdp_fsdp_limit_all_gathers": true,
507
+ "fsdp_fsdp_sync_module_states": true,
508
+ "use_git": true,
509
+ "random_seed": 0
510
+ }
511
+ }
512
+ }
513
+ }
01-ai/Yi-1.5-9B-Chat/results_2026-01-05-04-27-51.json ADDED
@@ -0,0 +1,513 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "model_name": "01-ai/Yi-1.5-9B-Chat",
4
+ "model_type": "instruction-tuned",
5
+ "params": 9.0
6
+ },
7
+ "results": {
8
+ "medmcqa": {
9
+ "acc_norm": 0.5264164475256993,
10
+ "acc_norm_stderr": 0.007720954911109498,
11
+ "args": {
12
+ "exp_name": "results/v1/medmcqa/01-ai/Yi-1.5-9B-Chat",
13
+ "task_name": "medmcqa",
14
+ "model_name": "01-ai/Yi-1.5-9B-Chat",
15
+ "dataset_param": {
16
+ "max_num_train": 50000,
17
+ "completion_only": false,
18
+ "apply_chat_template": true
19
+ },
20
+ "resume_from_checkpoint": false,
21
+ "train_param": {
22
+ "max_length": 1024,
23
+ "per_device_train_batch_size": 8,
24
+ "auto_find_batch_size": true,
25
+ "num_train_epochs": 1,
26
+ "max_steps": -1,
27
+ "logging_strategy": "steps",
28
+ "save_strategy": "steps",
29
+ "logging_steps": 0.2,
30
+ "save_steps": 0.2,
31
+ "gradient_accumulation_steps": 1,
32
+ "gradient_checkpointing": false,
33
+ "optim": "adamw_torch",
34
+ "learning_rate": 5e-05,
35
+ "weight_decay": 0.01,
36
+ "warmup_steps": 0,
37
+ "report_to": "tensorboard",
38
+ "save_safetensors": true,
39
+ "packing": false
40
+ },
41
+ "use_peft": true,
42
+ "peft_task_type": "CAUSAL_LM",
43
+ "peft_inference_mode": false,
44
+ "peft_r": 8,
45
+ "peft_lora_alpha": 32,
46
+ "peft_lora_dropout": 0.1,
47
+ "fsdp": "none",
48
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
49
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
50
+ "fsdp_fsdp_min_num_params": 100000000.0,
51
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
52
+ "fsdp_fsdp_forward_prefetch": false,
53
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
54
+ "fsdp_fsdp_offload_params": false,
55
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
56
+ "fsdp_fsdp_use_orig_params": true,
57
+ "fsdp_fsdp_activation_checkpointing": false,
58
+ "fsdp_fsdp_limit_all_gathers": true,
59
+ "fsdp_fsdp_sync_module_states": true,
60
+ "use_git": true,
61
+ "random_seed": 0
62
+ }
63
+ },
64
+ "nq_open": {
65
+ "exact_match,remove_whitespace": 0.10692520775623268,
66
+ "exact_match,remove_whitespace_stderr": 0.005143876416597367,
67
+ "args": {
68
+ "exp_name": "results/v1/nq_open/01-ai/Yi-1.5-9B-Chat",
69
+ "task_name": "nq_open",
70
+ "model_name": "01-ai/Yi-1.5-9B-Chat",
71
+ "dataset_param": {
72
+ "max_num_train": 50000,
73
+ "completion_only": false,
74
+ "apply_chat_template": true
75
+ },
76
+ "resume_from_checkpoint": false,
77
+ "train_param": {
78
+ "max_length": 1024,
79
+ "per_device_train_batch_size": 8,
80
+ "auto_find_batch_size": true,
81
+ "num_train_epochs": 1,
82
+ "max_steps": -1,
83
+ "logging_strategy": "steps",
84
+ "save_strategy": "steps",
85
+ "logging_steps": 0.2,
86
+ "save_steps": 0.2,
87
+ "gradient_accumulation_steps": 1,
88
+ "gradient_checkpointing": false,
89
+ "optim": "adamw_torch",
90
+ "learning_rate": 5e-05,
91
+ "weight_decay": 0.01,
92
+ "warmup_steps": 0,
93
+ "report_to": "tensorboard",
94
+ "save_safetensors": true,
95
+ "packing": false
96
+ },
97
+ "use_peft": true,
98
+ "peft_task_type": "CAUSAL_LM",
99
+ "peft_inference_mode": false,
100
+ "peft_r": 8,
101
+ "peft_lora_alpha": 32,
102
+ "peft_lora_dropout": 0.1,
103
+ "fsdp": "none",
104
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
105
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
106
+ "fsdp_fsdp_min_num_params": 100000000.0,
107
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
108
+ "fsdp_fsdp_forward_prefetch": false,
109
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
110
+ "fsdp_fsdp_offload_params": false,
111
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
112
+ "fsdp_fsdp_use_orig_params": true,
113
+ "fsdp_fsdp_activation_checkpointing": false,
114
+ "fsdp_fsdp_limit_all_gathers": true,
115
+ "fsdp_fsdp_sync_module_states": true,
116
+ "use_git": true,
117
+ "random_seed": 0
118
+ }
119
+ },
120
+ "winogrande": {
121
+ "acc": 0.8216258879242304,
122
+ "acc_stderr": 0.010759352014855925,
123
+ "args": {
124
+ "exp_name": "results/v1/winogrande/01-ai/Yi-1.5-9B-Chat",
125
+ "task_name": "winogrande",
126
+ "model_name": "01-ai/Yi-1.5-9B-Chat",
127
+ "dataset_param": {
128
+ "max_num_train": 50000,
129
+ "completion_only": false,
130
+ "apply_chat_template": true
131
+ },
132
+ "resume_from_checkpoint": false,
133
+ "train_param": {
134
+ "max_length": 1024,
135
+ "per_device_train_batch_size": 8,
136
+ "auto_find_batch_size": true,
137
+ "num_train_epochs": 1,
138
+ "max_steps": -1,
139
+ "logging_strategy": "steps",
140
+ "save_strategy": "steps",
141
+ "logging_steps": 0.2,
142
+ "save_steps": 0.2,
143
+ "gradient_accumulation_steps": 1,
144
+ "gradient_checkpointing": false,
145
+ "optim": "adamw_torch",
146
+ "learning_rate": 5e-05,
147
+ "weight_decay": 0.01,
148
+ "warmup_steps": 0,
149
+ "report_to": "tensorboard",
150
+ "save_safetensors": true,
151
+ "packing": false
152
+ },
153
+ "use_peft": true,
154
+ "peft_task_type": "CAUSAL_LM",
155
+ "peft_inference_mode": false,
156
+ "peft_r": 8,
157
+ "peft_lora_alpha": 32,
158
+ "peft_lora_dropout": 0.1,
159
+ "fsdp": "none",
160
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
161
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
162
+ "fsdp_fsdp_min_num_params": 100000000.0,
163
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
164
+ "fsdp_fsdp_forward_prefetch": false,
165
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
166
+ "fsdp_fsdp_offload_params": false,
167
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
168
+ "fsdp_fsdp_use_orig_params": true,
169
+ "fsdp_fsdp_activation_checkpointing": false,
170
+ "fsdp_fsdp_limit_all_gathers": true,
171
+ "fsdp_fsdp_sync_module_states": true,
172
+ "use_git": true,
173
+ "random_seed": 0
174
+ }
175
+ },
176
+ "hellaswag": {
177
+ "acc_norm": 0.8074,
178
+ "acc_norm_stderr": 0.003943612456608531,
179
+ "args": {
180
+ "exp_name": "results/v1/hellaswag/01-ai/Yi-1.5-9B-Chat",
181
+ "task_name": "hellaswag",
182
+ "model_name": "01-ai/Yi-1.5-9B-Chat",
183
+ "dataset_param": {
184
+ "max_num_train": 50000,
185
+ "completion_only": false,
186
+ "apply_chat_template": true
187
+ },
188
+ "resume_from_checkpoint": false,
189
+ "train_param": {
190
+ "max_length": 1024,
191
+ "per_device_train_batch_size": 8,
192
+ "auto_find_batch_size": true,
193
+ "num_train_epochs": 1,
194
+ "max_steps": -1,
195
+ "logging_strategy": "steps",
196
+ "save_strategy": "steps",
197
+ "logging_steps": 0.2,
198
+ "save_steps": 0.2,
199
+ "gradient_accumulation_steps": 1,
200
+ "gradient_checkpointing": false,
201
+ "optim": "adamw_torch",
202
+ "learning_rate": 5e-05,
203
+ "weight_decay": 0.01,
204
+ "warmup_steps": 0,
205
+ "report_to": "tensorboard",
206
+ "save_safetensors": true,
207
+ "packing": false
208
+ },
209
+ "use_peft": true,
210
+ "peft_task_type": "CAUSAL_LM",
211
+ "peft_inference_mode": false,
212
+ "peft_r": 8,
213
+ "peft_lora_alpha": 32,
214
+ "peft_lora_dropout": 0.1,
215
+ "fsdp": "none",
216
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
217
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
218
+ "fsdp_fsdp_min_num_params": 100000000.0,
219
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
220
+ "fsdp_fsdp_forward_prefetch": false,
221
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
222
+ "fsdp_fsdp_offload_params": false,
223
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
224
+ "fsdp_fsdp_use_orig_params": true,
225
+ "fsdp_fsdp_activation_checkpointing": false,
226
+ "fsdp_fsdp_limit_all_gathers": true,
227
+ "fsdp_fsdp_sync_module_states": true,
228
+ "use_git": true,
229
+ "random_seed": 0
230
+ }
231
+ },
232
+ "social_iqa": {
233
+ "acc": 0.5803480040941658,
234
+ "acc_stderr": 0.011167032303390535,
235
+ "args": {
236
+ "exp_name": "results/v1/social_iqa/01-ai/Yi-1.5-9B-Chat",
237
+ "task_name": "social_iqa",
238
+ "model_name": "01-ai/Yi-1.5-9B-Chat",
239
+ "dataset_param": {
240
+ "max_num_train": 50000,
241
+ "completion_only": false,
242
+ "apply_chat_template": true
243
+ },
244
+ "resume_from_checkpoint": false,
245
+ "train_param": {
246
+ "max_length": 1024,
247
+ "per_device_train_batch_size": 8,
248
+ "auto_find_batch_size": true,
249
+ "num_train_epochs": 1,
250
+ "max_steps": -1,
251
+ "logging_strategy": "steps",
252
+ "save_strategy": "steps",
253
+ "logging_steps": 0.2,
254
+ "save_steps": 0.2,
255
+ "gradient_accumulation_steps": 1,
256
+ "gradient_checkpointing": false,
257
+ "optim": "adamw_torch",
258
+ "learning_rate": 5e-05,
259
+ "weight_decay": 0.01,
260
+ "warmup_steps": 0,
261
+ "report_to": "tensorboard",
262
+ "save_safetensors": true,
263
+ "packing": false
264
+ },
265
+ "use_peft": true,
266
+ "peft_task_type": "CAUSAL_LM",
267
+ "peft_inference_mode": false,
268
+ "peft_r": 8,
269
+ "peft_lora_alpha": 32,
270
+ "peft_lora_dropout": 0.1,
271
+ "fsdp": "none",
272
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
273
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
274
+ "fsdp_fsdp_min_num_params": 100000000.0,
275
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
276
+ "fsdp_fsdp_forward_prefetch": false,
277
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
278
+ "fsdp_fsdp_offload_params": false,
279
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
280
+ "fsdp_fsdp_use_orig_params": true,
281
+ "fsdp_fsdp_activation_checkpointing": false,
282
+ "fsdp_fsdp_limit_all_gathers": true,
283
+ "fsdp_fsdp_sync_module_states": true,
284
+ "use_git": true,
285
+ "random_seed": 0
286
+ }
287
+ },
288
+ "piqa": {
289
+ "acc_norm": 0.8204570184983678,
290
+ "acc_norm_stderr": 0.008954834329201147,
291
+ "args": {
292
+ "exp_name": "results/v2/piqa/01-ai/Yi-1.5-9B-Chat",
293
+ "task_name": "piqa",
294
+ "model_name": "01-ai/Yi-1.5-9B-Chat",
295
+ "dataset_param": {
296
+ "max_num_train": 50000,
297
+ "completion_only": false,
298
+ "apply_chat_template": true
299
+ },
300
+ "resume_from_checkpoint": false,
301
+ "train_param": {
302
+ "max_length": 1024,
303
+ "per_device_train_batch_size": 8,
304
+ "auto_find_batch_size": true,
305
+ "num_train_epochs": 5,
306
+ "max_steps": -1,
307
+ "logging_strategy": "epoch",
308
+ "save_strategy": "epoch",
309
+ "logging_steps": 0.2,
310
+ "save_steps": 0.2,
311
+ "gradient_accumulation_steps": 1,
312
+ "gradient_checkpointing": false,
313
+ "optim": "adamw_torch",
314
+ "learning_rate": 5e-05,
315
+ "weight_decay": 0.01,
316
+ "warmup_steps": 0,
317
+ "report_to": "tensorboard",
318
+ "save_safetensors": true,
319
+ "packing": false
320
+ },
321
+ "use_peft": true,
322
+ "peft_task_type": "CAUSAL_LM",
323
+ "peft_inference_mode": false,
324
+ "peft_r": 8,
325
+ "peft_lora_alpha": 32,
326
+ "peft_lora_dropout": 0.1,
327
+ "fsdp": "none",
328
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
329
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
330
+ "fsdp_fsdp_min_num_params": 100000000.0,
331
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
332
+ "fsdp_fsdp_forward_prefetch": false,
333
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
334
+ "fsdp_fsdp_offload_params": false,
335
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
336
+ "fsdp_fsdp_use_orig_params": true,
337
+ "fsdp_fsdp_activation_checkpointing": false,
338
+ "fsdp_fsdp_limit_all_gathers": true,
339
+ "fsdp_fsdp_sync_module_states": true,
340
+ "use_git": true,
341
+ "random_seed": 0
342
+ }
343
+ },
344
+ "commonsense_qa": {
345
+ "acc": 0.8558558558558559,
346
+ "acc_stderr": 0.010055852624752136,
347
+ "args": {
348
+ "exp_name": "results/v2/commonsense_qa/01-ai/Yi-1.5-9B-Chat",
349
+ "task_name": "commonsense_qa",
350
+ "model_name": "01-ai/Yi-1.5-9B-Chat",
351
+ "dataset_param": {
352
+ "max_num_train": 50000,
353
+ "completion_only": false,
354
+ "apply_chat_template": true
355
+ },
356
+ "resume_from_checkpoint": false,
357
+ "train_param": {
358
+ "max_length": 1024,
359
+ "per_device_train_batch_size": 8,
360
+ "auto_find_batch_size": true,
361
+ "num_train_epochs": 5,
362
+ "max_steps": -1,
363
+ "logging_strategy": "epoch",
364
+ "save_strategy": "epoch",
365
+ "logging_steps": 0.2,
366
+ "save_steps": 0.2,
367
+ "gradient_accumulation_steps": 1,
368
+ "gradient_checkpointing": false,
369
+ "optim": "adamw_torch",
370
+ "learning_rate": 5e-05,
371
+ "weight_decay": 0.01,
372
+ "warmup_steps": 0,
373
+ "report_to": "tensorboard",
374
+ "save_safetensors": true,
375
+ "packing": false
376
+ },
377
+ "use_peft": true,
378
+ "peft_task_type": "CAUSAL_LM",
379
+ "peft_inference_mode": false,
380
+ "peft_r": 8,
381
+ "peft_lora_alpha": 32,
382
+ "peft_lora_dropout": 0.1,
383
+ "fsdp": "none",
384
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
385
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
386
+ "fsdp_fsdp_min_num_params": 100000000.0,
387
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
388
+ "fsdp_fsdp_forward_prefetch": false,
389
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
390
+ "fsdp_fsdp_offload_params": false,
391
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
392
+ "fsdp_fsdp_use_orig_params": true,
393
+ "fsdp_fsdp_activation_checkpointing": false,
394
+ "fsdp_fsdp_limit_all_gathers": true,
395
+ "fsdp_fsdp_sync_module_states": true,
396
+ "use_git": true,
397
+ "random_seed": 0
398
+ }
399
+ },
400
+ "gsm8k": {
401
+ "exact_match,flexible-extract": 0.6376042456406369,
402
+ "exact_match,flexible-extract_stderr": 0.013240654263574752,
403
+ "args": {
404
+ "exp_name": "results/v2/gsm8k/01-ai/Yi-1.5-9B-Chat",
405
+ "task_name": "gsm8k",
406
+ "model_name": "01-ai/Yi-1.5-9B-Chat",
407
+ "dataset_param": {
408
+ "max_num_train": 50000,
409
+ "completion_only": false,
410
+ "apply_chat_template": true
411
+ },
412
+ "resume_from_checkpoint": false,
413
+ "train_param": {
414
+ "max_length": 1024,
415
+ "per_device_train_batch_size": 8,
416
+ "auto_find_batch_size": true,
417
+ "num_train_epochs": 5,
418
+ "max_steps": -1,
419
+ "logging_strategy": "epoch",
420
+ "save_strategy": "epoch",
421
+ "logging_steps": 0.2,
422
+ "save_steps": 0.2,
423
+ "gradient_accumulation_steps": 1,
424
+ "gradient_checkpointing": false,
425
+ "optim": "adamw_torch",
426
+ "learning_rate": 5e-05,
427
+ "weight_decay": 0.01,
428
+ "warmup_steps": 0,
429
+ "report_to": "tensorboard",
430
+ "save_safetensors": true,
431
+ "packing": false
432
+ },
433
+ "use_peft": true,
434
+ "peft_task_type": "CAUSAL_LM",
435
+ "peft_inference_mode": false,
436
+ "peft_r": 8,
437
+ "peft_lora_alpha": 32,
438
+ "peft_lora_dropout": 0.1,
439
+ "fsdp": "none",
440
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
441
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
442
+ "fsdp_fsdp_min_num_params": 100000000.0,
443
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
444
+ "fsdp_fsdp_forward_prefetch": false,
445
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
446
+ "fsdp_fsdp_offload_params": false,
447
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
448
+ "fsdp_fsdp_use_orig_params": true,
449
+ "fsdp_fsdp_activation_checkpointing": false,
450
+ "fsdp_fsdp_limit_all_gathers": true,
451
+ "fsdp_fsdp_sync_module_states": true,
452
+ "use_git": true,
453
+ "random_seed": 0
454
+ }
455
+ },
456
+ "arc_challenge": {
457
+ "acc_norm": 0.5955631399317406,
458
+ "acc_norm_stderr": 0.01434203648343617,
459
+ "args": {
460
+ "exp_name": "results/v2/arc_challenge/01-ai/Yi-1.5-9B-Chat",
461
+ "task_name": "arc_challenge",
462
+ "model_name": "01-ai/Yi-1.5-9B-Chat",
463
+ "dataset_param": {
464
+ "max_num_train": 50000,
465
+ "completion_only": false,
466
+ "apply_chat_template": true
467
+ },
468
+ "resume_from_checkpoint": false,
469
+ "train_param": {
470
+ "max_length": 1024,
471
+ "per_device_train_batch_size": 8,
472
+ "auto_find_batch_size": true,
473
+ "num_train_epochs": 5,
474
+ "max_steps": -1,
475
+ "logging_strategy": "epoch",
476
+ "save_strategy": "epoch",
477
+ "logging_steps": 0.2,
478
+ "save_steps": 0.2,
479
+ "gradient_accumulation_steps": 1,
480
+ "gradient_checkpointing": false,
481
+ "optim": "adamw_torch",
482
+ "learning_rate": 5e-05,
483
+ "weight_decay": 0.01,
484
+ "warmup_steps": 0,
485
+ "report_to": "tensorboard",
486
+ "save_safetensors": true,
487
+ "packing": false
488
+ },
489
+ "use_peft": true,
490
+ "peft_task_type": "CAUSAL_LM",
491
+ "peft_inference_mode": false,
492
+ "peft_r": 8,
493
+ "peft_lora_alpha": 32,
494
+ "peft_lora_dropout": 0.1,
495
+ "fsdp": "none",
496
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
497
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
498
+ "fsdp_fsdp_min_num_params": 100000000.0,
499
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
500
+ "fsdp_fsdp_forward_prefetch": false,
501
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
502
+ "fsdp_fsdp_offload_params": false,
503
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
504
+ "fsdp_fsdp_use_orig_params": true,
505
+ "fsdp_fsdp_activation_checkpointing": false,
506
+ "fsdp_fsdp_limit_all_gathers": true,
507
+ "fsdp_fsdp_sync_module_states": true,
508
+ "use_git": true,
509
+ "random_seed": 0
510
+ }
511
+ }
512
+ }
513
+ }
01-ai/Yi-1.5-9B/results_2026-01-05-04-27-50.json ADDED
@@ -0,0 +1,513 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "model_name": "01-ai/Yi-1.5-9B",
4
+ "model_type": "pretrained",
5
+ "params": 9.0
6
+ },
7
+ "results": {
8
+ "medmcqa": {
9
+ "acc_norm": 0.5491274205115946,
10
+ "acc_norm_stderr": 0.0076943415916129534,
11
+ "args": {
12
+ "exp_name": "results/v1/medmcqa/01-ai/Yi-1.5-9B",
13
+ "task_name": "medmcqa",
14
+ "model_name": "01-ai/Yi-1.5-9B",
15
+ "dataset_param": {
16
+ "max_num_train": 50000,
17
+ "completion_only": false,
18
+ "apply_chat_template": false
19
+ },
20
+ "resume_from_checkpoint": false,
21
+ "train_param": {
22
+ "max_length": 1024,
23
+ "per_device_train_batch_size": 8,
24
+ "auto_find_batch_size": true,
25
+ "num_train_epochs": 1,
26
+ "max_steps": -1,
27
+ "logging_strategy": "steps",
28
+ "save_strategy": "steps",
29
+ "logging_steps": 0.2,
30
+ "save_steps": 0.2,
31
+ "gradient_accumulation_steps": 1,
32
+ "gradient_checkpointing": false,
33
+ "optim": "adamw_torch",
34
+ "learning_rate": 5e-05,
35
+ "weight_decay": 0.01,
36
+ "warmup_steps": 0,
37
+ "report_to": "tensorboard",
38
+ "save_safetensors": true,
39
+ "packing": false
40
+ },
41
+ "use_peft": true,
42
+ "peft_task_type": "CAUSAL_LM",
43
+ "peft_inference_mode": false,
44
+ "peft_r": 8,
45
+ "peft_lora_alpha": 32,
46
+ "peft_lora_dropout": 0.1,
47
+ "fsdp": "none",
48
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
49
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
50
+ "fsdp_fsdp_min_num_params": 100000000.0,
51
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
52
+ "fsdp_fsdp_forward_prefetch": false,
53
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
54
+ "fsdp_fsdp_offload_params": false,
55
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
56
+ "fsdp_fsdp_use_orig_params": true,
57
+ "fsdp_fsdp_activation_checkpointing": false,
58
+ "fsdp_fsdp_limit_all_gathers": true,
59
+ "fsdp_fsdp_sync_module_states": true,
60
+ "use_git": true,
61
+ "random_seed": 0
62
+ }
63
+ },
64
+ "nq_open": {
65
+ "exact_match,remove_whitespace": 0.28005540166204984,
66
+ "exact_match,remove_whitespace_stderr": 0.007474429925077295,
67
+ "args": {
68
+ "exp_name": "results/v1/nq_open/01-ai/Yi-1.5-9B",
69
+ "task_name": "nq_open",
70
+ "model_name": "01-ai/Yi-1.5-9B",
71
+ "dataset_param": {
72
+ "max_num_train": 50000,
73
+ "completion_only": false,
74
+ "apply_chat_template": false
75
+ },
76
+ "resume_from_checkpoint": false,
77
+ "train_param": {
78
+ "max_length": 1024,
79
+ "per_device_train_batch_size": 8,
80
+ "auto_find_batch_size": true,
81
+ "num_train_epochs": 1,
82
+ "max_steps": -1,
83
+ "logging_strategy": "steps",
84
+ "save_strategy": "steps",
85
+ "logging_steps": 0.2,
86
+ "save_steps": 0.2,
87
+ "gradient_accumulation_steps": 1,
88
+ "gradient_checkpointing": false,
89
+ "optim": "adamw_torch",
90
+ "learning_rate": 5e-05,
91
+ "weight_decay": 0.01,
92
+ "warmup_steps": 0,
93
+ "report_to": "tensorboard",
94
+ "save_safetensors": true,
95
+ "packing": false
96
+ },
97
+ "use_peft": true,
98
+ "peft_task_type": "CAUSAL_LM",
99
+ "peft_inference_mode": false,
100
+ "peft_r": 8,
101
+ "peft_lora_alpha": 32,
102
+ "peft_lora_dropout": 0.1,
103
+ "fsdp": "none",
104
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
105
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
106
+ "fsdp_fsdp_min_num_params": 100000000.0,
107
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
108
+ "fsdp_fsdp_forward_prefetch": false,
109
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
110
+ "fsdp_fsdp_offload_params": false,
111
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
112
+ "fsdp_fsdp_use_orig_params": true,
113
+ "fsdp_fsdp_activation_checkpointing": false,
114
+ "fsdp_fsdp_limit_all_gathers": true,
115
+ "fsdp_fsdp_sync_module_states": true,
116
+ "use_git": true,
117
+ "random_seed": 0
118
+ }
119
+ },
120
+ "winogrande": {
121
+ "acc": 0.8018942383583267,
122
+ "acc_stderr": 0.011201862744487052,
123
+ "args": {
124
+ "exp_name": "results/v1/winogrande/01-ai/Yi-1.5-9B",
125
+ "task_name": "winogrande",
126
+ "model_name": "01-ai/Yi-1.5-9B",
127
+ "dataset_param": {
128
+ "max_num_train": 50000,
129
+ "completion_only": false,
130
+ "apply_chat_template": false
131
+ },
132
+ "resume_from_checkpoint": false,
133
+ "train_param": {
134
+ "max_length": 1024,
135
+ "per_device_train_batch_size": 8,
136
+ "auto_find_batch_size": true,
137
+ "num_train_epochs": 1,
138
+ "max_steps": -1,
139
+ "logging_strategy": "steps",
140
+ "save_strategy": "steps",
141
+ "logging_steps": 0.2,
142
+ "save_steps": 0.2,
143
+ "gradient_accumulation_steps": 1,
144
+ "gradient_checkpointing": false,
145
+ "optim": "adamw_torch",
146
+ "learning_rate": 5e-05,
147
+ "weight_decay": 0.01,
148
+ "warmup_steps": 0,
149
+ "report_to": "tensorboard",
150
+ "save_safetensors": true,
151
+ "packing": false
152
+ },
153
+ "use_peft": true,
154
+ "peft_task_type": "CAUSAL_LM",
155
+ "peft_inference_mode": false,
156
+ "peft_r": 8,
157
+ "peft_lora_alpha": 32,
158
+ "peft_lora_dropout": 0.1,
159
+ "fsdp": "none",
160
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
161
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
162
+ "fsdp_fsdp_min_num_params": 100000000.0,
163
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
164
+ "fsdp_fsdp_forward_prefetch": false,
165
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
166
+ "fsdp_fsdp_offload_params": false,
167
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
168
+ "fsdp_fsdp_use_orig_params": true,
169
+ "fsdp_fsdp_activation_checkpointing": false,
170
+ "fsdp_fsdp_limit_all_gathers": true,
171
+ "fsdp_fsdp_sync_module_states": true,
172
+ "use_git": true,
173
+ "random_seed": 0
174
+ }
175
+ },
176
+ "hellaswag": {
177
+ "acc_norm": 0.8183,
178
+ "acc_norm_stderr": 0.0038561636375808567,
179
+ "args": {
180
+ "exp_name": "results/v1/hellaswag/01-ai/Yi-1.5-9B",
181
+ "task_name": "hellaswag",
182
+ "model_name": "01-ai/Yi-1.5-9B",
183
+ "dataset_param": {
184
+ "max_num_train": 50000,
185
+ "completion_only": false,
186
+ "apply_chat_template": false
187
+ },
188
+ "resume_from_checkpoint": false,
189
+ "train_param": {
190
+ "max_length": 1024,
191
+ "per_device_train_batch_size": 8,
192
+ "auto_find_batch_size": true,
193
+ "num_train_epochs": 1,
194
+ "max_steps": -1,
195
+ "logging_strategy": "steps",
196
+ "save_strategy": "steps",
197
+ "logging_steps": 0.2,
198
+ "save_steps": 0.2,
199
+ "gradient_accumulation_steps": 1,
200
+ "gradient_checkpointing": false,
201
+ "optim": "adamw_torch",
202
+ "learning_rate": 5e-05,
203
+ "weight_decay": 0.01,
204
+ "warmup_steps": 0,
205
+ "report_to": "tensorboard",
206
+ "save_safetensors": true,
207
+ "packing": false
208
+ },
209
+ "use_peft": true,
210
+ "peft_task_type": "CAUSAL_LM",
211
+ "peft_inference_mode": false,
212
+ "peft_r": 8,
213
+ "peft_lora_alpha": 32,
214
+ "peft_lora_dropout": 0.1,
215
+ "fsdp": "none",
216
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
217
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
218
+ "fsdp_fsdp_min_num_params": 100000000.0,
219
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
220
+ "fsdp_fsdp_forward_prefetch": false,
221
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
222
+ "fsdp_fsdp_offload_params": false,
223
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
224
+ "fsdp_fsdp_use_orig_params": true,
225
+ "fsdp_fsdp_activation_checkpointing": false,
226
+ "fsdp_fsdp_limit_all_gathers": true,
227
+ "fsdp_fsdp_sync_module_states": true,
228
+ "use_git": true,
229
+ "random_seed": 0
230
+ }
231
+ },
232
+ "social_iqa": {
233
+ "acc": 0.5823950870010235,
234
+ "acc_stderr": 0.011159391894922488,
235
+ "args": {
236
+ "exp_name": "results/v1/social_iqa/01-ai/Yi-1.5-9B",
237
+ "task_name": "social_iqa",
238
+ "model_name": "01-ai/Yi-1.5-9B",
239
+ "dataset_param": {
240
+ "max_num_train": 50000,
241
+ "completion_only": false,
242
+ "apply_chat_template": false
243
+ },
244
+ "resume_from_checkpoint": false,
245
+ "train_param": {
246
+ "max_length": 1024,
247
+ "per_device_train_batch_size": 8,
248
+ "auto_find_batch_size": true,
249
+ "num_train_epochs": 1,
250
+ "max_steps": -1,
251
+ "logging_strategy": "steps",
252
+ "save_strategy": "steps",
253
+ "logging_steps": 0.2,
254
+ "save_steps": 0.2,
255
+ "gradient_accumulation_steps": 1,
256
+ "gradient_checkpointing": false,
257
+ "optim": "adamw_torch",
258
+ "learning_rate": 5e-05,
259
+ "weight_decay": 0.01,
260
+ "warmup_steps": 0,
261
+ "report_to": "tensorboard",
262
+ "save_safetensors": true,
263
+ "packing": false
264
+ },
265
+ "use_peft": true,
266
+ "peft_task_type": "CAUSAL_LM",
267
+ "peft_inference_mode": false,
268
+ "peft_r": 8,
269
+ "peft_lora_alpha": 32,
270
+ "peft_lora_dropout": 0.1,
271
+ "fsdp": "none",
272
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
273
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
274
+ "fsdp_fsdp_min_num_params": 100000000.0,
275
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
276
+ "fsdp_fsdp_forward_prefetch": false,
277
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
278
+ "fsdp_fsdp_offload_params": false,
279
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
280
+ "fsdp_fsdp_use_orig_params": true,
281
+ "fsdp_fsdp_activation_checkpointing": false,
282
+ "fsdp_fsdp_limit_all_gathers": true,
283
+ "fsdp_fsdp_sync_module_states": true,
284
+ "use_git": true,
285
+ "random_seed": 0
286
+ }
287
+ },
288
+ "piqa": {
289
+ "acc_norm": 0.8307943416757345,
290
+ "acc_norm_stderr": 0.008747815763315783,
291
+ "args": {
292
+ "exp_name": "results/v2/piqa/01-ai/Yi-1.5-9B",
293
+ "task_name": "piqa",
294
+ "model_name": "01-ai/Yi-1.5-9B",
295
+ "dataset_param": {
296
+ "max_num_train": 50000,
297
+ "completion_only": false,
298
+ "apply_chat_template": false
299
+ },
300
+ "resume_from_checkpoint": false,
301
+ "train_param": {
302
+ "max_length": 1024,
303
+ "per_device_train_batch_size": 8,
304
+ "auto_find_batch_size": true,
305
+ "num_train_epochs": 5,
306
+ "max_steps": -1,
307
+ "logging_strategy": "epoch",
308
+ "save_strategy": "epoch",
309
+ "logging_steps": 0.2,
310
+ "save_steps": 0.2,
311
+ "gradient_accumulation_steps": 1,
312
+ "gradient_checkpointing": false,
313
+ "optim": "adamw_torch",
314
+ "learning_rate": 5e-05,
315
+ "weight_decay": 0.01,
316
+ "warmup_steps": 0,
317
+ "report_to": "tensorboard",
318
+ "save_safetensors": true,
319
+ "packing": false
320
+ },
321
+ "use_peft": true,
322
+ "peft_task_type": "CAUSAL_LM",
323
+ "peft_inference_mode": false,
324
+ "peft_r": 8,
325
+ "peft_lora_alpha": 32,
326
+ "peft_lora_dropout": 0.1,
327
+ "fsdp": "none",
328
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
329
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
330
+ "fsdp_fsdp_min_num_params": 100000000.0,
331
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
332
+ "fsdp_fsdp_forward_prefetch": false,
333
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
334
+ "fsdp_fsdp_offload_params": false,
335
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
336
+ "fsdp_fsdp_use_orig_params": true,
337
+ "fsdp_fsdp_activation_checkpointing": false,
338
+ "fsdp_fsdp_limit_all_gathers": true,
339
+ "fsdp_fsdp_sync_module_states": true,
340
+ "use_git": true,
341
+ "random_seed": 0
342
+ }
343
+ },
344
+ "commonsense_qa": {
345
+ "acc": 0.8337428337428338,
346
+ "acc_stderr": 0.010659242157930128,
347
+ "args": {
348
+ "exp_name": "results/v2/commonsense_qa/01-ai/Yi-1.5-9B",
349
+ "task_name": "commonsense_qa",
350
+ "model_name": "01-ai/Yi-1.5-9B",
351
+ "dataset_param": {
352
+ "max_num_train": 50000,
353
+ "completion_only": false,
354
+ "apply_chat_template": false
355
+ },
356
+ "resume_from_checkpoint": false,
357
+ "train_param": {
358
+ "max_length": 1024,
359
+ "per_device_train_batch_size": 8,
360
+ "auto_find_batch_size": true,
361
+ "num_train_epochs": 5,
362
+ "max_steps": -1,
363
+ "logging_strategy": "epoch",
364
+ "save_strategy": "epoch",
365
+ "logging_steps": 0.2,
366
+ "save_steps": 0.2,
367
+ "gradient_accumulation_steps": 1,
368
+ "gradient_checkpointing": false,
369
+ "optim": "adamw_torch",
370
+ "learning_rate": 5e-05,
371
+ "weight_decay": 0.01,
372
+ "warmup_steps": 0,
373
+ "report_to": "tensorboard",
374
+ "save_safetensors": true,
375
+ "packing": false
376
+ },
377
+ "use_peft": true,
378
+ "peft_task_type": "CAUSAL_LM",
379
+ "peft_inference_mode": false,
380
+ "peft_r": 8,
381
+ "peft_lora_alpha": 32,
382
+ "peft_lora_dropout": 0.1,
383
+ "fsdp": "none",
384
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
385
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
386
+ "fsdp_fsdp_min_num_params": 100000000.0,
387
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
388
+ "fsdp_fsdp_forward_prefetch": false,
389
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
390
+ "fsdp_fsdp_offload_params": false,
391
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
392
+ "fsdp_fsdp_use_orig_params": true,
393
+ "fsdp_fsdp_activation_checkpointing": false,
394
+ "fsdp_fsdp_limit_all_gathers": true,
395
+ "fsdp_fsdp_sync_module_states": true,
396
+ "use_git": true,
397
+ "random_seed": 0
398
+ }
399
+ },
400
+ "gsm8k": {
401
+ "exact_match,flexible-extract": 0.6618650492797574,
402
+ "exact_match,flexible-extract_stderr": 0.013030829145172212,
403
+ "args": {
404
+ "exp_name": "results/v2/gsm8k/01-ai/Yi-1.5-9B",
405
+ "task_name": "gsm8k",
406
+ "model_name": "01-ai/Yi-1.5-9B",
407
+ "dataset_param": {
408
+ "max_num_train": 50000,
409
+ "completion_only": false,
410
+ "apply_chat_template": false
411
+ },
412
+ "resume_from_checkpoint": false,
413
+ "train_param": {
414
+ "max_length": 1024,
415
+ "per_device_train_batch_size": 8,
416
+ "auto_find_batch_size": true,
417
+ "num_train_epochs": 5,
418
+ "max_steps": -1,
419
+ "logging_strategy": "epoch",
420
+ "save_strategy": "epoch",
421
+ "logging_steps": 0.2,
422
+ "save_steps": 0.2,
423
+ "gradient_accumulation_steps": 1,
424
+ "gradient_checkpointing": false,
425
+ "optim": "adamw_torch",
426
+ "learning_rate": 5e-05,
427
+ "weight_decay": 0.01,
428
+ "warmup_steps": 0,
429
+ "report_to": "tensorboard",
430
+ "save_safetensors": true,
431
+ "packing": false
432
+ },
433
+ "use_peft": true,
434
+ "peft_task_type": "CAUSAL_LM",
435
+ "peft_inference_mode": false,
436
+ "peft_r": 8,
437
+ "peft_lora_alpha": 32,
438
+ "peft_lora_dropout": 0.1,
439
+ "fsdp": "none",
440
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
441
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
442
+ "fsdp_fsdp_min_num_params": 100000000.0,
443
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
444
+ "fsdp_fsdp_forward_prefetch": false,
445
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
446
+ "fsdp_fsdp_offload_params": false,
447
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
448
+ "fsdp_fsdp_use_orig_params": true,
449
+ "fsdp_fsdp_activation_checkpointing": false,
450
+ "fsdp_fsdp_limit_all_gathers": true,
451
+ "fsdp_fsdp_sync_module_states": true,
452
+ "use_git": true,
453
+ "random_seed": 0
454
+ }
455
+ },
456
+ "arc_challenge": {
457
+ "acc_norm": 0.6151877133105802,
458
+ "acc_norm_stderr": 0.014218371065251102,
459
+ "args": {
460
+ "exp_name": "results/v2/arc_challenge/01-ai/Yi-1.5-9B",
461
+ "task_name": "arc_challenge",
462
+ "model_name": "01-ai/Yi-1.5-9B",
463
+ "dataset_param": {
464
+ "max_num_train": 50000,
465
+ "completion_only": false,
466
+ "apply_chat_template": false
467
+ },
468
+ "resume_from_checkpoint": false,
469
+ "train_param": {
470
+ "max_length": 1024,
471
+ "per_device_train_batch_size": 8,
472
+ "auto_find_batch_size": true,
473
+ "num_train_epochs": 5,
474
+ "max_steps": -1,
475
+ "logging_strategy": "epoch",
476
+ "save_strategy": "epoch",
477
+ "logging_steps": 0.2,
478
+ "save_steps": 0.2,
479
+ "gradient_accumulation_steps": 1,
480
+ "gradient_checkpointing": false,
481
+ "optim": "adamw_torch",
482
+ "learning_rate": 5e-05,
483
+ "weight_decay": 0.01,
484
+ "warmup_steps": 0,
485
+ "report_to": "tensorboard",
486
+ "save_safetensors": true,
487
+ "packing": false
488
+ },
489
+ "use_peft": true,
490
+ "peft_task_type": "CAUSAL_LM",
491
+ "peft_inference_mode": false,
492
+ "peft_r": 8,
493
+ "peft_lora_alpha": 32,
494
+ "peft_lora_dropout": 0.1,
495
+ "fsdp": "none",
496
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
497
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
498
+ "fsdp_fsdp_min_num_params": 100000000.0,
499
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
500
+ "fsdp_fsdp_forward_prefetch": false,
501
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
502
+ "fsdp_fsdp_offload_params": false,
503
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
504
+ "fsdp_fsdp_use_orig_params": true,
505
+ "fsdp_fsdp_activation_checkpointing": false,
506
+ "fsdp_fsdp_limit_all_gathers": true,
507
+ "fsdp_fsdp_sync_module_states": true,
508
+ "use_git": true,
509
+ "random_seed": 0
510
+ }
511
+ }
512
+ }
513
+ }
01-ai/Yi-6B-Chat/results_2026-01-05-04-27-51.json ADDED
@@ -0,0 +1,513 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "model_name": "01-ai/Yi-6B-Chat",
4
+ "model_type": "instruction-tuned",
5
+ "params": 6.0
6
+ },
7
+ "results": {
8
+ "medmcqa": {
9
+ "acc_norm": 0.510160172125269,
10
+ "acc_norm_stderr": 0.007730156881915955,
11
+ "args": {
12
+ "exp_name": "results/v1/medmcqa/01-ai/Yi-6B-Chat",
13
+ "task_name": "medmcqa",
14
+ "model_name": "01-ai/Yi-6B-Chat",
15
+ "dataset_param": {
16
+ "max_num_train": 50000,
17
+ "completion_only": false,
18
+ "apply_chat_template": true
19
+ },
20
+ "resume_from_checkpoint": false,
21
+ "train_param": {
22
+ "max_length": 1024,
23
+ "per_device_train_batch_size": 8,
24
+ "auto_find_batch_size": true,
25
+ "num_train_epochs": 1,
26
+ "max_steps": -1,
27
+ "logging_strategy": "steps",
28
+ "save_strategy": "steps",
29
+ "logging_steps": 0.2,
30
+ "save_steps": 0.2,
31
+ "gradient_accumulation_steps": 1,
32
+ "gradient_checkpointing": false,
33
+ "optim": "adamw_torch",
34
+ "learning_rate": 5e-05,
35
+ "weight_decay": 0.01,
36
+ "warmup_steps": 0,
37
+ "report_to": "tensorboard",
38
+ "save_safetensors": true,
39
+ "packing": false
40
+ },
41
+ "use_peft": true,
42
+ "peft_task_type": "CAUSAL_LM",
43
+ "peft_inference_mode": false,
44
+ "peft_r": 8,
45
+ "peft_lora_alpha": 32,
46
+ "peft_lora_dropout": 0.1,
47
+ "fsdp": "none",
48
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
49
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
50
+ "fsdp_fsdp_min_num_params": 100000000.0,
51
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
52
+ "fsdp_fsdp_forward_prefetch": false,
53
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
54
+ "fsdp_fsdp_offload_params": false,
55
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
56
+ "fsdp_fsdp_use_orig_params": true,
57
+ "fsdp_fsdp_activation_checkpointing": false,
58
+ "fsdp_fsdp_limit_all_gathers": true,
59
+ "fsdp_fsdp_sync_module_states": true,
60
+ "use_git": true,
61
+ "random_seed": 0
62
+ }
63
+ },
64
+ "nq_open": {
65
+ "exact_match,remove_whitespace": 0.27590027700831027,
66
+ "exact_match,remove_whitespace_stderr": 0.0074401521672125236,
67
+ "args": {
68
+ "exp_name": "results/v1/nq_open/01-ai/Yi-6B-Chat",
69
+ "task_name": "nq_open",
70
+ "model_name": "01-ai/Yi-6B-Chat",
71
+ "dataset_param": {
72
+ "max_num_train": 50000,
73
+ "completion_only": false,
74
+ "apply_chat_template": true
75
+ },
76
+ "resume_from_checkpoint": false,
77
+ "train_param": {
78
+ "max_length": 1024,
79
+ "per_device_train_batch_size": 8,
80
+ "auto_find_batch_size": true,
81
+ "num_train_epochs": 1,
82
+ "max_steps": -1,
83
+ "logging_strategy": "steps",
84
+ "save_strategy": "steps",
85
+ "logging_steps": 0.2,
86
+ "save_steps": 0.2,
87
+ "gradient_accumulation_steps": 1,
88
+ "gradient_checkpointing": false,
89
+ "optim": "adamw_torch",
90
+ "learning_rate": 5e-05,
91
+ "weight_decay": 0.01,
92
+ "warmup_steps": 0,
93
+ "report_to": "tensorboard",
94
+ "save_safetensors": true,
95
+ "packing": false
96
+ },
97
+ "use_peft": true,
98
+ "peft_task_type": "CAUSAL_LM",
99
+ "peft_inference_mode": false,
100
+ "peft_r": 8,
101
+ "peft_lora_alpha": 32,
102
+ "peft_lora_dropout": 0.1,
103
+ "fsdp": "none",
104
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
105
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
106
+ "fsdp_fsdp_min_num_params": 100000000.0,
107
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
108
+ "fsdp_fsdp_forward_prefetch": false,
109
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
110
+ "fsdp_fsdp_offload_params": false,
111
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
112
+ "fsdp_fsdp_use_orig_params": true,
113
+ "fsdp_fsdp_activation_checkpointing": false,
114
+ "fsdp_fsdp_limit_all_gathers": true,
115
+ "fsdp_fsdp_sync_module_states": true,
116
+ "use_git": true,
117
+ "random_seed": 0
118
+ }
119
+ },
120
+ "winogrande": {
121
+ "acc": 0.8058405682715075,
122
+ "acc_stderr": 0.011116983392392636,
123
+ "args": {
124
+ "exp_name": "results/v1/winogrande/01-ai/Yi-6B-Chat",
125
+ "task_name": "winogrande",
126
+ "model_name": "01-ai/Yi-6B-Chat",
127
+ "dataset_param": {
128
+ "max_num_train": 50000,
129
+ "completion_only": false,
130
+ "apply_chat_template": true
131
+ },
132
+ "resume_from_checkpoint": false,
133
+ "train_param": {
134
+ "max_length": 1024,
135
+ "per_device_train_batch_size": 8,
136
+ "auto_find_batch_size": true,
137
+ "num_train_epochs": 1,
138
+ "max_steps": -1,
139
+ "logging_strategy": "steps",
140
+ "save_strategy": "steps",
141
+ "logging_steps": 0.2,
142
+ "save_steps": 0.2,
143
+ "gradient_accumulation_steps": 1,
144
+ "gradient_checkpointing": false,
145
+ "optim": "adamw_torch",
146
+ "learning_rate": 5e-05,
147
+ "weight_decay": 0.01,
148
+ "warmup_steps": 0,
149
+ "report_to": "tensorboard",
150
+ "save_safetensors": true,
151
+ "packing": false
152
+ },
153
+ "use_peft": true,
154
+ "peft_task_type": "CAUSAL_LM",
155
+ "peft_inference_mode": false,
156
+ "peft_r": 8,
157
+ "peft_lora_alpha": 32,
158
+ "peft_lora_dropout": 0.1,
159
+ "fsdp": "none",
160
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
161
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
162
+ "fsdp_fsdp_min_num_params": 100000000.0,
163
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
164
+ "fsdp_fsdp_forward_prefetch": false,
165
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
166
+ "fsdp_fsdp_offload_params": false,
167
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
168
+ "fsdp_fsdp_use_orig_params": true,
169
+ "fsdp_fsdp_activation_checkpointing": false,
170
+ "fsdp_fsdp_limit_all_gathers": true,
171
+ "fsdp_fsdp_sync_module_states": true,
172
+ "use_git": true,
173
+ "random_seed": 0
174
+ }
175
+ },
176
+ "hellaswag": {
177
+ "acc_norm": 0.7794,
178
+ "acc_norm_stderr": 0.004146719610529641,
179
+ "args": {
180
+ "exp_name": "results/v1/hellaswag/01-ai/Yi-6B-Chat",
181
+ "task_name": "hellaswag",
182
+ "model_name": "01-ai/Yi-6B-Chat",
183
+ "dataset_param": {
184
+ "max_num_train": 50000,
185
+ "completion_only": false,
186
+ "apply_chat_template": true
187
+ },
188
+ "resume_from_checkpoint": false,
189
+ "train_param": {
190
+ "max_length": 1024,
191
+ "per_device_train_batch_size": 8,
192
+ "auto_find_batch_size": true,
193
+ "num_train_epochs": 1,
194
+ "max_steps": -1,
195
+ "logging_strategy": "steps",
196
+ "save_strategy": "steps",
197
+ "logging_steps": 0.2,
198
+ "save_steps": 0.2,
199
+ "gradient_accumulation_steps": 1,
200
+ "gradient_checkpointing": false,
201
+ "optim": "adamw_torch",
202
+ "learning_rate": 5e-05,
203
+ "weight_decay": 0.01,
204
+ "warmup_steps": 0,
205
+ "report_to": "tensorboard",
206
+ "save_safetensors": true,
207
+ "packing": false
208
+ },
209
+ "use_peft": true,
210
+ "peft_task_type": "CAUSAL_LM",
211
+ "peft_inference_mode": false,
212
+ "peft_r": 8,
213
+ "peft_lora_alpha": 32,
214
+ "peft_lora_dropout": 0.1,
215
+ "fsdp": "none",
216
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
217
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
218
+ "fsdp_fsdp_min_num_params": 100000000.0,
219
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
220
+ "fsdp_fsdp_forward_prefetch": false,
221
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
222
+ "fsdp_fsdp_offload_params": false,
223
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
224
+ "fsdp_fsdp_use_orig_params": true,
225
+ "fsdp_fsdp_activation_checkpointing": false,
226
+ "fsdp_fsdp_limit_all_gathers": true,
227
+ "fsdp_fsdp_sync_module_states": true,
228
+ "use_git": true,
229
+ "random_seed": 0
230
+ }
231
+ },
232
+ "social_iqa": {
233
+ "acc": 0.5649948822927329,
234
+ "acc_stderr": 0.0112180744655065,
235
+ "args": {
236
+ "exp_name": "results/v1/social_iqa/01-ai/Yi-6B-Chat",
237
+ "task_name": "social_iqa",
238
+ "model_name": "01-ai/Yi-6B-Chat",
239
+ "dataset_param": {
240
+ "max_num_train": 50000,
241
+ "completion_only": false,
242
+ "apply_chat_template": true
243
+ },
244
+ "resume_from_checkpoint": false,
245
+ "train_param": {
246
+ "max_length": 1024,
247
+ "per_device_train_batch_size": 8,
248
+ "auto_find_batch_size": true,
249
+ "num_train_epochs": 1,
250
+ "max_steps": -1,
251
+ "logging_strategy": "steps",
252
+ "save_strategy": "steps",
253
+ "logging_steps": 0.2,
254
+ "save_steps": 0.2,
255
+ "gradient_accumulation_steps": 1,
256
+ "gradient_checkpointing": false,
257
+ "optim": "adamw_torch",
258
+ "learning_rate": 5e-05,
259
+ "weight_decay": 0.01,
260
+ "warmup_steps": 0,
261
+ "report_to": "tensorboard",
262
+ "save_safetensors": true,
263
+ "packing": false
264
+ },
265
+ "use_peft": true,
266
+ "peft_task_type": "CAUSAL_LM",
267
+ "peft_inference_mode": false,
268
+ "peft_r": 8,
269
+ "peft_lora_alpha": 32,
270
+ "peft_lora_dropout": 0.1,
271
+ "fsdp": "none",
272
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
273
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
274
+ "fsdp_fsdp_min_num_params": 100000000.0,
275
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
276
+ "fsdp_fsdp_forward_prefetch": false,
277
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
278
+ "fsdp_fsdp_offload_params": false,
279
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
280
+ "fsdp_fsdp_use_orig_params": true,
281
+ "fsdp_fsdp_activation_checkpointing": false,
282
+ "fsdp_fsdp_limit_all_gathers": true,
283
+ "fsdp_fsdp_sync_module_states": true,
284
+ "use_git": true,
285
+ "random_seed": 0
286
+ }
287
+ },
288
+ "piqa": {
289
+ "acc_norm": 0.8128400435255713,
290
+ "acc_norm_stderr": 0.00910027329047355,
291
+ "args": {
292
+ "exp_name": "results/v2/piqa/01-ai/Yi-6B-Chat",
293
+ "task_name": "piqa",
294
+ "model_name": "01-ai/Yi-6B-Chat",
295
+ "dataset_param": {
296
+ "max_num_train": 50000,
297
+ "completion_only": false,
298
+ "apply_chat_template": true
299
+ },
300
+ "resume_from_checkpoint": false,
301
+ "train_param": {
302
+ "max_length": 1024,
303
+ "per_device_train_batch_size": 8,
304
+ "auto_find_batch_size": true,
305
+ "num_train_epochs": 5,
306
+ "max_steps": -1,
307
+ "logging_strategy": "epoch",
308
+ "save_strategy": "epoch",
309
+ "logging_steps": 0.2,
310
+ "save_steps": 0.2,
311
+ "gradient_accumulation_steps": 1,
312
+ "gradient_checkpointing": false,
313
+ "optim": "adamw_torch",
314
+ "learning_rate": 5e-05,
315
+ "weight_decay": 0.01,
316
+ "warmup_steps": 0,
317
+ "report_to": "tensorboard",
318
+ "save_safetensors": true,
319
+ "packing": false
320
+ },
321
+ "use_peft": true,
322
+ "peft_task_type": "CAUSAL_LM",
323
+ "peft_inference_mode": false,
324
+ "peft_r": 8,
325
+ "peft_lora_alpha": 32,
326
+ "peft_lora_dropout": 0.1,
327
+ "fsdp": "none",
328
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
329
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
330
+ "fsdp_fsdp_min_num_params": 100000000.0,
331
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
332
+ "fsdp_fsdp_forward_prefetch": false,
333
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
334
+ "fsdp_fsdp_offload_params": false,
335
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
336
+ "fsdp_fsdp_use_orig_params": true,
337
+ "fsdp_fsdp_activation_checkpointing": false,
338
+ "fsdp_fsdp_limit_all_gathers": true,
339
+ "fsdp_fsdp_sync_module_states": true,
340
+ "use_git": true,
341
+ "random_seed": 0
342
+ }
343
+ },
344
+ "commonsense_qa": {
345
+ "acc": 0.823914823914824,
346
+ "acc_stderr": 0.010904923946045584,
347
+ "args": {
348
+ "exp_name": "results/v2/commonsense_qa/01-ai/Yi-6B-Chat",
349
+ "task_name": "commonsense_qa",
350
+ "model_name": "01-ai/Yi-6B-Chat",
351
+ "dataset_param": {
352
+ "max_num_train": 50000,
353
+ "completion_only": false,
354
+ "apply_chat_template": true
355
+ },
356
+ "resume_from_checkpoint": false,
357
+ "train_param": {
358
+ "max_length": 1024,
359
+ "per_device_train_batch_size": 8,
360
+ "auto_find_batch_size": true,
361
+ "num_train_epochs": 5,
362
+ "max_steps": -1,
363
+ "logging_strategy": "epoch",
364
+ "save_strategy": "epoch",
365
+ "logging_steps": 0.2,
366
+ "save_steps": 0.2,
367
+ "gradient_accumulation_steps": 1,
368
+ "gradient_checkpointing": false,
369
+ "optim": "adamw_torch",
370
+ "learning_rate": 5e-05,
371
+ "weight_decay": 0.01,
372
+ "warmup_steps": 0,
373
+ "report_to": "tensorboard",
374
+ "save_safetensors": true,
375
+ "packing": false
376
+ },
377
+ "use_peft": true,
378
+ "peft_task_type": "CAUSAL_LM",
379
+ "peft_inference_mode": false,
380
+ "peft_r": 8,
381
+ "peft_lora_alpha": 32,
382
+ "peft_lora_dropout": 0.1,
383
+ "fsdp": "none",
384
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
385
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
386
+ "fsdp_fsdp_min_num_params": 100000000.0,
387
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
388
+ "fsdp_fsdp_forward_prefetch": false,
389
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
390
+ "fsdp_fsdp_offload_params": false,
391
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
392
+ "fsdp_fsdp_use_orig_params": true,
393
+ "fsdp_fsdp_activation_checkpointing": false,
394
+ "fsdp_fsdp_limit_all_gathers": true,
395
+ "fsdp_fsdp_sync_module_states": true,
396
+ "use_git": true,
397
+ "random_seed": 0
398
+ }
399
+ },
400
+ "gsm8k": {
401
+ "exact_match,flexible-extract": 0.41470811220621684,
402
+ "exact_match,flexible-extract_stderr": 0.01357062384230451,
403
+ "args": {
404
+ "exp_name": "results/v2/gsm8k/01-ai/Yi-6B-Chat",
405
+ "task_name": "gsm8k",
406
+ "model_name": "01-ai/Yi-6B-Chat",
407
+ "dataset_param": {
408
+ "max_num_train": 50000,
409
+ "completion_only": false,
410
+ "apply_chat_template": true
411
+ },
412
+ "resume_from_checkpoint": false,
413
+ "train_param": {
414
+ "max_length": 1024,
415
+ "per_device_train_batch_size": 8,
416
+ "auto_find_batch_size": true,
417
+ "num_train_epochs": 5,
418
+ "max_steps": -1,
419
+ "logging_strategy": "epoch",
420
+ "save_strategy": "epoch",
421
+ "logging_steps": 0.2,
422
+ "save_steps": 0.2,
423
+ "gradient_accumulation_steps": 1,
424
+ "gradient_checkpointing": false,
425
+ "optim": "adamw_torch",
426
+ "learning_rate": 5e-05,
427
+ "weight_decay": 0.01,
428
+ "warmup_steps": 0,
429
+ "report_to": "tensorboard",
430
+ "save_safetensors": true,
431
+ "packing": false
432
+ },
433
+ "use_peft": true,
434
+ "peft_task_type": "CAUSAL_LM",
435
+ "peft_inference_mode": false,
436
+ "peft_r": 8,
437
+ "peft_lora_alpha": 32,
438
+ "peft_lora_dropout": 0.1,
439
+ "fsdp": "none",
440
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
441
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
442
+ "fsdp_fsdp_min_num_params": 100000000.0,
443
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
444
+ "fsdp_fsdp_forward_prefetch": false,
445
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
446
+ "fsdp_fsdp_offload_params": false,
447
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
448
+ "fsdp_fsdp_use_orig_params": true,
449
+ "fsdp_fsdp_activation_checkpointing": false,
450
+ "fsdp_fsdp_limit_all_gathers": true,
451
+ "fsdp_fsdp_sync_module_states": true,
452
+ "use_git": true,
453
+ "random_seed": 0
454
+ }
455
+ },
456
+ "arc_challenge": {
457
+ "acc_norm": 0.5102389078498294,
458
+ "acc_norm_stderr": 0.014608326906285012,
459
+ "args": {
460
+ "exp_name": "results/v2/arc_challenge/01-ai/Yi-6B-Chat",
461
+ "task_name": "arc_challenge",
462
+ "model_name": "01-ai/Yi-6B-Chat",
463
+ "dataset_param": {
464
+ "max_num_train": 50000,
465
+ "completion_only": false,
466
+ "apply_chat_template": true
467
+ },
468
+ "resume_from_checkpoint": false,
469
+ "train_param": {
470
+ "max_length": 1024,
471
+ "per_device_train_batch_size": 8,
472
+ "auto_find_batch_size": true,
473
+ "num_train_epochs": 5,
474
+ "max_steps": -1,
475
+ "logging_strategy": "epoch",
476
+ "save_strategy": "epoch",
477
+ "logging_steps": 0.2,
478
+ "save_steps": 0.2,
479
+ "gradient_accumulation_steps": 1,
480
+ "gradient_checkpointing": false,
481
+ "optim": "adamw_torch",
482
+ "learning_rate": 5e-05,
483
+ "weight_decay": 0.01,
484
+ "warmup_steps": 0,
485
+ "report_to": "tensorboard",
486
+ "save_safetensors": true,
487
+ "packing": false
488
+ },
489
+ "use_peft": true,
490
+ "peft_task_type": "CAUSAL_LM",
491
+ "peft_inference_mode": false,
492
+ "peft_r": 8,
493
+ "peft_lora_alpha": 32,
494
+ "peft_lora_dropout": 0.1,
495
+ "fsdp": "none",
496
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
497
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
498
+ "fsdp_fsdp_min_num_params": 100000000.0,
499
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
500
+ "fsdp_fsdp_forward_prefetch": false,
501
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
502
+ "fsdp_fsdp_offload_params": false,
503
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
504
+ "fsdp_fsdp_use_orig_params": true,
505
+ "fsdp_fsdp_activation_checkpointing": false,
506
+ "fsdp_fsdp_limit_all_gathers": true,
507
+ "fsdp_fsdp_sync_module_states": true,
508
+ "use_git": true,
509
+ "random_seed": 0
510
+ }
511
+ }
512
+ }
513
+ }
01-ai/Yi-6B/results_2026-01-05-04-27-50.json ADDED
@@ -0,0 +1,513 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "model_name": "01-ai/Yi-6B",
4
+ "model_type": "pretrained",
5
+ "params": 6.0
6
+ },
7
+ "results": {
8
+ "medmcqa": {
9
+ "acc_norm": 0.5125508008606263,
10
+ "acc_norm_stderr": 0.007729317094939668,
11
+ "args": {
12
+ "exp_name": "results/v1/medmcqa/01-ai/Yi-6B",
13
+ "task_name": "medmcqa",
14
+ "model_name": "01-ai/Yi-6B",
15
+ "dataset_param": {
16
+ "max_num_train": 50000,
17
+ "completion_only": false,
18
+ "apply_chat_template": false
19
+ },
20
+ "resume_from_checkpoint": false,
21
+ "train_param": {
22
+ "max_length": 1024,
23
+ "per_device_train_batch_size": 8,
24
+ "auto_find_batch_size": true,
25
+ "num_train_epochs": 1,
26
+ "max_steps": -1,
27
+ "logging_strategy": "steps",
28
+ "save_strategy": "steps",
29
+ "logging_steps": 0.2,
30
+ "save_steps": 0.2,
31
+ "gradient_accumulation_steps": 1,
32
+ "gradient_checkpointing": false,
33
+ "optim": "adamw_torch",
34
+ "learning_rate": 5e-05,
35
+ "weight_decay": 0.01,
36
+ "warmup_steps": 0,
37
+ "report_to": "tensorboard",
38
+ "save_safetensors": true,
39
+ "packing": false
40
+ },
41
+ "use_peft": true,
42
+ "peft_task_type": "CAUSAL_LM",
43
+ "peft_inference_mode": false,
44
+ "peft_r": 8,
45
+ "peft_lora_alpha": 32,
46
+ "peft_lora_dropout": 0.1,
47
+ "fsdp": "none",
48
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
49
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
50
+ "fsdp_fsdp_min_num_params": 100000000.0,
51
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
52
+ "fsdp_fsdp_forward_prefetch": false,
53
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
54
+ "fsdp_fsdp_offload_params": false,
55
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
56
+ "fsdp_fsdp_use_orig_params": true,
57
+ "fsdp_fsdp_activation_checkpointing": false,
58
+ "fsdp_fsdp_limit_all_gathers": true,
59
+ "fsdp_fsdp_sync_module_states": true,
60
+ "use_git": true,
61
+ "random_seed": 0
62
+ }
63
+ },
64
+ "nq_open": {
65
+ "exact_match,remove_whitespace": 0.28310249307479224,
66
+ "exact_match,remove_whitespace_stderr": 0.007499061943546498,
67
+ "args": {
68
+ "exp_name": "results/v1/nq_open/01-ai/Yi-6B",
69
+ "task_name": "nq_open",
70
+ "model_name": "01-ai/Yi-6B",
71
+ "dataset_param": {
72
+ "max_num_train": 50000,
73
+ "completion_only": false,
74
+ "apply_chat_template": false
75
+ },
76
+ "resume_from_checkpoint": false,
77
+ "train_param": {
78
+ "max_length": 1024,
79
+ "per_device_train_batch_size": 8,
80
+ "auto_find_batch_size": true,
81
+ "num_train_epochs": 1,
82
+ "max_steps": -1,
83
+ "logging_strategy": "steps",
84
+ "save_strategy": "steps",
85
+ "logging_steps": 0.2,
86
+ "save_steps": 0.2,
87
+ "gradient_accumulation_steps": 1,
88
+ "gradient_checkpointing": false,
89
+ "optim": "adamw_torch",
90
+ "learning_rate": 5e-05,
91
+ "weight_decay": 0.01,
92
+ "warmup_steps": 0,
93
+ "report_to": "tensorboard",
94
+ "save_safetensors": true,
95
+ "packing": false
96
+ },
97
+ "use_peft": true,
98
+ "peft_task_type": "CAUSAL_LM",
99
+ "peft_inference_mode": false,
100
+ "peft_r": 8,
101
+ "peft_lora_alpha": 32,
102
+ "peft_lora_dropout": 0.1,
103
+ "fsdp": "none",
104
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
105
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
106
+ "fsdp_fsdp_min_num_params": 100000000.0,
107
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
108
+ "fsdp_fsdp_forward_prefetch": false,
109
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
110
+ "fsdp_fsdp_offload_params": false,
111
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
112
+ "fsdp_fsdp_use_orig_params": true,
113
+ "fsdp_fsdp_activation_checkpointing": false,
114
+ "fsdp_fsdp_limit_all_gathers": true,
115
+ "fsdp_fsdp_sync_module_states": true,
116
+ "use_git": true,
117
+ "random_seed": 0
118
+ }
119
+ },
120
+ "winogrande": {
121
+ "acc": 0.7742699289660616,
122
+ "acc_stderr": 0.011749626260902556,
123
+ "args": {
124
+ "exp_name": "results/v1/winogrande/01-ai/Yi-6B",
125
+ "task_name": "winogrande",
126
+ "model_name": "01-ai/Yi-6B",
127
+ "dataset_param": {
128
+ "max_num_train": 50000,
129
+ "completion_only": false,
130
+ "apply_chat_template": false
131
+ },
132
+ "resume_from_checkpoint": false,
133
+ "train_param": {
134
+ "max_length": 1024,
135
+ "per_device_train_batch_size": 8,
136
+ "auto_find_batch_size": true,
137
+ "num_train_epochs": 1,
138
+ "max_steps": -1,
139
+ "logging_strategy": "steps",
140
+ "save_strategy": "steps",
141
+ "logging_steps": 0.2,
142
+ "save_steps": 0.2,
143
+ "gradient_accumulation_steps": 1,
144
+ "gradient_checkpointing": false,
145
+ "optim": "adamw_torch",
146
+ "learning_rate": 5e-05,
147
+ "weight_decay": 0.01,
148
+ "warmup_steps": 0,
149
+ "report_to": "tensorboard",
150
+ "save_safetensors": true,
151
+ "packing": false
152
+ },
153
+ "use_peft": true,
154
+ "peft_task_type": "CAUSAL_LM",
155
+ "peft_inference_mode": false,
156
+ "peft_r": 8,
157
+ "peft_lora_alpha": 32,
158
+ "peft_lora_dropout": 0.1,
159
+ "fsdp": "none",
160
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
161
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
162
+ "fsdp_fsdp_min_num_params": 100000000.0,
163
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
164
+ "fsdp_fsdp_forward_prefetch": false,
165
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
166
+ "fsdp_fsdp_offload_params": false,
167
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
168
+ "fsdp_fsdp_use_orig_params": true,
169
+ "fsdp_fsdp_activation_checkpointing": false,
170
+ "fsdp_fsdp_limit_all_gathers": true,
171
+ "fsdp_fsdp_sync_module_states": true,
172
+ "use_git": true,
173
+ "random_seed": 0
174
+ }
175
+ },
176
+ "hellaswag": {
177
+ "acc_norm": 0.784,
178
+ "acc_norm_stderr": 0.004115348540447037,
179
+ "args": {
180
+ "exp_name": "results/v1/hellaswag/01-ai/Yi-6B",
181
+ "task_name": "hellaswag",
182
+ "model_name": "01-ai/Yi-6B",
183
+ "dataset_param": {
184
+ "max_num_train": 50000,
185
+ "completion_only": false,
186
+ "apply_chat_template": false
187
+ },
188
+ "resume_from_checkpoint": false,
189
+ "train_param": {
190
+ "max_length": 1024,
191
+ "per_device_train_batch_size": 8,
192
+ "auto_find_batch_size": true,
193
+ "num_train_epochs": 1,
194
+ "max_steps": -1,
195
+ "logging_strategy": "steps",
196
+ "save_strategy": "steps",
197
+ "logging_steps": 0.2,
198
+ "save_steps": 0.2,
199
+ "gradient_accumulation_steps": 1,
200
+ "gradient_checkpointing": false,
201
+ "optim": "adamw_torch",
202
+ "learning_rate": 5e-05,
203
+ "weight_decay": 0.01,
204
+ "warmup_steps": 0,
205
+ "report_to": "tensorboard",
206
+ "save_safetensors": true,
207
+ "packing": false
208
+ },
209
+ "use_peft": true,
210
+ "peft_task_type": "CAUSAL_LM",
211
+ "peft_inference_mode": false,
212
+ "peft_r": 8,
213
+ "peft_lora_alpha": 32,
214
+ "peft_lora_dropout": 0.1,
215
+ "fsdp": "none",
216
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
217
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
218
+ "fsdp_fsdp_min_num_params": 100000000.0,
219
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
220
+ "fsdp_fsdp_forward_prefetch": false,
221
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
222
+ "fsdp_fsdp_offload_params": false,
223
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
224
+ "fsdp_fsdp_use_orig_params": true,
225
+ "fsdp_fsdp_activation_checkpointing": false,
226
+ "fsdp_fsdp_limit_all_gathers": true,
227
+ "fsdp_fsdp_sync_module_states": true,
228
+ "use_git": true,
229
+ "random_seed": 0
230
+ }
231
+ },
232
+ "social_iqa": {
233
+ "acc": 0.5706243602865916,
234
+ "acc_stderr": 0.011200637273721802,
235
+ "args": {
236
+ "exp_name": "results/v1/social_iqa/01-ai/Yi-6B",
237
+ "task_name": "social_iqa",
238
+ "model_name": "01-ai/Yi-6B",
239
+ "dataset_param": {
240
+ "max_num_train": 50000,
241
+ "completion_only": false,
242
+ "apply_chat_template": false
243
+ },
244
+ "resume_from_checkpoint": false,
245
+ "train_param": {
246
+ "max_length": 1024,
247
+ "per_device_train_batch_size": 8,
248
+ "auto_find_batch_size": true,
249
+ "num_train_epochs": 1,
250
+ "max_steps": -1,
251
+ "logging_strategy": "steps",
252
+ "save_strategy": "steps",
253
+ "logging_steps": 0.2,
254
+ "save_steps": 0.2,
255
+ "gradient_accumulation_steps": 1,
256
+ "gradient_checkpointing": false,
257
+ "optim": "adamw_torch",
258
+ "learning_rate": 5e-05,
259
+ "weight_decay": 0.01,
260
+ "warmup_steps": 0,
261
+ "report_to": "tensorboard",
262
+ "save_safetensors": true,
263
+ "packing": false
264
+ },
265
+ "use_peft": true,
266
+ "peft_task_type": "CAUSAL_LM",
267
+ "peft_inference_mode": false,
268
+ "peft_r": 8,
269
+ "peft_lora_alpha": 32,
270
+ "peft_lora_dropout": 0.1,
271
+ "fsdp": "none",
272
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
273
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
274
+ "fsdp_fsdp_min_num_params": 100000000.0,
275
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
276
+ "fsdp_fsdp_forward_prefetch": false,
277
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
278
+ "fsdp_fsdp_offload_params": false,
279
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
280
+ "fsdp_fsdp_use_orig_params": true,
281
+ "fsdp_fsdp_activation_checkpointing": false,
282
+ "fsdp_fsdp_limit_all_gathers": true,
283
+ "fsdp_fsdp_sync_module_states": true,
284
+ "use_git": true,
285
+ "random_seed": 0
286
+ }
287
+ },
288
+ "piqa": {
289
+ "acc_norm": 0.8128400435255713,
290
+ "acc_norm_stderr": 0.009100273290473552,
291
+ "args": {
292
+ "exp_name": "results/v2/piqa/01-ai/Yi-6B",
293
+ "task_name": "piqa",
294
+ "model_name": "01-ai/Yi-6B",
295
+ "dataset_param": {
296
+ "max_num_train": 50000,
297
+ "completion_only": false,
298
+ "apply_chat_template": false
299
+ },
300
+ "resume_from_checkpoint": false,
301
+ "train_param": {
302
+ "max_length": 1024,
303
+ "per_device_train_batch_size": 8,
304
+ "auto_find_batch_size": true,
305
+ "num_train_epochs": 5,
306
+ "max_steps": -1,
307
+ "logging_strategy": "epoch",
308
+ "save_strategy": "epoch",
309
+ "logging_steps": 0.2,
310
+ "save_steps": 0.2,
311
+ "gradient_accumulation_steps": 1,
312
+ "gradient_checkpointing": false,
313
+ "optim": "adamw_torch",
314
+ "learning_rate": 5e-05,
315
+ "weight_decay": 0.01,
316
+ "warmup_steps": 0,
317
+ "report_to": "tensorboard",
318
+ "save_safetensors": true,
319
+ "packing": false
320
+ },
321
+ "use_peft": true,
322
+ "peft_task_type": "CAUSAL_LM",
323
+ "peft_inference_mode": false,
324
+ "peft_r": 8,
325
+ "peft_lora_alpha": 32,
326
+ "peft_lora_dropout": 0.1,
327
+ "fsdp": "none",
328
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
329
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
330
+ "fsdp_fsdp_min_num_params": 100000000.0,
331
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
332
+ "fsdp_fsdp_forward_prefetch": false,
333
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
334
+ "fsdp_fsdp_offload_params": false,
335
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
336
+ "fsdp_fsdp_use_orig_params": true,
337
+ "fsdp_fsdp_activation_checkpointing": false,
338
+ "fsdp_fsdp_limit_all_gathers": true,
339
+ "fsdp_fsdp_sync_module_states": true,
340
+ "use_git": true,
341
+ "random_seed": 0
342
+ }
343
+ },
344
+ "commonsense_qa": {
345
+ "acc": 0.8214578214578214,
346
+ "acc_stderr": 0.010964356148768764,
347
+ "args": {
348
+ "exp_name": "results/v2/commonsense_qa/01-ai/Yi-6B",
349
+ "task_name": "commonsense_qa",
350
+ "model_name": "01-ai/Yi-6B",
351
+ "dataset_param": {
352
+ "max_num_train": 50000,
353
+ "completion_only": false,
354
+ "apply_chat_template": false
355
+ },
356
+ "resume_from_checkpoint": false,
357
+ "train_param": {
358
+ "max_length": 1024,
359
+ "per_device_train_batch_size": 8,
360
+ "auto_find_batch_size": true,
361
+ "num_train_epochs": 5,
362
+ "max_steps": -1,
363
+ "logging_strategy": "epoch",
364
+ "save_strategy": "epoch",
365
+ "logging_steps": 0.2,
366
+ "save_steps": 0.2,
367
+ "gradient_accumulation_steps": 1,
368
+ "gradient_checkpointing": false,
369
+ "optim": "adamw_torch",
370
+ "learning_rate": 5e-05,
371
+ "weight_decay": 0.01,
372
+ "warmup_steps": 0,
373
+ "report_to": "tensorboard",
374
+ "save_safetensors": true,
375
+ "packing": false
376
+ },
377
+ "use_peft": true,
378
+ "peft_task_type": "CAUSAL_LM",
379
+ "peft_inference_mode": false,
380
+ "peft_r": 8,
381
+ "peft_lora_alpha": 32,
382
+ "peft_lora_dropout": 0.1,
383
+ "fsdp": "none",
384
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
385
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
386
+ "fsdp_fsdp_min_num_params": 100000000.0,
387
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
388
+ "fsdp_fsdp_forward_prefetch": false,
389
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
390
+ "fsdp_fsdp_offload_params": false,
391
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
392
+ "fsdp_fsdp_use_orig_params": true,
393
+ "fsdp_fsdp_activation_checkpointing": false,
394
+ "fsdp_fsdp_limit_all_gathers": true,
395
+ "fsdp_fsdp_sync_module_states": true,
396
+ "use_git": true,
397
+ "random_seed": 0
398
+ }
399
+ },
400
+ "gsm8k": {
401
+ "exact_match,flexible-extract": 0.43290371493555724,
402
+ "exact_match,flexible-extract_stderr": 0.013647916362576057,
403
+ "args": {
404
+ "exp_name": "results/v2/gsm8k/01-ai/Yi-6B",
405
+ "task_name": "gsm8k",
406
+ "model_name": "01-ai/Yi-6B",
407
+ "dataset_param": {
408
+ "max_num_train": 50000,
409
+ "completion_only": false,
410
+ "apply_chat_template": false
411
+ },
412
+ "resume_from_checkpoint": false,
413
+ "train_param": {
414
+ "max_length": 1024,
415
+ "per_device_train_batch_size": 8,
416
+ "auto_find_batch_size": true,
417
+ "num_train_epochs": 5,
418
+ "max_steps": -1,
419
+ "logging_strategy": "epoch",
420
+ "save_strategy": "epoch",
421
+ "logging_steps": 0.2,
422
+ "save_steps": 0.2,
423
+ "gradient_accumulation_steps": 1,
424
+ "gradient_checkpointing": false,
425
+ "optim": "adamw_torch",
426
+ "learning_rate": 5e-05,
427
+ "weight_decay": 0.01,
428
+ "warmup_steps": 0,
429
+ "report_to": "tensorboard",
430
+ "save_safetensors": true,
431
+ "packing": false
432
+ },
433
+ "use_peft": true,
434
+ "peft_task_type": "CAUSAL_LM",
435
+ "peft_inference_mode": false,
436
+ "peft_r": 8,
437
+ "peft_lora_alpha": 32,
438
+ "peft_lora_dropout": 0.1,
439
+ "fsdp": "none",
440
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
441
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
442
+ "fsdp_fsdp_min_num_params": 100000000.0,
443
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
444
+ "fsdp_fsdp_forward_prefetch": false,
445
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
446
+ "fsdp_fsdp_offload_params": false,
447
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
448
+ "fsdp_fsdp_use_orig_params": true,
449
+ "fsdp_fsdp_activation_checkpointing": false,
450
+ "fsdp_fsdp_limit_all_gathers": true,
451
+ "fsdp_fsdp_sync_module_states": true,
452
+ "use_git": true,
453
+ "random_seed": 0
454
+ }
455
+ },
456
+ "arc_challenge": {
457
+ "acc_norm": 0.560580204778157,
458
+ "acc_norm_stderr": 0.014503747823580125,
459
+ "args": {
460
+ "exp_name": "results/v2/arc_challenge/01-ai/Yi-6B",
461
+ "task_name": "arc_challenge",
462
+ "model_name": "01-ai/Yi-6B",
463
+ "dataset_param": {
464
+ "max_num_train": 50000,
465
+ "completion_only": false,
466
+ "apply_chat_template": false
467
+ },
468
+ "resume_from_checkpoint": false,
469
+ "train_param": {
470
+ "max_length": 1024,
471
+ "per_device_train_batch_size": 8,
472
+ "auto_find_batch_size": true,
473
+ "num_train_epochs": 5,
474
+ "max_steps": -1,
475
+ "logging_strategy": "epoch",
476
+ "save_strategy": "epoch",
477
+ "logging_steps": 0.2,
478
+ "save_steps": 0.2,
479
+ "gradient_accumulation_steps": 1,
480
+ "gradient_checkpointing": false,
481
+ "optim": "adamw_torch",
482
+ "learning_rate": 5e-05,
483
+ "weight_decay": 0.01,
484
+ "warmup_steps": 0,
485
+ "report_to": "tensorboard",
486
+ "save_safetensors": true,
487
+ "packing": false
488
+ },
489
+ "use_peft": true,
490
+ "peft_task_type": "CAUSAL_LM",
491
+ "peft_inference_mode": false,
492
+ "peft_r": 8,
493
+ "peft_lora_alpha": 32,
494
+ "peft_lora_dropout": 0.1,
495
+ "fsdp": "none",
496
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
497
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
498
+ "fsdp_fsdp_min_num_params": 100000000.0,
499
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
500
+ "fsdp_fsdp_forward_prefetch": false,
501
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
502
+ "fsdp_fsdp_offload_params": false,
503
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
504
+ "fsdp_fsdp_use_orig_params": true,
505
+ "fsdp_fsdp_activation_checkpointing": false,
506
+ "fsdp_fsdp_limit_all_gathers": true,
507
+ "fsdp_fsdp_sync_module_states": true,
508
+ "use_git": true,
509
+ "random_seed": 0
510
+ }
511
+ }
512
+ }
513
+ }
01-ai/Yi-9B/results_2026-01-05-04-27-50.json ADDED
@@ -0,0 +1,513 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "model_name": "01-ai/Yi-9B",
4
+ "model_type": "pretrained",
5
+ "params": 9.0
6
+ },
7
+ "results": {
8
+ "medmcqa": {
9
+ "acc_norm": 0.5378914654554148,
10
+ "acc_norm_stderr": 0.007709519428887803,
11
+ "args": {
12
+ "exp_name": "results/v1/medmcqa/01-ai/Yi-9B",
13
+ "task_name": "medmcqa",
14
+ "model_name": "01-ai/Yi-9B",
15
+ "dataset_param": {
16
+ "max_num_train": 50000,
17
+ "completion_only": false,
18
+ "apply_chat_template": false
19
+ },
20
+ "resume_from_checkpoint": false,
21
+ "train_param": {
22
+ "max_length": 1024,
23
+ "per_device_train_batch_size": 8,
24
+ "auto_find_batch_size": true,
25
+ "num_train_epochs": 1,
26
+ "max_steps": -1,
27
+ "logging_strategy": "steps",
28
+ "save_strategy": "steps",
29
+ "logging_steps": 0.2,
30
+ "save_steps": 0.2,
31
+ "gradient_accumulation_steps": 1,
32
+ "gradient_checkpointing": false,
33
+ "optim": "adamw_torch",
34
+ "learning_rate": 5e-05,
35
+ "weight_decay": 0.01,
36
+ "warmup_steps": 0,
37
+ "report_to": "tensorboard",
38
+ "save_safetensors": true,
39
+ "packing": false
40
+ },
41
+ "use_peft": true,
42
+ "peft_task_type": "CAUSAL_LM",
43
+ "peft_inference_mode": false,
44
+ "peft_r": 8,
45
+ "peft_lora_alpha": 32,
46
+ "peft_lora_dropout": 0.1,
47
+ "fsdp": "none",
48
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
49
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
50
+ "fsdp_fsdp_min_num_params": 100000000.0,
51
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
52
+ "fsdp_fsdp_forward_prefetch": false,
53
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
54
+ "fsdp_fsdp_offload_params": false,
55
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
56
+ "fsdp_fsdp_use_orig_params": true,
57
+ "fsdp_fsdp_activation_checkpointing": false,
58
+ "fsdp_fsdp_limit_all_gathers": true,
59
+ "fsdp_fsdp_sync_module_states": true,
60
+ "use_git": true,
61
+ "random_seed": 0
62
+ }
63
+ },
64
+ "nq_open": {
65
+ "exact_match,remove_whitespace": 0.2590027700831025,
66
+ "exact_match,remove_whitespace_stderr": 0.007292342460974554,
67
+ "args": {
68
+ "exp_name": "results/v1/nq_open/01-ai/Yi-9B",
69
+ "task_name": "nq_open",
70
+ "model_name": "01-ai/Yi-9B",
71
+ "dataset_param": {
72
+ "max_num_train": 50000,
73
+ "completion_only": false,
74
+ "apply_chat_template": false
75
+ },
76
+ "resume_from_checkpoint": false,
77
+ "train_param": {
78
+ "max_length": 1024,
79
+ "per_device_train_batch_size": 8,
80
+ "auto_find_batch_size": true,
81
+ "num_train_epochs": 1,
82
+ "max_steps": -1,
83
+ "logging_strategy": "steps",
84
+ "save_strategy": "steps",
85
+ "logging_steps": 0.2,
86
+ "save_steps": 0.2,
87
+ "gradient_accumulation_steps": 1,
88
+ "gradient_checkpointing": false,
89
+ "optim": "adamw_torch",
90
+ "learning_rate": 5e-05,
91
+ "weight_decay": 0.01,
92
+ "warmup_steps": 0,
93
+ "report_to": "tensorboard",
94
+ "save_safetensors": true,
95
+ "packing": false
96
+ },
97
+ "use_peft": true,
98
+ "peft_task_type": "CAUSAL_LM",
99
+ "peft_inference_mode": false,
100
+ "peft_r": 8,
101
+ "peft_lora_alpha": 32,
102
+ "peft_lora_dropout": 0.1,
103
+ "fsdp": "none",
104
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
105
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
106
+ "fsdp_fsdp_min_num_params": 100000000.0,
107
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
108
+ "fsdp_fsdp_forward_prefetch": false,
109
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
110
+ "fsdp_fsdp_offload_params": false,
111
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
112
+ "fsdp_fsdp_use_orig_params": true,
113
+ "fsdp_fsdp_activation_checkpointing": false,
114
+ "fsdp_fsdp_limit_all_gathers": true,
115
+ "fsdp_fsdp_sync_module_states": true,
116
+ "use_git": true,
117
+ "random_seed": 0
118
+ }
119
+ },
120
+ "winogrande": {
121
+ "acc": 0.8129439621152328,
122
+ "acc_stderr": 0.010959716435242912,
123
+ "args": {
124
+ "exp_name": "results/v1/winogrande/01-ai/Yi-9B",
125
+ "task_name": "winogrande",
126
+ "model_name": "01-ai/Yi-9B",
127
+ "dataset_param": {
128
+ "max_num_train": 50000,
129
+ "completion_only": false,
130
+ "apply_chat_template": false
131
+ },
132
+ "resume_from_checkpoint": false,
133
+ "train_param": {
134
+ "max_length": 1024,
135
+ "per_device_train_batch_size": 8,
136
+ "auto_find_batch_size": true,
137
+ "num_train_epochs": 1,
138
+ "max_steps": -1,
139
+ "logging_strategy": "steps",
140
+ "save_strategy": "steps",
141
+ "logging_steps": 0.2,
142
+ "save_steps": 0.2,
143
+ "gradient_accumulation_steps": 1,
144
+ "gradient_checkpointing": false,
145
+ "optim": "adamw_torch",
146
+ "learning_rate": 5e-05,
147
+ "weight_decay": 0.01,
148
+ "warmup_steps": 0,
149
+ "report_to": "tensorboard",
150
+ "save_safetensors": true,
151
+ "packing": false
152
+ },
153
+ "use_peft": true,
154
+ "peft_task_type": "CAUSAL_LM",
155
+ "peft_inference_mode": false,
156
+ "peft_r": 8,
157
+ "peft_lora_alpha": 32,
158
+ "peft_lora_dropout": 0.1,
159
+ "fsdp": "none",
160
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
161
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
162
+ "fsdp_fsdp_min_num_params": 100000000.0,
163
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
164
+ "fsdp_fsdp_forward_prefetch": false,
165
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
166
+ "fsdp_fsdp_offload_params": false,
167
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
168
+ "fsdp_fsdp_use_orig_params": true,
169
+ "fsdp_fsdp_activation_checkpointing": false,
170
+ "fsdp_fsdp_limit_all_gathers": true,
171
+ "fsdp_fsdp_sync_module_states": true,
172
+ "use_git": true,
173
+ "random_seed": 0
174
+ }
175
+ },
176
+ "hellaswag": {
177
+ "acc_norm": 0.8023,
178
+ "acc_norm_stderr": 0.003982845378084634,
179
+ "args": {
180
+ "exp_name": "results/v1/hellaswag/01-ai/Yi-9B",
181
+ "task_name": "hellaswag",
182
+ "model_name": "01-ai/Yi-9B",
183
+ "dataset_param": {
184
+ "max_num_train": 50000,
185
+ "completion_only": false,
186
+ "apply_chat_template": false
187
+ },
188
+ "resume_from_checkpoint": false,
189
+ "train_param": {
190
+ "max_length": 1024,
191
+ "per_device_train_batch_size": 8,
192
+ "auto_find_batch_size": true,
193
+ "num_train_epochs": 1,
194
+ "max_steps": -1,
195
+ "logging_strategy": "steps",
196
+ "save_strategy": "steps",
197
+ "logging_steps": 0.2,
198
+ "save_steps": 0.2,
199
+ "gradient_accumulation_steps": 1,
200
+ "gradient_checkpointing": false,
201
+ "optim": "adamw_torch",
202
+ "learning_rate": 5e-05,
203
+ "weight_decay": 0.01,
204
+ "warmup_steps": 0,
205
+ "report_to": "tensorboard",
206
+ "save_safetensors": true,
207
+ "packing": false
208
+ },
209
+ "use_peft": true,
210
+ "peft_task_type": "CAUSAL_LM",
211
+ "peft_inference_mode": false,
212
+ "peft_r": 8,
213
+ "peft_lora_alpha": 32,
214
+ "peft_lora_dropout": 0.1,
215
+ "fsdp": "none",
216
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
217
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
218
+ "fsdp_fsdp_min_num_params": 100000000.0,
219
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
220
+ "fsdp_fsdp_forward_prefetch": false,
221
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
222
+ "fsdp_fsdp_offload_params": false,
223
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
224
+ "fsdp_fsdp_use_orig_params": true,
225
+ "fsdp_fsdp_activation_checkpointing": false,
226
+ "fsdp_fsdp_limit_all_gathers": true,
227
+ "fsdp_fsdp_sync_module_states": true,
228
+ "use_git": true,
229
+ "random_seed": 0
230
+ }
231
+ },
232
+ "social_iqa": {
233
+ "acc": 0.5880245649948823,
234
+ "acc_stderr": 0.011137360400975266,
235
+ "args": {
236
+ "exp_name": "results/v1/social_iqa/01-ai/Yi-9B",
237
+ "task_name": "social_iqa",
238
+ "model_name": "01-ai/Yi-9B",
239
+ "dataset_param": {
240
+ "max_num_train": 50000,
241
+ "completion_only": false,
242
+ "apply_chat_template": false
243
+ },
244
+ "resume_from_checkpoint": false,
245
+ "train_param": {
246
+ "max_length": 1024,
247
+ "per_device_train_batch_size": 8,
248
+ "auto_find_batch_size": true,
249
+ "num_train_epochs": 1,
250
+ "max_steps": -1,
251
+ "logging_strategy": "steps",
252
+ "save_strategy": "steps",
253
+ "logging_steps": 0.2,
254
+ "save_steps": 0.2,
255
+ "gradient_accumulation_steps": 1,
256
+ "gradient_checkpointing": false,
257
+ "optim": "adamw_torch",
258
+ "learning_rate": 5e-05,
259
+ "weight_decay": 0.01,
260
+ "warmup_steps": 0,
261
+ "report_to": "tensorboard",
262
+ "save_safetensors": true,
263
+ "packing": false
264
+ },
265
+ "use_peft": true,
266
+ "peft_task_type": "CAUSAL_LM",
267
+ "peft_inference_mode": false,
268
+ "peft_r": 8,
269
+ "peft_lora_alpha": 32,
270
+ "peft_lora_dropout": 0.1,
271
+ "fsdp": "none",
272
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
273
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
274
+ "fsdp_fsdp_min_num_params": 100000000.0,
275
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
276
+ "fsdp_fsdp_forward_prefetch": false,
277
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
278
+ "fsdp_fsdp_offload_params": false,
279
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
280
+ "fsdp_fsdp_use_orig_params": true,
281
+ "fsdp_fsdp_activation_checkpointing": false,
282
+ "fsdp_fsdp_limit_all_gathers": true,
283
+ "fsdp_fsdp_sync_module_states": true,
284
+ "use_git": true,
285
+ "random_seed": 0
286
+ }
287
+ },
288
+ "piqa": {
289
+ "acc_norm": 0.823721436343852,
290
+ "acc_norm_stderr": 0.008890687746582126,
291
+ "args": {
292
+ "exp_name": "results/v2/piqa/01-ai/Yi-9B",
293
+ "task_name": "piqa",
294
+ "model_name": "01-ai/Yi-9B",
295
+ "dataset_param": {
296
+ "max_num_train": 50000,
297
+ "completion_only": false,
298
+ "apply_chat_template": false
299
+ },
300
+ "resume_from_checkpoint": false,
301
+ "train_param": {
302
+ "max_length": 1024,
303
+ "per_device_train_batch_size": 8,
304
+ "auto_find_batch_size": true,
305
+ "num_train_epochs": 5,
306
+ "max_steps": -1,
307
+ "logging_strategy": "epoch",
308
+ "save_strategy": "epoch",
309
+ "logging_steps": 0.2,
310
+ "save_steps": 0.2,
311
+ "gradient_accumulation_steps": 1,
312
+ "gradient_checkpointing": false,
313
+ "optim": "adamw_torch",
314
+ "learning_rate": 5e-05,
315
+ "weight_decay": 0.01,
316
+ "warmup_steps": 0,
317
+ "report_to": "tensorboard",
318
+ "save_safetensors": true,
319
+ "packing": false
320
+ },
321
+ "use_peft": true,
322
+ "peft_task_type": "CAUSAL_LM",
323
+ "peft_inference_mode": false,
324
+ "peft_r": 8,
325
+ "peft_lora_alpha": 32,
326
+ "peft_lora_dropout": 0.1,
327
+ "fsdp": "none",
328
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
329
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
330
+ "fsdp_fsdp_min_num_params": 100000000.0,
331
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
332
+ "fsdp_fsdp_forward_prefetch": false,
333
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
334
+ "fsdp_fsdp_offload_params": false,
335
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
336
+ "fsdp_fsdp_use_orig_params": true,
337
+ "fsdp_fsdp_activation_checkpointing": false,
338
+ "fsdp_fsdp_limit_all_gathers": true,
339
+ "fsdp_fsdp_sync_module_states": true,
340
+ "use_git": true,
341
+ "random_seed": 0
342
+ }
343
+ },
344
+ "commonsense_qa": {
345
+ "acc": 0.8271908271908271,
346
+ "acc_stderr": 0.010824462376464377,
347
+ "args": {
348
+ "exp_name": "results/v2/commonsense_qa/01-ai/Yi-9B",
349
+ "task_name": "commonsense_qa",
350
+ "model_name": "01-ai/Yi-9B",
351
+ "dataset_param": {
352
+ "max_num_train": 50000,
353
+ "completion_only": false,
354
+ "apply_chat_template": false
355
+ },
356
+ "resume_from_checkpoint": false,
357
+ "train_param": {
358
+ "max_length": 1024,
359
+ "per_device_train_batch_size": 8,
360
+ "auto_find_batch_size": true,
361
+ "num_train_epochs": 5,
362
+ "max_steps": -1,
363
+ "logging_strategy": "epoch",
364
+ "save_strategy": "epoch",
365
+ "logging_steps": 0.2,
366
+ "save_steps": 0.2,
367
+ "gradient_accumulation_steps": 1,
368
+ "gradient_checkpointing": false,
369
+ "optim": "adamw_torch",
370
+ "learning_rate": 5e-05,
371
+ "weight_decay": 0.01,
372
+ "warmup_steps": 0,
373
+ "report_to": "tensorboard",
374
+ "save_safetensors": true,
375
+ "packing": false
376
+ },
377
+ "use_peft": true,
378
+ "peft_task_type": "CAUSAL_LM",
379
+ "peft_inference_mode": false,
380
+ "peft_r": 8,
381
+ "peft_lora_alpha": 32,
382
+ "peft_lora_dropout": 0.1,
383
+ "fsdp": "none",
384
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
385
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
386
+ "fsdp_fsdp_min_num_params": 100000000.0,
387
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
388
+ "fsdp_fsdp_forward_prefetch": false,
389
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
390
+ "fsdp_fsdp_offload_params": false,
391
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
392
+ "fsdp_fsdp_use_orig_params": true,
393
+ "fsdp_fsdp_activation_checkpointing": false,
394
+ "fsdp_fsdp_limit_all_gathers": true,
395
+ "fsdp_fsdp_sync_module_states": true,
396
+ "use_git": true,
397
+ "random_seed": 0
398
+ }
399
+ },
400
+ "gsm8k": {
401
+ "exact_match,flexible-extract": 0.5443517816527672,
402
+ "exact_match,flexible-extract_stderr": 0.013718194542485596,
403
+ "args": {
404
+ "exp_name": "results/v2/gsm8k/01-ai/Yi-9B",
405
+ "task_name": "gsm8k",
406
+ "model_name": "01-ai/Yi-9B",
407
+ "dataset_param": {
408
+ "max_num_train": 50000,
409
+ "completion_only": false,
410
+ "apply_chat_template": false
411
+ },
412
+ "resume_from_checkpoint": false,
413
+ "train_param": {
414
+ "max_length": 1024,
415
+ "per_device_train_batch_size": 8,
416
+ "auto_find_batch_size": true,
417
+ "num_train_epochs": 5,
418
+ "max_steps": -1,
419
+ "logging_strategy": "epoch",
420
+ "save_strategy": "epoch",
421
+ "logging_steps": 0.2,
422
+ "save_steps": 0.2,
423
+ "gradient_accumulation_steps": 1,
424
+ "gradient_checkpointing": false,
425
+ "optim": "adamw_torch",
426
+ "learning_rate": 5e-05,
427
+ "weight_decay": 0.01,
428
+ "warmup_steps": 0,
429
+ "report_to": "tensorboard",
430
+ "save_safetensors": true,
431
+ "packing": false
432
+ },
433
+ "use_peft": true,
434
+ "peft_task_type": "CAUSAL_LM",
435
+ "peft_inference_mode": false,
436
+ "peft_r": 8,
437
+ "peft_lora_alpha": 32,
438
+ "peft_lora_dropout": 0.1,
439
+ "fsdp": "none",
440
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
441
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
442
+ "fsdp_fsdp_min_num_params": 100000000.0,
443
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
444
+ "fsdp_fsdp_forward_prefetch": false,
445
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
446
+ "fsdp_fsdp_offload_params": false,
447
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
448
+ "fsdp_fsdp_use_orig_params": true,
449
+ "fsdp_fsdp_activation_checkpointing": false,
450
+ "fsdp_fsdp_limit_all_gathers": true,
451
+ "fsdp_fsdp_sync_module_states": true,
452
+ "use_git": true,
453
+ "random_seed": 0
454
+ }
455
+ },
456
+ "arc_challenge": {
457
+ "acc_norm": 0.6186006825938567,
458
+ "acc_norm_stderr": 0.014194389086685253,
459
+ "args": {
460
+ "exp_name": "results/v2/arc_challenge/01-ai/Yi-9B",
461
+ "task_name": "arc_challenge",
462
+ "model_name": "01-ai/Yi-9B",
463
+ "dataset_param": {
464
+ "max_num_train": 50000,
465
+ "completion_only": false,
466
+ "apply_chat_template": false
467
+ },
468
+ "resume_from_checkpoint": false,
469
+ "train_param": {
470
+ "max_length": 1024,
471
+ "per_device_train_batch_size": 8,
472
+ "auto_find_batch_size": true,
473
+ "num_train_epochs": 5,
474
+ "max_steps": -1,
475
+ "logging_strategy": "epoch",
476
+ "save_strategy": "epoch",
477
+ "logging_steps": 0.2,
478
+ "save_steps": 0.2,
479
+ "gradient_accumulation_steps": 1,
480
+ "gradient_checkpointing": false,
481
+ "optim": "adamw_torch",
482
+ "learning_rate": 5e-05,
483
+ "weight_decay": 0.01,
484
+ "warmup_steps": 0,
485
+ "report_to": "tensorboard",
486
+ "save_safetensors": true,
487
+ "packing": false
488
+ },
489
+ "use_peft": true,
490
+ "peft_task_type": "CAUSAL_LM",
491
+ "peft_inference_mode": false,
492
+ "peft_r": 8,
493
+ "peft_lora_alpha": 32,
494
+ "peft_lora_dropout": 0.1,
495
+ "fsdp": "none",
496
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
497
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
498
+ "fsdp_fsdp_min_num_params": 100000000.0,
499
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
500
+ "fsdp_fsdp_forward_prefetch": false,
501
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
502
+ "fsdp_fsdp_offload_params": false,
503
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
504
+ "fsdp_fsdp_use_orig_params": true,
505
+ "fsdp_fsdp_activation_checkpointing": false,
506
+ "fsdp_fsdp_limit_all_gathers": true,
507
+ "fsdp_fsdp_sync_module_states": true,
508
+ "use_git": true,
509
+ "random_seed": 0
510
+ }
511
+ }
512
+ }
513
+ }
EleutherAI/pythia-1.4b-deduped/results_2026-01-05-04-27-50.json ADDED
@@ -0,0 +1,513 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "model_name": "EleutherAI/pythia-1.4b-deduped",
4
+ "model_type": "pretrained",
5
+ "params": 1.4
6
+ },
7
+ "results": {
8
+ "medmcqa": {
9
+ "acc_norm": 0.32273487927324884,
10
+ "acc_norm_stderr": 0.0072295338029876414,
11
+ "args": {
12
+ "exp_name": "results/v1/medmcqa/EleutherAI/pythia-1.4b-deduped",
13
+ "task_name": "medmcqa",
14
+ "model_name": "EleutherAI/pythia-1.4b-deduped",
15
+ "dataset_param": {
16
+ "max_num_train": 50000,
17
+ "completion_only": false,
18
+ "apply_chat_template": false
19
+ },
20
+ "resume_from_checkpoint": false,
21
+ "train_param": {
22
+ "max_length": 1024,
23
+ "per_device_train_batch_size": 8,
24
+ "auto_find_batch_size": true,
25
+ "num_train_epochs": 1,
26
+ "max_steps": -1,
27
+ "logging_strategy": "steps",
28
+ "save_strategy": "steps",
29
+ "logging_steps": 0.2,
30
+ "save_steps": 0.2,
31
+ "gradient_accumulation_steps": 1,
32
+ "gradient_checkpointing": false,
33
+ "optim": "adamw_torch",
34
+ "learning_rate": 5e-05,
35
+ "weight_decay": 0.01,
36
+ "warmup_steps": 0,
37
+ "report_to": "tensorboard",
38
+ "save_safetensors": true,
39
+ "packing": false
40
+ },
41
+ "use_peft": true,
42
+ "peft_task_type": "CAUSAL_LM",
43
+ "peft_inference_mode": false,
44
+ "peft_r": 8,
45
+ "peft_lora_alpha": 32,
46
+ "peft_lora_dropout": 0.1,
47
+ "fsdp": "none",
48
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
49
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
50
+ "fsdp_fsdp_min_num_params": 100000000.0,
51
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
52
+ "fsdp_fsdp_forward_prefetch": false,
53
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
54
+ "fsdp_fsdp_offload_params": false,
55
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
56
+ "fsdp_fsdp_use_orig_params": true,
57
+ "fsdp_fsdp_activation_checkpointing": false,
58
+ "fsdp_fsdp_limit_all_gathers": true,
59
+ "fsdp_fsdp_sync_module_states": true,
60
+ "use_git": true,
61
+ "random_seed": 0
62
+ }
63
+ },
64
+ "nq_open": {
65
+ "exact_match,remove_whitespace": 0.08587257617728532,
66
+ "exact_match,remove_whitespace_stderr": 0.004663770531026864,
67
+ "args": {
68
+ "exp_name": "results/v1/nq_open/EleutherAI/pythia-1.4b-deduped",
69
+ "task_name": "nq_open",
70
+ "model_name": "EleutherAI/pythia-1.4b-deduped",
71
+ "dataset_param": {
72
+ "max_num_train": 50000,
73
+ "completion_only": false,
74
+ "apply_chat_template": false
75
+ },
76
+ "resume_from_checkpoint": false,
77
+ "train_param": {
78
+ "max_length": 1024,
79
+ "per_device_train_batch_size": 8,
80
+ "auto_find_batch_size": true,
81
+ "num_train_epochs": 1,
82
+ "max_steps": -1,
83
+ "logging_strategy": "steps",
84
+ "save_strategy": "steps",
85
+ "logging_steps": 0.2,
86
+ "save_steps": 0.2,
87
+ "gradient_accumulation_steps": 1,
88
+ "gradient_checkpointing": false,
89
+ "optim": "adamw_torch",
90
+ "learning_rate": 5e-05,
91
+ "weight_decay": 0.01,
92
+ "warmup_steps": 0,
93
+ "report_to": "tensorboard",
94
+ "save_safetensors": true,
95
+ "packing": false
96
+ },
97
+ "use_peft": true,
98
+ "peft_task_type": "CAUSAL_LM",
99
+ "peft_inference_mode": false,
100
+ "peft_r": 8,
101
+ "peft_lora_alpha": 32,
102
+ "peft_lora_dropout": 0.1,
103
+ "fsdp": "none",
104
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
105
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
106
+ "fsdp_fsdp_min_num_params": 100000000.0,
107
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
108
+ "fsdp_fsdp_forward_prefetch": false,
109
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
110
+ "fsdp_fsdp_offload_params": false,
111
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
112
+ "fsdp_fsdp_use_orig_params": true,
113
+ "fsdp_fsdp_activation_checkpointing": false,
114
+ "fsdp_fsdp_limit_all_gathers": true,
115
+ "fsdp_fsdp_sync_module_states": true,
116
+ "use_git": true,
117
+ "random_seed": 0
118
+ }
119
+ },
120
+ "winogrande": {
121
+ "acc": 0.6211523283346487,
122
+ "acc_stderr": 0.013633724603180325,
123
+ "args": {
124
+ "exp_name": "results/v1/winogrande/EleutherAI/pythia-1.4b-deduped",
125
+ "task_name": "winogrande",
126
+ "model_name": "EleutherAI/pythia-1.4b-deduped",
127
+ "dataset_param": {
128
+ "max_num_train": 50000,
129
+ "completion_only": false,
130
+ "apply_chat_template": false
131
+ },
132
+ "resume_from_checkpoint": false,
133
+ "train_param": {
134
+ "max_length": 1024,
135
+ "per_device_train_batch_size": 8,
136
+ "auto_find_batch_size": true,
137
+ "num_train_epochs": 1,
138
+ "max_steps": -1,
139
+ "logging_strategy": "steps",
140
+ "save_strategy": "steps",
141
+ "logging_steps": 0.2,
142
+ "save_steps": 0.2,
143
+ "gradient_accumulation_steps": 1,
144
+ "gradient_checkpointing": false,
145
+ "optim": "adamw_torch",
146
+ "learning_rate": 5e-05,
147
+ "weight_decay": 0.01,
148
+ "warmup_steps": 0,
149
+ "report_to": "tensorboard",
150
+ "save_safetensors": true,
151
+ "packing": false
152
+ },
153
+ "use_peft": true,
154
+ "peft_task_type": "CAUSAL_LM",
155
+ "peft_inference_mode": false,
156
+ "peft_r": 8,
157
+ "peft_lora_alpha": 32,
158
+ "peft_lora_dropout": 0.1,
159
+ "fsdp": "none",
160
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
161
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
162
+ "fsdp_fsdp_min_num_params": 100000000.0,
163
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
164
+ "fsdp_fsdp_forward_prefetch": false,
165
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
166
+ "fsdp_fsdp_offload_params": false,
167
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
168
+ "fsdp_fsdp_use_orig_params": true,
169
+ "fsdp_fsdp_activation_checkpointing": false,
170
+ "fsdp_fsdp_limit_all_gathers": true,
171
+ "fsdp_fsdp_sync_module_states": true,
172
+ "use_git": true,
173
+ "random_seed": 0
174
+ }
175
+ },
176
+ "hellaswag": {
177
+ "acc_norm": 0.5674,
178
+ "acc_norm_stderr": 0.004954611873585999,
179
+ "args": {
180
+ "exp_name": "results/v1/hellaswag/EleutherAI/pythia-1.4b-deduped",
181
+ "task_name": "hellaswag",
182
+ "model_name": "EleutherAI/pythia-1.4b-deduped",
183
+ "dataset_param": {
184
+ "max_num_train": 50000,
185
+ "completion_only": false,
186
+ "apply_chat_template": false
187
+ },
188
+ "resume_from_checkpoint": false,
189
+ "train_param": {
190
+ "max_length": 1024,
191
+ "per_device_train_batch_size": 8,
192
+ "auto_find_batch_size": true,
193
+ "num_train_epochs": 1,
194
+ "max_steps": -1,
195
+ "logging_strategy": "steps",
196
+ "save_strategy": "steps",
197
+ "logging_steps": 0.2,
198
+ "save_steps": 0.2,
199
+ "gradient_accumulation_steps": 1,
200
+ "gradient_checkpointing": false,
201
+ "optim": "adamw_torch",
202
+ "learning_rate": 5e-05,
203
+ "weight_decay": 0.01,
204
+ "warmup_steps": 0,
205
+ "report_to": "tensorboard",
206
+ "save_safetensors": true,
207
+ "packing": false
208
+ },
209
+ "use_peft": true,
210
+ "peft_task_type": "CAUSAL_LM",
211
+ "peft_inference_mode": false,
212
+ "peft_r": 8,
213
+ "peft_lora_alpha": 32,
214
+ "peft_lora_dropout": 0.1,
215
+ "fsdp": "none",
216
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
217
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
218
+ "fsdp_fsdp_min_num_params": 100000000.0,
219
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
220
+ "fsdp_fsdp_forward_prefetch": false,
221
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
222
+ "fsdp_fsdp_offload_params": false,
223
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
224
+ "fsdp_fsdp_use_orig_params": true,
225
+ "fsdp_fsdp_activation_checkpointing": false,
226
+ "fsdp_fsdp_limit_all_gathers": true,
227
+ "fsdp_fsdp_sync_module_states": true,
228
+ "use_git": true,
229
+ "random_seed": 0
230
+ }
231
+ },
232
+ "social_iqa": {
233
+ "acc": 0.4953940634595701,
234
+ "acc_stderr": 0.011313590496725468,
235
+ "args": {
236
+ "exp_name": "results/v1/social_iqa/EleutherAI/pythia-1.4b-deduped",
237
+ "task_name": "social_iqa",
238
+ "model_name": "EleutherAI/pythia-1.4b-deduped",
239
+ "dataset_param": {
240
+ "max_num_train": 50000,
241
+ "completion_only": false,
242
+ "apply_chat_template": false
243
+ },
244
+ "resume_from_checkpoint": false,
245
+ "train_param": {
246
+ "max_length": 1024,
247
+ "per_device_train_batch_size": 8,
248
+ "auto_find_batch_size": true,
249
+ "num_train_epochs": 1,
250
+ "max_steps": -1,
251
+ "logging_strategy": "steps",
252
+ "save_strategy": "steps",
253
+ "logging_steps": 0.2,
254
+ "save_steps": 0.2,
255
+ "gradient_accumulation_steps": 1,
256
+ "gradient_checkpointing": false,
257
+ "optim": "adamw_torch",
258
+ "learning_rate": 5e-05,
259
+ "weight_decay": 0.01,
260
+ "warmup_steps": 0,
261
+ "report_to": "tensorboard",
262
+ "save_safetensors": true,
263
+ "packing": false
264
+ },
265
+ "use_peft": true,
266
+ "peft_task_type": "CAUSAL_LM",
267
+ "peft_inference_mode": false,
268
+ "peft_r": 8,
269
+ "peft_lora_alpha": 32,
270
+ "peft_lora_dropout": 0.1,
271
+ "fsdp": "none",
272
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
273
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
274
+ "fsdp_fsdp_min_num_params": 100000000.0,
275
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
276
+ "fsdp_fsdp_forward_prefetch": false,
277
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
278
+ "fsdp_fsdp_offload_params": false,
279
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
280
+ "fsdp_fsdp_use_orig_params": true,
281
+ "fsdp_fsdp_activation_checkpointing": false,
282
+ "fsdp_fsdp_limit_all_gathers": true,
283
+ "fsdp_fsdp_sync_module_states": true,
284
+ "use_git": true,
285
+ "random_seed": 0
286
+ }
287
+ },
288
+ "piqa": {
289
+ "acc_norm": 0.73449401523395,
290
+ "acc_norm_stderr": 0.010303308653024425,
291
+ "args": {
292
+ "exp_name": "results/v2/piqa/EleutherAI/pythia-1.4b-deduped",
293
+ "task_name": "piqa",
294
+ "model_name": "EleutherAI/pythia-1.4b-deduped",
295
+ "dataset_param": {
296
+ "max_num_train": 50000,
297
+ "completion_only": false,
298
+ "apply_chat_template": false
299
+ },
300
+ "resume_from_checkpoint": false,
301
+ "train_param": {
302
+ "max_length": 1024,
303
+ "per_device_train_batch_size": 8,
304
+ "auto_find_batch_size": true,
305
+ "num_train_epochs": 5,
306
+ "max_steps": -1,
307
+ "logging_strategy": "epoch",
308
+ "save_strategy": "epoch",
309
+ "logging_steps": 0.2,
310
+ "save_steps": 0.2,
311
+ "gradient_accumulation_steps": 1,
312
+ "gradient_checkpointing": false,
313
+ "optim": "adamw_torch",
314
+ "learning_rate": 5e-05,
315
+ "weight_decay": 0.01,
316
+ "warmup_steps": 0,
317
+ "report_to": "tensorboard",
318
+ "save_safetensors": true,
319
+ "packing": false
320
+ },
321
+ "use_peft": true,
322
+ "peft_task_type": "CAUSAL_LM",
323
+ "peft_inference_mode": false,
324
+ "peft_r": 8,
325
+ "peft_lora_alpha": 32,
326
+ "peft_lora_dropout": 0.1,
327
+ "fsdp": "none",
328
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
329
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
330
+ "fsdp_fsdp_min_num_params": 100000000.0,
331
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
332
+ "fsdp_fsdp_forward_prefetch": false,
333
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
334
+ "fsdp_fsdp_offload_params": false,
335
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
336
+ "fsdp_fsdp_use_orig_params": true,
337
+ "fsdp_fsdp_activation_checkpointing": false,
338
+ "fsdp_fsdp_limit_all_gathers": true,
339
+ "fsdp_fsdp_sync_module_states": true,
340
+ "use_git": true,
341
+ "random_seed": 0
342
+ }
343
+ },
344
+ "commonsense_qa": {
345
+ "acc": 0.21703521703521703,
346
+ "acc_stderr": 0.011802018846529995,
347
+ "args": {
348
+ "exp_name": "results/v2/commonsense_qa/EleutherAI/pythia-1.4b-deduped",
349
+ "task_name": "commonsense_qa",
350
+ "model_name": "EleutherAI/pythia-1.4b-deduped",
351
+ "dataset_param": {
352
+ "max_num_train": 50000,
353
+ "completion_only": false,
354
+ "apply_chat_template": false
355
+ },
356
+ "resume_from_checkpoint": false,
357
+ "train_param": {
358
+ "max_length": 1024,
359
+ "per_device_train_batch_size": 8,
360
+ "auto_find_batch_size": true,
361
+ "num_train_epochs": 5,
362
+ "max_steps": -1,
363
+ "logging_strategy": "epoch",
364
+ "save_strategy": "epoch",
365
+ "logging_steps": 0.2,
366
+ "save_steps": 0.2,
367
+ "gradient_accumulation_steps": 1,
368
+ "gradient_checkpointing": false,
369
+ "optim": "adamw_torch",
370
+ "learning_rate": 5e-05,
371
+ "weight_decay": 0.01,
372
+ "warmup_steps": 0,
373
+ "report_to": "tensorboard",
374
+ "save_safetensors": true,
375
+ "packing": false
376
+ },
377
+ "use_peft": true,
378
+ "peft_task_type": "CAUSAL_LM",
379
+ "peft_inference_mode": false,
380
+ "peft_r": 8,
381
+ "peft_lora_alpha": 32,
382
+ "peft_lora_dropout": 0.1,
383
+ "fsdp": "none",
384
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
385
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
386
+ "fsdp_fsdp_min_num_params": 100000000.0,
387
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
388
+ "fsdp_fsdp_forward_prefetch": false,
389
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
390
+ "fsdp_fsdp_offload_params": false,
391
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
392
+ "fsdp_fsdp_use_orig_params": true,
393
+ "fsdp_fsdp_activation_checkpointing": false,
394
+ "fsdp_fsdp_limit_all_gathers": true,
395
+ "fsdp_fsdp_sync_module_states": true,
396
+ "use_git": true,
397
+ "random_seed": 0
398
+ }
399
+ },
400
+ "gsm8k": {
401
+ "exact_match,flexible-extract": 0.058377558756633814,
402
+ "exact_match,flexible-extract_stderr": 0.006458083557832458,
403
+ "args": {
404
+ "exp_name": "results/v2/gsm8k/EleutherAI/pythia-1.4b-deduped",
405
+ "task_name": "gsm8k",
406
+ "model_name": "EleutherAI/pythia-1.4b-deduped",
407
+ "dataset_param": {
408
+ "max_num_train": 50000,
409
+ "completion_only": false,
410
+ "apply_chat_template": false
411
+ },
412
+ "resume_from_checkpoint": false,
413
+ "train_param": {
414
+ "max_length": 1024,
415
+ "per_device_train_batch_size": 8,
416
+ "auto_find_batch_size": true,
417
+ "num_train_epochs": 5,
418
+ "max_steps": -1,
419
+ "logging_strategy": "epoch",
420
+ "save_strategy": "epoch",
421
+ "logging_steps": 0.2,
422
+ "save_steps": 0.2,
423
+ "gradient_accumulation_steps": 1,
424
+ "gradient_checkpointing": false,
425
+ "optim": "adamw_torch",
426
+ "learning_rate": 5e-05,
427
+ "weight_decay": 0.01,
428
+ "warmup_steps": 0,
429
+ "report_to": "tensorboard",
430
+ "save_safetensors": true,
431
+ "packing": false
432
+ },
433
+ "use_peft": true,
434
+ "peft_task_type": "CAUSAL_LM",
435
+ "peft_inference_mode": false,
436
+ "peft_r": 8,
437
+ "peft_lora_alpha": 32,
438
+ "peft_lora_dropout": 0.1,
439
+ "fsdp": "none",
440
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
441
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
442
+ "fsdp_fsdp_min_num_params": 100000000.0,
443
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
444
+ "fsdp_fsdp_forward_prefetch": false,
445
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
446
+ "fsdp_fsdp_offload_params": false,
447
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
448
+ "fsdp_fsdp_use_orig_params": true,
449
+ "fsdp_fsdp_activation_checkpointing": false,
450
+ "fsdp_fsdp_limit_all_gathers": true,
451
+ "fsdp_fsdp_sync_module_states": true,
452
+ "use_git": true,
453
+ "random_seed": 0
454
+ }
455
+ },
456
+ "arc_challenge": {
457
+ "acc_norm": 0.3412969283276451,
458
+ "acc_norm_stderr": 0.013855831287497728,
459
+ "args": {
460
+ "exp_name": "results/v2/arc_challenge/EleutherAI/pythia-1.4b-deduped",
461
+ "task_name": "arc_challenge",
462
+ "model_name": "EleutherAI/pythia-1.4b-deduped",
463
+ "dataset_param": {
464
+ "max_num_train": 50000,
465
+ "completion_only": false,
466
+ "apply_chat_template": false
467
+ },
468
+ "resume_from_checkpoint": false,
469
+ "train_param": {
470
+ "max_length": 1024,
471
+ "per_device_train_batch_size": 8,
472
+ "auto_find_batch_size": true,
473
+ "num_train_epochs": 5,
474
+ "max_steps": -1,
475
+ "logging_strategy": "epoch",
476
+ "save_strategy": "epoch",
477
+ "logging_steps": 0.2,
478
+ "save_steps": 0.2,
479
+ "gradient_accumulation_steps": 1,
480
+ "gradient_checkpointing": false,
481
+ "optim": "adamw_torch",
482
+ "learning_rate": 5e-05,
483
+ "weight_decay": 0.01,
484
+ "warmup_steps": 0,
485
+ "report_to": "tensorboard",
486
+ "save_safetensors": true,
487
+ "packing": false
488
+ },
489
+ "use_peft": true,
490
+ "peft_task_type": "CAUSAL_LM",
491
+ "peft_inference_mode": false,
492
+ "peft_r": 8,
493
+ "peft_lora_alpha": 32,
494
+ "peft_lora_dropout": 0.1,
495
+ "fsdp": "none",
496
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
497
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
498
+ "fsdp_fsdp_min_num_params": 100000000.0,
499
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
500
+ "fsdp_fsdp_forward_prefetch": false,
501
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
502
+ "fsdp_fsdp_offload_params": false,
503
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
504
+ "fsdp_fsdp_use_orig_params": true,
505
+ "fsdp_fsdp_activation_checkpointing": false,
506
+ "fsdp_fsdp_limit_all_gathers": true,
507
+ "fsdp_fsdp_sync_module_states": true,
508
+ "use_git": true,
509
+ "random_seed": 0
510
+ }
511
+ }
512
+ }
513
+ }
EleutherAI/pythia-12b/results_2026-01-05-04-27-50.json ADDED
@@ -0,0 +1,513 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "model_name": "EleutherAI/pythia-12b",
4
+ "model_type": "pretrained",
5
+ "params": 12.0
6
+ },
7
+ "results": {
8
+ "medmcqa": {
9
+ "acc_norm": 0.33636146306478604,
10
+ "acc_norm_stderr": 0.007305953450283969,
11
+ "args": {
12
+ "exp_name": "results/v1/medmcqa/EleutherAI/pythia-12b",
13
+ "task_name": "medmcqa",
14
+ "model_name": "EleutherAI/pythia-12b",
15
+ "dataset_param": {
16
+ "max_num_train": 50000,
17
+ "completion_only": false,
18
+ "apply_chat_template": false
19
+ },
20
+ "resume_from_checkpoint": false,
21
+ "train_param": {
22
+ "max_length": 1024,
23
+ "per_device_train_batch_size": 8,
24
+ "auto_find_batch_size": true,
25
+ "num_train_epochs": 1,
26
+ "max_steps": -1,
27
+ "logging_strategy": "steps",
28
+ "save_strategy": "steps",
29
+ "logging_steps": 0.2,
30
+ "save_steps": 0.2,
31
+ "gradient_accumulation_steps": 1,
32
+ "gradient_checkpointing": false,
33
+ "optim": "adamw_torch",
34
+ "learning_rate": 5e-05,
35
+ "weight_decay": 0.01,
36
+ "warmup_steps": 0,
37
+ "report_to": "tensorboard",
38
+ "save_safetensors": true,
39
+ "packing": false
40
+ },
41
+ "use_peft": true,
42
+ "peft_task_type": "CAUSAL_LM",
43
+ "peft_inference_mode": false,
44
+ "peft_r": 8,
45
+ "peft_lora_alpha": 32,
46
+ "peft_lora_dropout": 0.1,
47
+ "fsdp": "none",
48
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
49
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
50
+ "fsdp_fsdp_min_num_params": 100000000.0,
51
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
52
+ "fsdp_fsdp_forward_prefetch": false,
53
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
54
+ "fsdp_fsdp_offload_params": false,
55
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
56
+ "fsdp_fsdp_use_orig_params": true,
57
+ "fsdp_fsdp_activation_checkpointing": false,
58
+ "fsdp_fsdp_limit_all_gathers": true,
59
+ "fsdp_fsdp_sync_module_states": true,
60
+ "use_git": true,
61
+ "random_seed": 0
62
+ }
63
+ },
64
+ "nq_open": {
65
+ "exact_match,remove_whitespace": 0.18421052631578946,
66
+ "exact_match,remove_whitespace_stderr": 0.006452868713443075,
67
+ "args": {
68
+ "exp_name": "results/v1/nq_open/EleutherAI/pythia-12b",
69
+ "task_name": "nq_open",
70
+ "model_name": "EleutherAI/pythia-12b",
71
+ "dataset_param": {
72
+ "max_num_train": 50000,
73
+ "completion_only": false,
74
+ "apply_chat_template": false
75
+ },
76
+ "resume_from_checkpoint": false,
77
+ "train_param": {
78
+ "max_length": 1024,
79
+ "per_device_train_batch_size": 8,
80
+ "auto_find_batch_size": true,
81
+ "num_train_epochs": 1,
82
+ "max_steps": -1,
83
+ "logging_strategy": "steps",
84
+ "save_strategy": "steps",
85
+ "logging_steps": 0.2,
86
+ "save_steps": 0.2,
87
+ "gradient_accumulation_steps": 1,
88
+ "gradient_checkpointing": false,
89
+ "optim": "adamw_torch",
90
+ "learning_rate": 5e-05,
91
+ "weight_decay": 0.01,
92
+ "warmup_steps": 0,
93
+ "report_to": "tensorboard",
94
+ "save_safetensors": true,
95
+ "packing": false
96
+ },
97
+ "use_peft": true,
98
+ "peft_task_type": "CAUSAL_LM",
99
+ "peft_inference_mode": false,
100
+ "peft_r": 8,
101
+ "peft_lora_alpha": 32,
102
+ "peft_lora_dropout": 0.1,
103
+ "fsdp": "none",
104
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
105
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
106
+ "fsdp_fsdp_min_num_params": 100000000.0,
107
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
108
+ "fsdp_fsdp_forward_prefetch": false,
109
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
110
+ "fsdp_fsdp_offload_params": false,
111
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
112
+ "fsdp_fsdp_use_orig_params": true,
113
+ "fsdp_fsdp_activation_checkpointing": false,
114
+ "fsdp_fsdp_limit_all_gathers": true,
115
+ "fsdp_fsdp_sync_module_states": true,
116
+ "use_git": true,
117
+ "random_seed": 0
118
+ }
119
+ },
120
+ "winogrande": {
121
+ "acc": 0.7048145224940805,
122
+ "acc_stderr": 0.012819410741754765,
123
+ "args": {
124
+ "exp_name": "results/v1/winogrande/EleutherAI/pythia-12b",
125
+ "task_name": "winogrande",
126
+ "model_name": "EleutherAI/pythia-12b",
127
+ "dataset_param": {
128
+ "max_num_train": 50000,
129
+ "completion_only": false,
130
+ "apply_chat_template": false
131
+ },
132
+ "resume_from_checkpoint": false,
133
+ "train_param": {
134
+ "max_length": 1024,
135
+ "per_device_train_batch_size": 8,
136
+ "auto_find_batch_size": true,
137
+ "num_train_epochs": 1,
138
+ "max_steps": -1,
139
+ "logging_strategy": "steps",
140
+ "save_strategy": "steps",
141
+ "logging_steps": 0.2,
142
+ "save_steps": 0.2,
143
+ "gradient_accumulation_steps": 1,
144
+ "gradient_checkpointing": false,
145
+ "optim": "adamw_torch",
146
+ "learning_rate": 5e-05,
147
+ "weight_decay": 0.01,
148
+ "warmup_steps": 0,
149
+ "report_to": "tensorboard",
150
+ "save_safetensors": true,
151
+ "packing": false
152
+ },
153
+ "use_peft": true,
154
+ "peft_task_type": "CAUSAL_LM",
155
+ "peft_inference_mode": false,
156
+ "peft_r": 8,
157
+ "peft_lora_alpha": 32,
158
+ "peft_lora_dropout": 0.1,
159
+ "fsdp": "none",
160
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
161
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
162
+ "fsdp_fsdp_min_num_params": 100000000.0,
163
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
164
+ "fsdp_fsdp_forward_prefetch": false,
165
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
166
+ "fsdp_fsdp_offload_params": false,
167
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
168
+ "fsdp_fsdp_use_orig_params": true,
169
+ "fsdp_fsdp_activation_checkpointing": false,
170
+ "fsdp_fsdp_limit_all_gathers": true,
171
+ "fsdp_fsdp_sync_module_states": true,
172
+ "use_git": true,
173
+ "random_seed": 0
174
+ }
175
+ },
176
+ "hellaswag": {
177
+ "acc_norm": 0.7129,
178
+ "acc_norm_stderr": 0.004524312758927125,
179
+ "args": {
180
+ "exp_name": "results/v1/hellaswag/EleutherAI/pythia-12b",
181
+ "task_name": "hellaswag",
182
+ "model_name": "EleutherAI/pythia-12b",
183
+ "dataset_param": {
184
+ "max_num_train": 50000,
185
+ "completion_only": false,
186
+ "apply_chat_template": false
187
+ },
188
+ "resume_from_checkpoint": false,
189
+ "train_param": {
190
+ "max_length": 1024,
191
+ "per_device_train_batch_size": 8,
192
+ "auto_find_batch_size": true,
193
+ "num_train_epochs": 1,
194
+ "max_steps": -1,
195
+ "logging_strategy": "steps",
196
+ "save_strategy": "steps",
197
+ "logging_steps": 0.2,
198
+ "save_steps": 0.2,
199
+ "gradient_accumulation_steps": 1,
200
+ "gradient_checkpointing": false,
201
+ "optim": "adamw_torch",
202
+ "learning_rate": 5e-05,
203
+ "weight_decay": 0.01,
204
+ "warmup_steps": 0,
205
+ "report_to": "tensorboard",
206
+ "save_safetensors": true,
207
+ "packing": false
208
+ },
209
+ "use_peft": true,
210
+ "peft_task_type": "CAUSAL_LM",
211
+ "peft_inference_mode": false,
212
+ "peft_r": 8,
213
+ "peft_lora_alpha": 32,
214
+ "peft_lora_dropout": 0.1,
215
+ "fsdp": "none",
216
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
217
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
218
+ "fsdp_fsdp_min_num_params": 100000000.0,
219
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
220
+ "fsdp_fsdp_forward_prefetch": false,
221
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
222
+ "fsdp_fsdp_offload_params": false,
223
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
224
+ "fsdp_fsdp_use_orig_params": true,
225
+ "fsdp_fsdp_activation_checkpointing": false,
226
+ "fsdp_fsdp_limit_all_gathers": true,
227
+ "fsdp_fsdp_sync_module_states": true,
228
+ "use_git": true,
229
+ "random_seed": 0
230
+ }
231
+ },
232
+ "social_iqa": {
233
+ "acc": 0.5368474923234391,
234
+ "acc_stderr": 0.01128330564744095,
235
+ "args": {
236
+ "exp_name": "results/v1/social_iqa/EleutherAI/pythia-12b",
237
+ "task_name": "social_iqa",
238
+ "model_name": "EleutherAI/pythia-12b",
239
+ "dataset_param": {
240
+ "max_num_train": 50000,
241
+ "completion_only": false,
242
+ "apply_chat_template": false
243
+ },
244
+ "resume_from_checkpoint": false,
245
+ "train_param": {
246
+ "max_length": 1024,
247
+ "per_device_train_batch_size": 8,
248
+ "auto_find_batch_size": true,
249
+ "num_train_epochs": 1,
250
+ "max_steps": -1,
251
+ "logging_strategy": "steps",
252
+ "save_strategy": "steps",
253
+ "logging_steps": 0.2,
254
+ "save_steps": 0.2,
255
+ "gradient_accumulation_steps": 1,
256
+ "gradient_checkpointing": false,
257
+ "optim": "adamw_torch",
258
+ "learning_rate": 5e-05,
259
+ "weight_decay": 0.01,
260
+ "warmup_steps": 0,
261
+ "report_to": "tensorboard",
262
+ "save_safetensors": true,
263
+ "packing": false
264
+ },
265
+ "use_peft": true,
266
+ "peft_task_type": "CAUSAL_LM",
267
+ "peft_inference_mode": false,
268
+ "peft_r": 8,
269
+ "peft_lora_alpha": 32,
270
+ "peft_lora_dropout": 0.1,
271
+ "fsdp": "none",
272
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
273
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
274
+ "fsdp_fsdp_min_num_params": 100000000.0,
275
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
276
+ "fsdp_fsdp_forward_prefetch": false,
277
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
278
+ "fsdp_fsdp_offload_params": false,
279
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
280
+ "fsdp_fsdp_use_orig_params": true,
281
+ "fsdp_fsdp_activation_checkpointing": false,
282
+ "fsdp_fsdp_limit_all_gathers": true,
283
+ "fsdp_fsdp_sync_module_states": true,
284
+ "use_git": true,
285
+ "random_seed": 0
286
+ }
287
+ },
288
+ "piqa": {
289
+ "acc_norm": 0.7850924918389554,
290
+ "acc_norm_stderr": 0.009583665082653311,
291
+ "args": {
292
+ "exp_name": "results/v2/piqa/EleutherAI/pythia-12b",
293
+ "task_name": "piqa",
294
+ "model_name": "EleutherAI/pythia-12b",
295
+ "dataset_param": {
296
+ "max_num_train": 50000,
297
+ "completion_only": false,
298
+ "apply_chat_template": false
299
+ },
300
+ "resume_from_checkpoint": false,
301
+ "train_param": {
302
+ "max_length": 1024,
303
+ "per_device_train_batch_size": 8,
304
+ "auto_find_batch_size": true,
305
+ "num_train_epochs": 5,
306
+ "max_steps": -1,
307
+ "logging_strategy": "epoch",
308
+ "save_strategy": "epoch",
309
+ "logging_steps": 0.2,
310
+ "save_steps": 0.2,
311
+ "gradient_accumulation_steps": 1,
312
+ "gradient_checkpointing": false,
313
+ "optim": "adamw_torch",
314
+ "learning_rate": 5e-05,
315
+ "weight_decay": 0.01,
316
+ "warmup_steps": 0,
317
+ "report_to": "tensorboard",
318
+ "save_safetensors": true,
319
+ "packing": false
320
+ },
321
+ "use_peft": true,
322
+ "peft_task_type": "CAUSAL_LM",
323
+ "peft_inference_mode": false,
324
+ "peft_r": 8,
325
+ "peft_lora_alpha": 32,
326
+ "peft_lora_dropout": 0.1,
327
+ "fsdp": "none",
328
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
329
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
330
+ "fsdp_fsdp_min_num_params": 100000000.0,
331
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
332
+ "fsdp_fsdp_forward_prefetch": false,
333
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
334
+ "fsdp_fsdp_offload_params": false,
335
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
336
+ "fsdp_fsdp_use_orig_params": true,
337
+ "fsdp_fsdp_activation_checkpointing": false,
338
+ "fsdp_fsdp_limit_all_gathers": true,
339
+ "fsdp_fsdp_sync_module_states": true,
340
+ "use_git": true,
341
+ "random_seed": 0
342
+ }
343
+ },
344
+ "commonsense_qa": {
345
+ "acc": 0.647010647010647,
346
+ "acc_stderr": 0.013682221013306819,
347
+ "args": {
348
+ "exp_name": "results/v2/commonsense_qa/EleutherAI/pythia-12b",
349
+ "task_name": "commonsense_qa",
350
+ "model_name": "EleutherAI/pythia-12b",
351
+ "dataset_param": {
352
+ "max_num_train": 50000,
353
+ "completion_only": false,
354
+ "apply_chat_template": false
355
+ },
356
+ "resume_from_checkpoint": false,
357
+ "train_param": {
358
+ "max_length": 1024,
359
+ "per_device_train_batch_size": 8,
360
+ "auto_find_batch_size": true,
361
+ "num_train_epochs": 5,
362
+ "max_steps": -1,
363
+ "logging_strategy": "epoch",
364
+ "save_strategy": "epoch",
365
+ "logging_steps": 0.2,
366
+ "save_steps": 0.2,
367
+ "gradient_accumulation_steps": 1,
368
+ "gradient_checkpointing": false,
369
+ "optim": "adamw_torch",
370
+ "learning_rate": 5e-05,
371
+ "weight_decay": 0.01,
372
+ "warmup_steps": 0,
373
+ "report_to": "tensorboard",
374
+ "save_safetensors": true,
375
+ "packing": false
376
+ },
377
+ "use_peft": true,
378
+ "peft_task_type": "CAUSAL_LM",
379
+ "peft_inference_mode": false,
380
+ "peft_r": 8,
381
+ "peft_lora_alpha": 32,
382
+ "peft_lora_dropout": 0.1,
383
+ "fsdp": "none",
384
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
385
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
386
+ "fsdp_fsdp_min_num_params": 100000000.0,
387
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
388
+ "fsdp_fsdp_forward_prefetch": false,
389
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
390
+ "fsdp_fsdp_offload_params": false,
391
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
392
+ "fsdp_fsdp_use_orig_params": true,
393
+ "fsdp_fsdp_activation_checkpointing": false,
394
+ "fsdp_fsdp_limit_all_gathers": true,
395
+ "fsdp_fsdp_sync_module_states": true,
396
+ "use_git": true,
397
+ "random_seed": 0
398
+ }
399
+ },
400
+ "gsm8k": {
401
+ "exact_match,flexible-extract": 0.14329037149355572,
402
+ "exact_match,flexible-extract_stderr": 0.009650895723357615,
403
+ "args": {
404
+ "exp_name": "results/v2/gsm8k/EleutherAI/pythia-12b",
405
+ "task_name": "gsm8k",
406
+ "model_name": "EleutherAI/pythia-12b",
407
+ "dataset_param": {
408
+ "max_num_train": 50000,
409
+ "completion_only": false,
410
+ "apply_chat_template": false
411
+ },
412
+ "resume_from_checkpoint": false,
413
+ "train_param": {
414
+ "max_length": 1024,
415
+ "per_device_train_batch_size": 8,
416
+ "auto_find_batch_size": true,
417
+ "num_train_epochs": 5,
418
+ "max_steps": -1,
419
+ "logging_strategy": "epoch",
420
+ "save_strategy": "epoch",
421
+ "logging_steps": 0.2,
422
+ "save_steps": 0.2,
423
+ "gradient_accumulation_steps": 1,
424
+ "gradient_checkpointing": false,
425
+ "optim": "adamw_torch",
426
+ "learning_rate": 5e-05,
427
+ "weight_decay": 0.01,
428
+ "warmup_steps": 0,
429
+ "report_to": "tensorboard",
430
+ "save_safetensors": true,
431
+ "packing": false
432
+ },
433
+ "use_peft": true,
434
+ "peft_task_type": "CAUSAL_LM",
435
+ "peft_inference_mode": false,
436
+ "peft_r": 8,
437
+ "peft_lora_alpha": 32,
438
+ "peft_lora_dropout": 0.1,
439
+ "fsdp": "none",
440
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
441
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
442
+ "fsdp_fsdp_min_num_params": 100000000.0,
443
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
444
+ "fsdp_fsdp_forward_prefetch": false,
445
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
446
+ "fsdp_fsdp_offload_params": false,
447
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
448
+ "fsdp_fsdp_use_orig_params": true,
449
+ "fsdp_fsdp_activation_checkpointing": false,
450
+ "fsdp_fsdp_limit_all_gathers": true,
451
+ "fsdp_fsdp_sync_module_states": true,
452
+ "use_git": true,
453
+ "random_seed": 0
454
+ }
455
+ },
456
+ "arc_challenge": {
457
+ "acc_norm": 0.4658703071672355,
458
+ "acc_norm_stderr": 0.0145773113152311,
459
+ "args": {
460
+ "exp_name": "results/v2/arc_challenge/EleutherAI/pythia-12b",
461
+ "task_name": "arc_challenge",
462
+ "model_name": "EleutherAI/pythia-12b",
463
+ "dataset_param": {
464
+ "max_num_train": 50000,
465
+ "completion_only": false,
466
+ "apply_chat_template": false
467
+ },
468
+ "resume_from_checkpoint": false,
469
+ "train_param": {
470
+ "max_length": 1024,
471
+ "per_device_train_batch_size": 8,
472
+ "auto_find_batch_size": true,
473
+ "num_train_epochs": 5,
474
+ "max_steps": -1,
475
+ "logging_strategy": "epoch",
476
+ "save_strategy": "epoch",
477
+ "logging_steps": 0.2,
478
+ "save_steps": 0.2,
479
+ "gradient_accumulation_steps": 1,
480
+ "gradient_checkpointing": false,
481
+ "optim": "adamw_torch",
482
+ "learning_rate": 5e-05,
483
+ "weight_decay": 0.01,
484
+ "warmup_steps": 0,
485
+ "report_to": "tensorboard",
486
+ "save_safetensors": true,
487
+ "packing": false
488
+ },
489
+ "use_peft": true,
490
+ "peft_task_type": "CAUSAL_LM",
491
+ "peft_inference_mode": false,
492
+ "peft_r": 8,
493
+ "peft_lora_alpha": 32,
494
+ "peft_lora_dropout": 0.1,
495
+ "fsdp": "none",
496
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
497
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
498
+ "fsdp_fsdp_min_num_params": 100000000.0,
499
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
500
+ "fsdp_fsdp_forward_prefetch": false,
501
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
502
+ "fsdp_fsdp_offload_params": false,
503
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
504
+ "fsdp_fsdp_use_orig_params": true,
505
+ "fsdp_fsdp_activation_checkpointing": false,
506
+ "fsdp_fsdp_limit_all_gathers": true,
507
+ "fsdp_fsdp_sync_module_states": true,
508
+ "use_git": true,
509
+ "random_seed": 0
510
+ }
511
+ }
512
+ }
513
+ }
EleutherAI/pythia-160m-deduped/results_2026-01-05-04-27-50.json ADDED
@@ -0,0 +1,513 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "model_name": "EleutherAI/pythia-160m-deduped",
4
+ "model_type": "pretrained",
5
+ "params": 0
6
+ },
7
+ "results": {
8
+ "medmcqa": {
9
+ "acc_norm": 0.3114989242170691,
10
+ "acc_norm_stderr": 0.007161245674121981,
11
+ "args": {
12
+ "exp_name": "results/v1/medmcqa/EleutherAI/pythia-160m-deduped",
13
+ "task_name": "medmcqa",
14
+ "model_name": "EleutherAI/pythia-160m-deduped",
15
+ "dataset_param": {
16
+ "max_num_train": 50000,
17
+ "completion_only": false,
18
+ "apply_chat_template": false
19
+ },
20
+ "resume_from_checkpoint": false,
21
+ "train_param": {
22
+ "max_length": 1024,
23
+ "per_device_train_batch_size": 8,
24
+ "auto_find_batch_size": true,
25
+ "num_train_epochs": 1,
26
+ "max_steps": -1,
27
+ "logging_strategy": "steps",
28
+ "save_strategy": "steps",
29
+ "logging_steps": 0.2,
30
+ "save_steps": 0.2,
31
+ "gradient_accumulation_steps": 1,
32
+ "gradient_checkpointing": false,
33
+ "optim": "adamw_torch",
34
+ "learning_rate": 5e-05,
35
+ "weight_decay": 0.01,
36
+ "warmup_steps": 0,
37
+ "report_to": "tensorboard",
38
+ "save_safetensors": true,
39
+ "packing": false
40
+ },
41
+ "use_peft": true,
42
+ "peft_task_type": "CAUSAL_LM",
43
+ "peft_inference_mode": false,
44
+ "peft_r": 8,
45
+ "peft_lora_alpha": 32,
46
+ "peft_lora_dropout": 0.1,
47
+ "fsdp": "none",
48
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
49
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
50
+ "fsdp_fsdp_min_num_params": 100000000.0,
51
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
52
+ "fsdp_fsdp_forward_prefetch": false,
53
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
54
+ "fsdp_fsdp_offload_params": false,
55
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
56
+ "fsdp_fsdp_use_orig_params": true,
57
+ "fsdp_fsdp_activation_checkpointing": false,
58
+ "fsdp_fsdp_limit_all_gathers": true,
59
+ "fsdp_fsdp_sync_module_states": true,
60
+ "use_git": true,
61
+ "random_seed": 0
62
+ }
63
+ },
64
+ "nq_open": {
65
+ "exact_match,remove_whitespace": 0.009972299168975069,
66
+ "exact_match,remove_whitespace_stderr": 0.0016539709656412614,
67
+ "args": {
68
+ "exp_name": "results/v1/nq_open/EleutherAI/pythia-160m-deduped",
69
+ "task_name": "nq_open",
70
+ "model_name": "EleutherAI/pythia-160m-deduped",
71
+ "dataset_param": {
72
+ "max_num_train": 50000,
73
+ "completion_only": false,
74
+ "apply_chat_template": false
75
+ },
76
+ "resume_from_checkpoint": false,
77
+ "train_param": {
78
+ "max_length": 1024,
79
+ "per_device_train_batch_size": 8,
80
+ "auto_find_batch_size": true,
81
+ "num_train_epochs": 1,
82
+ "max_steps": -1,
83
+ "logging_strategy": "steps",
84
+ "save_strategy": "steps",
85
+ "logging_steps": 0.2,
86
+ "save_steps": 0.2,
87
+ "gradient_accumulation_steps": 1,
88
+ "gradient_checkpointing": false,
89
+ "optim": "adamw_torch",
90
+ "learning_rate": 5e-05,
91
+ "weight_decay": 0.01,
92
+ "warmup_steps": 0,
93
+ "report_to": "tensorboard",
94
+ "save_safetensors": true,
95
+ "packing": false
96
+ },
97
+ "use_peft": true,
98
+ "peft_task_type": "CAUSAL_LM",
99
+ "peft_inference_mode": false,
100
+ "peft_r": 8,
101
+ "peft_lora_alpha": 32,
102
+ "peft_lora_dropout": 0.1,
103
+ "fsdp": "none",
104
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
105
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
106
+ "fsdp_fsdp_min_num_params": 100000000.0,
107
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
108
+ "fsdp_fsdp_forward_prefetch": false,
109
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
110
+ "fsdp_fsdp_offload_params": false,
111
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
112
+ "fsdp_fsdp_use_orig_params": true,
113
+ "fsdp_fsdp_activation_checkpointing": false,
114
+ "fsdp_fsdp_limit_all_gathers": true,
115
+ "fsdp_fsdp_sync_module_states": true,
116
+ "use_git": true,
117
+ "random_seed": 0
118
+ }
119
+ },
120
+ "winogrande": {
121
+ "acc": 0.5059194948697711,
122
+ "acc_stderr": 0.014051500838485807,
123
+ "args": {
124
+ "exp_name": "results/v1/winogrande/EleutherAI/pythia-160m-deduped",
125
+ "task_name": "winogrande",
126
+ "model_name": "EleutherAI/pythia-160m-deduped",
127
+ "dataset_param": {
128
+ "max_num_train": 50000,
129
+ "completion_only": false,
130
+ "apply_chat_template": false
131
+ },
132
+ "resume_from_checkpoint": false,
133
+ "train_param": {
134
+ "max_length": 1024,
135
+ "per_device_train_batch_size": 8,
136
+ "auto_find_batch_size": true,
137
+ "num_train_epochs": 1,
138
+ "max_steps": -1,
139
+ "logging_strategy": "steps",
140
+ "save_strategy": "steps",
141
+ "logging_steps": 0.2,
142
+ "save_steps": 0.2,
143
+ "gradient_accumulation_steps": 1,
144
+ "gradient_checkpointing": false,
145
+ "optim": "adamw_torch",
146
+ "learning_rate": 5e-05,
147
+ "weight_decay": 0.01,
148
+ "warmup_steps": 0,
149
+ "report_to": "tensorboard",
150
+ "save_safetensors": true,
151
+ "packing": false
152
+ },
153
+ "use_peft": true,
154
+ "peft_task_type": "CAUSAL_LM",
155
+ "peft_inference_mode": false,
156
+ "peft_r": 8,
157
+ "peft_lora_alpha": 32,
158
+ "peft_lora_dropout": 0.1,
159
+ "fsdp": "none",
160
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
161
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
162
+ "fsdp_fsdp_min_num_params": 100000000.0,
163
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
164
+ "fsdp_fsdp_forward_prefetch": false,
165
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
166
+ "fsdp_fsdp_offload_params": false,
167
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
168
+ "fsdp_fsdp_use_orig_params": true,
169
+ "fsdp_fsdp_activation_checkpointing": false,
170
+ "fsdp_fsdp_limit_all_gathers": true,
171
+ "fsdp_fsdp_sync_module_states": true,
172
+ "use_git": true,
173
+ "random_seed": 0
174
+ }
175
+ },
176
+ "hellaswag": {
177
+ "acc_norm": 0.3104,
178
+ "acc_norm_stderr": 0.004626805024255157,
179
+ "args": {
180
+ "exp_name": "results/v1/hellaswag/EleutherAI/pythia-160m-deduped",
181
+ "task_name": "hellaswag",
182
+ "model_name": "EleutherAI/pythia-160m-deduped",
183
+ "dataset_param": {
184
+ "max_num_train": 50000,
185
+ "completion_only": false,
186
+ "apply_chat_template": false
187
+ },
188
+ "resume_from_checkpoint": false,
189
+ "train_param": {
190
+ "max_length": 1024,
191
+ "per_device_train_batch_size": 8,
192
+ "auto_find_batch_size": true,
193
+ "num_train_epochs": 1,
194
+ "max_steps": -1,
195
+ "logging_strategy": "steps",
196
+ "save_strategy": "steps",
197
+ "logging_steps": 0.2,
198
+ "save_steps": 0.2,
199
+ "gradient_accumulation_steps": 1,
200
+ "gradient_checkpointing": false,
201
+ "optim": "adamw_torch",
202
+ "learning_rate": 5e-05,
203
+ "weight_decay": 0.01,
204
+ "warmup_steps": 0,
205
+ "report_to": "tensorboard",
206
+ "save_safetensors": true,
207
+ "packing": false
208
+ },
209
+ "use_peft": true,
210
+ "peft_task_type": "CAUSAL_LM",
211
+ "peft_inference_mode": false,
212
+ "peft_r": 8,
213
+ "peft_lora_alpha": 32,
214
+ "peft_lora_dropout": 0.1,
215
+ "fsdp": "none",
216
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
217
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
218
+ "fsdp_fsdp_min_num_params": 100000000.0,
219
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
220
+ "fsdp_fsdp_forward_prefetch": false,
221
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
222
+ "fsdp_fsdp_offload_params": false,
223
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
224
+ "fsdp_fsdp_use_orig_params": true,
225
+ "fsdp_fsdp_activation_checkpointing": false,
226
+ "fsdp_fsdp_limit_all_gathers": true,
227
+ "fsdp_fsdp_sync_module_states": true,
228
+ "use_git": true,
229
+ "random_seed": 0
230
+ }
231
+ },
232
+ "social_iqa": {
233
+ "acc": 0.37871033776867963,
234
+ "acc_stderr": 0.010976137078653211,
235
+ "args": {
236
+ "exp_name": "results/v1/social_iqa/EleutherAI/pythia-160m-deduped",
237
+ "task_name": "social_iqa",
238
+ "model_name": "EleutherAI/pythia-160m-deduped",
239
+ "dataset_param": {
240
+ "max_num_train": 50000,
241
+ "completion_only": false,
242
+ "apply_chat_template": false
243
+ },
244
+ "resume_from_checkpoint": false,
245
+ "train_param": {
246
+ "max_length": 1024,
247
+ "per_device_train_batch_size": 8,
248
+ "auto_find_batch_size": true,
249
+ "num_train_epochs": 1,
250
+ "max_steps": -1,
251
+ "logging_strategy": "steps",
252
+ "save_strategy": "steps",
253
+ "logging_steps": 0.2,
254
+ "save_steps": 0.2,
255
+ "gradient_accumulation_steps": 1,
256
+ "gradient_checkpointing": false,
257
+ "optim": "adamw_torch",
258
+ "learning_rate": 5e-05,
259
+ "weight_decay": 0.01,
260
+ "warmup_steps": 0,
261
+ "report_to": "tensorboard",
262
+ "save_safetensors": true,
263
+ "packing": false
264
+ },
265
+ "use_peft": true,
266
+ "peft_task_type": "CAUSAL_LM",
267
+ "peft_inference_mode": false,
268
+ "peft_r": 8,
269
+ "peft_lora_alpha": 32,
270
+ "peft_lora_dropout": 0.1,
271
+ "fsdp": "none",
272
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
273
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
274
+ "fsdp_fsdp_min_num_params": 100000000.0,
275
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
276
+ "fsdp_fsdp_forward_prefetch": false,
277
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
278
+ "fsdp_fsdp_offload_params": false,
279
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
280
+ "fsdp_fsdp_use_orig_params": true,
281
+ "fsdp_fsdp_activation_checkpointing": false,
282
+ "fsdp_fsdp_limit_all_gathers": true,
283
+ "fsdp_fsdp_sync_module_states": true,
284
+ "use_git": true,
285
+ "random_seed": 0
286
+ }
287
+ },
288
+ "piqa": {
289
+ "acc_norm": 0.6284004352557128,
290
+ "acc_norm_stderr": 0.011274603006724752,
291
+ "args": {
292
+ "exp_name": "results/v2/piqa/EleutherAI/pythia-160m-deduped",
293
+ "task_name": "piqa",
294
+ "model_name": "EleutherAI/pythia-160m-deduped",
295
+ "dataset_param": {
296
+ "max_num_train": 50000,
297
+ "completion_only": false,
298
+ "apply_chat_template": false
299
+ },
300
+ "resume_from_checkpoint": false,
301
+ "train_param": {
302
+ "max_length": 1024,
303
+ "per_device_train_batch_size": 8,
304
+ "auto_find_batch_size": true,
305
+ "num_train_epochs": 5,
306
+ "max_steps": -1,
307
+ "logging_strategy": "epoch",
308
+ "save_strategy": "epoch",
309
+ "logging_steps": 0.2,
310
+ "save_steps": 0.2,
311
+ "gradient_accumulation_steps": 1,
312
+ "gradient_checkpointing": false,
313
+ "optim": "adamw_torch",
314
+ "learning_rate": 5e-05,
315
+ "weight_decay": 0.01,
316
+ "warmup_steps": 0,
317
+ "report_to": "tensorboard",
318
+ "save_safetensors": true,
319
+ "packing": false
320
+ },
321
+ "use_peft": true,
322
+ "peft_task_type": "CAUSAL_LM",
323
+ "peft_inference_mode": false,
324
+ "peft_r": 8,
325
+ "peft_lora_alpha": 32,
326
+ "peft_lora_dropout": 0.1,
327
+ "fsdp": "none",
328
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
329
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
330
+ "fsdp_fsdp_min_num_params": 100000000.0,
331
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
332
+ "fsdp_fsdp_forward_prefetch": false,
333
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
334
+ "fsdp_fsdp_offload_params": false,
335
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
336
+ "fsdp_fsdp_use_orig_params": true,
337
+ "fsdp_fsdp_activation_checkpointing": false,
338
+ "fsdp_fsdp_limit_all_gathers": true,
339
+ "fsdp_fsdp_sync_module_states": true,
340
+ "use_git": true,
341
+ "random_seed": 0
342
+ }
343
+ },
344
+ "commonsense_qa": {
345
+ "acc": 0.19164619164619165,
346
+ "acc_stderr": 0.011268624978801631,
347
+ "args": {
348
+ "exp_name": "results/v2/commonsense_qa/EleutherAI/pythia-160m-deduped",
349
+ "task_name": "commonsense_qa",
350
+ "model_name": "EleutherAI/pythia-160m-deduped",
351
+ "dataset_param": {
352
+ "max_num_train": 50000,
353
+ "completion_only": false,
354
+ "apply_chat_template": false
355
+ },
356
+ "resume_from_checkpoint": false,
357
+ "train_param": {
358
+ "max_length": 1024,
359
+ "per_device_train_batch_size": 8,
360
+ "auto_find_batch_size": true,
361
+ "num_train_epochs": 5,
362
+ "max_steps": -1,
363
+ "logging_strategy": "epoch",
364
+ "save_strategy": "epoch",
365
+ "logging_steps": 0.2,
366
+ "save_steps": 0.2,
367
+ "gradient_accumulation_steps": 1,
368
+ "gradient_checkpointing": false,
369
+ "optim": "adamw_torch",
370
+ "learning_rate": 5e-05,
371
+ "weight_decay": 0.01,
372
+ "warmup_steps": 0,
373
+ "report_to": "tensorboard",
374
+ "save_safetensors": true,
375
+ "packing": false
376
+ },
377
+ "use_peft": true,
378
+ "peft_task_type": "CAUSAL_LM",
379
+ "peft_inference_mode": false,
380
+ "peft_r": 8,
381
+ "peft_lora_alpha": 32,
382
+ "peft_lora_dropout": 0.1,
383
+ "fsdp": "none",
384
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
385
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
386
+ "fsdp_fsdp_min_num_params": 100000000.0,
387
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
388
+ "fsdp_fsdp_forward_prefetch": false,
389
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
390
+ "fsdp_fsdp_offload_params": false,
391
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
392
+ "fsdp_fsdp_use_orig_params": true,
393
+ "fsdp_fsdp_activation_checkpointing": false,
394
+ "fsdp_fsdp_limit_all_gathers": true,
395
+ "fsdp_fsdp_sync_module_states": true,
396
+ "use_git": true,
397
+ "random_seed": 0
398
+ }
399
+ },
400
+ "gsm8k": {
401
+ "exact_match,flexible-extract": 0.01592115238817286,
402
+ "exact_match,flexible-extract_stderr": 0.003447819272389003,
403
+ "args": {
404
+ "exp_name": "results/v2/gsm8k/EleutherAI/pythia-160m-deduped",
405
+ "task_name": "gsm8k",
406
+ "model_name": "EleutherAI/pythia-160m-deduped",
407
+ "dataset_param": {
408
+ "max_num_train": 50000,
409
+ "completion_only": false,
410
+ "apply_chat_template": false
411
+ },
412
+ "resume_from_checkpoint": false,
413
+ "train_param": {
414
+ "max_length": 1024,
415
+ "per_device_train_batch_size": 8,
416
+ "auto_find_batch_size": true,
417
+ "num_train_epochs": 5,
418
+ "max_steps": -1,
419
+ "logging_strategy": "epoch",
420
+ "save_strategy": "epoch",
421
+ "logging_steps": 0.2,
422
+ "save_steps": 0.2,
423
+ "gradient_accumulation_steps": 1,
424
+ "gradient_checkpointing": false,
425
+ "optim": "adamw_torch",
426
+ "learning_rate": 5e-05,
427
+ "weight_decay": 0.01,
428
+ "warmup_steps": 0,
429
+ "report_to": "tensorboard",
430
+ "save_safetensors": true,
431
+ "packing": false
432
+ },
433
+ "use_peft": true,
434
+ "peft_task_type": "CAUSAL_LM",
435
+ "peft_inference_mode": false,
436
+ "peft_r": 8,
437
+ "peft_lora_alpha": 32,
438
+ "peft_lora_dropout": 0.1,
439
+ "fsdp": "none",
440
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
441
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
442
+ "fsdp_fsdp_min_num_params": 100000000.0,
443
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
444
+ "fsdp_fsdp_forward_prefetch": false,
445
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
446
+ "fsdp_fsdp_offload_params": false,
447
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
448
+ "fsdp_fsdp_use_orig_params": true,
449
+ "fsdp_fsdp_activation_checkpointing": false,
450
+ "fsdp_fsdp_limit_all_gathers": true,
451
+ "fsdp_fsdp_sync_module_states": true,
452
+ "use_git": true,
453
+ "random_seed": 0
454
+ }
455
+ },
456
+ "arc_challenge": {
457
+ "acc_norm": 0.24829351535836178,
458
+ "acc_norm_stderr": 0.012624912868089736,
459
+ "args": {
460
+ "exp_name": "results/v2/arc_challenge/EleutherAI/pythia-160m-deduped",
461
+ "task_name": "arc_challenge",
462
+ "model_name": "EleutherAI/pythia-160m-deduped",
463
+ "dataset_param": {
464
+ "max_num_train": 50000,
465
+ "completion_only": false,
466
+ "apply_chat_template": false
467
+ },
468
+ "resume_from_checkpoint": false,
469
+ "train_param": {
470
+ "max_length": 1024,
471
+ "per_device_train_batch_size": 8,
472
+ "auto_find_batch_size": true,
473
+ "num_train_epochs": 5,
474
+ "max_steps": -1,
475
+ "logging_strategy": "epoch",
476
+ "save_strategy": "epoch",
477
+ "logging_steps": 0.2,
478
+ "save_steps": 0.2,
479
+ "gradient_accumulation_steps": 1,
480
+ "gradient_checkpointing": false,
481
+ "optim": "adamw_torch",
482
+ "learning_rate": 5e-05,
483
+ "weight_decay": 0.01,
484
+ "warmup_steps": 0,
485
+ "report_to": "tensorboard",
486
+ "save_safetensors": true,
487
+ "packing": false
488
+ },
489
+ "use_peft": true,
490
+ "peft_task_type": "CAUSAL_LM",
491
+ "peft_inference_mode": false,
492
+ "peft_r": 8,
493
+ "peft_lora_alpha": 32,
494
+ "peft_lora_dropout": 0.1,
495
+ "fsdp": "none",
496
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
497
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
498
+ "fsdp_fsdp_min_num_params": 100000000.0,
499
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
500
+ "fsdp_fsdp_forward_prefetch": false,
501
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
502
+ "fsdp_fsdp_offload_params": false,
503
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
504
+ "fsdp_fsdp_use_orig_params": true,
505
+ "fsdp_fsdp_activation_checkpointing": false,
506
+ "fsdp_fsdp_limit_all_gathers": true,
507
+ "fsdp_fsdp_sync_module_states": true,
508
+ "use_git": true,
509
+ "random_seed": 0
510
+ }
511
+ }
512
+ }
513
+ }
EleutherAI/pythia-1b-deduped/results_2026-01-05-04-27-50.json ADDED
@@ -0,0 +1,513 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "model_name": "EleutherAI/pythia-1b-deduped",
4
+ "model_type": "pretrained",
5
+ "params": 1.0
6
+ },
7
+ "results": {
8
+ "medmcqa": {
9
+ "acc_norm": 0.33468802295003586,
10
+ "acc_norm_stderr": 0.00729693944483259,
11
+ "args": {
12
+ "exp_name": "results/v1/medmcqa/EleutherAI/pythia-1b-deduped",
13
+ "task_name": "medmcqa",
14
+ "model_name": "EleutherAI/pythia-1b-deduped",
15
+ "dataset_param": {
16
+ "max_num_train": 50000,
17
+ "completion_only": false,
18
+ "apply_chat_template": false
19
+ },
20
+ "resume_from_checkpoint": false,
21
+ "train_param": {
22
+ "max_length": 1024,
23
+ "per_device_train_batch_size": 8,
24
+ "auto_find_batch_size": true,
25
+ "num_train_epochs": 1,
26
+ "max_steps": -1,
27
+ "logging_strategy": "steps",
28
+ "save_strategy": "steps",
29
+ "logging_steps": 0.2,
30
+ "save_steps": 0.2,
31
+ "gradient_accumulation_steps": 1,
32
+ "gradient_checkpointing": false,
33
+ "optim": "adamw_torch",
34
+ "learning_rate": 5e-05,
35
+ "weight_decay": 0.01,
36
+ "warmup_steps": 0,
37
+ "report_to": "tensorboard",
38
+ "save_safetensors": true,
39
+ "packing": false
40
+ },
41
+ "use_peft": true,
42
+ "peft_task_type": "CAUSAL_LM",
43
+ "peft_inference_mode": false,
44
+ "peft_r": 8,
45
+ "peft_lora_alpha": 32,
46
+ "peft_lora_dropout": 0.1,
47
+ "fsdp": "none",
48
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
49
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
50
+ "fsdp_fsdp_min_num_params": 100000000.0,
51
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
52
+ "fsdp_fsdp_forward_prefetch": false,
53
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
54
+ "fsdp_fsdp_offload_params": false,
55
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
56
+ "fsdp_fsdp_use_orig_params": true,
57
+ "fsdp_fsdp_activation_checkpointing": false,
58
+ "fsdp_fsdp_limit_all_gathers": true,
59
+ "fsdp_fsdp_sync_module_states": true,
60
+ "use_git": true,
61
+ "random_seed": 0
62
+ }
63
+ },
64
+ "nq_open": {
65
+ "exact_match,remove_whitespace": 0.06177285318559557,
66
+ "exact_match,remove_whitespace_stderr": 0.004007369485749786,
67
+ "args": {
68
+ "exp_name": "results/v1/nq_open/EleutherAI/pythia-1b-deduped",
69
+ "task_name": "nq_open",
70
+ "model_name": "EleutherAI/pythia-1b-deduped",
71
+ "dataset_param": {
72
+ "max_num_train": 50000,
73
+ "completion_only": false,
74
+ "apply_chat_template": false
75
+ },
76
+ "resume_from_checkpoint": false,
77
+ "train_param": {
78
+ "max_length": 1024,
79
+ "per_device_train_batch_size": 8,
80
+ "auto_find_batch_size": true,
81
+ "num_train_epochs": 1,
82
+ "max_steps": -1,
83
+ "logging_strategy": "steps",
84
+ "save_strategy": "steps",
85
+ "logging_steps": 0.2,
86
+ "save_steps": 0.2,
87
+ "gradient_accumulation_steps": 1,
88
+ "gradient_checkpointing": false,
89
+ "optim": "adamw_torch",
90
+ "learning_rate": 5e-05,
91
+ "weight_decay": 0.01,
92
+ "warmup_steps": 0,
93
+ "report_to": "tensorboard",
94
+ "save_safetensors": true,
95
+ "packing": false
96
+ },
97
+ "use_peft": true,
98
+ "peft_task_type": "CAUSAL_LM",
99
+ "peft_inference_mode": false,
100
+ "peft_r": 8,
101
+ "peft_lora_alpha": 32,
102
+ "peft_lora_dropout": 0.1,
103
+ "fsdp": "none",
104
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
105
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
106
+ "fsdp_fsdp_min_num_params": 100000000.0,
107
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
108
+ "fsdp_fsdp_forward_prefetch": false,
109
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
110
+ "fsdp_fsdp_offload_params": false,
111
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
112
+ "fsdp_fsdp_use_orig_params": true,
113
+ "fsdp_fsdp_activation_checkpointing": false,
114
+ "fsdp_fsdp_limit_all_gathers": true,
115
+ "fsdp_fsdp_sync_module_states": true,
116
+ "use_git": true,
117
+ "random_seed": 0
118
+ }
119
+ },
120
+ "winogrande": {
121
+ "acc": 0.5816890292028414,
122
+ "acc_stderr": 0.013863669961195887,
123
+ "args": {
124
+ "exp_name": "results/v1/winogrande/EleutherAI/pythia-1b-deduped",
125
+ "task_name": "winogrande",
126
+ "model_name": "EleutherAI/pythia-1b-deduped",
127
+ "dataset_param": {
128
+ "max_num_train": 50000,
129
+ "completion_only": false,
130
+ "apply_chat_template": false
131
+ },
132
+ "resume_from_checkpoint": false,
133
+ "train_param": {
134
+ "max_length": 1024,
135
+ "per_device_train_batch_size": 8,
136
+ "auto_find_batch_size": true,
137
+ "num_train_epochs": 1,
138
+ "max_steps": -1,
139
+ "logging_strategy": "steps",
140
+ "save_strategy": "steps",
141
+ "logging_steps": 0.2,
142
+ "save_steps": 0.2,
143
+ "gradient_accumulation_steps": 1,
144
+ "gradient_checkpointing": false,
145
+ "optim": "adamw_torch",
146
+ "learning_rate": 5e-05,
147
+ "weight_decay": 0.01,
148
+ "warmup_steps": 0,
149
+ "report_to": "tensorboard",
150
+ "save_safetensors": true,
151
+ "packing": false
152
+ },
153
+ "use_peft": true,
154
+ "peft_task_type": "CAUSAL_LM",
155
+ "peft_inference_mode": false,
156
+ "peft_r": 8,
157
+ "peft_lora_alpha": 32,
158
+ "peft_lora_dropout": 0.1,
159
+ "fsdp": "none",
160
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
161
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
162
+ "fsdp_fsdp_min_num_params": 100000000.0,
163
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
164
+ "fsdp_fsdp_forward_prefetch": false,
165
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
166
+ "fsdp_fsdp_offload_params": false,
167
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
168
+ "fsdp_fsdp_use_orig_params": true,
169
+ "fsdp_fsdp_activation_checkpointing": false,
170
+ "fsdp_fsdp_limit_all_gathers": true,
171
+ "fsdp_fsdp_sync_module_states": true,
172
+ "use_git": true,
173
+ "random_seed": 0
174
+ }
175
+ },
176
+ "hellaswag": {
177
+ "acc_norm": 0.5076,
178
+ "acc_norm_stderr": 0.004999672356501502,
179
+ "args": {
180
+ "exp_name": "results/v1/hellaswag/EleutherAI/pythia-1b-deduped",
181
+ "task_name": "hellaswag",
182
+ "model_name": "EleutherAI/pythia-1b-deduped",
183
+ "dataset_param": {
184
+ "max_num_train": 50000,
185
+ "completion_only": false,
186
+ "apply_chat_template": false
187
+ },
188
+ "resume_from_checkpoint": false,
189
+ "train_param": {
190
+ "max_length": 1024,
191
+ "per_device_train_batch_size": 8,
192
+ "auto_find_batch_size": true,
193
+ "num_train_epochs": 1,
194
+ "max_steps": -1,
195
+ "logging_strategy": "steps",
196
+ "save_strategy": "steps",
197
+ "logging_steps": 0.2,
198
+ "save_steps": 0.2,
199
+ "gradient_accumulation_steps": 1,
200
+ "gradient_checkpointing": false,
201
+ "optim": "adamw_torch",
202
+ "learning_rate": 5e-05,
203
+ "weight_decay": 0.01,
204
+ "warmup_steps": 0,
205
+ "report_to": "tensorboard",
206
+ "save_safetensors": true,
207
+ "packing": false
208
+ },
209
+ "use_peft": true,
210
+ "peft_task_type": "CAUSAL_LM",
211
+ "peft_inference_mode": false,
212
+ "peft_r": 8,
213
+ "peft_lora_alpha": 32,
214
+ "peft_lora_dropout": 0.1,
215
+ "fsdp": "none",
216
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
217
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
218
+ "fsdp_fsdp_min_num_params": 100000000.0,
219
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
220
+ "fsdp_fsdp_forward_prefetch": false,
221
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
222
+ "fsdp_fsdp_offload_params": false,
223
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
224
+ "fsdp_fsdp_use_orig_params": true,
225
+ "fsdp_fsdp_activation_checkpointing": false,
226
+ "fsdp_fsdp_limit_all_gathers": true,
227
+ "fsdp_fsdp_sync_module_states": true,
228
+ "use_git": true,
229
+ "random_seed": 0
230
+ }
231
+ },
232
+ "social_iqa": {
233
+ "acc": 0.48464687819856705,
234
+ "acc_stderr": 0.011308735429087271,
235
+ "args": {
236
+ "exp_name": "results/v1/social_iqa/EleutherAI/pythia-1b-deduped",
237
+ "task_name": "social_iqa",
238
+ "model_name": "EleutherAI/pythia-1b-deduped",
239
+ "dataset_param": {
240
+ "max_num_train": 50000,
241
+ "completion_only": false,
242
+ "apply_chat_template": false
243
+ },
244
+ "resume_from_checkpoint": false,
245
+ "train_param": {
246
+ "max_length": 1024,
247
+ "per_device_train_batch_size": 8,
248
+ "auto_find_batch_size": true,
249
+ "num_train_epochs": 1,
250
+ "max_steps": -1,
251
+ "logging_strategy": "steps",
252
+ "save_strategy": "steps",
253
+ "logging_steps": 0.2,
254
+ "save_steps": 0.2,
255
+ "gradient_accumulation_steps": 1,
256
+ "gradient_checkpointing": false,
257
+ "optim": "adamw_torch",
258
+ "learning_rate": 5e-05,
259
+ "weight_decay": 0.01,
260
+ "warmup_steps": 0,
261
+ "report_to": "tensorboard",
262
+ "save_safetensors": true,
263
+ "packing": false
264
+ },
265
+ "use_peft": true,
266
+ "peft_task_type": "CAUSAL_LM",
267
+ "peft_inference_mode": false,
268
+ "peft_r": 8,
269
+ "peft_lora_alpha": 32,
270
+ "peft_lora_dropout": 0.1,
271
+ "fsdp": "none",
272
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
273
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
274
+ "fsdp_fsdp_min_num_params": 100000000.0,
275
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
276
+ "fsdp_fsdp_forward_prefetch": false,
277
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
278
+ "fsdp_fsdp_offload_params": false,
279
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
280
+ "fsdp_fsdp_use_orig_params": true,
281
+ "fsdp_fsdp_activation_checkpointing": false,
282
+ "fsdp_fsdp_limit_all_gathers": true,
283
+ "fsdp_fsdp_sync_module_states": true,
284
+ "use_git": true,
285
+ "random_seed": 0
286
+ }
287
+ },
288
+ "piqa": {
289
+ "acc_norm": 0.7170837867247007,
290
+ "acc_norm_stderr": 0.010508949177489683,
291
+ "args": {
292
+ "exp_name": "results/v2/piqa/EleutherAI/pythia-1b-deduped",
293
+ "task_name": "piqa",
294
+ "model_name": "EleutherAI/pythia-1b-deduped",
295
+ "dataset_param": {
296
+ "max_num_train": 50000,
297
+ "completion_only": false,
298
+ "apply_chat_template": false
299
+ },
300
+ "resume_from_checkpoint": false,
301
+ "train_param": {
302
+ "max_length": 1024,
303
+ "per_device_train_batch_size": 8,
304
+ "auto_find_batch_size": true,
305
+ "num_train_epochs": 5,
306
+ "max_steps": -1,
307
+ "logging_strategy": "epoch",
308
+ "save_strategy": "epoch",
309
+ "logging_steps": 0.2,
310
+ "save_steps": 0.2,
311
+ "gradient_accumulation_steps": 1,
312
+ "gradient_checkpointing": false,
313
+ "optim": "adamw_torch",
314
+ "learning_rate": 5e-05,
315
+ "weight_decay": 0.01,
316
+ "warmup_steps": 0,
317
+ "report_to": "tensorboard",
318
+ "save_safetensors": true,
319
+ "packing": false
320
+ },
321
+ "use_peft": true,
322
+ "peft_task_type": "CAUSAL_LM",
323
+ "peft_inference_mode": false,
324
+ "peft_r": 8,
325
+ "peft_lora_alpha": 32,
326
+ "peft_lora_dropout": 0.1,
327
+ "fsdp": "none",
328
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
329
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
330
+ "fsdp_fsdp_min_num_params": 100000000.0,
331
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
332
+ "fsdp_fsdp_forward_prefetch": false,
333
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
334
+ "fsdp_fsdp_offload_params": false,
335
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
336
+ "fsdp_fsdp_use_orig_params": true,
337
+ "fsdp_fsdp_activation_checkpointing": false,
338
+ "fsdp_fsdp_limit_all_gathers": true,
339
+ "fsdp_fsdp_sync_module_states": true,
340
+ "use_git": true,
341
+ "random_seed": 0
342
+ }
343
+ },
344
+ "commonsense_qa": {
345
+ "acc": 0.20638820638820637,
346
+ "acc_stderr": 0.011586881879177833,
347
+ "args": {
348
+ "exp_name": "results/v2/commonsense_qa/EleutherAI/pythia-1b-deduped",
349
+ "task_name": "commonsense_qa",
350
+ "model_name": "EleutherAI/pythia-1b-deduped",
351
+ "dataset_param": {
352
+ "max_num_train": 50000,
353
+ "completion_only": false,
354
+ "apply_chat_template": false
355
+ },
356
+ "resume_from_checkpoint": false,
357
+ "train_param": {
358
+ "max_length": 1024,
359
+ "per_device_train_batch_size": 8,
360
+ "auto_find_batch_size": true,
361
+ "num_train_epochs": 5,
362
+ "max_steps": -1,
363
+ "logging_strategy": "epoch",
364
+ "save_strategy": "epoch",
365
+ "logging_steps": 0.2,
366
+ "save_steps": 0.2,
367
+ "gradient_accumulation_steps": 1,
368
+ "gradient_checkpointing": false,
369
+ "optim": "adamw_torch",
370
+ "learning_rate": 5e-05,
371
+ "weight_decay": 0.01,
372
+ "warmup_steps": 0,
373
+ "report_to": "tensorboard",
374
+ "save_safetensors": true,
375
+ "packing": false
376
+ },
377
+ "use_peft": true,
378
+ "peft_task_type": "CAUSAL_LM",
379
+ "peft_inference_mode": false,
380
+ "peft_r": 8,
381
+ "peft_lora_alpha": 32,
382
+ "peft_lora_dropout": 0.1,
383
+ "fsdp": "none",
384
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
385
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
386
+ "fsdp_fsdp_min_num_params": 100000000.0,
387
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
388
+ "fsdp_fsdp_forward_prefetch": false,
389
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
390
+ "fsdp_fsdp_offload_params": false,
391
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
392
+ "fsdp_fsdp_use_orig_params": true,
393
+ "fsdp_fsdp_activation_checkpointing": false,
394
+ "fsdp_fsdp_limit_all_gathers": true,
395
+ "fsdp_fsdp_sync_module_states": true,
396
+ "use_git": true,
397
+ "random_seed": 0
398
+ }
399
+ },
400
+ "gsm8k": {
401
+ "exact_match,flexible-extract": 0.03866565579984837,
402
+ "exact_match,flexible-extract_stderr": 0.005310583162098035,
403
+ "args": {
404
+ "exp_name": "results/v2/gsm8k/EleutherAI/pythia-1b-deduped",
405
+ "task_name": "gsm8k",
406
+ "model_name": "EleutherAI/pythia-1b-deduped",
407
+ "dataset_param": {
408
+ "max_num_train": 50000,
409
+ "completion_only": false,
410
+ "apply_chat_template": false
411
+ },
412
+ "resume_from_checkpoint": false,
413
+ "train_param": {
414
+ "max_length": 1024,
415
+ "per_device_train_batch_size": 8,
416
+ "auto_find_batch_size": true,
417
+ "num_train_epochs": 5,
418
+ "max_steps": -1,
419
+ "logging_strategy": "epoch",
420
+ "save_strategy": "epoch",
421
+ "logging_steps": 0.2,
422
+ "save_steps": 0.2,
423
+ "gradient_accumulation_steps": 1,
424
+ "gradient_checkpointing": false,
425
+ "optim": "adamw_torch",
426
+ "learning_rate": 5e-05,
427
+ "weight_decay": 0.01,
428
+ "warmup_steps": 0,
429
+ "report_to": "tensorboard",
430
+ "save_safetensors": true,
431
+ "packing": false
432
+ },
433
+ "use_peft": true,
434
+ "peft_task_type": "CAUSAL_LM",
435
+ "peft_inference_mode": false,
436
+ "peft_r": 8,
437
+ "peft_lora_alpha": 32,
438
+ "peft_lora_dropout": 0.1,
439
+ "fsdp": "none",
440
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
441
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
442
+ "fsdp_fsdp_min_num_params": 100000000.0,
443
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
444
+ "fsdp_fsdp_forward_prefetch": false,
445
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
446
+ "fsdp_fsdp_offload_params": false,
447
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
448
+ "fsdp_fsdp_use_orig_params": true,
449
+ "fsdp_fsdp_activation_checkpointing": false,
450
+ "fsdp_fsdp_limit_all_gathers": true,
451
+ "fsdp_fsdp_sync_module_states": true,
452
+ "use_git": true,
453
+ "random_seed": 0
454
+ }
455
+ },
456
+ "arc_challenge": {
457
+ "acc_norm": 0.30119453924914674,
458
+ "acc_norm_stderr": 0.013406741767847624,
459
+ "args": {
460
+ "exp_name": "results/v2/arc_challenge/EleutherAI/pythia-1b-deduped",
461
+ "task_name": "arc_challenge",
462
+ "model_name": "EleutherAI/pythia-1b-deduped",
463
+ "dataset_param": {
464
+ "max_num_train": 50000,
465
+ "completion_only": false,
466
+ "apply_chat_template": false
467
+ },
468
+ "resume_from_checkpoint": false,
469
+ "train_param": {
470
+ "max_length": 1024,
471
+ "per_device_train_batch_size": 8,
472
+ "auto_find_batch_size": true,
473
+ "num_train_epochs": 5,
474
+ "max_steps": -1,
475
+ "logging_strategy": "epoch",
476
+ "save_strategy": "epoch",
477
+ "logging_steps": 0.2,
478
+ "save_steps": 0.2,
479
+ "gradient_accumulation_steps": 1,
480
+ "gradient_checkpointing": false,
481
+ "optim": "adamw_torch",
482
+ "learning_rate": 5e-05,
483
+ "weight_decay": 0.01,
484
+ "warmup_steps": 0,
485
+ "report_to": "tensorboard",
486
+ "save_safetensors": true,
487
+ "packing": false
488
+ },
489
+ "use_peft": true,
490
+ "peft_task_type": "CAUSAL_LM",
491
+ "peft_inference_mode": false,
492
+ "peft_r": 8,
493
+ "peft_lora_alpha": 32,
494
+ "peft_lora_dropout": 0.1,
495
+ "fsdp": "none",
496
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
497
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
498
+ "fsdp_fsdp_min_num_params": 100000000.0,
499
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
500
+ "fsdp_fsdp_forward_prefetch": false,
501
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
502
+ "fsdp_fsdp_offload_params": false,
503
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
504
+ "fsdp_fsdp_use_orig_params": true,
505
+ "fsdp_fsdp_activation_checkpointing": false,
506
+ "fsdp_fsdp_limit_all_gathers": true,
507
+ "fsdp_fsdp_sync_module_states": true,
508
+ "use_git": true,
509
+ "random_seed": 0
510
+ }
511
+ }
512
+ }
513
+ }
EleutherAI/pythia-2.8b-deduped/results_2026-01-05-04-27-50.json ADDED
@@ -0,0 +1,513 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "model_name": "EleutherAI/pythia-2.8b-deduped",
4
+ "model_type": "pretrained",
5
+ "params": 2.8
6
+ },
7
+ "results": {
8
+ "medmcqa": {
9
+ "acc_norm": 0.33110207984699974,
10
+ "acc_norm_stderr": 0.007277276277307049,
11
+ "args": {
12
+ "exp_name": "results/v1/medmcqa/EleutherAI/pythia-2.8b-deduped",
13
+ "task_name": "medmcqa",
14
+ "model_name": "EleutherAI/pythia-2.8b-deduped",
15
+ "dataset_param": {
16
+ "max_num_train": 50000,
17
+ "completion_only": false,
18
+ "apply_chat_template": false
19
+ },
20
+ "resume_from_checkpoint": false,
21
+ "train_param": {
22
+ "max_length": 1024,
23
+ "per_device_train_batch_size": 8,
24
+ "auto_find_batch_size": true,
25
+ "num_train_epochs": 1,
26
+ "max_steps": -1,
27
+ "logging_strategy": "steps",
28
+ "save_strategy": "steps",
29
+ "logging_steps": 0.2,
30
+ "save_steps": 0.2,
31
+ "gradient_accumulation_steps": 1,
32
+ "gradient_checkpointing": false,
33
+ "optim": "adamw_torch",
34
+ "learning_rate": 5e-05,
35
+ "weight_decay": 0.01,
36
+ "warmup_steps": 0,
37
+ "report_to": "tensorboard",
38
+ "save_safetensors": true,
39
+ "packing": false
40
+ },
41
+ "use_peft": true,
42
+ "peft_task_type": "CAUSAL_LM",
43
+ "peft_inference_mode": false,
44
+ "peft_r": 8,
45
+ "peft_lora_alpha": 32,
46
+ "peft_lora_dropout": 0.1,
47
+ "fsdp": "none",
48
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
49
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
50
+ "fsdp_fsdp_min_num_params": 100000000.0,
51
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
52
+ "fsdp_fsdp_forward_prefetch": false,
53
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
54
+ "fsdp_fsdp_offload_params": false,
55
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
56
+ "fsdp_fsdp_use_orig_params": true,
57
+ "fsdp_fsdp_activation_checkpointing": false,
58
+ "fsdp_fsdp_limit_all_gathers": true,
59
+ "fsdp_fsdp_sync_module_states": true,
60
+ "use_git": true,
61
+ "random_seed": 0
62
+ }
63
+ },
64
+ "nq_open": {
65
+ "exact_match,remove_whitespace": 0.12077562326869806,
66
+ "exact_match,remove_whitespace_stderr": 0.005424329501266528,
67
+ "args": {
68
+ "exp_name": "results/v1/nq_open/EleutherAI/pythia-2.8b-deduped",
69
+ "task_name": "nq_open",
70
+ "model_name": "EleutherAI/pythia-2.8b-deduped",
71
+ "dataset_param": {
72
+ "max_num_train": 50000,
73
+ "completion_only": false,
74
+ "apply_chat_template": false
75
+ },
76
+ "resume_from_checkpoint": false,
77
+ "train_param": {
78
+ "max_length": 1024,
79
+ "per_device_train_batch_size": 8,
80
+ "auto_find_batch_size": true,
81
+ "num_train_epochs": 1,
82
+ "max_steps": -1,
83
+ "logging_strategy": "steps",
84
+ "save_strategy": "steps",
85
+ "logging_steps": 0.2,
86
+ "save_steps": 0.2,
87
+ "gradient_accumulation_steps": 1,
88
+ "gradient_checkpointing": false,
89
+ "optim": "adamw_torch",
90
+ "learning_rate": 5e-05,
91
+ "weight_decay": 0.01,
92
+ "warmup_steps": 0,
93
+ "report_to": "tensorboard",
94
+ "save_safetensors": true,
95
+ "packing": false
96
+ },
97
+ "use_peft": true,
98
+ "peft_task_type": "CAUSAL_LM",
99
+ "peft_inference_mode": false,
100
+ "peft_r": 8,
101
+ "peft_lora_alpha": 32,
102
+ "peft_lora_dropout": 0.1,
103
+ "fsdp": "none",
104
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
105
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
106
+ "fsdp_fsdp_min_num_params": 100000000.0,
107
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
108
+ "fsdp_fsdp_forward_prefetch": false,
109
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
110
+ "fsdp_fsdp_offload_params": false,
111
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
112
+ "fsdp_fsdp_use_orig_params": true,
113
+ "fsdp_fsdp_activation_checkpointing": false,
114
+ "fsdp_fsdp_limit_all_gathers": true,
115
+ "fsdp_fsdp_sync_module_states": true,
116
+ "use_git": true,
117
+ "random_seed": 0
118
+ }
119
+ },
120
+ "winogrande": {
121
+ "acc": 0.6440410418310971,
122
+ "acc_stderr": 0.01345674065627396,
123
+ "args": {
124
+ "exp_name": "results/v1/winogrande/EleutherAI/pythia-2.8b-deduped",
125
+ "task_name": "winogrande",
126
+ "model_name": "EleutherAI/pythia-2.8b-deduped",
127
+ "dataset_param": {
128
+ "max_num_train": 50000,
129
+ "completion_only": false,
130
+ "apply_chat_template": false
131
+ },
132
+ "resume_from_checkpoint": false,
133
+ "train_param": {
134
+ "max_length": 1024,
135
+ "per_device_train_batch_size": 8,
136
+ "auto_find_batch_size": true,
137
+ "num_train_epochs": 1,
138
+ "max_steps": -1,
139
+ "logging_strategy": "steps",
140
+ "save_strategy": "steps",
141
+ "logging_steps": 0.2,
142
+ "save_steps": 0.2,
143
+ "gradient_accumulation_steps": 1,
144
+ "gradient_checkpointing": false,
145
+ "optim": "adamw_torch",
146
+ "learning_rate": 5e-05,
147
+ "weight_decay": 0.01,
148
+ "warmup_steps": 0,
149
+ "report_to": "tensorboard",
150
+ "save_safetensors": true,
151
+ "packing": false
152
+ },
153
+ "use_peft": true,
154
+ "peft_task_type": "CAUSAL_LM",
155
+ "peft_inference_mode": false,
156
+ "peft_r": 8,
157
+ "peft_lora_alpha": 32,
158
+ "peft_lora_dropout": 0.1,
159
+ "fsdp": "none",
160
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
161
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
162
+ "fsdp_fsdp_min_num_params": 100000000.0,
163
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
164
+ "fsdp_fsdp_forward_prefetch": false,
165
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
166
+ "fsdp_fsdp_offload_params": false,
167
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
168
+ "fsdp_fsdp_use_orig_params": true,
169
+ "fsdp_fsdp_activation_checkpointing": false,
170
+ "fsdp_fsdp_limit_all_gathers": true,
171
+ "fsdp_fsdp_sync_module_states": true,
172
+ "use_git": true,
173
+ "random_seed": 0
174
+ }
175
+ },
176
+ "hellaswag": {
177
+ "acc_norm": 0.6333,
178
+ "acc_norm_stderr": 0.004819277284340066,
179
+ "args": {
180
+ "exp_name": "results/v1/hellaswag/EleutherAI/pythia-2.8b-deduped",
181
+ "task_name": "hellaswag",
182
+ "model_name": "EleutherAI/pythia-2.8b-deduped",
183
+ "dataset_param": {
184
+ "max_num_train": 50000,
185
+ "completion_only": false,
186
+ "apply_chat_template": false
187
+ },
188
+ "resume_from_checkpoint": false,
189
+ "train_param": {
190
+ "max_length": 1024,
191
+ "per_device_train_batch_size": 8,
192
+ "auto_find_batch_size": true,
193
+ "num_train_epochs": 1,
194
+ "max_steps": -1,
195
+ "logging_strategy": "steps",
196
+ "save_strategy": "steps",
197
+ "logging_steps": 0.2,
198
+ "save_steps": 0.2,
199
+ "gradient_accumulation_steps": 1,
200
+ "gradient_checkpointing": false,
201
+ "optim": "adamw_torch",
202
+ "learning_rate": 5e-05,
203
+ "weight_decay": 0.01,
204
+ "warmup_steps": 0,
205
+ "report_to": "tensorboard",
206
+ "save_safetensors": true,
207
+ "packing": false
208
+ },
209
+ "use_peft": true,
210
+ "peft_task_type": "CAUSAL_LM",
211
+ "peft_inference_mode": false,
212
+ "peft_r": 8,
213
+ "peft_lora_alpha": 32,
214
+ "peft_lora_dropout": 0.1,
215
+ "fsdp": "none",
216
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
217
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
218
+ "fsdp_fsdp_min_num_params": 100000000.0,
219
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
220
+ "fsdp_fsdp_forward_prefetch": false,
221
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
222
+ "fsdp_fsdp_offload_params": false,
223
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
224
+ "fsdp_fsdp_use_orig_params": true,
225
+ "fsdp_fsdp_activation_checkpointing": false,
226
+ "fsdp_fsdp_limit_all_gathers": true,
227
+ "fsdp_fsdp_sync_module_states": true,
228
+ "use_git": true,
229
+ "random_seed": 0
230
+ }
231
+ },
232
+ "social_iqa": {
233
+ "acc": 0.5112589559877175,
234
+ "acc_stderr": 0.01131120175561399,
235
+ "args": {
236
+ "exp_name": "results/v1/social_iqa/EleutherAI/pythia-2.8b-deduped",
237
+ "task_name": "social_iqa",
238
+ "model_name": "EleutherAI/pythia-2.8b-deduped",
239
+ "dataset_param": {
240
+ "max_num_train": 50000,
241
+ "completion_only": false,
242
+ "apply_chat_template": false
243
+ },
244
+ "resume_from_checkpoint": false,
245
+ "train_param": {
246
+ "max_length": 1024,
247
+ "per_device_train_batch_size": 8,
248
+ "auto_find_batch_size": true,
249
+ "num_train_epochs": 1,
250
+ "max_steps": -1,
251
+ "logging_strategy": "steps",
252
+ "save_strategy": "steps",
253
+ "logging_steps": 0.2,
254
+ "save_steps": 0.2,
255
+ "gradient_accumulation_steps": 1,
256
+ "gradient_checkpointing": false,
257
+ "optim": "adamw_torch",
258
+ "learning_rate": 5e-05,
259
+ "weight_decay": 0.01,
260
+ "warmup_steps": 0,
261
+ "report_to": "tensorboard",
262
+ "save_safetensors": true,
263
+ "packing": false
264
+ },
265
+ "use_peft": true,
266
+ "peft_task_type": "CAUSAL_LM",
267
+ "peft_inference_mode": false,
268
+ "peft_r": 8,
269
+ "peft_lora_alpha": 32,
270
+ "peft_lora_dropout": 0.1,
271
+ "fsdp": "none",
272
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
273
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
274
+ "fsdp_fsdp_min_num_params": 100000000.0,
275
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
276
+ "fsdp_fsdp_forward_prefetch": false,
277
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
278
+ "fsdp_fsdp_offload_params": false,
279
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
280
+ "fsdp_fsdp_use_orig_params": true,
281
+ "fsdp_fsdp_activation_checkpointing": false,
282
+ "fsdp_fsdp_limit_all_gathers": true,
283
+ "fsdp_fsdp_sync_module_states": true,
284
+ "use_git": true,
285
+ "random_seed": 0
286
+ }
287
+ },
288
+ "piqa": {
289
+ "acc_norm": 0.7687704026115343,
290
+ "acc_norm_stderr": 0.009837063180625326,
291
+ "args": {
292
+ "exp_name": "results/v2/piqa/EleutherAI/pythia-2.8b-deduped",
293
+ "task_name": "piqa",
294
+ "model_name": "EleutherAI/pythia-2.8b-deduped",
295
+ "dataset_param": {
296
+ "max_num_train": 50000,
297
+ "completion_only": false,
298
+ "apply_chat_template": false
299
+ },
300
+ "resume_from_checkpoint": false,
301
+ "train_param": {
302
+ "max_length": 1024,
303
+ "per_device_train_batch_size": 8,
304
+ "auto_find_batch_size": true,
305
+ "num_train_epochs": 5,
306
+ "max_steps": -1,
307
+ "logging_strategy": "epoch",
308
+ "save_strategy": "epoch",
309
+ "logging_steps": 0.2,
310
+ "save_steps": 0.2,
311
+ "gradient_accumulation_steps": 1,
312
+ "gradient_checkpointing": false,
313
+ "optim": "adamw_torch",
314
+ "learning_rate": 5e-05,
315
+ "weight_decay": 0.01,
316
+ "warmup_steps": 0,
317
+ "report_to": "tensorboard",
318
+ "save_safetensors": true,
319
+ "packing": false
320
+ },
321
+ "use_peft": true,
322
+ "peft_task_type": "CAUSAL_LM",
323
+ "peft_inference_mode": false,
324
+ "peft_r": 8,
325
+ "peft_lora_alpha": 32,
326
+ "peft_lora_dropout": 0.1,
327
+ "fsdp": "none",
328
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
329
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
330
+ "fsdp_fsdp_min_num_params": 100000000.0,
331
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
332
+ "fsdp_fsdp_forward_prefetch": false,
333
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
334
+ "fsdp_fsdp_offload_params": false,
335
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
336
+ "fsdp_fsdp_use_orig_params": true,
337
+ "fsdp_fsdp_activation_checkpointing": false,
338
+ "fsdp_fsdp_limit_all_gathers": true,
339
+ "fsdp_fsdp_sync_module_states": true,
340
+ "use_git": true,
341
+ "random_seed": 0
342
+ }
343
+ },
344
+ "commonsense_qa": {
345
+ "acc": 0.20556920556920558,
346
+ "acc_stderr": 0.011569834551534302,
347
+ "args": {
348
+ "exp_name": "results/v2/commonsense_qa/EleutherAI/pythia-2.8b-deduped",
349
+ "task_name": "commonsense_qa",
350
+ "model_name": "EleutherAI/pythia-2.8b-deduped",
351
+ "dataset_param": {
352
+ "max_num_train": 50000,
353
+ "completion_only": false,
354
+ "apply_chat_template": false
355
+ },
356
+ "resume_from_checkpoint": false,
357
+ "train_param": {
358
+ "max_length": 1024,
359
+ "per_device_train_batch_size": 8,
360
+ "auto_find_batch_size": true,
361
+ "num_train_epochs": 5,
362
+ "max_steps": -1,
363
+ "logging_strategy": "epoch",
364
+ "save_strategy": "epoch",
365
+ "logging_steps": 0.2,
366
+ "save_steps": 0.2,
367
+ "gradient_accumulation_steps": 1,
368
+ "gradient_checkpointing": false,
369
+ "optim": "adamw_torch",
370
+ "learning_rate": 5e-05,
371
+ "weight_decay": 0.01,
372
+ "warmup_steps": 0,
373
+ "report_to": "tensorboard",
374
+ "save_safetensors": true,
375
+ "packing": false
376
+ },
377
+ "use_peft": true,
378
+ "peft_task_type": "CAUSAL_LM",
379
+ "peft_inference_mode": false,
380
+ "peft_r": 8,
381
+ "peft_lora_alpha": 32,
382
+ "peft_lora_dropout": 0.1,
383
+ "fsdp": "none",
384
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
385
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
386
+ "fsdp_fsdp_min_num_params": 100000000.0,
387
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
388
+ "fsdp_fsdp_forward_prefetch": false,
389
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
390
+ "fsdp_fsdp_offload_params": false,
391
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
392
+ "fsdp_fsdp_use_orig_params": true,
393
+ "fsdp_fsdp_activation_checkpointing": false,
394
+ "fsdp_fsdp_limit_all_gathers": true,
395
+ "fsdp_fsdp_sync_module_states": true,
396
+ "use_git": true,
397
+ "random_seed": 0
398
+ }
399
+ },
400
+ "gsm8k": {
401
+ "exact_match,flexible-extract": 0.09931766489764973,
402
+ "exact_match,flexible-extract_stderr": 0.008238371412683966,
403
+ "args": {
404
+ "exp_name": "results/v2/gsm8k/EleutherAI/pythia-2.8b-deduped",
405
+ "task_name": "gsm8k",
406
+ "model_name": "EleutherAI/pythia-2.8b-deduped",
407
+ "dataset_param": {
408
+ "max_num_train": 50000,
409
+ "completion_only": false,
410
+ "apply_chat_template": false
411
+ },
412
+ "resume_from_checkpoint": false,
413
+ "train_param": {
414
+ "max_length": 1024,
415
+ "per_device_train_batch_size": 8,
416
+ "auto_find_batch_size": true,
417
+ "num_train_epochs": 5,
418
+ "max_steps": -1,
419
+ "logging_strategy": "epoch",
420
+ "save_strategy": "epoch",
421
+ "logging_steps": 0.2,
422
+ "save_steps": 0.2,
423
+ "gradient_accumulation_steps": 1,
424
+ "gradient_checkpointing": false,
425
+ "optim": "adamw_torch",
426
+ "learning_rate": 5e-05,
427
+ "weight_decay": 0.01,
428
+ "warmup_steps": 0,
429
+ "report_to": "tensorboard",
430
+ "save_safetensors": true,
431
+ "packing": false
432
+ },
433
+ "use_peft": true,
434
+ "peft_task_type": "CAUSAL_LM",
435
+ "peft_inference_mode": false,
436
+ "peft_r": 8,
437
+ "peft_lora_alpha": 32,
438
+ "peft_lora_dropout": 0.1,
439
+ "fsdp": "none",
440
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
441
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
442
+ "fsdp_fsdp_min_num_params": 100000000.0,
443
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
444
+ "fsdp_fsdp_forward_prefetch": false,
445
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
446
+ "fsdp_fsdp_offload_params": false,
447
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
448
+ "fsdp_fsdp_use_orig_params": true,
449
+ "fsdp_fsdp_activation_checkpointing": false,
450
+ "fsdp_fsdp_limit_all_gathers": true,
451
+ "fsdp_fsdp_sync_module_states": true,
452
+ "use_git": true,
453
+ "random_seed": 0
454
+ }
455
+ },
456
+ "arc_challenge": {
457
+ "acc_norm": 0.39761092150170646,
458
+ "acc_norm_stderr": 0.014301752223279533,
459
+ "args": {
460
+ "exp_name": "results/v2/arc_challenge/EleutherAI/pythia-2.8b-deduped",
461
+ "task_name": "arc_challenge",
462
+ "model_name": "EleutherAI/pythia-2.8b-deduped",
463
+ "dataset_param": {
464
+ "max_num_train": 50000,
465
+ "completion_only": false,
466
+ "apply_chat_template": false
467
+ },
468
+ "resume_from_checkpoint": false,
469
+ "train_param": {
470
+ "max_length": 1024,
471
+ "per_device_train_batch_size": 8,
472
+ "auto_find_batch_size": true,
473
+ "num_train_epochs": 5,
474
+ "max_steps": -1,
475
+ "logging_strategy": "epoch",
476
+ "save_strategy": "epoch",
477
+ "logging_steps": 0.2,
478
+ "save_steps": 0.2,
479
+ "gradient_accumulation_steps": 1,
480
+ "gradient_checkpointing": false,
481
+ "optim": "adamw_torch",
482
+ "learning_rate": 5e-05,
483
+ "weight_decay": 0.01,
484
+ "warmup_steps": 0,
485
+ "report_to": "tensorboard",
486
+ "save_safetensors": true,
487
+ "packing": false
488
+ },
489
+ "use_peft": true,
490
+ "peft_task_type": "CAUSAL_LM",
491
+ "peft_inference_mode": false,
492
+ "peft_r": 8,
493
+ "peft_lora_alpha": 32,
494
+ "peft_lora_dropout": 0.1,
495
+ "fsdp": "none",
496
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
497
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
498
+ "fsdp_fsdp_min_num_params": 100000000.0,
499
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
500
+ "fsdp_fsdp_forward_prefetch": false,
501
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
502
+ "fsdp_fsdp_offload_params": false,
503
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
504
+ "fsdp_fsdp_use_orig_params": true,
505
+ "fsdp_fsdp_activation_checkpointing": false,
506
+ "fsdp_fsdp_limit_all_gathers": true,
507
+ "fsdp_fsdp_sync_module_states": true,
508
+ "use_git": true,
509
+ "random_seed": 0
510
+ }
511
+ }
512
+ }
513
+ }
EleutherAI/pythia-410m-deduped/results_2026-01-05-04-27-50.json ADDED
@@ -0,0 +1,513 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "model_name": "EleutherAI/pythia-410m-deduped",
4
+ "model_type": "pretrained",
5
+ "params": 0
6
+ },
7
+ "results": {
8
+ "medmcqa": {
9
+ "acc_norm": 0.31006454697585467,
10
+ "acc_norm_stderr": 0.007152177311002874,
11
+ "args": {
12
+ "exp_name": "results/v1/medmcqa/EleutherAI/pythia-410m-deduped",
13
+ "task_name": "medmcqa",
14
+ "model_name": "EleutherAI/pythia-410m-deduped",
15
+ "dataset_param": {
16
+ "max_num_train": 50000,
17
+ "completion_only": false,
18
+ "apply_chat_template": false
19
+ },
20
+ "resume_from_checkpoint": false,
21
+ "train_param": {
22
+ "max_length": 1024,
23
+ "per_device_train_batch_size": 8,
24
+ "auto_find_batch_size": true,
25
+ "num_train_epochs": 1,
26
+ "max_steps": -1,
27
+ "logging_strategy": "steps",
28
+ "save_strategy": "steps",
29
+ "logging_steps": 0.2,
30
+ "save_steps": 0.2,
31
+ "gradient_accumulation_steps": 1,
32
+ "gradient_checkpointing": false,
33
+ "optim": "adamw_torch",
34
+ "learning_rate": 5e-05,
35
+ "weight_decay": 0.01,
36
+ "warmup_steps": 0,
37
+ "report_to": "tensorboard",
38
+ "save_safetensors": true,
39
+ "packing": false
40
+ },
41
+ "use_peft": true,
42
+ "peft_task_type": "CAUSAL_LM",
43
+ "peft_inference_mode": false,
44
+ "peft_r": 8,
45
+ "peft_lora_alpha": 32,
46
+ "peft_lora_dropout": 0.1,
47
+ "fsdp": "none",
48
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
49
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
50
+ "fsdp_fsdp_min_num_params": 100000000.0,
51
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
52
+ "fsdp_fsdp_forward_prefetch": false,
53
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
54
+ "fsdp_fsdp_offload_params": false,
55
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
56
+ "fsdp_fsdp_use_orig_params": true,
57
+ "fsdp_fsdp_activation_checkpointing": false,
58
+ "fsdp_fsdp_limit_all_gathers": true,
59
+ "fsdp_fsdp_sync_module_states": true,
60
+ "use_git": true,
61
+ "random_seed": 0
62
+ }
63
+ },
64
+ "nq_open": {
65
+ "exact_match,remove_whitespace": 0.039889196675900275,
66
+ "exact_match,remove_whitespace_stderr": 0.003257578442335031,
67
+ "args": {
68
+ "exp_name": "results/v1/nq_open/EleutherAI/pythia-410m-deduped",
69
+ "task_name": "nq_open",
70
+ "model_name": "EleutherAI/pythia-410m-deduped",
71
+ "dataset_param": {
72
+ "max_num_train": 50000,
73
+ "completion_only": false,
74
+ "apply_chat_template": false
75
+ },
76
+ "resume_from_checkpoint": false,
77
+ "train_param": {
78
+ "max_length": 1024,
79
+ "per_device_train_batch_size": 8,
80
+ "auto_find_batch_size": true,
81
+ "num_train_epochs": 1,
82
+ "max_steps": -1,
83
+ "logging_strategy": "steps",
84
+ "save_strategy": "steps",
85
+ "logging_steps": 0.2,
86
+ "save_steps": 0.2,
87
+ "gradient_accumulation_steps": 1,
88
+ "gradient_checkpointing": false,
89
+ "optim": "adamw_torch",
90
+ "learning_rate": 5e-05,
91
+ "weight_decay": 0.01,
92
+ "warmup_steps": 0,
93
+ "report_to": "tensorboard",
94
+ "save_safetensors": true,
95
+ "packing": false
96
+ },
97
+ "use_peft": true,
98
+ "peft_task_type": "CAUSAL_LM",
99
+ "peft_inference_mode": false,
100
+ "peft_r": 8,
101
+ "peft_lora_alpha": 32,
102
+ "peft_lora_dropout": 0.1,
103
+ "fsdp": "none",
104
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
105
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
106
+ "fsdp_fsdp_min_num_params": 100000000.0,
107
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
108
+ "fsdp_fsdp_forward_prefetch": false,
109
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
110
+ "fsdp_fsdp_offload_params": false,
111
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
112
+ "fsdp_fsdp_use_orig_params": true,
113
+ "fsdp_fsdp_activation_checkpointing": false,
114
+ "fsdp_fsdp_limit_all_gathers": true,
115
+ "fsdp_fsdp_sync_module_states": true,
116
+ "use_git": true,
117
+ "random_seed": 0
118
+ }
119
+ },
120
+ "winogrande": {
121
+ "acc": 0.5524861878453039,
122
+ "acc_stderr": 0.013974847640536197,
123
+ "args": {
124
+ "exp_name": "results/v1/winogrande/EleutherAI/pythia-410m-deduped",
125
+ "task_name": "winogrande",
126
+ "model_name": "EleutherAI/pythia-410m-deduped",
127
+ "dataset_param": {
128
+ "max_num_train": 50000,
129
+ "completion_only": false,
130
+ "apply_chat_template": false
131
+ },
132
+ "resume_from_checkpoint": false,
133
+ "train_param": {
134
+ "max_length": 1024,
135
+ "per_device_train_batch_size": 8,
136
+ "auto_find_batch_size": true,
137
+ "num_train_epochs": 1,
138
+ "max_steps": -1,
139
+ "logging_strategy": "steps",
140
+ "save_strategy": "steps",
141
+ "logging_steps": 0.2,
142
+ "save_steps": 0.2,
143
+ "gradient_accumulation_steps": 1,
144
+ "gradient_checkpointing": false,
145
+ "optim": "adamw_torch",
146
+ "learning_rate": 5e-05,
147
+ "weight_decay": 0.01,
148
+ "warmup_steps": 0,
149
+ "report_to": "tensorboard",
150
+ "save_safetensors": true,
151
+ "packing": false
152
+ },
153
+ "use_peft": true,
154
+ "peft_task_type": "CAUSAL_LM",
155
+ "peft_inference_mode": false,
156
+ "peft_r": 8,
157
+ "peft_lora_alpha": 32,
158
+ "peft_lora_dropout": 0.1,
159
+ "fsdp": "none",
160
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
161
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
162
+ "fsdp_fsdp_min_num_params": 100000000.0,
163
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
164
+ "fsdp_fsdp_forward_prefetch": false,
165
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
166
+ "fsdp_fsdp_offload_params": false,
167
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
168
+ "fsdp_fsdp_use_orig_params": true,
169
+ "fsdp_fsdp_activation_checkpointing": false,
170
+ "fsdp_fsdp_limit_all_gathers": true,
171
+ "fsdp_fsdp_sync_module_states": true,
172
+ "use_git": true,
173
+ "random_seed": 0
174
+ }
175
+ },
176
+ "hellaswag": {
177
+ "acc_norm": 0.4274,
178
+ "acc_norm_stderr": 0.004947258992326414,
179
+ "args": {
180
+ "exp_name": "results/v1/hellaswag/EleutherAI/pythia-410m-deduped",
181
+ "task_name": "hellaswag",
182
+ "model_name": "EleutherAI/pythia-410m-deduped",
183
+ "dataset_param": {
184
+ "max_num_train": 50000,
185
+ "completion_only": false,
186
+ "apply_chat_template": false
187
+ },
188
+ "resume_from_checkpoint": false,
189
+ "train_param": {
190
+ "max_length": 1024,
191
+ "per_device_train_batch_size": 8,
192
+ "auto_find_batch_size": true,
193
+ "num_train_epochs": 1,
194
+ "max_steps": -1,
195
+ "logging_strategy": "steps",
196
+ "save_strategy": "steps",
197
+ "logging_steps": 0.2,
198
+ "save_steps": 0.2,
199
+ "gradient_accumulation_steps": 1,
200
+ "gradient_checkpointing": false,
201
+ "optim": "adamw_torch",
202
+ "learning_rate": 5e-05,
203
+ "weight_decay": 0.01,
204
+ "warmup_steps": 0,
205
+ "report_to": "tensorboard",
206
+ "save_safetensors": true,
207
+ "packing": false
208
+ },
209
+ "use_peft": true,
210
+ "peft_task_type": "CAUSAL_LM",
211
+ "peft_inference_mode": false,
212
+ "peft_r": 8,
213
+ "peft_lora_alpha": 32,
214
+ "peft_lora_dropout": 0.1,
215
+ "fsdp": "none",
216
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
217
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
218
+ "fsdp_fsdp_min_num_params": 100000000.0,
219
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
220
+ "fsdp_fsdp_forward_prefetch": false,
221
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
222
+ "fsdp_fsdp_offload_params": false,
223
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
224
+ "fsdp_fsdp_use_orig_params": true,
225
+ "fsdp_fsdp_activation_checkpointing": false,
226
+ "fsdp_fsdp_limit_all_gathers": true,
227
+ "fsdp_fsdp_sync_module_states": true,
228
+ "use_git": true,
229
+ "random_seed": 0
230
+ }
231
+ },
232
+ "social_iqa": {
233
+ "acc": 0.45291709314227224,
234
+ "acc_stderr": 0.01126379679411243,
235
+ "args": {
236
+ "exp_name": "results/v1/social_iqa/EleutherAI/pythia-410m-deduped",
237
+ "task_name": "social_iqa",
238
+ "model_name": "EleutherAI/pythia-410m-deduped",
239
+ "dataset_param": {
240
+ "max_num_train": 50000,
241
+ "completion_only": false,
242
+ "apply_chat_template": false
243
+ },
244
+ "resume_from_checkpoint": false,
245
+ "train_param": {
246
+ "max_length": 1024,
247
+ "per_device_train_batch_size": 8,
248
+ "auto_find_batch_size": true,
249
+ "num_train_epochs": 1,
250
+ "max_steps": -1,
251
+ "logging_strategy": "steps",
252
+ "save_strategy": "steps",
253
+ "logging_steps": 0.2,
254
+ "save_steps": 0.2,
255
+ "gradient_accumulation_steps": 1,
256
+ "gradient_checkpointing": false,
257
+ "optim": "adamw_torch",
258
+ "learning_rate": 5e-05,
259
+ "weight_decay": 0.01,
260
+ "warmup_steps": 0,
261
+ "report_to": "tensorboard",
262
+ "save_safetensors": true,
263
+ "packing": false
264
+ },
265
+ "use_peft": true,
266
+ "peft_task_type": "CAUSAL_LM",
267
+ "peft_inference_mode": false,
268
+ "peft_r": 8,
269
+ "peft_lora_alpha": 32,
270
+ "peft_lora_dropout": 0.1,
271
+ "fsdp": "none",
272
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
273
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
274
+ "fsdp_fsdp_min_num_params": 100000000.0,
275
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
276
+ "fsdp_fsdp_forward_prefetch": false,
277
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
278
+ "fsdp_fsdp_offload_params": false,
279
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
280
+ "fsdp_fsdp_use_orig_params": true,
281
+ "fsdp_fsdp_activation_checkpointing": false,
282
+ "fsdp_fsdp_limit_all_gathers": true,
283
+ "fsdp_fsdp_sync_module_states": true,
284
+ "use_git": true,
285
+ "random_seed": 0
286
+ }
287
+ },
288
+ "piqa": {
289
+ "acc_norm": 0.6866158868335147,
290
+ "acc_norm_stderr": 0.01082282992919548,
291
+ "args": {
292
+ "exp_name": "results/v2/piqa/EleutherAI/pythia-410m-deduped",
293
+ "task_name": "piqa",
294
+ "model_name": "EleutherAI/pythia-410m-deduped",
295
+ "dataset_param": {
296
+ "max_num_train": 50000,
297
+ "completion_only": false,
298
+ "apply_chat_template": false
299
+ },
300
+ "resume_from_checkpoint": false,
301
+ "train_param": {
302
+ "max_length": 1024,
303
+ "per_device_train_batch_size": 8,
304
+ "auto_find_batch_size": true,
305
+ "num_train_epochs": 5,
306
+ "max_steps": -1,
307
+ "logging_strategy": "epoch",
308
+ "save_strategy": "epoch",
309
+ "logging_steps": 0.2,
310
+ "save_steps": 0.2,
311
+ "gradient_accumulation_steps": 1,
312
+ "gradient_checkpointing": false,
313
+ "optim": "adamw_torch",
314
+ "learning_rate": 5e-05,
315
+ "weight_decay": 0.01,
316
+ "warmup_steps": 0,
317
+ "report_to": "tensorboard",
318
+ "save_safetensors": true,
319
+ "packing": false
320
+ },
321
+ "use_peft": true,
322
+ "peft_task_type": "CAUSAL_LM",
323
+ "peft_inference_mode": false,
324
+ "peft_r": 8,
325
+ "peft_lora_alpha": 32,
326
+ "peft_lora_dropout": 0.1,
327
+ "fsdp": "none",
328
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
329
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
330
+ "fsdp_fsdp_min_num_params": 100000000.0,
331
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
332
+ "fsdp_fsdp_forward_prefetch": false,
333
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
334
+ "fsdp_fsdp_offload_params": false,
335
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
336
+ "fsdp_fsdp_use_orig_params": true,
337
+ "fsdp_fsdp_activation_checkpointing": false,
338
+ "fsdp_fsdp_limit_all_gathers": true,
339
+ "fsdp_fsdp_sync_module_states": true,
340
+ "use_git": true,
341
+ "random_seed": 0
342
+ }
343
+ },
344
+ "commonsense_qa": {
345
+ "acc": 0.20475020475020475,
346
+ "acc_stderr": 0.01155271447787666,
347
+ "args": {
348
+ "exp_name": "results/v2/commonsense_qa/EleutherAI/pythia-410m-deduped",
349
+ "task_name": "commonsense_qa",
350
+ "model_name": "EleutherAI/pythia-410m-deduped",
351
+ "dataset_param": {
352
+ "max_num_train": 50000,
353
+ "completion_only": false,
354
+ "apply_chat_template": false
355
+ },
356
+ "resume_from_checkpoint": false,
357
+ "train_param": {
358
+ "max_length": 1024,
359
+ "per_device_train_batch_size": 8,
360
+ "auto_find_batch_size": true,
361
+ "num_train_epochs": 5,
362
+ "max_steps": -1,
363
+ "logging_strategy": "epoch",
364
+ "save_strategy": "epoch",
365
+ "logging_steps": 0.2,
366
+ "save_steps": 0.2,
367
+ "gradient_accumulation_steps": 1,
368
+ "gradient_checkpointing": false,
369
+ "optim": "adamw_torch",
370
+ "learning_rate": 5e-05,
371
+ "weight_decay": 0.01,
372
+ "warmup_steps": 0,
373
+ "report_to": "tensorboard",
374
+ "save_safetensors": true,
375
+ "packing": false
376
+ },
377
+ "use_peft": true,
378
+ "peft_task_type": "CAUSAL_LM",
379
+ "peft_inference_mode": false,
380
+ "peft_r": 8,
381
+ "peft_lora_alpha": 32,
382
+ "peft_lora_dropout": 0.1,
383
+ "fsdp": "none",
384
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
385
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
386
+ "fsdp_fsdp_min_num_params": 100000000.0,
387
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
388
+ "fsdp_fsdp_forward_prefetch": false,
389
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
390
+ "fsdp_fsdp_offload_params": false,
391
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
392
+ "fsdp_fsdp_use_orig_params": true,
393
+ "fsdp_fsdp_activation_checkpointing": false,
394
+ "fsdp_fsdp_limit_all_gathers": true,
395
+ "fsdp_fsdp_sync_module_states": true,
396
+ "use_git": true,
397
+ "random_seed": 0
398
+ }
399
+ },
400
+ "gsm8k": {
401
+ "exact_match,flexible-extract": 0.025018953752843062,
402
+ "exact_match,flexible-extract_stderr": 0.004302045046564306,
403
+ "args": {
404
+ "exp_name": "results/v2/gsm8k/EleutherAI/pythia-410m-deduped",
405
+ "task_name": "gsm8k",
406
+ "model_name": "EleutherAI/pythia-410m-deduped",
407
+ "dataset_param": {
408
+ "max_num_train": 50000,
409
+ "completion_only": false,
410
+ "apply_chat_template": false
411
+ },
412
+ "resume_from_checkpoint": false,
413
+ "train_param": {
414
+ "max_length": 1024,
415
+ "per_device_train_batch_size": 8,
416
+ "auto_find_batch_size": true,
417
+ "num_train_epochs": 5,
418
+ "max_steps": -1,
419
+ "logging_strategy": "epoch",
420
+ "save_strategy": "epoch",
421
+ "logging_steps": 0.2,
422
+ "save_steps": 0.2,
423
+ "gradient_accumulation_steps": 1,
424
+ "gradient_checkpointing": false,
425
+ "optim": "adamw_torch",
426
+ "learning_rate": 5e-05,
427
+ "weight_decay": 0.01,
428
+ "warmup_steps": 0,
429
+ "report_to": "tensorboard",
430
+ "save_safetensors": true,
431
+ "packing": false
432
+ },
433
+ "use_peft": true,
434
+ "peft_task_type": "CAUSAL_LM",
435
+ "peft_inference_mode": false,
436
+ "peft_r": 8,
437
+ "peft_lora_alpha": 32,
438
+ "peft_lora_dropout": 0.1,
439
+ "fsdp": "none",
440
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
441
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
442
+ "fsdp_fsdp_min_num_params": 100000000.0,
443
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
444
+ "fsdp_fsdp_forward_prefetch": false,
445
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
446
+ "fsdp_fsdp_offload_params": false,
447
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
448
+ "fsdp_fsdp_use_orig_params": true,
449
+ "fsdp_fsdp_activation_checkpointing": false,
450
+ "fsdp_fsdp_limit_all_gathers": true,
451
+ "fsdp_fsdp_sync_module_states": true,
452
+ "use_git": true,
453
+ "random_seed": 0
454
+ }
455
+ },
456
+ "arc_challenge": {
457
+ "acc_norm": 0.2636518771331058,
458
+ "acc_norm_stderr": 0.012875929151297054,
459
+ "args": {
460
+ "exp_name": "results/v2/arc_challenge/EleutherAI/pythia-410m-deduped",
461
+ "task_name": "arc_challenge",
462
+ "model_name": "EleutherAI/pythia-410m-deduped",
463
+ "dataset_param": {
464
+ "max_num_train": 50000,
465
+ "completion_only": false,
466
+ "apply_chat_template": false
467
+ },
468
+ "resume_from_checkpoint": false,
469
+ "train_param": {
470
+ "max_length": 1024,
471
+ "per_device_train_batch_size": 8,
472
+ "auto_find_batch_size": true,
473
+ "num_train_epochs": 5,
474
+ "max_steps": -1,
475
+ "logging_strategy": "epoch",
476
+ "save_strategy": "epoch",
477
+ "logging_steps": 0.2,
478
+ "save_steps": 0.2,
479
+ "gradient_accumulation_steps": 1,
480
+ "gradient_checkpointing": false,
481
+ "optim": "adamw_torch",
482
+ "learning_rate": 5e-05,
483
+ "weight_decay": 0.01,
484
+ "warmup_steps": 0,
485
+ "report_to": "tensorboard",
486
+ "save_safetensors": true,
487
+ "packing": false
488
+ },
489
+ "use_peft": true,
490
+ "peft_task_type": "CAUSAL_LM",
491
+ "peft_inference_mode": false,
492
+ "peft_r": 8,
493
+ "peft_lora_alpha": 32,
494
+ "peft_lora_dropout": 0.1,
495
+ "fsdp": "none",
496
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
497
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
498
+ "fsdp_fsdp_min_num_params": 100000000.0,
499
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
500
+ "fsdp_fsdp_forward_prefetch": false,
501
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
502
+ "fsdp_fsdp_offload_params": false,
503
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
504
+ "fsdp_fsdp_use_orig_params": true,
505
+ "fsdp_fsdp_activation_checkpointing": false,
506
+ "fsdp_fsdp_limit_all_gathers": true,
507
+ "fsdp_fsdp_sync_module_states": true,
508
+ "use_git": true,
509
+ "random_seed": 0
510
+ }
511
+ }
512
+ }
513
+ }
EleutherAI/pythia-6.9b-deduped/results_2026-01-05-04-27-50.json ADDED
@@ -0,0 +1,513 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "model_name": "EleutherAI/pythia-6.9b-deduped",
4
+ "model_type": "pretrained",
5
+ "params": 6.9
6
+ },
7
+ "results": {
8
+ "medmcqa": {
9
+ "acc_norm": 0.3258426966292135,
10
+ "acc_norm_stderr": 0.007247573080870266,
11
+ "args": {
12
+ "exp_name": "results/v1/medmcqa/EleutherAI/pythia-6.9b-deduped",
13
+ "task_name": "medmcqa",
14
+ "model_name": "EleutherAI/pythia-6.9b-deduped",
15
+ "dataset_param": {
16
+ "max_num_train": 50000,
17
+ "completion_only": false,
18
+ "apply_chat_template": false
19
+ },
20
+ "resume_from_checkpoint": false,
21
+ "train_param": {
22
+ "max_length": 1024,
23
+ "per_device_train_batch_size": 8,
24
+ "auto_find_batch_size": true,
25
+ "num_train_epochs": 1,
26
+ "max_steps": -1,
27
+ "logging_strategy": "steps",
28
+ "save_strategy": "steps",
29
+ "logging_steps": 0.2,
30
+ "save_steps": 0.2,
31
+ "gradient_accumulation_steps": 1,
32
+ "gradient_checkpointing": false,
33
+ "optim": "adamw_torch",
34
+ "learning_rate": 5e-05,
35
+ "weight_decay": 0.01,
36
+ "warmup_steps": 0,
37
+ "report_to": "tensorboard",
38
+ "save_safetensors": true,
39
+ "packing": false
40
+ },
41
+ "use_peft": true,
42
+ "peft_task_type": "CAUSAL_LM",
43
+ "peft_inference_mode": false,
44
+ "peft_r": 8,
45
+ "peft_lora_alpha": 32,
46
+ "peft_lora_dropout": 0.1,
47
+ "fsdp": "none",
48
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
49
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
50
+ "fsdp_fsdp_min_num_params": 100000000.0,
51
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
52
+ "fsdp_fsdp_forward_prefetch": false,
53
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
54
+ "fsdp_fsdp_offload_params": false,
55
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
56
+ "fsdp_fsdp_use_orig_params": true,
57
+ "fsdp_fsdp_activation_checkpointing": false,
58
+ "fsdp_fsdp_limit_all_gathers": true,
59
+ "fsdp_fsdp_sync_module_states": true,
60
+ "use_git": true,
61
+ "random_seed": 0
62
+ }
63
+ },
64
+ "nq_open": {
65
+ "exact_match,remove_whitespace": 0.14681440443213298,
66
+ "exact_match,remove_whitespace_stderr": 0.00589132050298358,
67
+ "args": {
68
+ "exp_name": "results/v1/nq_open/EleutherAI/pythia-6.9b-deduped",
69
+ "task_name": "nq_open",
70
+ "model_name": "EleutherAI/pythia-6.9b-deduped",
71
+ "dataset_param": {
72
+ "max_num_train": 50000,
73
+ "completion_only": false,
74
+ "apply_chat_template": false
75
+ },
76
+ "resume_from_checkpoint": false,
77
+ "train_param": {
78
+ "max_length": 1024,
79
+ "per_device_train_batch_size": 8,
80
+ "auto_find_batch_size": true,
81
+ "num_train_epochs": 1,
82
+ "max_steps": -1,
83
+ "logging_strategy": "steps",
84
+ "save_strategy": "steps",
85
+ "logging_steps": 0.2,
86
+ "save_steps": 0.2,
87
+ "gradient_accumulation_steps": 1,
88
+ "gradient_checkpointing": false,
89
+ "optim": "adamw_torch",
90
+ "learning_rate": 5e-05,
91
+ "weight_decay": 0.01,
92
+ "warmup_steps": 0,
93
+ "report_to": "tensorboard",
94
+ "save_safetensors": true,
95
+ "packing": false
96
+ },
97
+ "use_peft": true,
98
+ "peft_task_type": "CAUSAL_LM",
99
+ "peft_inference_mode": false,
100
+ "peft_r": 8,
101
+ "peft_lora_alpha": 32,
102
+ "peft_lora_dropout": 0.1,
103
+ "fsdp": "none",
104
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
105
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
106
+ "fsdp_fsdp_min_num_params": 100000000.0,
107
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
108
+ "fsdp_fsdp_forward_prefetch": false,
109
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
110
+ "fsdp_fsdp_offload_params": false,
111
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
112
+ "fsdp_fsdp_use_orig_params": true,
113
+ "fsdp_fsdp_activation_checkpointing": false,
114
+ "fsdp_fsdp_limit_all_gathers": true,
115
+ "fsdp_fsdp_sync_module_states": true,
116
+ "use_git": true,
117
+ "random_seed": 0
118
+ }
119
+ },
120
+ "winogrande": {
121
+ "acc": 0.7142857142857143,
122
+ "acc_stderr": 0.01269653187003862,
123
+ "args": {
124
+ "exp_name": "results/v1/winogrande/EleutherAI/pythia-6.9b-deduped",
125
+ "task_name": "winogrande",
126
+ "model_name": "EleutherAI/pythia-6.9b-deduped",
127
+ "dataset_param": {
128
+ "max_num_train": 50000,
129
+ "completion_only": false,
130
+ "apply_chat_template": false
131
+ },
132
+ "resume_from_checkpoint": false,
133
+ "train_param": {
134
+ "max_length": 1024,
135
+ "per_device_train_batch_size": 8,
136
+ "auto_find_batch_size": true,
137
+ "num_train_epochs": 1,
138
+ "max_steps": -1,
139
+ "logging_strategy": "steps",
140
+ "save_strategy": "steps",
141
+ "logging_steps": 0.2,
142
+ "save_steps": 0.2,
143
+ "gradient_accumulation_steps": 1,
144
+ "gradient_checkpointing": false,
145
+ "optim": "adamw_torch",
146
+ "learning_rate": 5e-05,
147
+ "weight_decay": 0.01,
148
+ "warmup_steps": 0,
149
+ "report_to": "tensorboard",
150
+ "save_safetensors": true,
151
+ "packing": false
152
+ },
153
+ "use_peft": true,
154
+ "peft_task_type": "CAUSAL_LM",
155
+ "peft_inference_mode": false,
156
+ "peft_r": 8,
157
+ "peft_lora_alpha": 32,
158
+ "peft_lora_dropout": 0.1,
159
+ "fsdp": "none",
160
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
161
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
162
+ "fsdp_fsdp_min_num_params": 100000000.0,
163
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
164
+ "fsdp_fsdp_forward_prefetch": false,
165
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
166
+ "fsdp_fsdp_offload_params": false,
167
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
168
+ "fsdp_fsdp_use_orig_params": true,
169
+ "fsdp_fsdp_activation_checkpointing": false,
170
+ "fsdp_fsdp_limit_all_gathers": true,
171
+ "fsdp_fsdp_sync_module_states": true,
172
+ "use_git": true,
173
+ "random_seed": 0
174
+ }
175
+ },
176
+ "hellaswag": {
177
+ "acc_norm": 0.6915,
178
+ "acc_norm_stderr": 0.004618972666172099,
179
+ "args": {
180
+ "exp_name": "results/v1/hellaswag/EleutherAI/pythia-6.9b-deduped",
181
+ "task_name": "hellaswag",
182
+ "model_name": "EleutherAI/pythia-6.9b-deduped",
183
+ "dataset_param": {
184
+ "max_num_train": 50000,
185
+ "completion_only": false,
186
+ "apply_chat_template": false
187
+ },
188
+ "resume_from_checkpoint": false,
189
+ "train_param": {
190
+ "max_length": 1024,
191
+ "per_device_train_batch_size": 8,
192
+ "auto_find_batch_size": true,
193
+ "num_train_epochs": 1,
194
+ "max_steps": -1,
195
+ "logging_strategy": "steps",
196
+ "save_strategy": "steps",
197
+ "logging_steps": 0.2,
198
+ "save_steps": 0.2,
199
+ "gradient_accumulation_steps": 1,
200
+ "gradient_checkpointing": false,
201
+ "optim": "adamw_torch",
202
+ "learning_rate": 5e-05,
203
+ "weight_decay": 0.01,
204
+ "warmup_steps": 0,
205
+ "report_to": "tensorboard",
206
+ "save_safetensors": true,
207
+ "packing": false
208
+ },
209
+ "use_peft": true,
210
+ "peft_task_type": "CAUSAL_LM",
211
+ "peft_inference_mode": false,
212
+ "peft_r": 8,
213
+ "peft_lora_alpha": 32,
214
+ "peft_lora_dropout": 0.1,
215
+ "fsdp": "none",
216
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
217
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
218
+ "fsdp_fsdp_min_num_params": 100000000.0,
219
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
220
+ "fsdp_fsdp_forward_prefetch": false,
221
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
222
+ "fsdp_fsdp_offload_params": false,
223
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
224
+ "fsdp_fsdp_use_orig_params": true,
225
+ "fsdp_fsdp_activation_checkpointing": false,
226
+ "fsdp_fsdp_limit_all_gathers": true,
227
+ "fsdp_fsdp_sync_module_states": true,
228
+ "use_git": true,
229
+ "random_seed": 0
230
+ }
231
+ },
232
+ "social_iqa": {
233
+ "acc": 0.5327533265097236,
234
+ "acc_stderr": 0.011289769431024872,
235
+ "args": {
236
+ "exp_name": "results/v1/social_iqa/EleutherAI/pythia-6.9b-deduped",
237
+ "task_name": "social_iqa",
238
+ "model_name": "EleutherAI/pythia-6.9b-deduped",
239
+ "dataset_param": {
240
+ "max_num_train": 50000,
241
+ "completion_only": false,
242
+ "apply_chat_template": false
243
+ },
244
+ "resume_from_checkpoint": false,
245
+ "train_param": {
246
+ "max_length": 1024,
247
+ "per_device_train_batch_size": 8,
248
+ "auto_find_batch_size": true,
249
+ "num_train_epochs": 1,
250
+ "max_steps": -1,
251
+ "logging_strategy": "steps",
252
+ "save_strategy": "steps",
253
+ "logging_steps": 0.2,
254
+ "save_steps": 0.2,
255
+ "gradient_accumulation_steps": 1,
256
+ "gradient_checkpointing": false,
257
+ "optim": "adamw_torch",
258
+ "learning_rate": 5e-05,
259
+ "weight_decay": 0.01,
260
+ "warmup_steps": 0,
261
+ "report_to": "tensorboard",
262
+ "save_safetensors": true,
263
+ "packing": false
264
+ },
265
+ "use_peft": true,
266
+ "peft_task_type": "CAUSAL_LM",
267
+ "peft_inference_mode": false,
268
+ "peft_r": 8,
269
+ "peft_lora_alpha": 32,
270
+ "peft_lora_dropout": 0.1,
271
+ "fsdp": "none",
272
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
273
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
274
+ "fsdp_fsdp_min_num_params": 100000000.0,
275
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
276
+ "fsdp_fsdp_forward_prefetch": false,
277
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
278
+ "fsdp_fsdp_offload_params": false,
279
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
280
+ "fsdp_fsdp_use_orig_params": true,
281
+ "fsdp_fsdp_activation_checkpointing": false,
282
+ "fsdp_fsdp_limit_all_gathers": true,
283
+ "fsdp_fsdp_sync_module_states": true,
284
+ "use_git": true,
285
+ "random_seed": 0
286
+ }
287
+ },
288
+ "piqa": {
289
+ "acc_norm": 0.7867247007616975,
290
+ "acc_norm_stderr": 0.00955712122586133,
291
+ "args": {
292
+ "exp_name": "results/v2/piqa/EleutherAI/pythia-6.9b-deduped",
293
+ "task_name": "piqa",
294
+ "model_name": "EleutherAI/pythia-6.9b-deduped",
295
+ "dataset_param": {
296
+ "max_num_train": 50000,
297
+ "completion_only": false,
298
+ "apply_chat_template": false
299
+ },
300
+ "resume_from_checkpoint": false,
301
+ "train_param": {
302
+ "max_length": 1024,
303
+ "per_device_train_batch_size": 8,
304
+ "auto_find_batch_size": true,
305
+ "num_train_epochs": 5,
306
+ "max_steps": -1,
307
+ "logging_strategy": "epoch",
308
+ "save_strategy": "epoch",
309
+ "logging_steps": 0.2,
310
+ "save_steps": 0.2,
311
+ "gradient_accumulation_steps": 1,
312
+ "gradient_checkpointing": false,
313
+ "optim": "adamw_torch",
314
+ "learning_rate": 5e-05,
315
+ "weight_decay": 0.01,
316
+ "warmup_steps": 0,
317
+ "report_to": "tensorboard",
318
+ "save_safetensors": true,
319
+ "packing": false
320
+ },
321
+ "use_peft": true,
322
+ "peft_task_type": "CAUSAL_LM",
323
+ "peft_inference_mode": false,
324
+ "peft_r": 8,
325
+ "peft_lora_alpha": 32,
326
+ "peft_lora_dropout": 0.1,
327
+ "fsdp": "none",
328
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
329
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
330
+ "fsdp_fsdp_min_num_params": 100000000.0,
331
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
332
+ "fsdp_fsdp_forward_prefetch": false,
333
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
334
+ "fsdp_fsdp_offload_params": false,
335
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
336
+ "fsdp_fsdp_use_orig_params": true,
337
+ "fsdp_fsdp_activation_checkpointing": false,
338
+ "fsdp_fsdp_limit_all_gathers": true,
339
+ "fsdp_fsdp_sync_module_states": true,
340
+ "use_git": true,
341
+ "random_seed": 0
342
+ }
343
+ },
344
+ "commonsense_qa": {
345
+ "acc": 0.6396396396396397,
346
+ "acc_stderr": 0.013745364972621882,
347
+ "args": {
348
+ "exp_name": "results/v2/commonsense_qa/EleutherAI/pythia-6.9b-deduped",
349
+ "task_name": "commonsense_qa",
350
+ "model_name": "EleutherAI/pythia-6.9b-deduped",
351
+ "dataset_param": {
352
+ "max_num_train": 50000,
353
+ "completion_only": false,
354
+ "apply_chat_template": false
355
+ },
356
+ "resume_from_checkpoint": false,
357
+ "train_param": {
358
+ "max_length": 1024,
359
+ "per_device_train_batch_size": 8,
360
+ "auto_find_batch_size": true,
361
+ "num_train_epochs": 5,
362
+ "max_steps": -1,
363
+ "logging_strategy": "epoch",
364
+ "save_strategy": "epoch",
365
+ "logging_steps": 0.2,
366
+ "save_steps": 0.2,
367
+ "gradient_accumulation_steps": 1,
368
+ "gradient_checkpointing": false,
369
+ "optim": "adamw_torch",
370
+ "learning_rate": 5e-05,
371
+ "weight_decay": 0.01,
372
+ "warmup_steps": 0,
373
+ "report_to": "tensorboard",
374
+ "save_safetensors": true,
375
+ "packing": false
376
+ },
377
+ "use_peft": true,
378
+ "peft_task_type": "CAUSAL_LM",
379
+ "peft_inference_mode": false,
380
+ "peft_r": 8,
381
+ "peft_lora_alpha": 32,
382
+ "peft_lora_dropout": 0.1,
383
+ "fsdp": "none",
384
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
385
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
386
+ "fsdp_fsdp_min_num_params": 100000000.0,
387
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
388
+ "fsdp_fsdp_forward_prefetch": false,
389
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
390
+ "fsdp_fsdp_offload_params": false,
391
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
392
+ "fsdp_fsdp_use_orig_params": true,
393
+ "fsdp_fsdp_activation_checkpointing": false,
394
+ "fsdp_fsdp_limit_all_gathers": true,
395
+ "fsdp_fsdp_sync_module_states": true,
396
+ "use_git": true,
397
+ "random_seed": 0
398
+ }
399
+ },
400
+ "gsm8k": {
401
+ "exact_match,flexible-extract": 0.12130401819560273,
402
+ "exact_match,flexible-extract_stderr": 0.008992888497275595,
403
+ "args": {
404
+ "exp_name": "results/v2/gsm8k/EleutherAI/pythia-6.9b-deduped",
405
+ "task_name": "gsm8k",
406
+ "model_name": "EleutherAI/pythia-6.9b-deduped",
407
+ "dataset_param": {
408
+ "max_num_train": 50000,
409
+ "completion_only": false,
410
+ "apply_chat_template": false
411
+ },
412
+ "resume_from_checkpoint": false,
413
+ "train_param": {
414
+ "max_length": 1024,
415
+ "per_device_train_batch_size": 8,
416
+ "auto_find_batch_size": true,
417
+ "num_train_epochs": 5,
418
+ "max_steps": -1,
419
+ "logging_strategy": "epoch",
420
+ "save_strategy": "epoch",
421
+ "logging_steps": 0.2,
422
+ "save_steps": 0.2,
423
+ "gradient_accumulation_steps": 1,
424
+ "gradient_checkpointing": false,
425
+ "optim": "adamw_torch",
426
+ "learning_rate": 5e-05,
427
+ "weight_decay": 0.01,
428
+ "warmup_steps": 0,
429
+ "report_to": "tensorboard",
430
+ "save_safetensors": true,
431
+ "packing": false
432
+ },
433
+ "use_peft": true,
434
+ "peft_task_type": "CAUSAL_LM",
435
+ "peft_inference_mode": false,
436
+ "peft_r": 8,
437
+ "peft_lora_alpha": 32,
438
+ "peft_lora_dropout": 0.1,
439
+ "fsdp": "none",
440
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
441
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
442
+ "fsdp_fsdp_min_num_params": 100000000.0,
443
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
444
+ "fsdp_fsdp_forward_prefetch": false,
445
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
446
+ "fsdp_fsdp_offload_params": false,
447
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
448
+ "fsdp_fsdp_use_orig_params": true,
449
+ "fsdp_fsdp_activation_checkpointing": false,
450
+ "fsdp_fsdp_limit_all_gathers": true,
451
+ "fsdp_fsdp_sync_module_states": true,
452
+ "use_git": true,
453
+ "random_seed": 0
454
+ }
455
+ },
456
+ "arc_challenge": {
457
+ "acc_norm": 0.4112627986348123,
458
+ "acc_norm_stderr": 0.014379441068522078,
459
+ "args": {
460
+ "exp_name": "results/v2/arc_challenge/EleutherAI/pythia-6.9b-deduped",
461
+ "task_name": "arc_challenge",
462
+ "model_name": "EleutherAI/pythia-6.9b-deduped",
463
+ "dataset_param": {
464
+ "max_num_train": 50000,
465
+ "completion_only": false,
466
+ "apply_chat_template": false
467
+ },
468
+ "resume_from_checkpoint": false,
469
+ "train_param": {
470
+ "max_length": 1024,
471
+ "per_device_train_batch_size": 8,
472
+ "auto_find_batch_size": true,
473
+ "num_train_epochs": 5,
474
+ "max_steps": -1,
475
+ "logging_strategy": "epoch",
476
+ "save_strategy": "epoch",
477
+ "logging_steps": 0.2,
478
+ "save_steps": 0.2,
479
+ "gradient_accumulation_steps": 1,
480
+ "gradient_checkpointing": false,
481
+ "optim": "adamw_torch",
482
+ "learning_rate": 5e-05,
483
+ "weight_decay": 0.01,
484
+ "warmup_steps": 0,
485
+ "report_to": "tensorboard",
486
+ "save_safetensors": true,
487
+ "packing": false
488
+ },
489
+ "use_peft": true,
490
+ "peft_task_type": "CAUSAL_LM",
491
+ "peft_inference_mode": false,
492
+ "peft_r": 8,
493
+ "peft_lora_alpha": 32,
494
+ "peft_lora_dropout": 0.1,
495
+ "fsdp": "none",
496
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
497
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
498
+ "fsdp_fsdp_min_num_params": 100000000.0,
499
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
500
+ "fsdp_fsdp_forward_prefetch": false,
501
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
502
+ "fsdp_fsdp_offload_params": false,
503
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
504
+ "fsdp_fsdp_use_orig_params": true,
505
+ "fsdp_fsdp_activation_checkpointing": false,
506
+ "fsdp_fsdp_limit_all_gathers": true,
507
+ "fsdp_fsdp_sync_module_states": true,
508
+ "use_git": true,
509
+ "random_seed": 0
510
+ }
511
+ }
512
+ }
513
+ }
EleutherAI/pythia-70m-deduped/results_2026-01-05-04-27-50.json ADDED
@@ -0,0 +1,513 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "model_name": "EleutherAI/pythia-70m-deduped",
4
+ "model_type": "pretrained",
5
+ "params": 0
6
+ },
7
+ "results": {
8
+ "medmcqa": {
9
+ "acc_norm": 0.31962706191728424,
10
+ "acc_norm_stderr": 0.0072111291320445595,
11
+ "args": {
12
+ "exp_name": "results/v1/medmcqa/EleutherAI/pythia-70m-deduped",
13
+ "task_name": "medmcqa",
14
+ "model_name": "EleutherAI/pythia-70m-deduped",
15
+ "dataset_param": {
16
+ "max_num_train": 50000,
17
+ "completion_only": false,
18
+ "apply_chat_template": false
19
+ },
20
+ "resume_from_checkpoint": false,
21
+ "train_param": {
22
+ "max_length": 1024,
23
+ "per_device_train_batch_size": 8,
24
+ "auto_find_batch_size": true,
25
+ "num_train_epochs": 1,
26
+ "max_steps": -1,
27
+ "logging_strategy": "steps",
28
+ "save_strategy": "steps",
29
+ "logging_steps": 0.2,
30
+ "save_steps": 0.2,
31
+ "gradient_accumulation_steps": 1,
32
+ "gradient_checkpointing": false,
33
+ "optim": "adamw_torch",
34
+ "learning_rate": 5e-05,
35
+ "weight_decay": 0.01,
36
+ "warmup_steps": 0,
37
+ "report_to": "tensorboard",
38
+ "save_safetensors": true,
39
+ "packing": false
40
+ },
41
+ "use_peft": true,
42
+ "peft_task_type": "CAUSAL_LM",
43
+ "peft_inference_mode": false,
44
+ "peft_r": 8,
45
+ "peft_lora_alpha": 32,
46
+ "peft_lora_dropout": 0.1,
47
+ "fsdp": "none",
48
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
49
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
50
+ "fsdp_fsdp_min_num_params": 100000000.0,
51
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
52
+ "fsdp_fsdp_forward_prefetch": false,
53
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
54
+ "fsdp_fsdp_offload_params": false,
55
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
56
+ "fsdp_fsdp_use_orig_params": true,
57
+ "fsdp_fsdp_activation_checkpointing": false,
58
+ "fsdp_fsdp_limit_all_gathers": true,
59
+ "fsdp_fsdp_sync_module_states": true,
60
+ "use_git": true,
61
+ "random_seed": 0
62
+ }
63
+ },
64
+ "nq_open": {
65
+ "exact_match,remove_whitespace": 0.00110803324099723,
66
+ "exact_match,remove_whitespace_stderr": 0.000553786308029499,
67
+ "args": {
68
+ "exp_name": "results/v1/nq_open/EleutherAI/pythia-70m-deduped",
69
+ "task_name": "nq_open",
70
+ "model_name": "EleutherAI/pythia-70m-deduped",
71
+ "dataset_param": {
72
+ "max_num_train": 50000,
73
+ "completion_only": false,
74
+ "apply_chat_template": false
75
+ },
76
+ "resume_from_checkpoint": false,
77
+ "train_param": {
78
+ "max_length": 1024,
79
+ "per_device_train_batch_size": 8,
80
+ "auto_find_batch_size": true,
81
+ "num_train_epochs": 1,
82
+ "max_steps": -1,
83
+ "logging_strategy": "steps",
84
+ "save_strategy": "steps",
85
+ "logging_steps": 0.2,
86
+ "save_steps": 0.2,
87
+ "gradient_accumulation_steps": 1,
88
+ "gradient_checkpointing": false,
89
+ "optim": "adamw_torch",
90
+ "learning_rate": 5e-05,
91
+ "weight_decay": 0.01,
92
+ "warmup_steps": 0,
93
+ "report_to": "tensorboard",
94
+ "save_safetensors": true,
95
+ "packing": false
96
+ },
97
+ "use_peft": true,
98
+ "peft_task_type": "CAUSAL_LM",
99
+ "peft_inference_mode": false,
100
+ "peft_r": 8,
101
+ "peft_lora_alpha": 32,
102
+ "peft_lora_dropout": 0.1,
103
+ "fsdp": "none",
104
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
105
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
106
+ "fsdp_fsdp_min_num_params": 100000000.0,
107
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
108
+ "fsdp_fsdp_forward_prefetch": false,
109
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
110
+ "fsdp_fsdp_offload_params": false,
111
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
112
+ "fsdp_fsdp_use_orig_params": true,
113
+ "fsdp_fsdp_activation_checkpointing": false,
114
+ "fsdp_fsdp_limit_all_gathers": true,
115
+ "fsdp_fsdp_sync_module_states": true,
116
+ "use_git": true,
117
+ "random_seed": 0
118
+ }
119
+ },
120
+ "winogrande": {
121
+ "acc": 0.5019731649565904,
122
+ "acc_stderr": 0.014052376259225636,
123
+ "args": {
124
+ "exp_name": "results/v1/winogrande/EleutherAI/pythia-70m-deduped",
125
+ "task_name": "winogrande",
126
+ "model_name": "EleutherAI/pythia-70m-deduped",
127
+ "dataset_param": {
128
+ "max_num_train": 50000,
129
+ "completion_only": false,
130
+ "apply_chat_template": false
131
+ },
132
+ "resume_from_checkpoint": false,
133
+ "train_param": {
134
+ "max_length": 1024,
135
+ "per_device_train_batch_size": 8,
136
+ "auto_find_batch_size": true,
137
+ "num_train_epochs": 1,
138
+ "max_steps": -1,
139
+ "logging_strategy": "steps",
140
+ "save_strategy": "steps",
141
+ "logging_steps": 0.2,
142
+ "save_steps": 0.2,
143
+ "gradient_accumulation_steps": 1,
144
+ "gradient_checkpointing": false,
145
+ "optim": "adamw_torch",
146
+ "learning_rate": 5e-05,
147
+ "weight_decay": 0.01,
148
+ "warmup_steps": 0,
149
+ "report_to": "tensorboard",
150
+ "save_safetensors": true,
151
+ "packing": false
152
+ },
153
+ "use_peft": true,
154
+ "peft_task_type": "CAUSAL_LM",
155
+ "peft_inference_mode": false,
156
+ "peft_r": 8,
157
+ "peft_lora_alpha": 32,
158
+ "peft_lora_dropout": 0.1,
159
+ "fsdp": "none",
160
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
161
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
162
+ "fsdp_fsdp_min_num_params": 100000000.0,
163
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
164
+ "fsdp_fsdp_forward_prefetch": false,
165
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
166
+ "fsdp_fsdp_offload_params": false,
167
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
168
+ "fsdp_fsdp_use_orig_params": true,
169
+ "fsdp_fsdp_activation_checkpointing": false,
170
+ "fsdp_fsdp_limit_all_gathers": true,
171
+ "fsdp_fsdp_sync_module_states": true,
172
+ "use_git": true,
173
+ "random_seed": 0
174
+ }
175
+ },
176
+ "hellaswag": {
177
+ "acc_norm": 0.2679,
178
+ "acc_norm_stderr": 0.004428873501472784,
179
+ "args": {
180
+ "exp_name": "results/v1/hellaswag/EleutherAI/pythia-70m-deduped",
181
+ "task_name": "hellaswag",
182
+ "model_name": "EleutherAI/pythia-70m-deduped",
183
+ "dataset_param": {
184
+ "max_num_train": 50000,
185
+ "completion_only": false,
186
+ "apply_chat_template": false
187
+ },
188
+ "resume_from_checkpoint": false,
189
+ "train_param": {
190
+ "max_length": 1024,
191
+ "per_device_train_batch_size": 8,
192
+ "auto_find_batch_size": true,
193
+ "num_train_epochs": 1,
194
+ "max_steps": -1,
195
+ "logging_strategy": "steps",
196
+ "save_strategy": "steps",
197
+ "logging_steps": 0.2,
198
+ "save_steps": 0.2,
199
+ "gradient_accumulation_steps": 1,
200
+ "gradient_checkpointing": false,
201
+ "optim": "adamw_torch",
202
+ "learning_rate": 5e-05,
203
+ "weight_decay": 0.01,
204
+ "warmup_steps": 0,
205
+ "report_to": "tensorboard",
206
+ "save_safetensors": true,
207
+ "packing": false
208
+ },
209
+ "use_peft": true,
210
+ "peft_task_type": "CAUSAL_LM",
211
+ "peft_inference_mode": false,
212
+ "peft_r": 8,
213
+ "peft_lora_alpha": 32,
214
+ "peft_lora_dropout": 0.1,
215
+ "fsdp": "none",
216
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
217
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
218
+ "fsdp_fsdp_min_num_params": 100000000.0,
219
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
220
+ "fsdp_fsdp_forward_prefetch": false,
221
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
222
+ "fsdp_fsdp_offload_params": false,
223
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
224
+ "fsdp_fsdp_use_orig_params": true,
225
+ "fsdp_fsdp_activation_checkpointing": false,
226
+ "fsdp_fsdp_limit_all_gathers": true,
227
+ "fsdp_fsdp_sync_module_states": true,
228
+ "use_git": true,
229
+ "random_seed": 0
230
+ }
231
+ },
232
+ "social_iqa": {
233
+ "acc": 0.35056294779938585,
234
+ "acc_stderr": 0.010796933233333233,
235
+ "args": {
236
+ "exp_name": "results/v1/social_iqa/EleutherAI/pythia-70m-deduped",
237
+ "task_name": "social_iqa",
238
+ "model_name": "EleutherAI/pythia-70m-deduped",
239
+ "dataset_param": {
240
+ "max_num_train": 50000,
241
+ "completion_only": false,
242
+ "apply_chat_template": false
243
+ },
244
+ "resume_from_checkpoint": false,
245
+ "train_param": {
246
+ "max_length": 1024,
247
+ "per_device_train_batch_size": 8,
248
+ "auto_find_batch_size": true,
249
+ "num_train_epochs": 1,
250
+ "max_steps": -1,
251
+ "logging_strategy": "steps",
252
+ "save_strategy": "steps",
253
+ "logging_steps": 0.2,
254
+ "save_steps": 0.2,
255
+ "gradient_accumulation_steps": 1,
256
+ "gradient_checkpointing": false,
257
+ "optim": "adamw_torch",
258
+ "learning_rate": 5e-05,
259
+ "weight_decay": 0.01,
260
+ "warmup_steps": 0,
261
+ "report_to": "tensorboard",
262
+ "save_safetensors": true,
263
+ "packing": false
264
+ },
265
+ "use_peft": true,
266
+ "peft_task_type": "CAUSAL_LM",
267
+ "peft_inference_mode": false,
268
+ "peft_r": 8,
269
+ "peft_lora_alpha": 32,
270
+ "peft_lora_dropout": 0.1,
271
+ "fsdp": "none",
272
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
273
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
274
+ "fsdp_fsdp_min_num_params": 100000000.0,
275
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
276
+ "fsdp_fsdp_forward_prefetch": false,
277
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
278
+ "fsdp_fsdp_offload_params": false,
279
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
280
+ "fsdp_fsdp_use_orig_params": true,
281
+ "fsdp_fsdp_activation_checkpointing": false,
282
+ "fsdp_fsdp_limit_all_gathers": true,
283
+ "fsdp_fsdp_sync_module_states": true,
284
+ "use_git": true,
285
+ "random_seed": 0
286
+ }
287
+ },
288
+ "piqa": {
289
+ "acc_norm": 0.5674646354733406,
290
+ "acc_norm_stderr": 0.011559142916063149,
291
+ "args": {
292
+ "exp_name": "results/v2/piqa/EleutherAI/pythia-70m-deduped",
293
+ "task_name": "piqa",
294
+ "model_name": "EleutherAI/pythia-70m-deduped",
295
+ "dataset_param": {
296
+ "max_num_train": 50000,
297
+ "completion_only": false,
298
+ "apply_chat_template": false
299
+ },
300
+ "resume_from_checkpoint": false,
301
+ "train_param": {
302
+ "max_length": 1024,
303
+ "per_device_train_batch_size": 8,
304
+ "auto_find_batch_size": true,
305
+ "num_train_epochs": 5,
306
+ "max_steps": -1,
307
+ "logging_strategy": "epoch",
308
+ "save_strategy": "epoch",
309
+ "logging_steps": 0.2,
310
+ "save_steps": 0.2,
311
+ "gradient_accumulation_steps": 1,
312
+ "gradient_checkpointing": false,
313
+ "optim": "adamw_torch",
314
+ "learning_rate": 5e-05,
315
+ "weight_decay": 0.01,
316
+ "warmup_steps": 0,
317
+ "report_to": "tensorboard",
318
+ "save_safetensors": true,
319
+ "packing": false
320
+ },
321
+ "use_peft": true,
322
+ "peft_task_type": "CAUSAL_LM",
323
+ "peft_inference_mode": false,
324
+ "peft_r": 8,
325
+ "peft_lora_alpha": 32,
326
+ "peft_lora_dropout": 0.1,
327
+ "fsdp": "none",
328
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
329
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
330
+ "fsdp_fsdp_min_num_params": 100000000.0,
331
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
332
+ "fsdp_fsdp_forward_prefetch": false,
333
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
334
+ "fsdp_fsdp_offload_params": false,
335
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
336
+ "fsdp_fsdp_use_orig_params": true,
337
+ "fsdp_fsdp_activation_checkpointing": false,
338
+ "fsdp_fsdp_limit_all_gathers": true,
339
+ "fsdp_fsdp_sync_module_states": true,
340
+ "use_git": true,
341
+ "random_seed": 0
342
+ }
343
+ },
344
+ "commonsense_qa": {
345
+ "acc": 0.20147420147420148,
346
+ "acc_stderr": 0.011483500195202903,
347
+ "args": {
348
+ "exp_name": "results/v2/commonsense_qa/EleutherAI/pythia-70m-deduped",
349
+ "task_name": "commonsense_qa",
350
+ "model_name": "EleutherAI/pythia-70m-deduped",
351
+ "dataset_param": {
352
+ "max_num_train": 50000,
353
+ "completion_only": false,
354
+ "apply_chat_template": false
355
+ },
356
+ "resume_from_checkpoint": false,
357
+ "train_param": {
358
+ "max_length": 1024,
359
+ "per_device_train_batch_size": 8,
360
+ "auto_find_batch_size": true,
361
+ "num_train_epochs": 5,
362
+ "max_steps": -1,
363
+ "logging_strategy": "epoch",
364
+ "save_strategy": "epoch",
365
+ "logging_steps": 0.2,
366
+ "save_steps": 0.2,
367
+ "gradient_accumulation_steps": 1,
368
+ "gradient_checkpointing": false,
369
+ "optim": "adamw_torch",
370
+ "learning_rate": 5e-05,
371
+ "weight_decay": 0.01,
372
+ "warmup_steps": 0,
373
+ "report_to": "tensorboard",
374
+ "save_safetensors": true,
375
+ "packing": false
376
+ },
377
+ "use_peft": true,
378
+ "peft_task_type": "CAUSAL_LM",
379
+ "peft_inference_mode": false,
380
+ "peft_r": 8,
381
+ "peft_lora_alpha": 32,
382
+ "peft_lora_dropout": 0.1,
383
+ "fsdp": "none",
384
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
385
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
386
+ "fsdp_fsdp_min_num_params": 100000000.0,
387
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
388
+ "fsdp_fsdp_forward_prefetch": false,
389
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
390
+ "fsdp_fsdp_offload_params": false,
391
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
392
+ "fsdp_fsdp_use_orig_params": true,
393
+ "fsdp_fsdp_activation_checkpointing": false,
394
+ "fsdp_fsdp_limit_all_gathers": true,
395
+ "fsdp_fsdp_sync_module_states": true,
396
+ "use_git": true,
397
+ "random_seed": 0
398
+ }
399
+ },
400
+ "gsm8k": {
401
+ "exact_match,flexible-extract": 0.01592115238817286,
402
+ "exact_match,flexible-extract_stderr": 0.0034478192723890106,
403
+ "args": {
404
+ "exp_name": "results/v2/gsm8k/EleutherAI/pythia-70m-deduped",
405
+ "task_name": "gsm8k",
406
+ "model_name": "EleutherAI/pythia-70m-deduped",
407
+ "dataset_param": {
408
+ "max_num_train": 50000,
409
+ "completion_only": false,
410
+ "apply_chat_template": false
411
+ },
412
+ "resume_from_checkpoint": false,
413
+ "train_param": {
414
+ "max_length": 1024,
415
+ "per_device_train_batch_size": 8,
416
+ "auto_find_batch_size": true,
417
+ "num_train_epochs": 5,
418
+ "max_steps": -1,
419
+ "logging_strategy": "epoch",
420
+ "save_strategy": "epoch",
421
+ "logging_steps": 0.2,
422
+ "save_steps": 0.2,
423
+ "gradient_accumulation_steps": 1,
424
+ "gradient_checkpointing": false,
425
+ "optim": "adamw_torch",
426
+ "learning_rate": 5e-05,
427
+ "weight_decay": 0.01,
428
+ "warmup_steps": 0,
429
+ "report_to": "tensorboard",
430
+ "save_safetensors": true,
431
+ "packing": false
432
+ },
433
+ "use_peft": true,
434
+ "peft_task_type": "CAUSAL_LM",
435
+ "peft_inference_mode": false,
436
+ "peft_r": 8,
437
+ "peft_lora_alpha": 32,
438
+ "peft_lora_dropout": 0.1,
439
+ "fsdp": "none",
440
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
441
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
442
+ "fsdp_fsdp_min_num_params": 100000000.0,
443
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
444
+ "fsdp_fsdp_forward_prefetch": false,
445
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
446
+ "fsdp_fsdp_offload_params": false,
447
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
448
+ "fsdp_fsdp_use_orig_params": true,
449
+ "fsdp_fsdp_activation_checkpointing": false,
450
+ "fsdp_fsdp_limit_all_gathers": true,
451
+ "fsdp_fsdp_sync_module_states": true,
452
+ "use_git": true,
453
+ "random_seed": 0
454
+ }
455
+ },
456
+ "arc_challenge": {
457
+ "acc_norm": 0.23122866894197952,
458
+ "acc_norm_stderr": 0.012320858834772297,
459
+ "args": {
460
+ "exp_name": "results/v2/arc_challenge/EleutherAI/pythia-70m-deduped",
461
+ "task_name": "arc_challenge",
462
+ "model_name": "EleutherAI/pythia-70m-deduped",
463
+ "dataset_param": {
464
+ "max_num_train": 50000,
465
+ "completion_only": false,
466
+ "apply_chat_template": false
467
+ },
468
+ "resume_from_checkpoint": false,
469
+ "train_param": {
470
+ "max_length": 1024,
471
+ "per_device_train_batch_size": 8,
472
+ "auto_find_batch_size": true,
473
+ "num_train_epochs": 5,
474
+ "max_steps": -1,
475
+ "logging_strategy": "epoch",
476
+ "save_strategy": "epoch",
477
+ "logging_steps": 0.2,
478
+ "save_steps": 0.2,
479
+ "gradient_accumulation_steps": 1,
480
+ "gradient_checkpointing": false,
481
+ "optim": "adamw_torch",
482
+ "learning_rate": 5e-05,
483
+ "weight_decay": 0.01,
484
+ "warmup_steps": 0,
485
+ "report_to": "tensorboard",
486
+ "save_safetensors": true,
487
+ "packing": false
488
+ },
489
+ "use_peft": true,
490
+ "peft_task_type": "CAUSAL_LM",
491
+ "peft_inference_mode": false,
492
+ "peft_r": 8,
493
+ "peft_lora_alpha": 32,
494
+ "peft_lora_dropout": 0.1,
495
+ "fsdp": "none",
496
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
497
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
498
+ "fsdp_fsdp_min_num_params": 100000000.0,
499
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
500
+ "fsdp_fsdp_forward_prefetch": false,
501
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
502
+ "fsdp_fsdp_offload_params": false,
503
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
504
+ "fsdp_fsdp_use_orig_params": true,
505
+ "fsdp_fsdp_activation_checkpointing": false,
506
+ "fsdp_fsdp_limit_all_gathers": true,
507
+ "fsdp_fsdp_sync_module_states": true,
508
+ "use_git": true,
509
+ "random_seed": 0
510
+ }
511
+ }
512
+ }
513
+ }
HuggingFaceTB/SmolLM3-3B-Base/results_2026-01-05-04-27-50.json ADDED
@@ -0,0 +1,513 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "model_name": "HuggingFaceTB/SmolLM3-3B-Base",
4
+ "model_type": "pretrained",
5
+ "params": 3.0
6
+ },
7
+ "results": {
8
+ "medmcqa": {
9
+ "acc_norm": 0.4884054506335166,
10
+ "acc_norm_stderr": 0.007729674236551038,
11
+ "args": {
12
+ "exp_name": "results/v1/medmcqa/HuggingFaceTB/SmolLM3-3B-Base",
13
+ "task_name": "medmcqa",
14
+ "model_name": "HuggingFaceTB/SmolLM3-3B-Base",
15
+ "dataset_param": {
16
+ "max_num_train": 50000,
17
+ "completion_only": false,
18
+ "apply_chat_template": false
19
+ },
20
+ "resume_from_checkpoint": false,
21
+ "train_param": {
22
+ "max_length": 1024,
23
+ "per_device_train_batch_size": 8,
24
+ "auto_find_batch_size": true,
25
+ "num_train_epochs": 1,
26
+ "max_steps": -1,
27
+ "logging_strategy": "steps",
28
+ "save_strategy": "steps",
29
+ "logging_steps": 0.2,
30
+ "save_steps": 0.2,
31
+ "gradient_accumulation_steps": 1,
32
+ "gradient_checkpointing": false,
33
+ "optim": "adamw_torch",
34
+ "learning_rate": 5e-05,
35
+ "weight_decay": 0.01,
36
+ "warmup_steps": 0,
37
+ "report_to": "tensorboard",
38
+ "save_safetensors": true,
39
+ "packing": false
40
+ },
41
+ "use_peft": true,
42
+ "peft_task_type": "CAUSAL_LM",
43
+ "peft_inference_mode": false,
44
+ "peft_r": 8,
45
+ "peft_lora_alpha": 32,
46
+ "peft_lora_dropout": 0.1,
47
+ "fsdp": "none",
48
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
49
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
50
+ "fsdp_fsdp_min_num_params": 100000000.0,
51
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
52
+ "fsdp_fsdp_forward_prefetch": false,
53
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
54
+ "fsdp_fsdp_offload_params": false,
55
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
56
+ "fsdp_fsdp_use_orig_params": true,
57
+ "fsdp_fsdp_activation_checkpointing": false,
58
+ "fsdp_fsdp_limit_all_gathers": true,
59
+ "fsdp_fsdp_sync_module_states": true,
60
+ "use_git": true,
61
+ "random_seed": 0
62
+ }
63
+ },
64
+ "nq_open": {
65
+ "exact_match,remove_whitespace": 0.24598337950138505,
66
+ "exact_match,remove_whitespace_stderr": 0.007168856718678337,
67
+ "args": {
68
+ "exp_name": "results/v1/nq_open/HuggingFaceTB/SmolLM3-3B-Base",
69
+ "task_name": "nq_open",
70
+ "model_name": "HuggingFaceTB/SmolLM3-3B-Base",
71
+ "dataset_param": {
72
+ "max_num_train": 50000,
73
+ "completion_only": false,
74
+ "apply_chat_template": false
75
+ },
76
+ "resume_from_checkpoint": false,
77
+ "train_param": {
78
+ "max_length": 1024,
79
+ "per_device_train_batch_size": 8,
80
+ "auto_find_batch_size": true,
81
+ "num_train_epochs": 1,
82
+ "max_steps": -1,
83
+ "logging_strategy": "steps",
84
+ "save_strategy": "steps",
85
+ "logging_steps": 0.2,
86
+ "save_steps": 0.2,
87
+ "gradient_accumulation_steps": 1,
88
+ "gradient_checkpointing": false,
89
+ "optim": "adamw_torch",
90
+ "learning_rate": 5e-05,
91
+ "weight_decay": 0.01,
92
+ "warmup_steps": 0,
93
+ "report_to": "tensorboard",
94
+ "save_safetensors": true,
95
+ "packing": false
96
+ },
97
+ "use_peft": true,
98
+ "peft_task_type": "CAUSAL_LM",
99
+ "peft_inference_mode": false,
100
+ "peft_r": 8,
101
+ "peft_lora_alpha": 32,
102
+ "peft_lora_dropout": 0.1,
103
+ "fsdp": "none",
104
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
105
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
106
+ "fsdp_fsdp_min_num_params": 100000000.0,
107
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
108
+ "fsdp_fsdp_forward_prefetch": false,
109
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
110
+ "fsdp_fsdp_offload_params": false,
111
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
112
+ "fsdp_fsdp_use_orig_params": true,
113
+ "fsdp_fsdp_activation_checkpointing": false,
114
+ "fsdp_fsdp_limit_all_gathers": true,
115
+ "fsdp_fsdp_sync_module_states": true,
116
+ "use_git": true,
117
+ "random_seed": 0
118
+ }
119
+ },
120
+ "winogrande": {
121
+ "acc": 0.7947908445146015,
122
+ "acc_stderr": 0.011350315707462052,
123
+ "args": {
124
+ "exp_name": "results/v1/winogrande/HuggingFaceTB/SmolLM3-3B-Base",
125
+ "task_name": "winogrande",
126
+ "model_name": "HuggingFaceTB/SmolLM3-3B-Base",
127
+ "dataset_param": {
128
+ "max_num_train": 50000,
129
+ "completion_only": false,
130
+ "apply_chat_template": false
131
+ },
132
+ "resume_from_checkpoint": false,
133
+ "train_param": {
134
+ "max_length": 1024,
135
+ "per_device_train_batch_size": 8,
136
+ "auto_find_batch_size": true,
137
+ "num_train_epochs": 1,
138
+ "max_steps": -1,
139
+ "logging_strategy": "steps",
140
+ "save_strategy": "steps",
141
+ "logging_steps": 0.2,
142
+ "save_steps": 0.2,
143
+ "gradient_accumulation_steps": 1,
144
+ "gradient_checkpointing": false,
145
+ "optim": "adamw_torch",
146
+ "learning_rate": 5e-05,
147
+ "weight_decay": 0.01,
148
+ "warmup_steps": 0,
149
+ "report_to": "tensorboard",
150
+ "save_safetensors": true,
151
+ "packing": false
152
+ },
153
+ "use_peft": true,
154
+ "peft_task_type": "CAUSAL_LM",
155
+ "peft_inference_mode": false,
156
+ "peft_r": 8,
157
+ "peft_lora_alpha": 32,
158
+ "peft_lora_dropout": 0.1,
159
+ "fsdp": "none",
160
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
161
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
162
+ "fsdp_fsdp_min_num_params": 100000000.0,
163
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
164
+ "fsdp_fsdp_forward_prefetch": false,
165
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
166
+ "fsdp_fsdp_offload_params": false,
167
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
168
+ "fsdp_fsdp_use_orig_params": true,
169
+ "fsdp_fsdp_activation_checkpointing": false,
170
+ "fsdp_fsdp_limit_all_gathers": true,
171
+ "fsdp_fsdp_sync_module_states": true,
172
+ "use_git": true,
173
+ "random_seed": 0
174
+ }
175
+ },
176
+ "hellaswag": {
177
+ "acc_norm": 0.7937,
178
+ "acc_norm_stderr": 0.004046686121612992,
179
+ "args": {
180
+ "exp_name": "results/v1/hellaswag/HuggingFaceTB/SmolLM3-3B-Base",
181
+ "task_name": "hellaswag",
182
+ "model_name": "HuggingFaceTB/SmolLM3-3B-Base",
183
+ "dataset_param": {
184
+ "max_num_train": 50000,
185
+ "completion_only": false,
186
+ "apply_chat_template": false
187
+ },
188
+ "resume_from_checkpoint": false,
189
+ "train_param": {
190
+ "max_length": 1024,
191
+ "per_device_train_batch_size": 8,
192
+ "auto_find_batch_size": true,
193
+ "num_train_epochs": 1,
194
+ "max_steps": -1,
195
+ "logging_strategy": "steps",
196
+ "save_strategy": "steps",
197
+ "logging_steps": 0.2,
198
+ "save_steps": 0.2,
199
+ "gradient_accumulation_steps": 1,
200
+ "gradient_checkpointing": false,
201
+ "optim": "adamw_torch",
202
+ "learning_rate": 5e-05,
203
+ "weight_decay": 0.01,
204
+ "warmup_steps": 0,
205
+ "report_to": "tensorboard",
206
+ "save_safetensors": true,
207
+ "packing": false
208
+ },
209
+ "use_peft": true,
210
+ "peft_task_type": "CAUSAL_LM",
211
+ "peft_inference_mode": false,
212
+ "peft_r": 8,
213
+ "peft_lora_alpha": 32,
214
+ "peft_lora_dropout": 0.1,
215
+ "fsdp": "none",
216
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
217
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
218
+ "fsdp_fsdp_min_num_params": 100000000.0,
219
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
220
+ "fsdp_fsdp_forward_prefetch": false,
221
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
222
+ "fsdp_fsdp_offload_params": false,
223
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
224
+ "fsdp_fsdp_use_orig_params": true,
225
+ "fsdp_fsdp_activation_checkpointing": false,
226
+ "fsdp_fsdp_limit_all_gathers": true,
227
+ "fsdp_fsdp_sync_module_states": true,
228
+ "use_git": true,
229
+ "random_seed": 0
230
+ }
231
+ },
232
+ "social_iqa": {
233
+ "acc": 0.5706243602865916,
234
+ "acc_stderr": 0.011200637273721807,
235
+ "args": {
236
+ "exp_name": "results/v1/social_iqa/HuggingFaceTB/SmolLM3-3B-Base",
237
+ "task_name": "social_iqa",
238
+ "model_name": "HuggingFaceTB/SmolLM3-3B-Base",
239
+ "dataset_param": {
240
+ "max_num_train": 50000,
241
+ "completion_only": false,
242
+ "apply_chat_template": false
243
+ },
244
+ "resume_from_checkpoint": false,
245
+ "train_param": {
246
+ "max_length": 1024,
247
+ "per_device_train_batch_size": 8,
248
+ "auto_find_batch_size": true,
249
+ "num_train_epochs": 1,
250
+ "max_steps": -1,
251
+ "logging_strategy": "steps",
252
+ "save_strategy": "steps",
253
+ "logging_steps": 0.2,
254
+ "save_steps": 0.2,
255
+ "gradient_accumulation_steps": 1,
256
+ "gradient_checkpointing": false,
257
+ "optim": "adamw_torch",
258
+ "learning_rate": 5e-05,
259
+ "weight_decay": 0.01,
260
+ "warmup_steps": 0,
261
+ "report_to": "tensorboard",
262
+ "save_safetensors": true,
263
+ "packing": false
264
+ },
265
+ "use_peft": true,
266
+ "peft_task_type": "CAUSAL_LM",
267
+ "peft_inference_mode": false,
268
+ "peft_r": 8,
269
+ "peft_lora_alpha": 32,
270
+ "peft_lora_dropout": 0.1,
271
+ "fsdp": "none",
272
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
273
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
274
+ "fsdp_fsdp_min_num_params": 100000000.0,
275
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
276
+ "fsdp_fsdp_forward_prefetch": false,
277
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
278
+ "fsdp_fsdp_offload_params": false,
279
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
280
+ "fsdp_fsdp_use_orig_params": true,
281
+ "fsdp_fsdp_activation_checkpointing": false,
282
+ "fsdp_fsdp_limit_all_gathers": true,
283
+ "fsdp_fsdp_sync_module_states": true,
284
+ "use_git": true,
285
+ "random_seed": 0
286
+ }
287
+ },
288
+ "piqa": {
289
+ "acc_norm": 0.8166485310119695,
290
+ "acc_norm_stderr": 0.009028283984689424,
291
+ "args": {
292
+ "exp_name": "results/v2/piqa/HuggingFaceTB/SmolLM3-3B-Base",
293
+ "task_name": "piqa",
294
+ "model_name": "HuggingFaceTB/SmolLM3-3B-Base",
295
+ "dataset_param": {
296
+ "max_num_train": 50000,
297
+ "completion_only": false,
298
+ "apply_chat_template": false
299
+ },
300
+ "resume_from_checkpoint": false,
301
+ "train_param": {
302
+ "max_length": 1024,
303
+ "per_device_train_batch_size": 8,
304
+ "auto_find_batch_size": true,
305
+ "num_train_epochs": 5,
306
+ "max_steps": -1,
307
+ "logging_strategy": "epoch",
308
+ "save_strategy": "epoch",
309
+ "logging_steps": 0.2,
310
+ "save_steps": 0.2,
311
+ "gradient_accumulation_steps": 1,
312
+ "gradient_checkpointing": false,
313
+ "optim": "adamw_torch",
314
+ "learning_rate": 5e-05,
315
+ "weight_decay": 0.01,
316
+ "warmup_steps": 0,
317
+ "report_to": "tensorboard",
318
+ "save_safetensors": true,
319
+ "packing": false
320
+ },
321
+ "use_peft": true,
322
+ "peft_task_type": "CAUSAL_LM",
323
+ "peft_inference_mode": false,
324
+ "peft_r": 8,
325
+ "peft_lora_alpha": 32,
326
+ "peft_lora_dropout": 0.1,
327
+ "fsdp": "none",
328
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
329
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
330
+ "fsdp_fsdp_min_num_params": 100000000.0,
331
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
332
+ "fsdp_fsdp_forward_prefetch": false,
333
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
334
+ "fsdp_fsdp_offload_params": false,
335
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
336
+ "fsdp_fsdp_use_orig_params": true,
337
+ "fsdp_fsdp_activation_checkpointing": false,
338
+ "fsdp_fsdp_limit_all_gathers": true,
339
+ "fsdp_fsdp_sync_module_states": true,
340
+ "use_git": true,
341
+ "random_seed": 0
342
+ }
343
+ },
344
+ "commonsense_qa": {
345
+ "acc": 0.7927927927927928,
346
+ "acc_stderr": 0.011603856781422558,
347
+ "args": {
348
+ "exp_name": "results/v2/commonsense_qa/HuggingFaceTB/SmolLM3-3B-Base",
349
+ "task_name": "commonsense_qa",
350
+ "model_name": "HuggingFaceTB/SmolLM3-3B-Base",
351
+ "dataset_param": {
352
+ "max_num_train": 50000,
353
+ "completion_only": false,
354
+ "apply_chat_template": false
355
+ },
356
+ "resume_from_checkpoint": false,
357
+ "train_param": {
358
+ "max_length": 1024,
359
+ "per_device_train_batch_size": 8,
360
+ "auto_find_batch_size": true,
361
+ "num_train_epochs": 5,
362
+ "max_steps": -1,
363
+ "logging_strategy": "epoch",
364
+ "save_strategy": "epoch",
365
+ "logging_steps": 0.2,
366
+ "save_steps": 0.2,
367
+ "gradient_accumulation_steps": 1,
368
+ "gradient_checkpointing": false,
369
+ "optim": "adamw_torch",
370
+ "learning_rate": 5e-05,
371
+ "weight_decay": 0.01,
372
+ "warmup_steps": 0,
373
+ "report_to": "tensorboard",
374
+ "save_safetensors": true,
375
+ "packing": false
376
+ },
377
+ "use_peft": true,
378
+ "peft_task_type": "CAUSAL_LM",
379
+ "peft_inference_mode": false,
380
+ "peft_r": 8,
381
+ "peft_lora_alpha": 32,
382
+ "peft_lora_dropout": 0.1,
383
+ "fsdp": "none",
384
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
385
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
386
+ "fsdp_fsdp_min_num_params": 100000000.0,
387
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
388
+ "fsdp_fsdp_forward_prefetch": false,
389
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
390
+ "fsdp_fsdp_offload_params": false,
391
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
392
+ "fsdp_fsdp_use_orig_params": true,
393
+ "fsdp_fsdp_activation_checkpointing": false,
394
+ "fsdp_fsdp_limit_all_gathers": true,
395
+ "fsdp_fsdp_sync_module_states": true,
396
+ "use_git": true,
397
+ "random_seed": 0
398
+ }
399
+ },
400
+ "gsm8k": {
401
+ "exact_match,flexible-extract": 0.6777862016679302,
402
+ "exact_match,flexible-extract_stderr": 0.012872435481188778,
403
+ "args": {
404
+ "exp_name": "results/v2/gsm8k/HuggingFaceTB/SmolLM3-3B-Base",
405
+ "task_name": "gsm8k",
406
+ "model_name": "HuggingFaceTB/SmolLM3-3B-Base",
407
+ "dataset_param": {
408
+ "max_num_train": 50000,
409
+ "completion_only": false,
410
+ "apply_chat_template": false
411
+ },
412
+ "resume_from_checkpoint": false,
413
+ "train_param": {
414
+ "max_length": 1024,
415
+ "per_device_train_batch_size": 8,
416
+ "auto_find_batch_size": true,
417
+ "num_train_epochs": 5,
418
+ "max_steps": -1,
419
+ "logging_strategy": "epoch",
420
+ "save_strategy": "epoch",
421
+ "logging_steps": 0.2,
422
+ "save_steps": 0.2,
423
+ "gradient_accumulation_steps": 1,
424
+ "gradient_checkpointing": false,
425
+ "optim": "adamw_torch",
426
+ "learning_rate": 5e-05,
427
+ "weight_decay": 0.01,
428
+ "warmup_steps": 0,
429
+ "report_to": "tensorboard",
430
+ "save_safetensors": true,
431
+ "packing": false
432
+ },
433
+ "use_peft": true,
434
+ "peft_task_type": "CAUSAL_LM",
435
+ "peft_inference_mode": false,
436
+ "peft_r": 8,
437
+ "peft_lora_alpha": 32,
438
+ "peft_lora_dropout": 0.1,
439
+ "fsdp": "none",
440
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
441
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
442
+ "fsdp_fsdp_min_num_params": 100000000.0,
443
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
444
+ "fsdp_fsdp_forward_prefetch": false,
445
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
446
+ "fsdp_fsdp_offload_params": false,
447
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
448
+ "fsdp_fsdp_use_orig_params": true,
449
+ "fsdp_fsdp_activation_checkpointing": false,
450
+ "fsdp_fsdp_limit_all_gathers": true,
451
+ "fsdp_fsdp_sync_module_states": true,
452
+ "use_git": true,
453
+ "random_seed": 0
454
+ }
455
+ },
456
+ "arc_challenge": {
457
+ "acc_norm": 0.6040955631399317,
458
+ "acc_norm_stderr": 0.01429122839353659,
459
+ "args": {
460
+ "exp_name": "results/v2/arc_challenge/HuggingFaceTB/SmolLM3-3B-Base",
461
+ "task_name": "arc_challenge",
462
+ "model_name": "HuggingFaceTB/SmolLM3-3B-Base",
463
+ "dataset_param": {
464
+ "max_num_train": 50000,
465
+ "completion_only": false,
466
+ "apply_chat_template": false
467
+ },
468
+ "resume_from_checkpoint": false,
469
+ "train_param": {
470
+ "max_length": 1024,
471
+ "per_device_train_batch_size": 8,
472
+ "auto_find_batch_size": true,
473
+ "num_train_epochs": 5,
474
+ "max_steps": -1,
475
+ "logging_strategy": "epoch",
476
+ "save_strategy": "epoch",
477
+ "logging_steps": 0.2,
478
+ "save_steps": 0.2,
479
+ "gradient_accumulation_steps": 1,
480
+ "gradient_checkpointing": false,
481
+ "optim": "adamw_torch",
482
+ "learning_rate": 5e-05,
483
+ "weight_decay": 0.01,
484
+ "warmup_steps": 0,
485
+ "report_to": "tensorboard",
486
+ "save_safetensors": true,
487
+ "packing": false
488
+ },
489
+ "use_peft": true,
490
+ "peft_task_type": "CAUSAL_LM",
491
+ "peft_inference_mode": false,
492
+ "peft_r": 8,
493
+ "peft_lora_alpha": 32,
494
+ "peft_lora_dropout": 0.1,
495
+ "fsdp": "none",
496
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
497
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
498
+ "fsdp_fsdp_min_num_params": 100000000.0,
499
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
500
+ "fsdp_fsdp_forward_prefetch": false,
501
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
502
+ "fsdp_fsdp_offload_params": false,
503
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
504
+ "fsdp_fsdp_use_orig_params": true,
505
+ "fsdp_fsdp_activation_checkpointing": false,
506
+ "fsdp_fsdp_limit_all_gathers": true,
507
+ "fsdp_fsdp_sync_module_states": true,
508
+ "use_git": true,
509
+ "random_seed": 0
510
+ }
511
+ }
512
+ }
513
+ }
HuggingFaceTB/SmolLM3-3B/results_2026-01-05-04-27-51.json ADDED
@@ -0,0 +1,513 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "model_name": "HuggingFaceTB/SmolLM3-3B",
4
+ "model_type": "instruction-tuned",
5
+ "params": 3.0
6
+ },
7
+ "results": {
8
+ "medmcqa": {
9
+ "acc_norm": 0.47334448960076503,
10
+ "acc_norm_stderr": 0.00772075844043794,
11
+ "args": {
12
+ "exp_name": "results/v1/medmcqa/HuggingFaceTB/SmolLM3-3B",
13
+ "task_name": "medmcqa",
14
+ "model_name": "HuggingFaceTB/SmolLM3-3B",
15
+ "dataset_param": {
16
+ "max_num_train": 50000,
17
+ "completion_only": false,
18
+ "apply_chat_template": true
19
+ },
20
+ "resume_from_checkpoint": false,
21
+ "train_param": {
22
+ "max_length": 1024,
23
+ "per_device_train_batch_size": 8,
24
+ "auto_find_batch_size": true,
25
+ "num_train_epochs": 1,
26
+ "max_steps": -1,
27
+ "logging_strategy": "steps",
28
+ "save_strategy": "steps",
29
+ "logging_steps": 0.2,
30
+ "save_steps": 0.2,
31
+ "gradient_accumulation_steps": 1,
32
+ "gradient_checkpointing": false,
33
+ "optim": "adamw_torch",
34
+ "learning_rate": 5e-05,
35
+ "weight_decay": 0.01,
36
+ "warmup_steps": 0,
37
+ "report_to": "tensorboard",
38
+ "save_safetensors": true,
39
+ "packing": false
40
+ },
41
+ "use_peft": true,
42
+ "peft_task_type": "CAUSAL_LM",
43
+ "peft_inference_mode": false,
44
+ "peft_r": 8,
45
+ "peft_lora_alpha": 32,
46
+ "peft_lora_dropout": 0.1,
47
+ "fsdp": "none",
48
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
49
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
50
+ "fsdp_fsdp_min_num_params": 100000000.0,
51
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
52
+ "fsdp_fsdp_forward_prefetch": false,
53
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
54
+ "fsdp_fsdp_offload_params": false,
55
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
56
+ "fsdp_fsdp_use_orig_params": true,
57
+ "fsdp_fsdp_activation_checkpointing": false,
58
+ "fsdp_fsdp_limit_all_gathers": true,
59
+ "fsdp_fsdp_sync_module_states": true,
60
+ "use_git": true,
61
+ "random_seed": 0
62
+ }
63
+ },
64
+ "nq_open": {
65
+ "exact_match,remove_whitespace": 0.20443213296398893,
66
+ "exact_match,remove_whitespace_stderr": 0.006713048050224865,
67
+ "args": {
68
+ "exp_name": "results/v1/nq_open/HuggingFaceTB/SmolLM3-3B",
69
+ "task_name": "nq_open",
70
+ "model_name": "HuggingFaceTB/SmolLM3-3B",
71
+ "dataset_param": {
72
+ "max_num_train": 50000,
73
+ "completion_only": false,
74
+ "apply_chat_template": true
75
+ },
76
+ "resume_from_checkpoint": false,
77
+ "train_param": {
78
+ "max_length": 1024,
79
+ "per_device_train_batch_size": 8,
80
+ "auto_find_batch_size": true,
81
+ "num_train_epochs": 1,
82
+ "max_steps": -1,
83
+ "logging_strategy": "steps",
84
+ "save_strategy": "steps",
85
+ "logging_steps": 0.2,
86
+ "save_steps": 0.2,
87
+ "gradient_accumulation_steps": 1,
88
+ "gradient_checkpointing": false,
89
+ "optim": "adamw_torch",
90
+ "learning_rate": 5e-05,
91
+ "weight_decay": 0.01,
92
+ "warmup_steps": 0,
93
+ "report_to": "tensorboard",
94
+ "save_safetensors": true,
95
+ "packing": false
96
+ },
97
+ "use_peft": true,
98
+ "peft_task_type": "CAUSAL_LM",
99
+ "peft_inference_mode": false,
100
+ "peft_r": 8,
101
+ "peft_lora_alpha": 32,
102
+ "peft_lora_dropout": 0.1,
103
+ "fsdp": "none",
104
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
105
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
106
+ "fsdp_fsdp_min_num_params": 100000000.0,
107
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
108
+ "fsdp_fsdp_forward_prefetch": false,
109
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
110
+ "fsdp_fsdp_offload_params": false,
111
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
112
+ "fsdp_fsdp_use_orig_params": true,
113
+ "fsdp_fsdp_activation_checkpointing": false,
114
+ "fsdp_fsdp_limit_all_gathers": true,
115
+ "fsdp_fsdp_sync_module_states": true,
116
+ "use_git": true,
117
+ "random_seed": 0
118
+ }
119
+ },
120
+ "winogrande": {
121
+ "acc": 0.7813733228097869,
122
+ "acc_stderr": 0.011616198215773232,
123
+ "args": {
124
+ "exp_name": "results/v1/winogrande/HuggingFaceTB/SmolLM3-3B",
125
+ "task_name": "winogrande",
126
+ "model_name": "HuggingFaceTB/SmolLM3-3B",
127
+ "dataset_param": {
128
+ "max_num_train": 50000,
129
+ "completion_only": false,
130
+ "apply_chat_template": true
131
+ },
132
+ "resume_from_checkpoint": false,
133
+ "train_param": {
134
+ "max_length": 1024,
135
+ "per_device_train_batch_size": 8,
136
+ "auto_find_batch_size": true,
137
+ "num_train_epochs": 1,
138
+ "max_steps": -1,
139
+ "logging_strategy": "steps",
140
+ "save_strategy": "steps",
141
+ "logging_steps": 0.2,
142
+ "save_steps": 0.2,
143
+ "gradient_accumulation_steps": 1,
144
+ "gradient_checkpointing": false,
145
+ "optim": "adamw_torch",
146
+ "learning_rate": 5e-05,
147
+ "weight_decay": 0.01,
148
+ "warmup_steps": 0,
149
+ "report_to": "tensorboard",
150
+ "save_safetensors": true,
151
+ "packing": false
152
+ },
153
+ "use_peft": true,
154
+ "peft_task_type": "CAUSAL_LM",
155
+ "peft_inference_mode": false,
156
+ "peft_r": 8,
157
+ "peft_lora_alpha": 32,
158
+ "peft_lora_dropout": 0.1,
159
+ "fsdp": "none",
160
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
161
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
162
+ "fsdp_fsdp_min_num_params": 100000000.0,
163
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
164
+ "fsdp_fsdp_forward_prefetch": false,
165
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
166
+ "fsdp_fsdp_offload_params": false,
167
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
168
+ "fsdp_fsdp_use_orig_params": true,
169
+ "fsdp_fsdp_activation_checkpointing": false,
170
+ "fsdp_fsdp_limit_all_gathers": true,
171
+ "fsdp_fsdp_sync_module_states": true,
172
+ "use_git": true,
173
+ "random_seed": 0
174
+ }
175
+ },
176
+ "hellaswag": {
177
+ "acc_norm": 0.7679,
178
+ "acc_norm_stderr": 0.004221935749646755,
179
+ "args": {
180
+ "exp_name": "results/v1/hellaswag/HuggingFaceTB/SmolLM3-3B",
181
+ "task_name": "hellaswag",
182
+ "model_name": "HuggingFaceTB/SmolLM3-3B",
183
+ "dataset_param": {
184
+ "max_num_train": 50000,
185
+ "completion_only": false,
186
+ "apply_chat_template": true
187
+ },
188
+ "resume_from_checkpoint": false,
189
+ "train_param": {
190
+ "max_length": 1024,
191
+ "per_device_train_batch_size": 8,
192
+ "auto_find_batch_size": true,
193
+ "num_train_epochs": 1,
194
+ "max_steps": -1,
195
+ "logging_strategy": "steps",
196
+ "save_strategy": "steps",
197
+ "logging_steps": 0.2,
198
+ "save_steps": 0.2,
199
+ "gradient_accumulation_steps": 1,
200
+ "gradient_checkpointing": false,
201
+ "optim": "adamw_torch",
202
+ "learning_rate": 5e-05,
203
+ "weight_decay": 0.01,
204
+ "warmup_steps": 0,
205
+ "report_to": "tensorboard",
206
+ "save_safetensors": true,
207
+ "packing": false
208
+ },
209
+ "use_peft": true,
210
+ "peft_task_type": "CAUSAL_LM",
211
+ "peft_inference_mode": false,
212
+ "peft_r": 8,
213
+ "peft_lora_alpha": 32,
214
+ "peft_lora_dropout": 0.1,
215
+ "fsdp": "none",
216
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
217
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
218
+ "fsdp_fsdp_min_num_params": 100000000.0,
219
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
220
+ "fsdp_fsdp_forward_prefetch": false,
221
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
222
+ "fsdp_fsdp_offload_params": false,
223
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
224
+ "fsdp_fsdp_use_orig_params": true,
225
+ "fsdp_fsdp_activation_checkpointing": false,
226
+ "fsdp_fsdp_limit_all_gathers": true,
227
+ "fsdp_fsdp_sync_module_states": true,
228
+ "use_git": true,
229
+ "random_seed": 0
230
+ }
231
+ },
232
+ "social_iqa": {
233
+ "acc": 0.5583418628454453,
234
+ "acc_stderr": 0.011236785537985829,
235
+ "args": {
236
+ "exp_name": "results/v1/social_iqa/HuggingFaceTB/SmolLM3-3B",
237
+ "task_name": "social_iqa",
238
+ "model_name": "HuggingFaceTB/SmolLM3-3B",
239
+ "dataset_param": {
240
+ "max_num_train": 50000,
241
+ "completion_only": false,
242
+ "apply_chat_template": true
243
+ },
244
+ "resume_from_checkpoint": false,
245
+ "train_param": {
246
+ "max_length": 1024,
247
+ "per_device_train_batch_size": 8,
248
+ "auto_find_batch_size": true,
249
+ "num_train_epochs": 1,
250
+ "max_steps": -1,
251
+ "logging_strategy": "steps",
252
+ "save_strategy": "steps",
253
+ "logging_steps": 0.2,
254
+ "save_steps": 0.2,
255
+ "gradient_accumulation_steps": 1,
256
+ "gradient_checkpointing": false,
257
+ "optim": "adamw_torch",
258
+ "learning_rate": 5e-05,
259
+ "weight_decay": 0.01,
260
+ "warmup_steps": 0,
261
+ "report_to": "tensorboard",
262
+ "save_safetensors": true,
263
+ "packing": false
264
+ },
265
+ "use_peft": true,
266
+ "peft_task_type": "CAUSAL_LM",
267
+ "peft_inference_mode": false,
268
+ "peft_r": 8,
269
+ "peft_lora_alpha": 32,
270
+ "peft_lora_dropout": 0.1,
271
+ "fsdp": "none",
272
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
273
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
274
+ "fsdp_fsdp_min_num_params": 100000000.0,
275
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
276
+ "fsdp_fsdp_forward_prefetch": false,
277
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
278
+ "fsdp_fsdp_offload_params": false,
279
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
280
+ "fsdp_fsdp_use_orig_params": true,
281
+ "fsdp_fsdp_activation_checkpointing": false,
282
+ "fsdp_fsdp_limit_all_gathers": true,
283
+ "fsdp_fsdp_sync_module_states": true,
284
+ "use_git": true,
285
+ "random_seed": 0
286
+ }
287
+ },
288
+ "piqa": {
289
+ "acc_norm": 0.8014145810663765,
290
+ "acc_norm_stderr": 0.009307814521717892,
291
+ "args": {
292
+ "exp_name": "results/v2/piqa/HuggingFaceTB/SmolLM3-3B",
293
+ "task_name": "piqa",
294
+ "model_name": "HuggingFaceTB/SmolLM3-3B",
295
+ "dataset_param": {
296
+ "max_num_train": 50000,
297
+ "completion_only": false,
298
+ "apply_chat_template": true
299
+ },
300
+ "resume_from_checkpoint": false,
301
+ "train_param": {
302
+ "max_length": 1024,
303
+ "per_device_train_batch_size": 8,
304
+ "auto_find_batch_size": true,
305
+ "num_train_epochs": 5,
306
+ "max_steps": -1,
307
+ "logging_strategy": "epoch",
308
+ "save_strategy": "epoch",
309
+ "logging_steps": 0.2,
310
+ "save_steps": 0.2,
311
+ "gradient_accumulation_steps": 1,
312
+ "gradient_checkpointing": false,
313
+ "optim": "adamw_torch",
314
+ "learning_rate": 5e-05,
315
+ "weight_decay": 0.01,
316
+ "warmup_steps": 0,
317
+ "report_to": "tensorboard",
318
+ "save_safetensors": true,
319
+ "packing": false
320
+ },
321
+ "use_peft": true,
322
+ "peft_task_type": "CAUSAL_LM",
323
+ "peft_inference_mode": false,
324
+ "peft_r": 8,
325
+ "peft_lora_alpha": 32,
326
+ "peft_lora_dropout": 0.1,
327
+ "fsdp": "none",
328
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
329
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
330
+ "fsdp_fsdp_min_num_params": 100000000.0,
331
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
332
+ "fsdp_fsdp_forward_prefetch": false,
333
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
334
+ "fsdp_fsdp_offload_params": false,
335
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
336
+ "fsdp_fsdp_use_orig_params": true,
337
+ "fsdp_fsdp_activation_checkpointing": false,
338
+ "fsdp_fsdp_limit_all_gathers": true,
339
+ "fsdp_fsdp_sync_module_states": true,
340
+ "use_git": true,
341
+ "random_seed": 0
342
+ }
343
+ },
344
+ "commonsense_qa": {
345
+ "acc": 0.7854217854217854,
346
+ "acc_stderr": 0.011753423094216855,
347
+ "args": {
348
+ "exp_name": "results/v2/commonsense_qa/HuggingFaceTB/SmolLM3-3B",
349
+ "task_name": "commonsense_qa",
350
+ "model_name": "HuggingFaceTB/SmolLM3-3B",
351
+ "dataset_param": {
352
+ "max_num_train": 50000,
353
+ "completion_only": false,
354
+ "apply_chat_template": true
355
+ },
356
+ "resume_from_checkpoint": false,
357
+ "train_param": {
358
+ "max_length": 1024,
359
+ "per_device_train_batch_size": 8,
360
+ "auto_find_batch_size": true,
361
+ "num_train_epochs": 5,
362
+ "max_steps": -1,
363
+ "logging_strategy": "epoch",
364
+ "save_strategy": "epoch",
365
+ "logging_steps": 0.2,
366
+ "save_steps": 0.2,
367
+ "gradient_accumulation_steps": 1,
368
+ "gradient_checkpointing": false,
369
+ "optim": "adamw_torch",
370
+ "learning_rate": 5e-05,
371
+ "weight_decay": 0.01,
372
+ "warmup_steps": 0,
373
+ "report_to": "tensorboard",
374
+ "save_safetensors": true,
375
+ "packing": false
376
+ },
377
+ "use_peft": true,
378
+ "peft_task_type": "CAUSAL_LM",
379
+ "peft_inference_mode": false,
380
+ "peft_r": 8,
381
+ "peft_lora_alpha": 32,
382
+ "peft_lora_dropout": 0.1,
383
+ "fsdp": "none",
384
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
385
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
386
+ "fsdp_fsdp_min_num_params": 100000000.0,
387
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
388
+ "fsdp_fsdp_forward_prefetch": false,
389
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
390
+ "fsdp_fsdp_offload_params": false,
391
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
392
+ "fsdp_fsdp_use_orig_params": true,
393
+ "fsdp_fsdp_activation_checkpointing": false,
394
+ "fsdp_fsdp_limit_all_gathers": true,
395
+ "fsdp_fsdp_sync_module_states": true,
396
+ "use_git": true,
397
+ "random_seed": 0
398
+ }
399
+ },
400
+ "gsm8k": {
401
+ "exact_match,flexible-extract": 0.6929492039423806,
402
+ "exact_match,flexible-extract_stderr": 0.01270568572313171,
403
+ "args": {
404
+ "exp_name": "results/v2/gsm8k/HuggingFaceTB/SmolLM3-3B",
405
+ "task_name": "gsm8k",
406
+ "model_name": "HuggingFaceTB/SmolLM3-3B",
407
+ "dataset_param": {
408
+ "max_num_train": 50000,
409
+ "completion_only": false,
410
+ "apply_chat_template": true
411
+ },
412
+ "resume_from_checkpoint": false,
413
+ "train_param": {
414
+ "max_length": 1024,
415
+ "per_device_train_batch_size": 8,
416
+ "auto_find_batch_size": true,
417
+ "num_train_epochs": 5,
418
+ "max_steps": -1,
419
+ "logging_strategy": "epoch",
420
+ "save_strategy": "epoch",
421
+ "logging_steps": 0.2,
422
+ "save_steps": 0.2,
423
+ "gradient_accumulation_steps": 1,
424
+ "gradient_checkpointing": false,
425
+ "optim": "adamw_torch",
426
+ "learning_rate": 5e-05,
427
+ "weight_decay": 0.01,
428
+ "warmup_steps": 0,
429
+ "report_to": "tensorboard",
430
+ "save_safetensors": true,
431
+ "packing": false
432
+ },
433
+ "use_peft": true,
434
+ "peft_task_type": "CAUSAL_LM",
435
+ "peft_inference_mode": false,
436
+ "peft_r": 8,
437
+ "peft_lora_alpha": 32,
438
+ "peft_lora_dropout": 0.1,
439
+ "fsdp": "none",
440
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
441
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
442
+ "fsdp_fsdp_min_num_params": 100000000.0,
443
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
444
+ "fsdp_fsdp_forward_prefetch": false,
445
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
446
+ "fsdp_fsdp_offload_params": false,
447
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
448
+ "fsdp_fsdp_use_orig_params": true,
449
+ "fsdp_fsdp_activation_checkpointing": false,
450
+ "fsdp_fsdp_limit_all_gathers": true,
451
+ "fsdp_fsdp_sync_module_states": true,
452
+ "use_git": true,
453
+ "random_seed": 0
454
+ }
455
+ },
456
+ "arc_challenge": {
457
+ "acc_norm": 0.5503412969283277,
458
+ "acc_norm_stderr": 0.014537144444284736,
459
+ "args": {
460
+ "exp_name": "results/v2/arc_challenge/HuggingFaceTB/SmolLM3-3B",
461
+ "task_name": "arc_challenge",
462
+ "model_name": "HuggingFaceTB/SmolLM3-3B",
463
+ "dataset_param": {
464
+ "max_num_train": 50000,
465
+ "completion_only": false,
466
+ "apply_chat_template": true
467
+ },
468
+ "resume_from_checkpoint": false,
469
+ "train_param": {
470
+ "max_length": 1024,
471
+ "per_device_train_batch_size": 8,
472
+ "auto_find_batch_size": true,
473
+ "num_train_epochs": 5,
474
+ "max_steps": -1,
475
+ "logging_strategy": "epoch",
476
+ "save_strategy": "epoch",
477
+ "logging_steps": 0.2,
478
+ "save_steps": 0.2,
479
+ "gradient_accumulation_steps": 1,
480
+ "gradient_checkpointing": false,
481
+ "optim": "adamw_torch",
482
+ "learning_rate": 5e-05,
483
+ "weight_decay": 0.01,
484
+ "warmup_steps": 0,
485
+ "report_to": "tensorboard",
486
+ "save_safetensors": true,
487
+ "packing": false
488
+ },
489
+ "use_peft": true,
490
+ "peft_task_type": "CAUSAL_LM",
491
+ "peft_inference_mode": false,
492
+ "peft_r": 8,
493
+ "peft_lora_alpha": 32,
494
+ "peft_lora_dropout": 0.1,
495
+ "fsdp": "none",
496
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
497
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
498
+ "fsdp_fsdp_min_num_params": 100000000.0,
499
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
500
+ "fsdp_fsdp_forward_prefetch": false,
501
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
502
+ "fsdp_fsdp_offload_params": false,
503
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
504
+ "fsdp_fsdp_use_orig_params": true,
505
+ "fsdp_fsdp_activation_checkpointing": false,
506
+ "fsdp_fsdp_limit_all_gathers": true,
507
+ "fsdp_fsdp_sync_module_states": true,
508
+ "use_git": true,
509
+ "random_seed": 0
510
+ }
511
+ }
512
+ }
513
+ }
LumiOpen/Llama-Poro-2-8B-Instruct/results_2026-01-05-04-27-51.json ADDED
@@ -0,0 +1,513 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "model_name": "LumiOpen/Llama-Poro-2-8B-Instruct",
4
+ "model_type": "instruction-tuned",
5
+ "params": 8.0
6
+ },
7
+ "results": {
8
+ "medmcqa": {
9
+ "acc_norm": 0.5118336122400191,
10
+ "acc_norm_stderr": 0.0077295876047059325,
11
+ "args": {
12
+ "exp_name": "results/v1/medmcqa/LumiOpen/Llama-Poro-2-8B-Instruct",
13
+ "task_name": "medmcqa",
14
+ "model_name": "LumiOpen/Llama-Poro-2-8B-Instruct",
15
+ "dataset_param": {
16
+ "max_num_train": 50000,
17
+ "completion_only": false,
18
+ "apply_chat_template": true
19
+ },
20
+ "resume_from_checkpoint": false,
21
+ "train_param": {
22
+ "max_length": 1024,
23
+ "per_device_train_batch_size": 8,
24
+ "auto_find_batch_size": true,
25
+ "num_train_epochs": 1,
26
+ "max_steps": -1,
27
+ "logging_strategy": "steps",
28
+ "save_strategy": "steps",
29
+ "logging_steps": 0.2,
30
+ "save_steps": 0.2,
31
+ "gradient_accumulation_steps": 1,
32
+ "gradient_checkpointing": false,
33
+ "optim": "adamw_torch",
34
+ "learning_rate": 5e-05,
35
+ "weight_decay": 0.01,
36
+ "warmup_steps": 0,
37
+ "report_to": "tensorboard",
38
+ "save_safetensors": true,
39
+ "packing": false
40
+ },
41
+ "use_peft": true,
42
+ "peft_task_type": "CAUSAL_LM",
43
+ "peft_inference_mode": false,
44
+ "peft_r": 8,
45
+ "peft_lora_alpha": 32,
46
+ "peft_lora_dropout": 0.1,
47
+ "fsdp": "none",
48
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
49
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
50
+ "fsdp_fsdp_min_num_params": 100000000.0,
51
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
52
+ "fsdp_fsdp_forward_prefetch": false,
53
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
54
+ "fsdp_fsdp_offload_params": false,
55
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
56
+ "fsdp_fsdp_use_orig_params": true,
57
+ "fsdp_fsdp_activation_checkpointing": false,
58
+ "fsdp_fsdp_limit_all_gathers": true,
59
+ "fsdp_fsdp_sync_module_states": true,
60
+ "use_git": true,
61
+ "random_seed": 0
62
+ }
63
+ },
64
+ "nq_open": {
65
+ "exact_match,remove_whitespace": 0.2698060941828255,
66
+ "exact_match,remove_whitespace_stderr": 0.007388419374207356,
67
+ "args": {
68
+ "exp_name": "results/v1/nq_open/LumiOpen/Llama-Poro-2-8B-Instruct",
69
+ "task_name": "nq_open",
70
+ "model_name": "LumiOpen/Llama-Poro-2-8B-Instruct",
71
+ "dataset_param": {
72
+ "max_num_train": 50000,
73
+ "completion_only": false,
74
+ "apply_chat_template": true
75
+ },
76
+ "resume_from_checkpoint": false,
77
+ "train_param": {
78
+ "max_length": 1024,
79
+ "per_device_train_batch_size": 8,
80
+ "auto_find_batch_size": true,
81
+ "num_train_epochs": 1,
82
+ "max_steps": -1,
83
+ "logging_strategy": "steps",
84
+ "save_strategy": "steps",
85
+ "logging_steps": 0.2,
86
+ "save_steps": 0.2,
87
+ "gradient_accumulation_steps": 1,
88
+ "gradient_checkpointing": false,
89
+ "optim": "adamw_torch",
90
+ "learning_rate": 5e-05,
91
+ "weight_decay": 0.01,
92
+ "warmup_steps": 0,
93
+ "report_to": "tensorboard",
94
+ "save_safetensors": true,
95
+ "packing": false
96
+ },
97
+ "use_peft": true,
98
+ "peft_task_type": "CAUSAL_LM",
99
+ "peft_inference_mode": false,
100
+ "peft_r": 8,
101
+ "peft_lora_alpha": 32,
102
+ "peft_lora_dropout": 0.1,
103
+ "fsdp": "none",
104
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
105
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
106
+ "fsdp_fsdp_min_num_params": 100000000.0,
107
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
108
+ "fsdp_fsdp_forward_prefetch": false,
109
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
110
+ "fsdp_fsdp_offload_params": false,
111
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
112
+ "fsdp_fsdp_use_orig_params": true,
113
+ "fsdp_fsdp_activation_checkpointing": false,
114
+ "fsdp_fsdp_limit_all_gathers": true,
115
+ "fsdp_fsdp_sync_module_states": true,
116
+ "use_git": true,
117
+ "random_seed": 0
118
+ }
119
+ },
120
+ "winogrande": {
121
+ "acc": 0.8050513022888713,
122
+ "acc_stderr": 0.011134099415938284,
123
+ "args": {
124
+ "exp_name": "results/v1/winogrande/LumiOpen/Llama-Poro-2-8B-Instruct",
125
+ "task_name": "winogrande",
126
+ "model_name": "LumiOpen/Llama-Poro-2-8B-Instruct",
127
+ "dataset_param": {
128
+ "max_num_train": 50000,
129
+ "completion_only": false,
130
+ "apply_chat_template": true
131
+ },
132
+ "resume_from_checkpoint": false,
133
+ "train_param": {
134
+ "max_length": 1024,
135
+ "per_device_train_batch_size": 8,
136
+ "auto_find_batch_size": true,
137
+ "num_train_epochs": 1,
138
+ "max_steps": -1,
139
+ "logging_strategy": "steps",
140
+ "save_strategy": "steps",
141
+ "logging_steps": 0.2,
142
+ "save_steps": 0.2,
143
+ "gradient_accumulation_steps": 1,
144
+ "gradient_checkpointing": false,
145
+ "optim": "adamw_torch",
146
+ "learning_rate": 5e-05,
147
+ "weight_decay": 0.01,
148
+ "warmup_steps": 0,
149
+ "report_to": "tensorboard",
150
+ "save_safetensors": true,
151
+ "packing": false
152
+ },
153
+ "use_peft": true,
154
+ "peft_task_type": "CAUSAL_LM",
155
+ "peft_inference_mode": false,
156
+ "peft_r": 8,
157
+ "peft_lora_alpha": 32,
158
+ "peft_lora_dropout": 0.1,
159
+ "fsdp": "none",
160
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
161
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
162
+ "fsdp_fsdp_min_num_params": 100000000.0,
163
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
164
+ "fsdp_fsdp_forward_prefetch": false,
165
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
166
+ "fsdp_fsdp_offload_params": false,
167
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
168
+ "fsdp_fsdp_use_orig_params": true,
169
+ "fsdp_fsdp_activation_checkpointing": false,
170
+ "fsdp_fsdp_limit_all_gathers": true,
171
+ "fsdp_fsdp_sync_module_states": true,
172
+ "use_git": true,
173
+ "random_seed": 0
174
+ }
175
+ },
176
+ "hellaswag": {
177
+ "acc_norm": 0.8117,
178
+ "acc_norm_stderr": 0.003909710933554639,
179
+ "args": {
180
+ "exp_name": "results/v1/hellaswag/LumiOpen/Llama-Poro-2-8B-Instruct",
181
+ "task_name": "hellaswag",
182
+ "model_name": "LumiOpen/Llama-Poro-2-8B-Instruct",
183
+ "dataset_param": {
184
+ "max_num_train": 50000,
185
+ "completion_only": false,
186
+ "apply_chat_template": true
187
+ },
188
+ "resume_from_checkpoint": false,
189
+ "train_param": {
190
+ "max_length": 1024,
191
+ "per_device_train_batch_size": 8,
192
+ "auto_find_batch_size": true,
193
+ "num_train_epochs": 1,
194
+ "max_steps": -1,
195
+ "logging_strategy": "steps",
196
+ "save_strategy": "steps",
197
+ "logging_steps": 0.2,
198
+ "save_steps": 0.2,
199
+ "gradient_accumulation_steps": 1,
200
+ "gradient_checkpointing": false,
201
+ "optim": "adamw_torch",
202
+ "learning_rate": 5e-05,
203
+ "weight_decay": 0.01,
204
+ "warmup_steps": 0,
205
+ "report_to": "tensorboard",
206
+ "save_safetensors": true,
207
+ "packing": false
208
+ },
209
+ "use_peft": true,
210
+ "peft_task_type": "CAUSAL_LM",
211
+ "peft_inference_mode": false,
212
+ "peft_r": 8,
213
+ "peft_lora_alpha": 32,
214
+ "peft_lora_dropout": 0.1,
215
+ "fsdp": "none",
216
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
217
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
218
+ "fsdp_fsdp_min_num_params": 100000000.0,
219
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
220
+ "fsdp_fsdp_forward_prefetch": false,
221
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
222
+ "fsdp_fsdp_offload_params": false,
223
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
224
+ "fsdp_fsdp_use_orig_params": true,
225
+ "fsdp_fsdp_activation_checkpointing": false,
226
+ "fsdp_fsdp_limit_all_gathers": true,
227
+ "fsdp_fsdp_sync_module_states": true,
228
+ "use_git": true,
229
+ "random_seed": 0
230
+ }
231
+ },
232
+ "social_iqa": {
233
+ "acc": 0.5701125895598772,
234
+ "acc_stderr": 0.011202283451328801,
235
+ "args": {
236
+ "exp_name": "results/v1/social_iqa/LumiOpen/Llama-Poro-2-8B-Instruct",
237
+ "task_name": "social_iqa",
238
+ "model_name": "LumiOpen/Llama-Poro-2-8B-Instruct",
239
+ "dataset_param": {
240
+ "max_num_train": 50000,
241
+ "completion_only": false,
242
+ "apply_chat_template": true
243
+ },
244
+ "resume_from_checkpoint": false,
245
+ "train_param": {
246
+ "max_length": 1024,
247
+ "per_device_train_batch_size": 8,
248
+ "auto_find_batch_size": true,
249
+ "num_train_epochs": 1,
250
+ "max_steps": -1,
251
+ "logging_strategy": "steps",
252
+ "save_strategy": "steps",
253
+ "logging_steps": 0.2,
254
+ "save_steps": 0.2,
255
+ "gradient_accumulation_steps": 1,
256
+ "gradient_checkpointing": false,
257
+ "optim": "adamw_torch",
258
+ "learning_rate": 5e-05,
259
+ "weight_decay": 0.01,
260
+ "warmup_steps": 0,
261
+ "report_to": "tensorboard",
262
+ "save_safetensors": true,
263
+ "packing": false
264
+ },
265
+ "use_peft": true,
266
+ "peft_task_type": "CAUSAL_LM",
267
+ "peft_inference_mode": false,
268
+ "peft_r": 8,
269
+ "peft_lora_alpha": 32,
270
+ "peft_lora_dropout": 0.1,
271
+ "fsdp": "none",
272
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
273
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
274
+ "fsdp_fsdp_min_num_params": 100000000.0,
275
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
276
+ "fsdp_fsdp_forward_prefetch": false,
277
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
278
+ "fsdp_fsdp_offload_params": false,
279
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
280
+ "fsdp_fsdp_use_orig_params": true,
281
+ "fsdp_fsdp_activation_checkpointing": false,
282
+ "fsdp_fsdp_limit_all_gathers": true,
283
+ "fsdp_fsdp_sync_module_states": true,
284
+ "use_git": true,
285
+ "random_seed": 0
286
+ }
287
+ },
288
+ "piqa": {
289
+ "acc_norm": 0.824265505984766,
290
+ "acc_norm_stderr": 0.008879888086402025,
291
+ "args": {
292
+ "exp_name": "results/v2/piqa/LumiOpen/Llama-Poro-2-8B-Instruct",
293
+ "task_name": "piqa",
294
+ "model_name": "LumiOpen/Llama-Poro-2-8B-Instruct",
295
+ "dataset_param": {
296
+ "max_num_train": 50000,
297
+ "completion_only": false,
298
+ "apply_chat_template": true
299
+ },
300
+ "resume_from_checkpoint": false,
301
+ "train_param": {
302
+ "max_length": 1024,
303
+ "per_device_train_batch_size": 8,
304
+ "auto_find_batch_size": true,
305
+ "num_train_epochs": 5,
306
+ "max_steps": -1,
307
+ "logging_strategy": "epoch",
308
+ "save_strategy": "epoch",
309
+ "logging_steps": 0.2,
310
+ "save_steps": 0.2,
311
+ "gradient_accumulation_steps": 1,
312
+ "gradient_checkpointing": false,
313
+ "optim": "adamw_torch",
314
+ "learning_rate": 5e-05,
315
+ "weight_decay": 0.01,
316
+ "warmup_steps": 0,
317
+ "report_to": "tensorboard",
318
+ "save_safetensors": true,
319
+ "packing": false
320
+ },
321
+ "use_peft": true,
322
+ "peft_task_type": "CAUSAL_LM",
323
+ "peft_inference_mode": false,
324
+ "peft_r": 8,
325
+ "peft_lora_alpha": 32,
326
+ "peft_lora_dropout": 0.1,
327
+ "fsdp": "none",
328
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
329
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
330
+ "fsdp_fsdp_min_num_params": 100000000.0,
331
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
332
+ "fsdp_fsdp_forward_prefetch": false,
333
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
334
+ "fsdp_fsdp_offload_params": false,
335
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
336
+ "fsdp_fsdp_use_orig_params": true,
337
+ "fsdp_fsdp_activation_checkpointing": false,
338
+ "fsdp_fsdp_limit_all_gathers": true,
339
+ "fsdp_fsdp_sync_module_states": true,
340
+ "use_git": true,
341
+ "random_seed": 0
342
+ }
343
+ },
344
+ "commonsense_qa": {
345
+ "acc": 0.8222768222768223,
346
+ "acc_stderr": 0.010944631509166454,
347
+ "args": {
348
+ "exp_name": "results/v2/commonsense_qa/LumiOpen/Llama-Poro-2-8B-Instruct",
349
+ "task_name": "commonsense_qa",
350
+ "model_name": "LumiOpen/Llama-Poro-2-8B-Instruct",
351
+ "dataset_param": {
352
+ "max_num_train": 50000,
353
+ "completion_only": false,
354
+ "apply_chat_template": true
355
+ },
356
+ "resume_from_checkpoint": false,
357
+ "train_param": {
358
+ "max_length": 1024,
359
+ "per_device_train_batch_size": 8,
360
+ "auto_find_batch_size": true,
361
+ "num_train_epochs": 5,
362
+ "max_steps": -1,
363
+ "logging_strategy": "epoch",
364
+ "save_strategy": "epoch",
365
+ "logging_steps": 0.2,
366
+ "save_steps": 0.2,
367
+ "gradient_accumulation_steps": 1,
368
+ "gradient_checkpointing": false,
369
+ "optim": "adamw_torch",
370
+ "learning_rate": 5e-05,
371
+ "weight_decay": 0.01,
372
+ "warmup_steps": 0,
373
+ "report_to": "tensorboard",
374
+ "save_safetensors": true,
375
+ "packing": false
376
+ },
377
+ "use_peft": true,
378
+ "peft_task_type": "CAUSAL_LM",
379
+ "peft_inference_mode": false,
380
+ "peft_r": 8,
381
+ "peft_lora_alpha": 32,
382
+ "peft_lora_dropout": 0.1,
383
+ "fsdp": "none",
384
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
385
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
386
+ "fsdp_fsdp_min_num_params": 100000000.0,
387
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
388
+ "fsdp_fsdp_forward_prefetch": false,
389
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
390
+ "fsdp_fsdp_offload_params": false,
391
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
392
+ "fsdp_fsdp_use_orig_params": true,
393
+ "fsdp_fsdp_activation_checkpointing": false,
394
+ "fsdp_fsdp_limit_all_gathers": true,
395
+ "fsdp_fsdp_sync_module_states": true,
396
+ "use_git": true,
397
+ "random_seed": 0
398
+ }
399
+ },
400
+ "gsm8k": {
401
+ "exact_match,flexible-extract": 0.7414708112206216,
402
+ "exact_match,flexible-extract_stderr": 0.012059911372516118,
403
+ "args": {
404
+ "exp_name": "results/v2/gsm8k/LumiOpen/Llama-Poro-2-8B-Instruct",
405
+ "task_name": "gsm8k",
406
+ "model_name": "LumiOpen/Llama-Poro-2-8B-Instruct",
407
+ "dataset_param": {
408
+ "max_num_train": 50000,
409
+ "completion_only": false,
410
+ "apply_chat_template": true
411
+ },
412
+ "resume_from_checkpoint": false,
413
+ "train_param": {
414
+ "max_length": 1024,
415
+ "per_device_train_batch_size": 8,
416
+ "auto_find_batch_size": true,
417
+ "num_train_epochs": 5,
418
+ "max_steps": -1,
419
+ "logging_strategy": "epoch",
420
+ "save_strategy": "epoch",
421
+ "logging_steps": 0.2,
422
+ "save_steps": 0.2,
423
+ "gradient_accumulation_steps": 1,
424
+ "gradient_checkpointing": false,
425
+ "optim": "adamw_torch",
426
+ "learning_rate": 5e-05,
427
+ "weight_decay": 0.01,
428
+ "warmup_steps": 0,
429
+ "report_to": "tensorboard",
430
+ "save_safetensors": true,
431
+ "packing": false
432
+ },
433
+ "use_peft": true,
434
+ "peft_task_type": "CAUSAL_LM",
435
+ "peft_inference_mode": false,
436
+ "peft_r": 8,
437
+ "peft_lora_alpha": 32,
438
+ "peft_lora_dropout": 0.1,
439
+ "fsdp": "none",
440
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
441
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
442
+ "fsdp_fsdp_min_num_params": 100000000.0,
443
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
444
+ "fsdp_fsdp_forward_prefetch": false,
445
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
446
+ "fsdp_fsdp_offload_params": false,
447
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
448
+ "fsdp_fsdp_use_orig_params": true,
449
+ "fsdp_fsdp_activation_checkpointing": false,
450
+ "fsdp_fsdp_limit_all_gathers": true,
451
+ "fsdp_fsdp_sync_module_states": true,
452
+ "use_git": true,
453
+ "random_seed": 0
454
+ }
455
+ },
456
+ "arc_challenge": {
457
+ "acc_norm": 0.5921501706484642,
458
+ "acc_norm_stderr": 0.014361097288449705,
459
+ "args": {
460
+ "exp_name": "results/v2/arc_challenge/LumiOpen/Llama-Poro-2-8B-Instruct",
461
+ "task_name": "arc_challenge",
462
+ "model_name": "LumiOpen/Llama-Poro-2-8B-Instruct",
463
+ "dataset_param": {
464
+ "max_num_train": 50000,
465
+ "completion_only": false,
466
+ "apply_chat_template": true
467
+ },
468
+ "resume_from_checkpoint": false,
469
+ "train_param": {
470
+ "max_length": 1024,
471
+ "per_device_train_batch_size": 8,
472
+ "auto_find_batch_size": true,
473
+ "num_train_epochs": 5,
474
+ "max_steps": -1,
475
+ "logging_strategy": "epoch",
476
+ "save_strategy": "epoch",
477
+ "logging_steps": 0.2,
478
+ "save_steps": 0.2,
479
+ "gradient_accumulation_steps": 1,
480
+ "gradient_checkpointing": false,
481
+ "optim": "adamw_torch",
482
+ "learning_rate": 5e-05,
483
+ "weight_decay": 0.01,
484
+ "warmup_steps": 0,
485
+ "report_to": "tensorboard",
486
+ "save_safetensors": true,
487
+ "packing": false
488
+ },
489
+ "use_peft": true,
490
+ "peft_task_type": "CAUSAL_LM",
491
+ "peft_inference_mode": false,
492
+ "peft_r": 8,
493
+ "peft_lora_alpha": 32,
494
+ "peft_lora_dropout": 0.1,
495
+ "fsdp": "none",
496
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
497
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
498
+ "fsdp_fsdp_min_num_params": 100000000.0,
499
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
500
+ "fsdp_fsdp_forward_prefetch": false,
501
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
502
+ "fsdp_fsdp_offload_params": false,
503
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
504
+ "fsdp_fsdp_use_orig_params": true,
505
+ "fsdp_fsdp_activation_checkpointing": false,
506
+ "fsdp_fsdp_limit_all_gathers": true,
507
+ "fsdp_fsdp_sync_module_states": true,
508
+ "use_git": true,
509
+ "random_seed": 0
510
+ }
511
+ }
512
+ }
513
+ }
LumiOpen/Llama-Poro-2-8B-base/results_2026-01-05-04-27-51.json ADDED
@@ -0,0 +1,513 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "model_name": "LumiOpen/Llama-Poro-2-8B-base",
4
+ "model_type": "pretrained",
5
+ "params": 8.0
6
+ },
7
+ "results": {
8
+ "medmcqa": {
9
+ "acc_norm": 0.5319148936170213,
10
+ "acc_norm_stderr": 0.007715986738889257,
11
+ "args": {
12
+ "exp_name": "results/v1/medmcqa/LumiOpen/Llama-Poro-2-8B-base",
13
+ "task_name": "medmcqa",
14
+ "model_name": "LumiOpen/Llama-Poro-2-8B-base",
15
+ "dataset_param": {
16
+ "max_num_train": 50000,
17
+ "completion_only": false,
18
+ "apply_chat_template": false
19
+ },
20
+ "resume_from_checkpoint": false,
21
+ "train_param": {
22
+ "max_length": 1024,
23
+ "per_device_train_batch_size": 8,
24
+ "auto_find_batch_size": true,
25
+ "num_train_epochs": 1,
26
+ "max_steps": -1,
27
+ "logging_strategy": "steps",
28
+ "save_strategy": "steps",
29
+ "logging_steps": 0.2,
30
+ "save_steps": 0.2,
31
+ "gradient_accumulation_steps": 1,
32
+ "gradient_checkpointing": false,
33
+ "optim": "adamw_torch",
34
+ "learning_rate": 5e-05,
35
+ "weight_decay": 0.01,
36
+ "warmup_steps": 0,
37
+ "report_to": "tensorboard",
38
+ "save_safetensors": true,
39
+ "packing": false
40
+ },
41
+ "use_peft": true,
42
+ "peft_task_type": "CAUSAL_LM",
43
+ "peft_inference_mode": false,
44
+ "peft_r": 8,
45
+ "peft_lora_alpha": 32,
46
+ "peft_lora_dropout": 0.1,
47
+ "fsdp": "none",
48
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
49
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
50
+ "fsdp_fsdp_min_num_params": 100000000.0,
51
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
52
+ "fsdp_fsdp_forward_prefetch": false,
53
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
54
+ "fsdp_fsdp_offload_params": false,
55
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
56
+ "fsdp_fsdp_use_orig_params": true,
57
+ "fsdp_fsdp_activation_checkpointing": false,
58
+ "fsdp_fsdp_limit_all_gathers": true,
59
+ "fsdp_fsdp_sync_module_states": true,
60
+ "use_git": true,
61
+ "random_seed": 0
62
+ }
63
+ },
64
+ "nq_open": {
65
+ "exact_match,remove_whitespace": 0.1185595567867036,
66
+ "exact_match,remove_whitespace_stderr": 0.005381103300405394,
67
+ "args": {
68
+ "exp_name": "results/v1/nq_open/LumiOpen/Llama-Poro-2-8B-base",
69
+ "task_name": "nq_open",
70
+ "model_name": "LumiOpen/Llama-Poro-2-8B-base",
71
+ "dataset_param": {
72
+ "max_num_train": 50000,
73
+ "completion_only": false,
74
+ "apply_chat_template": false
75
+ },
76
+ "resume_from_checkpoint": false,
77
+ "train_param": {
78
+ "max_length": 1024,
79
+ "per_device_train_batch_size": 8,
80
+ "auto_find_batch_size": true,
81
+ "num_train_epochs": 1,
82
+ "max_steps": -1,
83
+ "logging_strategy": "steps",
84
+ "save_strategy": "steps",
85
+ "logging_steps": 0.2,
86
+ "save_steps": 0.2,
87
+ "gradient_accumulation_steps": 1,
88
+ "gradient_checkpointing": false,
89
+ "optim": "adamw_torch",
90
+ "learning_rate": 5e-05,
91
+ "weight_decay": 0.01,
92
+ "warmup_steps": 0,
93
+ "report_to": "tensorboard",
94
+ "save_safetensors": true,
95
+ "packing": false
96
+ },
97
+ "use_peft": true,
98
+ "peft_task_type": "CAUSAL_LM",
99
+ "peft_inference_mode": false,
100
+ "peft_r": 8,
101
+ "peft_lora_alpha": 32,
102
+ "peft_lora_dropout": 0.1,
103
+ "fsdp": "none",
104
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
105
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
106
+ "fsdp_fsdp_min_num_params": 100000000.0,
107
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
108
+ "fsdp_fsdp_forward_prefetch": false,
109
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
110
+ "fsdp_fsdp_offload_params": false,
111
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
112
+ "fsdp_fsdp_use_orig_params": true,
113
+ "fsdp_fsdp_activation_checkpointing": false,
114
+ "fsdp_fsdp_limit_all_gathers": true,
115
+ "fsdp_fsdp_sync_module_states": true,
116
+ "use_git": true,
117
+ "random_seed": 0
118
+ }
119
+ },
120
+ "winogrande": {
121
+ "acc": 0.8271507498026835,
122
+ "acc_stderr": 0.010626964529971864,
123
+ "args": {
124
+ "exp_name": "results/v1/winogrande/LumiOpen/Llama-Poro-2-8B-base",
125
+ "task_name": "winogrande",
126
+ "model_name": "LumiOpen/Llama-Poro-2-8B-base",
127
+ "dataset_param": {
128
+ "max_num_train": 50000,
129
+ "completion_only": false,
130
+ "apply_chat_template": false
131
+ },
132
+ "resume_from_checkpoint": false,
133
+ "train_param": {
134
+ "max_length": 1024,
135
+ "per_device_train_batch_size": 8,
136
+ "auto_find_batch_size": true,
137
+ "num_train_epochs": 1,
138
+ "max_steps": -1,
139
+ "logging_strategy": "steps",
140
+ "save_strategy": "steps",
141
+ "logging_steps": 0.2,
142
+ "save_steps": 0.2,
143
+ "gradient_accumulation_steps": 1,
144
+ "gradient_checkpointing": false,
145
+ "optim": "adamw_torch",
146
+ "learning_rate": 5e-05,
147
+ "weight_decay": 0.01,
148
+ "warmup_steps": 0,
149
+ "report_to": "tensorboard",
150
+ "save_safetensors": true,
151
+ "packing": false
152
+ },
153
+ "use_peft": true,
154
+ "peft_task_type": "CAUSAL_LM",
155
+ "peft_inference_mode": false,
156
+ "peft_r": 8,
157
+ "peft_lora_alpha": 32,
158
+ "peft_lora_dropout": 0.1,
159
+ "fsdp": "none",
160
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
161
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
162
+ "fsdp_fsdp_min_num_params": 100000000.0,
163
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
164
+ "fsdp_fsdp_forward_prefetch": false,
165
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
166
+ "fsdp_fsdp_offload_params": false,
167
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
168
+ "fsdp_fsdp_use_orig_params": true,
169
+ "fsdp_fsdp_activation_checkpointing": false,
170
+ "fsdp_fsdp_limit_all_gathers": true,
171
+ "fsdp_fsdp_sync_module_states": true,
172
+ "use_git": true,
173
+ "random_seed": 0
174
+ }
175
+ },
176
+ "hellaswag": {
177
+ "acc_norm": 0.8243,
178
+ "acc_norm_stderr": 0.0038058375477608357,
179
+ "args": {
180
+ "exp_name": "results/v1/hellaswag/LumiOpen/Llama-Poro-2-8B-base",
181
+ "task_name": "hellaswag",
182
+ "model_name": "LumiOpen/Llama-Poro-2-8B-base",
183
+ "dataset_param": {
184
+ "max_num_train": 50000,
185
+ "completion_only": false,
186
+ "apply_chat_template": false
187
+ },
188
+ "resume_from_checkpoint": false,
189
+ "train_param": {
190
+ "max_length": 1024,
191
+ "per_device_train_batch_size": 8,
192
+ "auto_find_batch_size": true,
193
+ "num_train_epochs": 1,
194
+ "max_steps": -1,
195
+ "logging_strategy": "steps",
196
+ "save_strategy": "steps",
197
+ "logging_steps": 0.2,
198
+ "save_steps": 0.2,
199
+ "gradient_accumulation_steps": 1,
200
+ "gradient_checkpointing": false,
201
+ "optim": "adamw_torch",
202
+ "learning_rate": 5e-05,
203
+ "weight_decay": 0.01,
204
+ "warmup_steps": 0,
205
+ "report_to": "tensorboard",
206
+ "save_safetensors": true,
207
+ "packing": false
208
+ },
209
+ "use_peft": true,
210
+ "peft_task_type": "CAUSAL_LM",
211
+ "peft_inference_mode": false,
212
+ "peft_r": 8,
213
+ "peft_lora_alpha": 32,
214
+ "peft_lora_dropout": 0.1,
215
+ "fsdp": "none",
216
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
217
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
218
+ "fsdp_fsdp_min_num_params": 100000000.0,
219
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
220
+ "fsdp_fsdp_forward_prefetch": false,
221
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
222
+ "fsdp_fsdp_offload_params": false,
223
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
224
+ "fsdp_fsdp_use_orig_params": true,
225
+ "fsdp_fsdp_activation_checkpointing": false,
226
+ "fsdp_fsdp_limit_all_gathers": true,
227
+ "fsdp_fsdp_sync_module_states": true,
228
+ "use_git": true,
229
+ "random_seed": 0
230
+ }
231
+ },
232
+ "social_iqa": {
233
+ "acc": 0.5736949846468782,
234
+ "acc_stderr": 0.011190503463264756,
235
+ "args": {
236
+ "exp_name": "results/v1/social_iqa/LumiOpen/Llama-Poro-2-8B-base",
237
+ "task_name": "social_iqa",
238
+ "model_name": "LumiOpen/Llama-Poro-2-8B-base",
239
+ "dataset_param": {
240
+ "max_num_train": 50000,
241
+ "completion_only": false,
242
+ "apply_chat_template": false
243
+ },
244
+ "resume_from_checkpoint": false,
245
+ "train_param": {
246
+ "max_length": 1024,
247
+ "per_device_train_batch_size": 8,
248
+ "auto_find_batch_size": true,
249
+ "num_train_epochs": 1,
250
+ "max_steps": -1,
251
+ "logging_strategy": "steps",
252
+ "save_strategy": "steps",
253
+ "logging_steps": 0.2,
254
+ "save_steps": 0.2,
255
+ "gradient_accumulation_steps": 1,
256
+ "gradient_checkpointing": false,
257
+ "optim": "adamw_torch",
258
+ "learning_rate": 5e-05,
259
+ "weight_decay": 0.01,
260
+ "warmup_steps": 0,
261
+ "report_to": "tensorboard",
262
+ "save_safetensors": true,
263
+ "packing": false
264
+ },
265
+ "use_peft": true,
266
+ "peft_task_type": "CAUSAL_LM",
267
+ "peft_inference_mode": false,
268
+ "peft_r": 8,
269
+ "peft_lora_alpha": 32,
270
+ "peft_lora_dropout": 0.1,
271
+ "fsdp": "none",
272
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
273
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
274
+ "fsdp_fsdp_min_num_params": 100000000.0,
275
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
276
+ "fsdp_fsdp_forward_prefetch": false,
277
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
278
+ "fsdp_fsdp_offload_params": false,
279
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
280
+ "fsdp_fsdp_use_orig_params": true,
281
+ "fsdp_fsdp_activation_checkpointing": false,
282
+ "fsdp_fsdp_limit_all_gathers": true,
283
+ "fsdp_fsdp_sync_module_states": true,
284
+ "use_git": true,
285
+ "random_seed": 0
286
+ }
287
+ },
288
+ "piqa": {
289
+ "acc_norm": 0.8362350380848749,
290
+ "acc_norm_stderr": 0.00863416007989679,
291
+ "args": {
292
+ "exp_name": "results/v2/piqa/LumiOpen/Llama-Poro-2-8B-base",
293
+ "task_name": "piqa",
294
+ "model_name": "LumiOpen/Llama-Poro-2-8B-base",
295
+ "dataset_param": {
296
+ "max_num_train": 50000,
297
+ "completion_only": false,
298
+ "apply_chat_template": false
299
+ },
300
+ "resume_from_checkpoint": false,
301
+ "train_param": {
302
+ "max_length": 1024,
303
+ "per_device_train_batch_size": 8,
304
+ "auto_find_batch_size": true,
305
+ "num_train_epochs": 5,
306
+ "max_steps": -1,
307
+ "logging_strategy": "epoch",
308
+ "save_strategy": "epoch",
309
+ "logging_steps": 0.2,
310
+ "save_steps": 0.2,
311
+ "gradient_accumulation_steps": 1,
312
+ "gradient_checkpointing": false,
313
+ "optim": "adamw_torch",
314
+ "learning_rate": 5e-05,
315
+ "weight_decay": 0.01,
316
+ "warmup_steps": 0,
317
+ "report_to": "tensorboard",
318
+ "save_safetensors": true,
319
+ "packing": false
320
+ },
321
+ "use_peft": true,
322
+ "peft_task_type": "CAUSAL_LM",
323
+ "peft_inference_mode": false,
324
+ "peft_r": 8,
325
+ "peft_lora_alpha": 32,
326
+ "peft_lora_dropout": 0.1,
327
+ "fsdp": "none",
328
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
329
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
330
+ "fsdp_fsdp_min_num_params": 100000000.0,
331
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
332
+ "fsdp_fsdp_forward_prefetch": false,
333
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
334
+ "fsdp_fsdp_offload_params": false,
335
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
336
+ "fsdp_fsdp_use_orig_params": true,
337
+ "fsdp_fsdp_activation_checkpointing": false,
338
+ "fsdp_fsdp_limit_all_gathers": true,
339
+ "fsdp_fsdp_sync_module_states": true,
340
+ "use_git": true,
341
+ "random_seed": 0
342
+ }
343
+ },
344
+ "commonsense_qa": {
345
+ "acc": 0.800982800982801,
346
+ "acc_stderr": 0.011430809442838396,
347
+ "args": {
348
+ "exp_name": "results/v2/commonsense_qa/LumiOpen/Llama-Poro-2-8B-base",
349
+ "task_name": "commonsense_qa",
350
+ "model_name": "LumiOpen/Llama-Poro-2-8B-base",
351
+ "dataset_param": {
352
+ "max_num_train": 50000,
353
+ "completion_only": false,
354
+ "apply_chat_template": false
355
+ },
356
+ "resume_from_checkpoint": false,
357
+ "train_param": {
358
+ "max_length": 1024,
359
+ "per_device_train_batch_size": 8,
360
+ "auto_find_batch_size": true,
361
+ "num_train_epochs": 5,
362
+ "max_steps": -1,
363
+ "logging_strategy": "epoch",
364
+ "save_strategy": "epoch",
365
+ "logging_steps": 0.2,
366
+ "save_steps": 0.2,
367
+ "gradient_accumulation_steps": 1,
368
+ "gradient_checkpointing": false,
369
+ "optim": "adamw_torch",
370
+ "learning_rate": 5e-05,
371
+ "weight_decay": 0.01,
372
+ "warmup_steps": 0,
373
+ "report_to": "tensorboard",
374
+ "save_safetensors": true,
375
+ "packing": false
376
+ },
377
+ "use_peft": true,
378
+ "peft_task_type": "CAUSAL_LM",
379
+ "peft_inference_mode": false,
380
+ "peft_r": 8,
381
+ "peft_lora_alpha": 32,
382
+ "peft_lora_dropout": 0.1,
383
+ "fsdp": "none",
384
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
385
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
386
+ "fsdp_fsdp_min_num_params": 100000000.0,
387
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
388
+ "fsdp_fsdp_forward_prefetch": false,
389
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
390
+ "fsdp_fsdp_offload_params": false,
391
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
392
+ "fsdp_fsdp_use_orig_params": true,
393
+ "fsdp_fsdp_activation_checkpointing": false,
394
+ "fsdp_fsdp_limit_all_gathers": true,
395
+ "fsdp_fsdp_sync_module_states": true,
396
+ "use_git": true,
397
+ "random_seed": 0
398
+ }
399
+ },
400
+ "gsm8k": {
401
+ "exact_match,flexible-extract": 0.5837755875663382,
402
+ "exact_match,flexible-extract_stderr": 0.013577788334652676,
403
+ "args": {
404
+ "exp_name": "results/v2/gsm8k/LumiOpen/Llama-Poro-2-8B-base",
405
+ "task_name": "gsm8k",
406
+ "model_name": "LumiOpen/Llama-Poro-2-8B-base",
407
+ "dataset_param": {
408
+ "max_num_train": 50000,
409
+ "completion_only": false,
410
+ "apply_chat_template": false
411
+ },
412
+ "resume_from_checkpoint": false,
413
+ "train_param": {
414
+ "max_length": 1024,
415
+ "per_device_train_batch_size": 8,
416
+ "auto_find_batch_size": true,
417
+ "num_train_epochs": 5,
418
+ "max_steps": -1,
419
+ "logging_strategy": "epoch",
420
+ "save_strategy": "epoch",
421
+ "logging_steps": 0.2,
422
+ "save_steps": 0.2,
423
+ "gradient_accumulation_steps": 1,
424
+ "gradient_checkpointing": false,
425
+ "optim": "adamw_torch",
426
+ "learning_rate": 5e-05,
427
+ "weight_decay": 0.01,
428
+ "warmup_steps": 0,
429
+ "report_to": "tensorboard",
430
+ "save_safetensors": true,
431
+ "packing": false
432
+ },
433
+ "use_peft": true,
434
+ "peft_task_type": "CAUSAL_LM",
435
+ "peft_inference_mode": false,
436
+ "peft_r": 8,
437
+ "peft_lora_alpha": 32,
438
+ "peft_lora_dropout": 0.1,
439
+ "fsdp": "none",
440
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
441
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
442
+ "fsdp_fsdp_min_num_params": 100000000.0,
443
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
444
+ "fsdp_fsdp_forward_prefetch": false,
445
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
446
+ "fsdp_fsdp_offload_params": false,
447
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
448
+ "fsdp_fsdp_use_orig_params": true,
449
+ "fsdp_fsdp_activation_checkpointing": false,
450
+ "fsdp_fsdp_limit_all_gathers": true,
451
+ "fsdp_fsdp_sync_module_states": true,
452
+ "use_git": true,
453
+ "random_seed": 0
454
+ }
455
+ },
456
+ "arc_challenge": {
457
+ "acc_norm": 0.6348122866894198,
458
+ "acc_norm_stderr": 0.0140702655192688,
459
+ "args": {
460
+ "exp_name": "results/v2/arc_challenge/LumiOpen/Llama-Poro-2-8B-base",
461
+ "task_name": "arc_challenge",
462
+ "model_name": "LumiOpen/Llama-Poro-2-8B-base",
463
+ "dataset_param": {
464
+ "max_num_train": 50000,
465
+ "completion_only": false,
466
+ "apply_chat_template": false
467
+ },
468
+ "resume_from_checkpoint": false,
469
+ "train_param": {
470
+ "max_length": 1024,
471
+ "per_device_train_batch_size": 8,
472
+ "auto_find_batch_size": true,
473
+ "num_train_epochs": 5,
474
+ "max_steps": -1,
475
+ "logging_strategy": "epoch",
476
+ "save_strategy": "epoch",
477
+ "logging_steps": 0.2,
478
+ "save_steps": 0.2,
479
+ "gradient_accumulation_steps": 1,
480
+ "gradient_checkpointing": false,
481
+ "optim": "adamw_torch",
482
+ "learning_rate": 5e-05,
483
+ "weight_decay": 0.01,
484
+ "warmup_steps": 0,
485
+ "report_to": "tensorboard",
486
+ "save_safetensors": true,
487
+ "packing": false
488
+ },
489
+ "use_peft": true,
490
+ "peft_task_type": "CAUSAL_LM",
491
+ "peft_inference_mode": false,
492
+ "peft_r": 8,
493
+ "peft_lora_alpha": 32,
494
+ "peft_lora_dropout": 0.1,
495
+ "fsdp": "none",
496
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
497
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
498
+ "fsdp_fsdp_min_num_params": 100000000.0,
499
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
500
+ "fsdp_fsdp_forward_prefetch": false,
501
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
502
+ "fsdp_fsdp_offload_params": false,
503
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
504
+ "fsdp_fsdp_use_orig_params": true,
505
+ "fsdp_fsdp_activation_checkpointing": false,
506
+ "fsdp_fsdp_limit_all_gathers": true,
507
+ "fsdp_fsdp_sync_module_states": true,
508
+ "use_git": true,
509
+ "random_seed": 0
510
+ }
511
+ }
512
+ }
513
+ }
Qwen/Qwen1.5-0.5B-Chat/results_2026-01-05-04-27-51.json ADDED
@@ -0,0 +1,513 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "model_name": "Qwen/Qwen1.5-0.5B-Chat",
4
+ "model_type": "instruction-tuned",
5
+ "params": 0.5
6
+ },
7
+ "results": {
8
+ "medmcqa": {
9
+ "acc_norm": 0.3559646186947167,
10
+ "acc_norm_stderr": 0.00740399753936656,
11
+ "args": {
12
+ "exp_name": "results/v1/medmcqa/Qwen/Qwen1.5-0.5B-Chat",
13
+ "task_name": "medmcqa",
14
+ "model_name": "Qwen/Qwen1.5-0.5B-Chat",
15
+ "dataset_param": {
16
+ "max_num_train": 50000,
17
+ "completion_only": false,
18
+ "apply_chat_template": true
19
+ },
20
+ "resume_from_checkpoint": false,
21
+ "train_param": {
22
+ "max_length": 1024,
23
+ "per_device_train_batch_size": 8,
24
+ "auto_find_batch_size": true,
25
+ "num_train_epochs": 1,
26
+ "max_steps": -1,
27
+ "logging_strategy": "steps",
28
+ "save_strategy": "steps",
29
+ "logging_steps": 0.2,
30
+ "save_steps": 0.2,
31
+ "gradient_accumulation_steps": 1,
32
+ "gradient_checkpointing": false,
33
+ "optim": "adamw_torch",
34
+ "learning_rate": 5e-05,
35
+ "weight_decay": 0.01,
36
+ "warmup_steps": 0,
37
+ "report_to": "tensorboard",
38
+ "save_safetensors": true,
39
+ "packing": false
40
+ },
41
+ "use_peft": true,
42
+ "peft_task_type": "CAUSAL_LM",
43
+ "peft_inference_mode": false,
44
+ "peft_r": 8,
45
+ "peft_lora_alpha": 32,
46
+ "peft_lora_dropout": 0.1,
47
+ "fsdp": "none",
48
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
49
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
50
+ "fsdp_fsdp_min_num_params": 100000000.0,
51
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
52
+ "fsdp_fsdp_forward_prefetch": false,
53
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
54
+ "fsdp_fsdp_offload_params": false,
55
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
56
+ "fsdp_fsdp_use_orig_params": true,
57
+ "fsdp_fsdp_activation_checkpointing": false,
58
+ "fsdp_fsdp_limit_all_gathers": true,
59
+ "fsdp_fsdp_sync_module_states": true,
60
+ "use_git": true,
61
+ "random_seed": 0
62
+ }
63
+ },
64
+ "nq_open": {
65
+ "exact_match,remove_whitespace": 0.05789473684210526,
66
+ "exact_match,remove_whitespace_stderr": 0.003887548463099522,
67
+ "args": {
68
+ "exp_name": "results/v1/nq_open/Qwen/Qwen1.5-0.5B-Chat",
69
+ "task_name": "nq_open",
70
+ "model_name": "Qwen/Qwen1.5-0.5B-Chat",
71
+ "dataset_param": {
72
+ "max_num_train": 50000,
73
+ "completion_only": false,
74
+ "apply_chat_template": true
75
+ },
76
+ "resume_from_checkpoint": false,
77
+ "train_param": {
78
+ "max_length": 1024,
79
+ "per_device_train_batch_size": 8,
80
+ "auto_find_batch_size": true,
81
+ "num_train_epochs": 1,
82
+ "max_steps": -1,
83
+ "logging_strategy": "steps",
84
+ "save_strategy": "steps",
85
+ "logging_steps": 0.2,
86
+ "save_steps": 0.2,
87
+ "gradient_accumulation_steps": 1,
88
+ "gradient_checkpointing": false,
89
+ "optim": "adamw_torch",
90
+ "learning_rate": 5e-05,
91
+ "weight_decay": 0.01,
92
+ "warmup_steps": 0,
93
+ "report_to": "tensorboard",
94
+ "save_safetensors": true,
95
+ "packing": false
96
+ },
97
+ "use_peft": true,
98
+ "peft_task_type": "CAUSAL_LM",
99
+ "peft_inference_mode": false,
100
+ "peft_r": 8,
101
+ "peft_lora_alpha": 32,
102
+ "peft_lora_dropout": 0.1,
103
+ "fsdp": "none",
104
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
105
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
106
+ "fsdp_fsdp_min_num_params": 100000000.0,
107
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
108
+ "fsdp_fsdp_forward_prefetch": false,
109
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
110
+ "fsdp_fsdp_offload_params": false,
111
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
112
+ "fsdp_fsdp_use_orig_params": true,
113
+ "fsdp_fsdp_activation_checkpointing": false,
114
+ "fsdp_fsdp_limit_all_gathers": true,
115
+ "fsdp_fsdp_sync_module_states": true,
116
+ "use_git": true,
117
+ "random_seed": 0
118
+ }
119
+ },
120
+ "winogrande": {
121
+ "acc": 0.5864246250986582,
122
+ "acc_stderr": 0.013840971763195304,
123
+ "args": {
124
+ "exp_name": "results/v1/winogrande/Qwen/Qwen1.5-0.5B-Chat",
125
+ "task_name": "winogrande",
126
+ "model_name": "Qwen/Qwen1.5-0.5B-Chat",
127
+ "dataset_param": {
128
+ "max_num_train": 50000,
129
+ "completion_only": false,
130
+ "apply_chat_template": true
131
+ },
132
+ "resume_from_checkpoint": false,
133
+ "train_param": {
134
+ "max_length": 1024,
135
+ "per_device_train_batch_size": 8,
136
+ "auto_find_batch_size": true,
137
+ "num_train_epochs": 1,
138
+ "max_steps": -1,
139
+ "logging_strategy": "steps",
140
+ "save_strategy": "steps",
141
+ "logging_steps": 0.2,
142
+ "save_steps": 0.2,
143
+ "gradient_accumulation_steps": 1,
144
+ "gradient_checkpointing": false,
145
+ "optim": "adamw_torch",
146
+ "learning_rate": 5e-05,
147
+ "weight_decay": 0.01,
148
+ "warmup_steps": 0,
149
+ "report_to": "tensorboard",
150
+ "save_safetensors": true,
151
+ "packing": false
152
+ },
153
+ "use_peft": true,
154
+ "peft_task_type": "CAUSAL_LM",
155
+ "peft_inference_mode": false,
156
+ "peft_r": 8,
157
+ "peft_lora_alpha": 32,
158
+ "peft_lora_dropout": 0.1,
159
+ "fsdp": "none",
160
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
161
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
162
+ "fsdp_fsdp_min_num_params": 100000000.0,
163
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
164
+ "fsdp_fsdp_forward_prefetch": false,
165
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
166
+ "fsdp_fsdp_offload_params": false,
167
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
168
+ "fsdp_fsdp_use_orig_params": true,
169
+ "fsdp_fsdp_activation_checkpointing": false,
170
+ "fsdp_fsdp_limit_all_gathers": true,
171
+ "fsdp_fsdp_sync_module_states": true,
172
+ "use_git": true,
173
+ "random_seed": 0
174
+ }
175
+ },
176
+ "hellaswag": {
177
+ "acc_norm": 0.4782,
178
+ "acc_norm_stderr": 0.004995495120320984,
179
+ "args": {
180
+ "exp_name": "results/v1/hellaswag/Qwen/Qwen1.5-0.5B-Chat",
181
+ "task_name": "hellaswag",
182
+ "model_name": "Qwen/Qwen1.5-0.5B-Chat",
183
+ "dataset_param": {
184
+ "max_num_train": 50000,
185
+ "completion_only": false,
186
+ "apply_chat_template": true
187
+ },
188
+ "resume_from_checkpoint": false,
189
+ "train_param": {
190
+ "max_length": 1024,
191
+ "per_device_train_batch_size": 8,
192
+ "auto_find_batch_size": true,
193
+ "num_train_epochs": 1,
194
+ "max_steps": -1,
195
+ "logging_strategy": "steps",
196
+ "save_strategy": "steps",
197
+ "logging_steps": 0.2,
198
+ "save_steps": 0.2,
199
+ "gradient_accumulation_steps": 1,
200
+ "gradient_checkpointing": false,
201
+ "optim": "adamw_torch",
202
+ "learning_rate": 5e-05,
203
+ "weight_decay": 0.01,
204
+ "warmup_steps": 0,
205
+ "report_to": "tensorboard",
206
+ "save_safetensors": true,
207
+ "packing": false
208
+ },
209
+ "use_peft": true,
210
+ "peft_task_type": "CAUSAL_LM",
211
+ "peft_inference_mode": false,
212
+ "peft_r": 8,
213
+ "peft_lora_alpha": 32,
214
+ "peft_lora_dropout": 0.1,
215
+ "fsdp": "none",
216
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
217
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
218
+ "fsdp_fsdp_min_num_params": 100000000.0,
219
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
220
+ "fsdp_fsdp_forward_prefetch": false,
221
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
222
+ "fsdp_fsdp_offload_params": false,
223
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
224
+ "fsdp_fsdp_use_orig_params": true,
225
+ "fsdp_fsdp_activation_checkpointing": false,
226
+ "fsdp_fsdp_limit_all_gathers": true,
227
+ "fsdp_fsdp_sync_module_states": true,
228
+ "use_git": true,
229
+ "random_seed": 0
230
+ }
231
+ },
232
+ "social_iqa": {
233
+ "acc": 0.458546571136131,
234
+ "acc_stderr": 0.01127511960864091,
235
+ "args": {
236
+ "exp_name": "results/v1/social_iqa/Qwen/Qwen1.5-0.5B-Chat",
237
+ "task_name": "social_iqa",
238
+ "model_name": "Qwen/Qwen1.5-0.5B-Chat",
239
+ "dataset_param": {
240
+ "max_num_train": 50000,
241
+ "completion_only": false,
242
+ "apply_chat_template": true
243
+ },
244
+ "resume_from_checkpoint": false,
245
+ "train_param": {
246
+ "max_length": 1024,
247
+ "per_device_train_batch_size": 8,
248
+ "auto_find_batch_size": true,
249
+ "num_train_epochs": 1,
250
+ "max_steps": -1,
251
+ "logging_strategy": "steps",
252
+ "save_strategy": "steps",
253
+ "logging_steps": 0.2,
254
+ "save_steps": 0.2,
255
+ "gradient_accumulation_steps": 1,
256
+ "gradient_checkpointing": false,
257
+ "optim": "adamw_torch",
258
+ "learning_rate": 5e-05,
259
+ "weight_decay": 0.01,
260
+ "warmup_steps": 0,
261
+ "report_to": "tensorboard",
262
+ "save_safetensors": true,
263
+ "packing": false
264
+ },
265
+ "use_peft": true,
266
+ "peft_task_type": "CAUSAL_LM",
267
+ "peft_inference_mode": false,
268
+ "peft_r": 8,
269
+ "peft_lora_alpha": 32,
270
+ "peft_lora_dropout": 0.1,
271
+ "fsdp": "none",
272
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
273
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
274
+ "fsdp_fsdp_min_num_params": 100000000.0,
275
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
276
+ "fsdp_fsdp_forward_prefetch": false,
277
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
278
+ "fsdp_fsdp_offload_params": false,
279
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
280
+ "fsdp_fsdp_use_orig_params": true,
281
+ "fsdp_fsdp_activation_checkpointing": false,
282
+ "fsdp_fsdp_limit_all_gathers": true,
283
+ "fsdp_fsdp_sync_module_states": true,
284
+ "use_git": true,
285
+ "random_seed": 0
286
+ }
287
+ },
288
+ "piqa": {
289
+ "acc_norm": 0.7013057671381937,
290
+ "acc_norm_stderr": 0.010678556398149236,
291
+ "args": {
292
+ "exp_name": "results/v2/piqa/Qwen/Qwen1.5-0.5B-Chat",
293
+ "task_name": "piqa",
294
+ "model_name": "Qwen/Qwen1.5-0.5B-Chat",
295
+ "dataset_param": {
296
+ "max_num_train": 50000,
297
+ "completion_only": false,
298
+ "apply_chat_template": true
299
+ },
300
+ "resume_from_checkpoint": false,
301
+ "train_param": {
302
+ "max_length": 1024,
303
+ "per_device_train_batch_size": 8,
304
+ "auto_find_batch_size": true,
305
+ "num_train_epochs": 5,
306
+ "max_steps": -1,
307
+ "logging_strategy": "epoch",
308
+ "save_strategy": "epoch",
309
+ "logging_steps": 0.2,
310
+ "save_steps": 0.2,
311
+ "gradient_accumulation_steps": 1,
312
+ "gradient_checkpointing": false,
313
+ "optim": "adamw_torch",
314
+ "learning_rate": 5e-05,
315
+ "weight_decay": 0.01,
316
+ "warmup_steps": 0,
317
+ "report_to": "tensorboard",
318
+ "save_safetensors": true,
319
+ "packing": false
320
+ },
321
+ "use_peft": true,
322
+ "peft_task_type": "CAUSAL_LM",
323
+ "peft_inference_mode": false,
324
+ "peft_r": 8,
325
+ "peft_lora_alpha": 32,
326
+ "peft_lora_dropout": 0.1,
327
+ "fsdp": "none",
328
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
329
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
330
+ "fsdp_fsdp_min_num_params": 100000000.0,
331
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
332
+ "fsdp_fsdp_forward_prefetch": false,
333
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
334
+ "fsdp_fsdp_offload_params": false,
335
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
336
+ "fsdp_fsdp_use_orig_params": true,
337
+ "fsdp_fsdp_activation_checkpointing": false,
338
+ "fsdp_fsdp_limit_all_gathers": true,
339
+ "fsdp_fsdp_sync_module_states": true,
340
+ "use_git": true,
341
+ "random_seed": 0
342
+ }
343
+ },
344
+ "commonsense_qa": {
345
+ "acc": 0.5724815724815725,
346
+ "acc_stderr": 0.01416375024578871,
347
+ "args": {
348
+ "exp_name": "results/v2/commonsense_qa/Qwen/Qwen1.5-0.5B-Chat",
349
+ "task_name": "commonsense_qa",
350
+ "model_name": "Qwen/Qwen1.5-0.5B-Chat",
351
+ "dataset_param": {
352
+ "max_num_train": 50000,
353
+ "completion_only": false,
354
+ "apply_chat_template": true
355
+ },
356
+ "resume_from_checkpoint": false,
357
+ "train_param": {
358
+ "max_length": 1024,
359
+ "per_device_train_batch_size": 8,
360
+ "auto_find_batch_size": true,
361
+ "num_train_epochs": 5,
362
+ "max_steps": -1,
363
+ "logging_strategy": "epoch",
364
+ "save_strategy": "epoch",
365
+ "logging_steps": 0.2,
366
+ "save_steps": 0.2,
367
+ "gradient_accumulation_steps": 1,
368
+ "gradient_checkpointing": false,
369
+ "optim": "adamw_torch",
370
+ "learning_rate": 5e-05,
371
+ "weight_decay": 0.01,
372
+ "warmup_steps": 0,
373
+ "report_to": "tensorboard",
374
+ "save_safetensors": true,
375
+ "packing": false
376
+ },
377
+ "use_peft": true,
378
+ "peft_task_type": "CAUSAL_LM",
379
+ "peft_inference_mode": false,
380
+ "peft_r": 8,
381
+ "peft_lora_alpha": 32,
382
+ "peft_lora_dropout": 0.1,
383
+ "fsdp": "none",
384
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
385
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
386
+ "fsdp_fsdp_min_num_params": 100000000.0,
387
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
388
+ "fsdp_fsdp_forward_prefetch": false,
389
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
390
+ "fsdp_fsdp_offload_params": false,
391
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
392
+ "fsdp_fsdp_use_orig_params": true,
393
+ "fsdp_fsdp_activation_checkpointing": false,
394
+ "fsdp_fsdp_limit_all_gathers": true,
395
+ "fsdp_fsdp_sync_module_states": true,
396
+ "use_git": true,
397
+ "random_seed": 0
398
+ }
399
+ },
400
+ "gsm8k": {
401
+ "exact_match,flexible-extract": 0.15314632297194844,
402
+ "exact_match,flexible-extract_stderr": 0.009919728152791484,
403
+ "args": {
404
+ "exp_name": "results/v2/gsm8k/Qwen/Qwen1.5-0.5B-Chat",
405
+ "task_name": "gsm8k",
406
+ "model_name": "Qwen/Qwen1.5-0.5B-Chat",
407
+ "dataset_param": {
408
+ "max_num_train": 50000,
409
+ "completion_only": false,
410
+ "apply_chat_template": true
411
+ },
412
+ "resume_from_checkpoint": false,
413
+ "train_param": {
414
+ "max_length": 1024,
415
+ "per_device_train_batch_size": 8,
416
+ "auto_find_batch_size": true,
417
+ "num_train_epochs": 5,
418
+ "max_steps": -1,
419
+ "logging_strategy": "epoch",
420
+ "save_strategy": "epoch",
421
+ "logging_steps": 0.2,
422
+ "save_steps": 0.2,
423
+ "gradient_accumulation_steps": 1,
424
+ "gradient_checkpointing": false,
425
+ "optim": "adamw_torch",
426
+ "learning_rate": 5e-05,
427
+ "weight_decay": 0.01,
428
+ "warmup_steps": 0,
429
+ "report_to": "tensorboard",
430
+ "save_safetensors": true,
431
+ "packing": false
432
+ },
433
+ "use_peft": true,
434
+ "peft_task_type": "CAUSAL_LM",
435
+ "peft_inference_mode": false,
436
+ "peft_r": 8,
437
+ "peft_lora_alpha": 32,
438
+ "peft_lora_dropout": 0.1,
439
+ "fsdp": "none",
440
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
441
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
442
+ "fsdp_fsdp_min_num_params": 100000000.0,
443
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
444
+ "fsdp_fsdp_forward_prefetch": false,
445
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
446
+ "fsdp_fsdp_offload_params": false,
447
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
448
+ "fsdp_fsdp_use_orig_params": true,
449
+ "fsdp_fsdp_activation_checkpointing": false,
450
+ "fsdp_fsdp_limit_all_gathers": true,
451
+ "fsdp_fsdp_sync_module_states": true,
452
+ "use_git": true,
453
+ "random_seed": 0
454
+ }
455
+ },
456
+ "arc_challenge": {
457
+ "acc_norm": 0.28668941979522183,
458
+ "acc_norm_stderr": 0.013214986329274774,
459
+ "args": {
460
+ "exp_name": "results/v2/arc_challenge/Qwen/Qwen1.5-0.5B-Chat",
461
+ "task_name": "arc_challenge",
462
+ "model_name": "Qwen/Qwen1.5-0.5B-Chat",
463
+ "dataset_param": {
464
+ "max_num_train": 50000,
465
+ "completion_only": false,
466
+ "apply_chat_template": true
467
+ },
468
+ "resume_from_checkpoint": false,
469
+ "train_param": {
470
+ "max_length": 1024,
471
+ "per_device_train_batch_size": 8,
472
+ "auto_find_batch_size": true,
473
+ "num_train_epochs": 5,
474
+ "max_steps": -1,
475
+ "logging_strategy": "epoch",
476
+ "save_strategy": "epoch",
477
+ "logging_steps": 0.2,
478
+ "save_steps": 0.2,
479
+ "gradient_accumulation_steps": 1,
480
+ "gradient_checkpointing": false,
481
+ "optim": "adamw_torch",
482
+ "learning_rate": 5e-05,
483
+ "weight_decay": 0.01,
484
+ "warmup_steps": 0,
485
+ "report_to": "tensorboard",
486
+ "save_safetensors": true,
487
+ "packing": false
488
+ },
489
+ "use_peft": true,
490
+ "peft_task_type": "CAUSAL_LM",
491
+ "peft_inference_mode": false,
492
+ "peft_r": 8,
493
+ "peft_lora_alpha": 32,
494
+ "peft_lora_dropout": 0.1,
495
+ "fsdp": "none",
496
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
497
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
498
+ "fsdp_fsdp_min_num_params": 100000000.0,
499
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
500
+ "fsdp_fsdp_forward_prefetch": false,
501
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
502
+ "fsdp_fsdp_offload_params": false,
503
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
504
+ "fsdp_fsdp_use_orig_params": true,
505
+ "fsdp_fsdp_activation_checkpointing": false,
506
+ "fsdp_fsdp_limit_all_gathers": true,
507
+ "fsdp_fsdp_sync_module_states": true,
508
+ "use_git": true,
509
+ "random_seed": 0
510
+ }
511
+ }
512
+ }
513
+ }
Qwen/Qwen1.5-0.5B/results_2026-01-05-04-27-50.json ADDED
@@ -0,0 +1,513 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "model_name": "Qwen/Qwen1.5-0.5B",
4
+ "model_type": "pretrained",
5
+ "params": 0.5
6
+ },
7
+ "results": {
8
+ "medmcqa": {
9
+ "acc_norm": 0.3662443222567535,
10
+ "acc_norm_stderr": 0.007449967320266241,
11
+ "args": {
12
+ "exp_name": "results/v1/medmcqa/Qwen/Qwen1.5-0.5B",
13
+ "task_name": "medmcqa",
14
+ "model_name": "Qwen/Qwen1.5-0.5B",
15
+ "dataset_param": {
16
+ "max_num_train": 50000,
17
+ "completion_only": false,
18
+ "apply_chat_template": false
19
+ },
20
+ "resume_from_checkpoint": false,
21
+ "train_param": {
22
+ "max_length": 1024,
23
+ "per_device_train_batch_size": 8,
24
+ "auto_find_batch_size": true,
25
+ "num_train_epochs": 1,
26
+ "max_steps": -1,
27
+ "logging_strategy": "steps",
28
+ "save_strategy": "steps",
29
+ "logging_steps": 0.2,
30
+ "save_steps": 0.2,
31
+ "gradient_accumulation_steps": 1,
32
+ "gradient_checkpointing": false,
33
+ "optim": "adamw_torch",
34
+ "learning_rate": 5e-05,
35
+ "weight_decay": 0.01,
36
+ "warmup_steps": 0,
37
+ "report_to": "tensorboard",
38
+ "save_safetensors": true,
39
+ "packing": false
40
+ },
41
+ "use_peft": true,
42
+ "peft_task_type": "CAUSAL_LM",
43
+ "peft_inference_mode": false,
44
+ "peft_r": 8,
45
+ "peft_lora_alpha": 32,
46
+ "peft_lora_dropout": 0.1,
47
+ "fsdp": "none",
48
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
49
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
50
+ "fsdp_fsdp_min_num_params": 100000000.0,
51
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
52
+ "fsdp_fsdp_forward_prefetch": false,
53
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
54
+ "fsdp_fsdp_offload_params": false,
55
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
56
+ "fsdp_fsdp_use_orig_params": true,
57
+ "fsdp_fsdp_activation_checkpointing": false,
58
+ "fsdp_fsdp_limit_all_gathers": true,
59
+ "fsdp_fsdp_sync_module_states": true,
60
+ "use_git": true,
61
+ "random_seed": 0
62
+ }
63
+ },
64
+ "nq_open": {
65
+ "exact_match,remove_whitespace": 0.07340720221606649,
66
+ "exact_match,remove_whitespace_stderr": 0.004341305566323645,
67
+ "args": {
68
+ "exp_name": "results/v1/nq_open/Qwen/Qwen1.5-0.5B",
69
+ "task_name": "nq_open",
70
+ "model_name": "Qwen/Qwen1.5-0.5B",
71
+ "dataset_param": {
72
+ "max_num_train": 50000,
73
+ "completion_only": false,
74
+ "apply_chat_template": false
75
+ },
76
+ "resume_from_checkpoint": false,
77
+ "train_param": {
78
+ "max_length": 1024,
79
+ "per_device_train_batch_size": 8,
80
+ "auto_find_batch_size": true,
81
+ "num_train_epochs": 1,
82
+ "max_steps": -1,
83
+ "logging_strategy": "steps",
84
+ "save_strategy": "steps",
85
+ "logging_steps": 0.2,
86
+ "save_steps": 0.2,
87
+ "gradient_accumulation_steps": 1,
88
+ "gradient_checkpointing": false,
89
+ "optim": "adamw_torch",
90
+ "learning_rate": 5e-05,
91
+ "weight_decay": 0.01,
92
+ "warmup_steps": 0,
93
+ "report_to": "tensorboard",
94
+ "save_safetensors": true,
95
+ "packing": false
96
+ },
97
+ "use_peft": true,
98
+ "peft_task_type": "CAUSAL_LM",
99
+ "peft_inference_mode": false,
100
+ "peft_r": 8,
101
+ "peft_lora_alpha": 32,
102
+ "peft_lora_dropout": 0.1,
103
+ "fsdp": "none",
104
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
105
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
106
+ "fsdp_fsdp_min_num_params": 100000000.0,
107
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
108
+ "fsdp_fsdp_forward_prefetch": false,
109
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
110
+ "fsdp_fsdp_offload_params": false,
111
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
112
+ "fsdp_fsdp_use_orig_params": true,
113
+ "fsdp_fsdp_activation_checkpointing": false,
114
+ "fsdp_fsdp_limit_all_gathers": true,
115
+ "fsdp_fsdp_sync_module_states": true,
116
+ "use_git": true,
117
+ "random_seed": 0
118
+ }
119
+ },
120
+ "winogrande": {
121
+ "acc": 0.5951065509076559,
122
+ "acc_stderr": 0.01379592700312494,
123
+ "args": {
124
+ "exp_name": "results/v1/winogrande/Qwen/Qwen1.5-0.5B",
125
+ "task_name": "winogrande",
126
+ "model_name": "Qwen/Qwen1.5-0.5B",
127
+ "dataset_param": {
128
+ "max_num_train": 50000,
129
+ "completion_only": false,
130
+ "apply_chat_template": false
131
+ },
132
+ "resume_from_checkpoint": false,
133
+ "train_param": {
134
+ "max_length": 1024,
135
+ "per_device_train_batch_size": 8,
136
+ "auto_find_batch_size": true,
137
+ "num_train_epochs": 1,
138
+ "max_steps": -1,
139
+ "logging_strategy": "steps",
140
+ "save_strategy": "steps",
141
+ "logging_steps": 0.2,
142
+ "save_steps": 0.2,
143
+ "gradient_accumulation_steps": 1,
144
+ "gradient_checkpointing": false,
145
+ "optim": "adamw_torch",
146
+ "learning_rate": 5e-05,
147
+ "weight_decay": 0.01,
148
+ "warmup_steps": 0,
149
+ "report_to": "tensorboard",
150
+ "save_safetensors": true,
151
+ "packing": false
152
+ },
153
+ "use_peft": true,
154
+ "peft_task_type": "CAUSAL_LM",
155
+ "peft_inference_mode": false,
156
+ "peft_r": 8,
157
+ "peft_lora_alpha": 32,
158
+ "peft_lora_dropout": 0.1,
159
+ "fsdp": "none",
160
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
161
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
162
+ "fsdp_fsdp_min_num_params": 100000000.0,
163
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
164
+ "fsdp_fsdp_forward_prefetch": false,
165
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
166
+ "fsdp_fsdp_offload_params": false,
167
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
168
+ "fsdp_fsdp_use_orig_params": true,
169
+ "fsdp_fsdp_activation_checkpointing": false,
170
+ "fsdp_fsdp_limit_all_gathers": true,
171
+ "fsdp_fsdp_sync_module_states": true,
172
+ "use_git": true,
173
+ "random_seed": 0
174
+ }
175
+ },
176
+ "hellaswag": {
177
+ "acc_norm": 0.5139,
178
+ "acc_norm_stderr": 0.004998317448675236,
179
+ "args": {
180
+ "exp_name": "results/v1/hellaswag/Qwen/Qwen1.5-0.5B",
181
+ "task_name": "hellaswag",
182
+ "model_name": "Qwen/Qwen1.5-0.5B",
183
+ "dataset_param": {
184
+ "max_num_train": 50000,
185
+ "completion_only": false,
186
+ "apply_chat_template": false
187
+ },
188
+ "resume_from_checkpoint": false,
189
+ "train_param": {
190
+ "max_length": 1024,
191
+ "per_device_train_batch_size": 8,
192
+ "auto_find_batch_size": true,
193
+ "num_train_epochs": 1,
194
+ "max_steps": -1,
195
+ "logging_strategy": "steps",
196
+ "save_strategy": "steps",
197
+ "logging_steps": 0.2,
198
+ "save_steps": 0.2,
199
+ "gradient_accumulation_steps": 1,
200
+ "gradient_checkpointing": false,
201
+ "optim": "adamw_torch",
202
+ "learning_rate": 5e-05,
203
+ "weight_decay": 0.01,
204
+ "warmup_steps": 0,
205
+ "report_to": "tensorboard",
206
+ "save_safetensors": true,
207
+ "packing": false
208
+ },
209
+ "use_peft": true,
210
+ "peft_task_type": "CAUSAL_LM",
211
+ "peft_inference_mode": false,
212
+ "peft_r": 8,
213
+ "peft_lora_alpha": 32,
214
+ "peft_lora_dropout": 0.1,
215
+ "fsdp": "none",
216
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
217
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
218
+ "fsdp_fsdp_min_num_params": 100000000.0,
219
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
220
+ "fsdp_fsdp_forward_prefetch": false,
221
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
222
+ "fsdp_fsdp_offload_params": false,
223
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
224
+ "fsdp_fsdp_use_orig_params": true,
225
+ "fsdp_fsdp_activation_checkpointing": false,
226
+ "fsdp_fsdp_limit_all_gathers": true,
227
+ "fsdp_fsdp_sync_module_states": true,
228
+ "use_git": true,
229
+ "random_seed": 0
230
+ }
231
+ },
232
+ "social_iqa": {
233
+ "acc": 0.4759467758444217,
234
+ "acc_stderr": 0.011300971289127737,
235
+ "args": {
236
+ "exp_name": "results/v1/social_iqa/Qwen/Qwen1.5-0.5B",
237
+ "task_name": "social_iqa",
238
+ "model_name": "Qwen/Qwen1.5-0.5B",
239
+ "dataset_param": {
240
+ "max_num_train": 50000,
241
+ "completion_only": false,
242
+ "apply_chat_template": false
243
+ },
244
+ "resume_from_checkpoint": false,
245
+ "train_param": {
246
+ "max_length": 1024,
247
+ "per_device_train_batch_size": 8,
248
+ "auto_find_batch_size": true,
249
+ "num_train_epochs": 1,
250
+ "max_steps": -1,
251
+ "logging_strategy": "steps",
252
+ "save_strategy": "steps",
253
+ "logging_steps": 0.2,
254
+ "save_steps": 0.2,
255
+ "gradient_accumulation_steps": 1,
256
+ "gradient_checkpointing": false,
257
+ "optim": "adamw_torch",
258
+ "learning_rate": 5e-05,
259
+ "weight_decay": 0.01,
260
+ "warmup_steps": 0,
261
+ "report_to": "tensorboard",
262
+ "save_safetensors": true,
263
+ "packing": false
264
+ },
265
+ "use_peft": true,
266
+ "peft_task_type": "CAUSAL_LM",
267
+ "peft_inference_mode": false,
268
+ "peft_r": 8,
269
+ "peft_lora_alpha": 32,
270
+ "peft_lora_dropout": 0.1,
271
+ "fsdp": "none",
272
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
273
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
274
+ "fsdp_fsdp_min_num_params": 100000000.0,
275
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
276
+ "fsdp_fsdp_forward_prefetch": false,
277
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
278
+ "fsdp_fsdp_offload_params": false,
279
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
280
+ "fsdp_fsdp_use_orig_params": true,
281
+ "fsdp_fsdp_activation_checkpointing": false,
282
+ "fsdp_fsdp_limit_all_gathers": true,
283
+ "fsdp_fsdp_sync_module_states": true,
284
+ "use_git": true,
285
+ "random_seed": 0
286
+ }
287
+ },
288
+ "piqa": {
289
+ "acc_norm": 0.7138193688792165,
290
+ "acc_norm_stderr": 0.010545318576106645,
291
+ "args": {
292
+ "exp_name": "results/v2/piqa/Qwen/Qwen1.5-0.5B",
293
+ "task_name": "piqa",
294
+ "model_name": "Qwen/Qwen1.5-0.5B",
295
+ "dataset_param": {
296
+ "max_num_train": 50000,
297
+ "completion_only": false,
298
+ "apply_chat_template": false
299
+ },
300
+ "resume_from_checkpoint": false,
301
+ "train_param": {
302
+ "max_length": 1024,
303
+ "per_device_train_batch_size": 8,
304
+ "auto_find_batch_size": true,
305
+ "num_train_epochs": 5,
306
+ "max_steps": -1,
307
+ "logging_strategy": "epoch",
308
+ "save_strategy": "epoch",
309
+ "logging_steps": 0.2,
310
+ "save_steps": 0.2,
311
+ "gradient_accumulation_steps": 1,
312
+ "gradient_checkpointing": false,
313
+ "optim": "adamw_torch",
314
+ "learning_rate": 5e-05,
315
+ "weight_decay": 0.01,
316
+ "warmup_steps": 0,
317
+ "report_to": "tensorboard",
318
+ "save_safetensors": true,
319
+ "packing": false
320
+ },
321
+ "use_peft": true,
322
+ "peft_task_type": "CAUSAL_LM",
323
+ "peft_inference_mode": false,
324
+ "peft_r": 8,
325
+ "peft_lora_alpha": 32,
326
+ "peft_lora_dropout": 0.1,
327
+ "fsdp": "none",
328
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
329
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
330
+ "fsdp_fsdp_min_num_params": 100000000.0,
331
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
332
+ "fsdp_fsdp_forward_prefetch": false,
333
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
334
+ "fsdp_fsdp_offload_params": false,
335
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
336
+ "fsdp_fsdp_use_orig_params": true,
337
+ "fsdp_fsdp_activation_checkpointing": false,
338
+ "fsdp_fsdp_limit_all_gathers": true,
339
+ "fsdp_fsdp_sync_module_states": true,
340
+ "use_git": true,
341
+ "random_seed": 0
342
+ }
343
+ },
344
+ "commonsense_qa": {
345
+ "acc": 0.5864045864045864,
346
+ "acc_stderr": 0.014099594443889911,
347
+ "args": {
348
+ "exp_name": "results/v2/commonsense_qa/Qwen/Qwen1.5-0.5B",
349
+ "task_name": "commonsense_qa",
350
+ "model_name": "Qwen/Qwen1.5-0.5B",
351
+ "dataset_param": {
352
+ "max_num_train": 50000,
353
+ "completion_only": false,
354
+ "apply_chat_template": false
355
+ },
356
+ "resume_from_checkpoint": false,
357
+ "train_param": {
358
+ "max_length": 1024,
359
+ "per_device_train_batch_size": 8,
360
+ "auto_find_batch_size": true,
361
+ "num_train_epochs": 5,
362
+ "max_steps": -1,
363
+ "logging_strategy": "epoch",
364
+ "save_strategy": "epoch",
365
+ "logging_steps": 0.2,
366
+ "save_steps": 0.2,
367
+ "gradient_accumulation_steps": 1,
368
+ "gradient_checkpointing": false,
369
+ "optim": "adamw_torch",
370
+ "learning_rate": 5e-05,
371
+ "weight_decay": 0.01,
372
+ "warmup_steps": 0,
373
+ "report_to": "tensorboard",
374
+ "save_safetensors": true,
375
+ "packing": false
376
+ },
377
+ "use_peft": true,
378
+ "peft_task_type": "CAUSAL_LM",
379
+ "peft_inference_mode": false,
380
+ "peft_r": 8,
381
+ "peft_lora_alpha": 32,
382
+ "peft_lora_dropout": 0.1,
383
+ "fsdp": "none",
384
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
385
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
386
+ "fsdp_fsdp_min_num_params": 100000000.0,
387
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
388
+ "fsdp_fsdp_forward_prefetch": false,
389
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
390
+ "fsdp_fsdp_offload_params": false,
391
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
392
+ "fsdp_fsdp_use_orig_params": true,
393
+ "fsdp_fsdp_activation_checkpointing": false,
394
+ "fsdp_fsdp_limit_all_gathers": true,
395
+ "fsdp_fsdp_sync_module_states": true,
396
+ "use_git": true,
397
+ "random_seed": 0
398
+ }
399
+ },
400
+ "gsm8k": {
401
+ "exact_match,flexible-extract": 0.22744503411675512,
402
+ "exact_match,flexible-extract_stderr": 0.011546363312548094,
403
+ "args": {
404
+ "exp_name": "results/v2/gsm8k/Qwen/Qwen1.5-0.5B",
405
+ "task_name": "gsm8k",
406
+ "model_name": "Qwen/Qwen1.5-0.5B",
407
+ "dataset_param": {
408
+ "max_num_train": 50000,
409
+ "completion_only": false,
410
+ "apply_chat_template": false
411
+ },
412
+ "resume_from_checkpoint": false,
413
+ "train_param": {
414
+ "max_length": 1024,
415
+ "per_device_train_batch_size": 8,
416
+ "auto_find_batch_size": true,
417
+ "num_train_epochs": 5,
418
+ "max_steps": -1,
419
+ "logging_strategy": "epoch",
420
+ "save_strategy": "epoch",
421
+ "logging_steps": 0.2,
422
+ "save_steps": 0.2,
423
+ "gradient_accumulation_steps": 1,
424
+ "gradient_checkpointing": false,
425
+ "optim": "adamw_torch",
426
+ "learning_rate": 5e-05,
427
+ "weight_decay": 0.01,
428
+ "warmup_steps": 0,
429
+ "report_to": "tensorboard",
430
+ "save_safetensors": true,
431
+ "packing": false
432
+ },
433
+ "use_peft": true,
434
+ "peft_task_type": "CAUSAL_LM",
435
+ "peft_inference_mode": false,
436
+ "peft_r": 8,
437
+ "peft_lora_alpha": 32,
438
+ "peft_lora_dropout": 0.1,
439
+ "fsdp": "none",
440
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
441
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
442
+ "fsdp_fsdp_min_num_params": 100000000.0,
443
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
444
+ "fsdp_fsdp_forward_prefetch": false,
445
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
446
+ "fsdp_fsdp_offload_params": false,
447
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
448
+ "fsdp_fsdp_use_orig_params": true,
449
+ "fsdp_fsdp_activation_checkpointing": false,
450
+ "fsdp_fsdp_limit_all_gathers": true,
451
+ "fsdp_fsdp_sync_module_states": true,
452
+ "use_git": true,
453
+ "random_seed": 0
454
+ }
455
+ },
456
+ "arc_challenge": {
457
+ "acc_norm": 0.30716723549488056,
458
+ "acc_norm_stderr": 0.013481034054980943,
459
+ "args": {
460
+ "exp_name": "results/v2/arc_challenge/Qwen/Qwen1.5-0.5B",
461
+ "task_name": "arc_challenge",
462
+ "model_name": "Qwen/Qwen1.5-0.5B",
463
+ "dataset_param": {
464
+ "max_num_train": 50000,
465
+ "completion_only": false,
466
+ "apply_chat_template": false
467
+ },
468
+ "resume_from_checkpoint": false,
469
+ "train_param": {
470
+ "max_length": 1024,
471
+ "per_device_train_batch_size": 8,
472
+ "auto_find_batch_size": true,
473
+ "num_train_epochs": 5,
474
+ "max_steps": -1,
475
+ "logging_strategy": "epoch",
476
+ "save_strategy": "epoch",
477
+ "logging_steps": 0.2,
478
+ "save_steps": 0.2,
479
+ "gradient_accumulation_steps": 1,
480
+ "gradient_checkpointing": false,
481
+ "optim": "adamw_torch",
482
+ "learning_rate": 5e-05,
483
+ "weight_decay": 0.01,
484
+ "warmup_steps": 0,
485
+ "report_to": "tensorboard",
486
+ "save_safetensors": true,
487
+ "packing": false
488
+ },
489
+ "use_peft": true,
490
+ "peft_task_type": "CAUSAL_LM",
491
+ "peft_inference_mode": false,
492
+ "peft_r": 8,
493
+ "peft_lora_alpha": 32,
494
+ "peft_lora_dropout": 0.1,
495
+ "fsdp": "none",
496
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
497
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
498
+ "fsdp_fsdp_min_num_params": 100000000.0,
499
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
500
+ "fsdp_fsdp_forward_prefetch": false,
501
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
502
+ "fsdp_fsdp_offload_params": false,
503
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
504
+ "fsdp_fsdp_use_orig_params": true,
505
+ "fsdp_fsdp_activation_checkpointing": false,
506
+ "fsdp_fsdp_limit_all_gathers": true,
507
+ "fsdp_fsdp_sync_module_states": true,
508
+ "use_git": true,
509
+ "random_seed": 0
510
+ }
511
+ }
512
+ }
513
+ }
Qwen/Qwen1.5-1.8B-Chat/results_2026-01-05-04-27-51.json ADDED
@@ -0,0 +1,513 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "model_name": "Qwen/Qwen1.5-1.8B-Chat",
4
+ "model_type": "instruction-tuned",
5
+ "params": 1.8
6
+ },
7
+ "results": {
8
+ "medmcqa": {
9
+ "acc_norm": 0.41381783409036577,
10
+ "acc_norm_stderr": 0.0076160342429284125,
11
+ "args": {
12
+ "exp_name": "results/v1/medmcqa/Qwen/Qwen1.5-1.8B-Chat",
13
+ "task_name": "medmcqa",
14
+ "model_name": "Qwen/Qwen1.5-1.8B-Chat",
15
+ "dataset_param": {
16
+ "max_num_train": 50000,
17
+ "completion_only": false,
18
+ "apply_chat_template": true
19
+ },
20
+ "resume_from_checkpoint": false,
21
+ "train_param": {
22
+ "max_length": 1024,
23
+ "per_device_train_batch_size": 8,
24
+ "auto_find_batch_size": true,
25
+ "num_train_epochs": 1,
26
+ "max_steps": -1,
27
+ "logging_strategy": "steps",
28
+ "save_strategy": "steps",
29
+ "logging_steps": 0.2,
30
+ "save_steps": 0.2,
31
+ "gradient_accumulation_steps": 1,
32
+ "gradient_checkpointing": false,
33
+ "optim": "adamw_torch",
34
+ "learning_rate": 5e-05,
35
+ "weight_decay": 0.01,
36
+ "warmup_steps": 0,
37
+ "report_to": "tensorboard",
38
+ "save_safetensors": true,
39
+ "packing": false
40
+ },
41
+ "use_peft": true,
42
+ "peft_task_type": "CAUSAL_LM",
43
+ "peft_inference_mode": false,
44
+ "peft_r": 8,
45
+ "peft_lora_alpha": 32,
46
+ "peft_lora_dropout": 0.1,
47
+ "fsdp": "none",
48
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
49
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
50
+ "fsdp_fsdp_min_num_params": 100000000.0,
51
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
52
+ "fsdp_fsdp_forward_prefetch": false,
53
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
54
+ "fsdp_fsdp_offload_params": false,
55
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
56
+ "fsdp_fsdp_use_orig_params": true,
57
+ "fsdp_fsdp_activation_checkpointing": false,
58
+ "fsdp_fsdp_limit_all_gathers": true,
59
+ "fsdp_fsdp_sync_module_states": true,
60
+ "use_git": true,
61
+ "random_seed": 0
62
+ }
63
+ },
64
+ "nq_open": {
65
+ "exact_match,remove_whitespace": 0.12409972299168975,
66
+ "exact_match,remove_whitespace_stderr": 0.005488065636885807,
67
+ "args": {
68
+ "exp_name": "results/v1/nq_open/Qwen/Qwen1.5-1.8B-Chat",
69
+ "task_name": "nq_open",
70
+ "model_name": "Qwen/Qwen1.5-1.8B-Chat",
71
+ "dataset_param": {
72
+ "max_num_train": 50000,
73
+ "completion_only": false,
74
+ "apply_chat_template": true
75
+ },
76
+ "resume_from_checkpoint": false,
77
+ "train_param": {
78
+ "max_length": 1024,
79
+ "per_device_train_batch_size": 8,
80
+ "auto_find_batch_size": true,
81
+ "num_train_epochs": 1,
82
+ "max_steps": -1,
83
+ "logging_strategy": "steps",
84
+ "save_strategy": "steps",
85
+ "logging_steps": 0.2,
86
+ "save_steps": 0.2,
87
+ "gradient_accumulation_steps": 1,
88
+ "gradient_checkpointing": false,
89
+ "optim": "adamw_torch",
90
+ "learning_rate": 5e-05,
91
+ "weight_decay": 0.01,
92
+ "warmup_steps": 0,
93
+ "report_to": "tensorboard",
94
+ "save_safetensors": true,
95
+ "packing": false
96
+ },
97
+ "use_peft": true,
98
+ "peft_task_type": "CAUSAL_LM",
99
+ "peft_inference_mode": false,
100
+ "peft_r": 8,
101
+ "peft_lora_alpha": 32,
102
+ "peft_lora_dropout": 0.1,
103
+ "fsdp": "none",
104
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
105
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
106
+ "fsdp_fsdp_min_num_params": 100000000.0,
107
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
108
+ "fsdp_fsdp_forward_prefetch": false,
109
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
110
+ "fsdp_fsdp_offload_params": false,
111
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
112
+ "fsdp_fsdp_use_orig_params": true,
113
+ "fsdp_fsdp_activation_checkpointing": false,
114
+ "fsdp_fsdp_limit_all_gathers": true,
115
+ "fsdp_fsdp_sync_module_states": true,
116
+ "use_git": true,
117
+ "random_seed": 0
118
+ }
119
+ },
120
+ "winogrande": {
121
+ "acc": 0.6621941594317285,
122
+ "acc_stderr": 0.013292583502910892,
123
+ "args": {
124
+ "exp_name": "results/v1/winogrande/Qwen/Qwen1.5-1.8B-Chat",
125
+ "task_name": "winogrande",
126
+ "model_name": "Qwen/Qwen1.5-1.8B-Chat",
127
+ "dataset_param": {
128
+ "max_num_train": 50000,
129
+ "completion_only": false,
130
+ "apply_chat_template": true
131
+ },
132
+ "resume_from_checkpoint": false,
133
+ "train_param": {
134
+ "max_length": 1024,
135
+ "per_device_train_batch_size": 8,
136
+ "auto_find_batch_size": true,
137
+ "num_train_epochs": 1,
138
+ "max_steps": -1,
139
+ "logging_strategy": "steps",
140
+ "save_strategy": "steps",
141
+ "logging_steps": 0.2,
142
+ "save_steps": 0.2,
143
+ "gradient_accumulation_steps": 1,
144
+ "gradient_checkpointing": false,
145
+ "optim": "adamw_torch",
146
+ "learning_rate": 5e-05,
147
+ "weight_decay": 0.01,
148
+ "warmup_steps": 0,
149
+ "report_to": "tensorboard",
150
+ "save_safetensors": true,
151
+ "packing": false
152
+ },
153
+ "use_peft": true,
154
+ "peft_task_type": "CAUSAL_LM",
155
+ "peft_inference_mode": false,
156
+ "peft_r": 8,
157
+ "peft_lora_alpha": 32,
158
+ "peft_lora_dropout": 0.1,
159
+ "fsdp": "none",
160
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
161
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
162
+ "fsdp_fsdp_min_num_params": 100000000.0,
163
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
164
+ "fsdp_fsdp_forward_prefetch": false,
165
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
166
+ "fsdp_fsdp_offload_params": false,
167
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
168
+ "fsdp_fsdp_use_orig_params": true,
169
+ "fsdp_fsdp_activation_checkpointing": false,
170
+ "fsdp_fsdp_limit_all_gathers": true,
171
+ "fsdp_fsdp_sync_module_states": true,
172
+ "use_git": true,
173
+ "random_seed": 0
174
+ }
175
+ },
176
+ "hellaswag": {
177
+ "acc_norm": 0.621,
178
+ "acc_norm_stderr": 0.004851623833870553,
179
+ "args": {
180
+ "exp_name": "results/v1/hellaswag/Qwen/Qwen1.5-1.8B-Chat",
181
+ "task_name": "hellaswag",
182
+ "model_name": "Qwen/Qwen1.5-1.8B-Chat",
183
+ "dataset_param": {
184
+ "max_num_train": 50000,
185
+ "completion_only": false,
186
+ "apply_chat_template": true
187
+ },
188
+ "resume_from_checkpoint": false,
189
+ "train_param": {
190
+ "max_length": 1024,
191
+ "per_device_train_batch_size": 8,
192
+ "auto_find_batch_size": true,
193
+ "num_train_epochs": 1,
194
+ "max_steps": -1,
195
+ "logging_strategy": "steps",
196
+ "save_strategy": "steps",
197
+ "logging_steps": 0.2,
198
+ "save_steps": 0.2,
199
+ "gradient_accumulation_steps": 1,
200
+ "gradient_checkpointing": false,
201
+ "optim": "adamw_torch",
202
+ "learning_rate": 5e-05,
203
+ "weight_decay": 0.01,
204
+ "warmup_steps": 0,
205
+ "report_to": "tensorboard",
206
+ "save_safetensors": true,
207
+ "packing": false
208
+ },
209
+ "use_peft": true,
210
+ "peft_task_type": "CAUSAL_LM",
211
+ "peft_inference_mode": false,
212
+ "peft_r": 8,
213
+ "peft_lora_alpha": 32,
214
+ "peft_lora_dropout": 0.1,
215
+ "fsdp": "none",
216
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
217
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
218
+ "fsdp_fsdp_min_num_params": 100000000.0,
219
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
220
+ "fsdp_fsdp_forward_prefetch": false,
221
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
222
+ "fsdp_fsdp_offload_params": false,
223
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
224
+ "fsdp_fsdp_use_orig_params": true,
225
+ "fsdp_fsdp_activation_checkpointing": false,
226
+ "fsdp_fsdp_limit_all_gathers": true,
227
+ "fsdp_fsdp_sync_module_states": true,
228
+ "use_git": true,
229
+ "random_seed": 0
230
+ }
231
+ },
232
+ "social_iqa": {
233
+ "acc": 0.49590583418628453,
234
+ "acc_stderr": 0.011313691251392989,
235
+ "args": {
236
+ "exp_name": "results/v1/social_iqa/Qwen/Qwen1.5-1.8B-Chat",
237
+ "task_name": "social_iqa",
238
+ "model_name": "Qwen/Qwen1.5-1.8B-Chat",
239
+ "dataset_param": {
240
+ "max_num_train": 50000,
241
+ "completion_only": false,
242
+ "apply_chat_template": true
243
+ },
244
+ "resume_from_checkpoint": false,
245
+ "train_param": {
246
+ "max_length": 1024,
247
+ "per_device_train_batch_size": 8,
248
+ "auto_find_batch_size": true,
249
+ "num_train_epochs": 1,
250
+ "max_steps": -1,
251
+ "logging_strategy": "steps",
252
+ "save_strategy": "steps",
253
+ "logging_steps": 0.2,
254
+ "save_steps": 0.2,
255
+ "gradient_accumulation_steps": 1,
256
+ "gradient_checkpointing": false,
257
+ "optim": "adamw_torch",
258
+ "learning_rate": 5e-05,
259
+ "weight_decay": 0.01,
260
+ "warmup_steps": 0,
261
+ "report_to": "tensorboard",
262
+ "save_safetensors": true,
263
+ "packing": false
264
+ },
265
+ "use_peft": true,
266
+ "peft_task_type": "CAUSAL_LM",
267
+ "peft_inference_mode": false,
268
+ "peft_r": 8,
269
+ "peft_lora_alpha": 32,
270
+ "peft_lora_dropout": 0.1,
271
+ "fsdp": "none",
272
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
273
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
274
+ "fsdp_fsdp_min_num_params": 100000000.0,
275
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
276
+ "fsdp_fsdp_forward_prefetch": false,
277
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
278
+ "fsdp_fsdp_offload_params": false,
279
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
280
+ "fsdp_fsdp_use_orig_params": true,
281
+ "fsdp_fsdp_activation_checkpointing": false,
282
+ "fsdp_fsdp_limit_all_gathers": true,
283
+ "fsdp_fsdp_sync_module_states": true,
284
+ "use_git": true,
285
+ "random_seed": 0
286
+ }
287
+ },
288
+ "piqa": {
289
+ "acc_norm": 0.7578890097932536,
290
+ "acc_norm_stderr": 0.00999437126910437,
291
+ "args": {
292
+ "exp_name": "results/v2/piqa/Qwen/Qwen1.5-1.8B-Chat",
293
+ "task_name": "piqa",
294
+ "model_name": "Qwen/Qwen1.5-1.8B-Chat",
295
+ "dataset_param": {
296
+ "max_num_train": 50000,
297
+ "completion_only": false,
298
+ "apply_chat_template": true
299
+ },
300
+ "resume_from_checkpoint": false,
301
+ "train_param": {
302
+ "max_length": 1024,
303
+ "per_device_train_batch_size": 8,
304
+ "auto_find_batch_size": true,
305
+ "num_train_epochs": 5,
306
+ "max_steps": -1,
307
+ "logging_strategy": "epoch",
308
+ "save_strategy": "epoch",
309
+ "logging_steps": 0.2,
310
+ "save_steps": 0.2,
311
+ "gradient_accumulation_steps": 1,
312
+ "gradient_checkpointing": false,
313
+ "optim": "adamw_torch",
314
+ "learning_rate": 5e-05,
315
+ "weight_decay": 0.01,
316
+ "warmup_steps": 0,
317
+ "report_to": "tensorboard",
318
+ "save_safetensors": true,
319
+ "packing": false
320
+ },
321
+ "use_peft": true,
322
+ "peft_task_type": "CAUSAL_LM",
323
+ "peft_inference_mode": false,
324
+ "peft_r": 8,
325
+ "peft_lora_alpha": 32,
326
+ "peft_lora_dropout": 0.1,
327
+ "fsdp": "none",
328
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
329
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
330
+ "fsdp_fsdp_min_num_params": 100000000.0,
331
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
332
+ "fsdp_fsdp_forward_prefetch": false,
333
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
334
+ "fsdp_fsdp_offload_params": false,
335
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
336
+ "fsdp_fsdp_use_orig_params": true,
337
+ "fsdp_fsdp_activation_checkpointing": false,
338
+ "fsdp_fsdp_limit_all_gathers": true,
339
+ "fsdp_fsdp_sync_module_states": true,
340
+ "use_git": true,
341
+ "random_seed": 0
342
+ }
343
+ },
344
+ "commonsense_qa": {
345
+ "acc": 0.7346437346437347,
346
+ "acc_stderr": 0.012640758880532864,
347
+ "args": {
348
+ "exp_name": "results/v2/commonsense_qa/Qwen/Qwen1.5-1.8B-Chat",
349
+ "task_name": "commonsense_qa",
350
+ "model_name": "Qwen/Qwen1.5-1.8B-Chat",
351
+ "dataset_param": {
352
+ "max_num_train": 50000,
353
+ "completion_only": false,
354
+ "apply_chat_template": true
355
+ },
356
+ "resume_from_checkpoint": false,
357
+ "train_param": {
358
+ "max_length": 1024,
359
+ "per_device_train_batch_size": 8,
360
+ "auto_find_batch_size": true,
361
+ "num_train_epochs": 5,
362
+ "max_steps": -1,
363
+ "logging_strategy": "epoch",
364
+ "save_strategy": "epoch",
365
+ "logging_steps": 0.2,
366
+ "save_steps": 0.2,
367
+ "gradient_accumulation_steps": 1,
368
+ "gradient_checkpointing": false,
369
+ "optim": "adamw_torch",
370
+ "learning_rate": 5e-05,
371
+ "weight_decay": 0.01,
372
+ "warmup_steps": 0,
373
+ "report_to": "tensorboard",
374
+ "save_safetensors": true,
375
+ "packing": false
376
+ },
377
+ "use_peft": true,
378
+ "peft_task_type": "CAUSAL_LM",
379
+ "peft_inference_mode": false,
380
+ "peft_r": 8,
381
+ "peft_lora_alpha": 32,
382
+ "peft_lora_dropout": 0.1,
383
+ "fsdp": "none",
384
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
385
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
386
+ "fsdp_fsdp_min_num_params": 100000000.0,
387
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
388
+ "fsdp_fsdp_forward_prefetch": false,
389
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
390
+ "fsdp_fsdp_offload_params": false,
391
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
392
+ "fsdp_fsdp_use_orig_params": true,
393
+ "fsdp_fsdp_activation_checkpointing": false,
394
+ "fsdp_fsdp_limit_all_gathers": true,
395
+ "fsdp_fsdp_sync_module_states": true,
396
+ "use_git": true,
397
+ "random_seed": 0
398
+ }
399
+ },
400
+ "gsm8k": {
401
+ "exact_match,flexible-extract": 0.32676269901440486,
402
+ "exact_match,flexible-extract_stderr": 0.012919408108656421,
403
+ "args": {
404
+ "exp_name": "results/v2/gsm8k/Qwen/Qwen1.5-1.8B-Chat",
405
+ "task_name": "gsm8k",
406
+ "model_name": "Qwen/Qwen1.5-1.8B-Chat",
407
+ "dataset_param": {
408
+ "max_num_train": 50000,
409
+ "completion_only": false,
410
+ "apply_chat_template": true
411
+ },
412
+ "resume_from_checkpoint": false,
413
+ "train_param": {
414
+ "max_length": 1024,
415
+ "per_device_train_batch_size": 8,
416
+ "auto_find_batch_size": true,
417
+ "num_train_epochs": 5,
418
+ "max_steps": -1,
419
+ "logging_strategy": "epoch",
420
+ "save_strategy": "epoch",
421
+ "logging_steps": 0.2,
422
+ "save_steps": 0.2,
423
+ "gradient_accumulation_steps": 1,
424
+ "gradient_checkpointing": false,
425
+ "optim": "adamw_torch",
426
+ "learning_rate": 5e-05,
427
+ "weight_decay": 0.01,
428
+ "warmup_steps": 0,
429
+ "report_to": "tensorboard",
430
+ "save_safetensors": true,
431
+ "packing": false
432
+ },
433
+ "use_peft": true,
434
+ "peft_task_type": "CAUSAL_LM",
435
+ "peft_inference_mode": false,
436
+ "peft_r": 8,
437
+ "peft_lora_alpha": 32,
438
+ "peft_lora_dropout": 0.1,
439
+ "fsdp": "none",
440
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
441
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
442
+ "fsdp_fsdp_min_num_params": 100000000.0,
443
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
444
+ "fsdp_fsdp_forward_prefetch": false,
445
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
446
+ "fsdp_fsdp_offload_params": false,
447
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
448
+ "fsdp_fsdp_use_orig_params": true,
449
+ "fsdp_fsdp_activation_checkpointing": false,
450
+ "fsdp_fsdp_limit_all_gathers": true,
451
+ "fsdp_fsdp_sync_module_states": true,
452
+ "use_git": true,
453
+ "random_seed": 0
454
+ }
455
+ },
456
+ "arc_challenge": {
457
+ "acc_norm": 0.3319112627986348,
458
+ "acc_norm_stderr": 0.013760988200880536,
459
+ "args": {
460
+ "exp_name": "results/v2/arc_challenge/Qwen/Qwen1.5-1.8B-Chat",
461
+ "task_name": "arc_challenge",
462
+ "model_name": "Qwen/Qwen1.5-1.8B-Chat",
463
+ "dataset_param": {
464
+ "max_num_train": 50000,
465
+ "completion_only": false,
466
+ "apply_chat_template": true
467
+ },
468
+ "resume_from_checkpoint": false,
469
+ "train_param": {
470
+ "max_length": 1024,
471
+ "per_device_train_batch_size": 8,
472
+ "auto_find_batch_size": true,
473
+ "num_train_epochs": 5,
474
+ "max_steps": -1,
475
+ "logging_strategy": "epoch",
476
+ "save_strategy": "epoch",
477
+ "logging_steps": 0.2,
478
+ "save_steps": 0.2,
479
+ "gradient_accumulation_steps": 1,
480
+ "gradient_checkpointing": false,
481
+ "optim": "adamw_torch",
482
+ "learning_rate": 5e-05,
483
+ "weight_decay": 0.01,
484
+ "warmup_steps": 0,
485
+ "report_to": "tensorboard",
486
+ "save_safetensors": true,
487
+ "packing": false
488
+ },
489
+ "use_peft": true,
490
+ "peft_task_type": "CAUSAL_LM",
491
+ "peft_inference_mode": false,
492
+ "peft_r": 8,
493
+ "peft_lora_alpha": 32,
494
+ "peft_lora_dropout": 0.1,
495
+ "fsdp": "none",
496
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
497
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
498
+ "fsdp_fsdp_min_num_params": 100000000.0,
499
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
500
+ "fsdp_fsdp_forward_prefetch": false,
501
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
502
+ "fsdp_fsdp_offload_params": false,
503
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
504
+ "fsdp_fsdp_use_orig_params": true,
505
+ "fsdp_fsdp_activation_checkpointing": false,
506
+ "fsdp_fsdp_limit_all_gathers": true,
507
+ "fsdp_fsdp_sync_module_states": true,
508
+ "use_git": true,
509
+ "random_seed": 0
510
+ }
511
+ }
512
+ }
513
+ }
Qwen/Qwen1.5-1.8B/results_2026-01-05-04-27-50.json ADDED
@@ -0,0 +1,513 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "model_name": "Qwen/Qwen1.5-1.8B",
4
+ "model_type": "pretrained",
5
+ "params": 1.8
6
+ },
7
+ "results": {
8
+ "medmcqa": {
9
+ "acc_norm": 0.4133397083432943,
10
+ "acc_norm_stderr": 0.007614736804833202,
11
+ "args": {
12
+ "exp_name": "results/v1/medmcqa/Qwen/Qwen1.5-1.8B",
13
+ "task_name": "medmcqa",
14
+ "model_name": "Qwen/Qwen1.5-1.8B",
15
+ "dataset_param": {
16
+ "max_num_train": 50000,
17
+ "completion_only": false,
18
+ "apply_chat_template": false
19
+ },
20
+ "resume_from_checkpoint": false,
21
+ "train_param": {
22
+ "max_length": 1024,
23
+ "per_device_train_batch_size": 8,
24
+ "auto_find_batch_size": true,
25
+ "num_train_epochs": 1,
26
+ "max_steps": -1,
27
+ "logging_strategy": "steps",
28
+ "save_strategy": "steps",
29
+ "logging_steps": 0.2,
30
+ "save_steps": 0.2,
31
+ "gradient_accumulation_steps": 1,
32
+ "gradient_checkpointing": false,
33
+ "optim": "adamw_torch",
34
+ "learning_rate": 5e-05,
35
+ "weight_decay": 0.01,
36
+ "warmup_steps": 0,
37
+ "report_to": "tensorboard",
38
+ "save_safetensors": true,
39
+ "packing": false
40
+ },
41
+ "use_peft": true,
42
+ "peft_task_type": "CAUSAL_LM",
43
+ "peft_inference_mode": false,
44
+ "peft_r": 8,
45
+ "peft_lora_alpha": 32,
46
+ "peft_lora_dropout": 0.1,
47
+ "fsdp": "none",
48
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
49
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
50
+ "fsdp_fsdp_min_num_params": 100000000.0,
51
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
52
+ "fsdp_fsdp_forward_prefetch": false,
53
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
54
+ "fsdp_fsdp_offload_params": false,
55
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
56
+ "fsdp_fsdp_use_orig_params": true,
57
+ "fsdp_fsdp_activation_checkpointing": false,
58
+ "fsdp_fsdp_limit_all_gathers": true,
59
+ "fsdp_fsdp_sync_module_states": true,
60
+ "use_git": true,
61
+ "random_seed": 0
62
+ }
63
+ },
64
+ "nq_open": {
65
+ "exact_match,remove_whitespace": 0.1299168975069252,
66
+ "exact_match,remove_whitespace_stderr": 0.005596541733139245,
67
+ "args": {
68
+ "exp_name": "results/v1/nq_open/Qwen/Qwen1.5-1.8B",
69
+ "task_name": "nq_open",
70
+ "model_name": "Qwen/Qwen1.5-1.8B",
71
+ "dataset_param": {
72
+ "max_num_train": 50000,
73
+ "completion_only": false,
74
+ "apply_chat_template": false
75
+ },
76
+ "resume_from_checkpoint": false,
77
+ "train_param": {
78
+ "max_length": 1024,
79
+ "per_device_train_batch_size": 8,
80
+ "auto_find_batch_size": true,
81
+ "num_train_epochs": 1,
82
+ "max_steps": -1,
83
+ "logging_strategy": "steps",
84
+ "save_strategy": "steps",
85
+ "logging_steps": 0.2,
86
+ "save_steps": 0.2,
87
+ "gradient_accumulation_steps": 1,
88
+ "gradient_checkpointing": false,
89
+ "optim": "adamw_torch",
90
+ "learning_rate": 5e-05,
91
+ "weight_decay": 0.01,
92
+ "warmup_steps": 0,
93
+ "report_to": "tensorboard",
94
+ "save_safetensors": true,
95
+ "packing": false
96
+ },
97
+ "use_peft": true,
98
+ "peft_task_type": "CAUSAL_LM",
99
+ "peft_inference_mode": false,
100
+ "peft_r": 8,
101
+ "peft_lora_alpha": 32,
102
+ "peft_lora_dropout": 0.1,
103
+ "fsdp": "none",
104
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
105
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
106
+ "fsdp_fsdp_min_num_params": 100000000.0,
107
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
108
+ "fsdp_fsdp_forward_prefetch": false,
109
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
110
+ "fsdp_fsdp_offload_params": false,
111
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
112
+ "fsdp_fsdp_use_orig_params": true,
113
+ "fsdp_fsdp_activation_checkpointing": false,
114
+ "fsdp_fsdp_limit_all_gathers": true,
115
+ "fsdp_fsdp_sync_module_states": true,
116
+ "use_git": true,
117
+ "random_seed": 0
118
+ }
119
+ },
120
+ "winogrande": {
121
+ "acc": 0.659037095501184,
122
+ "acc_stderr": 0.013322681435934786,
123
+ "args": {
124
+ "exp_name": "results/v1/winogrande/Qwen/Qwen1.5-1.8B",
125
+ "task_name": "winogrande",
126
+ "model_name": "Qwen/Qwen1.5-1.8B",
127
+ "dataset_param": {
128
+ "max_num_train": 50000,
129
+ "completion_only": false,
130
+ "apply_chat_template": false
131
+ },
132
+ "resume_from_checkpoint": false,
133
+ "train_param": {
134
+ "max_length": 1024,
135
+ "per_device_train_batch_size": 8,
136
+ "auto_find_batch_size": true,
137
+ "num_train_epochs": 1,
138
+ "max_steps": -1,
139
+ "logging_strategy": "steps",
140
+ "save_strategy": "steps",
141
+ "logging_steps": 0.2,
142
+ "save_steps": 0.2,
143
+ "gradient_accumulation_steps": 1,
144
+ "gradient_checkpointing": false,
145
+ "optim": "adamw_torch",
146
+ "learning_rate": 5e-05,
147
+ "weight_decay": 0.01,
148
+ "warmup_steps": 0,
149
+ "report_to": "tensorboard",
150
+ "save_safetensors": true,
151
+ "packing": false
152
+ },
153
+ "use_peft": true,
154
+ "peft_task_type": "CAUSAL_LM",
155
+ "peft_inference_mode": false,
156
+ "peft_r": 8,
157
+ "peft_lora_alpha": 32,
158
+ "peft_lora_dropout": 0.1,
159
+ "fsdp": "none",
160
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
161
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
162
+ "fsdp_fsdp_min_num_params": 100000000.0,
163
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
164
+ "fsdp_fsdp_forward_prefetch": false,
165
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
166
+ "fsdp_fsdp_offload_params": false,
167
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
168
+ "fsdp_fsdp_use_orig_params": true,
169
+ "fsdp_fsdp_activation_checkpointing": false,
170
+ "fsdp_fsdp_limit_all_gathers": true,
171
+ "fsdp_fsdp_sync_module_states": true,
172
+ "use_git": true,
173
+ "random_seed": 0
174
+ }
175
+ },
176
+ "hellaswag": {
177
+ "acc_norm": 0.6304,
178
+ "acc_norm_stderr": 0.004827205629700884,
179
+ "args": {
180
+ "exp_name": "results/v1/hellaswag/Qwen/Qwen1.5-1.8B",
181
+ "task_name": "hellaswag",
182
+ "model_name": "Qwen/Qwen1.5-1.8B",
183
+ "dataset_param": {
184
+ "max_num_train": 50000,
185
+ "completion_only": false,
186
+ "apply_chat_template": false
187
+ },
188
+ "resume_from_checkpoint": false,
189
+ "train_param": {
190
+ "max_length": 1024,
191
+ "per_device_train_batch_size": 8,
192
+ "auto_find_batch_size": true,
193
+ "num_train_epochs": 1,
194
+ "max_steps": -1,
195
+ "logging_strategy": "steps",
196
+ "save_strategy": "steps",
197
+ "logging_steps": 0.2,
198
+ "save_steps": 0.2,
199
+ "gradient_accumulation_steps": 1,
200
+ "gradient_checkpointing": false,
201
+ "optim": "adamw_torch",
202
+ "learning_rate": 5e-05,
203
+ "weight_decay": 0.01,
204
+ "warmup_steps": 0,
205
+ "report_to": "tensorboard",
206
+ "save_safetensors": true,
207
+ "packing": false
208
+ },
209
+ "use_peft": true,
210
+ "peft_task_type": "CAUSAL_LM",
211
+ "peft_inference_mode": false,
212
+ "peft_r": 8,
213
+ "peft_lora_alpha": 32,
214
+ "peft_lora_dropout": 0.1,
215
+ "fsdp": "none",
216
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
217
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
218
+ "fsdp_fsdp_min_num_params": 100000000.0,
219
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
220
+ "fsdp_fsdp_forward_prefetch": false,
221
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
222
+ "fsdp_fsdp_offload_params": false,
223
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
224
+ "fsdp_fsdp_use_orig_params": true,
225
+ "fsdp_fsdp_activation_checkpointing": false,
226
+ "fsdp_fsdp_limit_all_gathers": true,
227
+ "fsdp_fsdp_sync_module_states": true,
228
+ "use_git": true,
229
+ "random_seed": 0
230
+ }
231
+ },
232
+ "social_iqa": {
233
+ "acc": 0.5199590583418628,
234
+ "acc_stderr": 0.011305052724195018,
235
+ "args": {
236
+ "exp_name": "results/v1/social_iqa/Qwen/Qwen1.5-1.8B",
237
+ "task_name": "social_iqa",
238
+ "model_name": "Qwen/Qwen1.5-1.8B",
239
+ "dataset_param": {
240
+ "max_num_train": 50000,
241
+ "completion_only": false,
242
+ "apply_chat_template": false
243
+ },
244
+ "resume_from_checkpoint": false,
245
+ "train_param": {
246
+ "max_length": 1024,
247
+ "per_device_train_batch_size": 8,
248
+ "auto_find_batch_size": true,
249
+ "num_train_epochs": 1,
250
+ "max_steps": -1,
251
+ "logging_strategy": "steps",
252
+ "save_strategy": "steps",
253
+ "logging_steps": 0.2,
254
+ "save_steps": 0.2,
255
+ "gradient_accumulation_steps": 1,
256
+ "gradient_checkpointing": false,
257
+ "optim": "adamw_torch",
258
+ "learning_rate": 5e-05,
259
+ "weight_decay": 0.01,
260
+ "warmup_steps": 0,
261
+ "report_to": "tensorboard",
262
+ "save_safetensors": true,
263
+ "packing": false
264
+ },
265
+ "use_peft": true,
266
+ "peft_task_type": "CAUSAL_LM",
267
+ "peft_inference_mode": false,
268
+ "peft_r": 8,
269
+ "peft_lora_alpha": 32,
270
+ "peft_lora_dropout": 0.1,
271
+ "fsdp": "none",
272
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
273
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
274
+ "fsdp_fsdp_min_num_params": 100000000.0,
275
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
276
+ "fsdp_fsdp_forward_prefetch": false,
277
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
278
+ "fsdp_fsdp_offload_params": false,
279
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
280
+ "fsdp_fsdp_use_orig_params": true,
281
+ "fsdp_fsdp_activation_checkpointing": false,
282
+ "fsdp_fsdp_limit_all_gathers": true,
283
+ "fsdp_fsdp_sync_module_states": true,
284
+ "use_git": true,
285
+ "random_seed": 0
286
+ }
287
+ },
288
+ "piqa": {
289
+ "acc_norm": 0.7589771490750816,
290
+ "acc_norm_stderr": 0.009979042717267312,
291
+ "args": {
292
+ "exp_name": "results/v2/piqa/Qwen/Qwen1.5-1.8B",
293
+ "task_name": "piqa",
294
+ "model_name": "Qwen/Qwen1.5-1.8B",
295
+ "dataset_param": {
296
+ "max_num_train": 50000,
297
+ "completion_only": false,
298
+ "apply_chat_template": false
299
+ },
300
+ "resume_from_checkpoint": false,
301
+ "train_param": {
302
+ "max_length": 1024,
303
+ "per_device_train_batch_size": 8,
304
+ "auto_find_batch_size": true,
305
+ "num_train_epochs": 5,
306
+ "max_steps": -1,
307
+ "logging_strategy": "epoch",
308
+ "save_strategy": "epoch",
309
+ "logging_steps": 0.2,
310
+ "save_steps": 0.2,
311
+ "gradient_accumulation_steps": 1,
312
+ "gradient_checkpointing": false,
313
+ "optim": "adamw_torch",
314
+ "learning_rate": 5e-05,
315
+ "weight_decay": 0.01,
316
+ "warmup_steps": 0,
317
+ "report_to": "tensorboard",
318
+ "save_safetensors": true,
319
+ "packing": false
320
+ },
321
+ "use_peft": true,
322
+ "peft_task_type": "CAUSAL_LM",
323
+ "peft_inference_mode": false,
324
+ "peft_r": 8,
325
+ "peft_lora_alpha": 32,
326
+ "peft_lora_dropout": 0.1,
327
+ "fsdp": "none",
328
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
329
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
330
+ "fsdp_fsdp_min_num_params": 100000000.0,
331
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
332
+ "fsdp_fsdp_forward_prefetch": false,
333
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
334
+ "fsdp_fsdp_offload_params": false,
335
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
336
+ "fsdp_fsdp_use_orig_params": true,
337
+ "fsdp_fsdp_activation_checkpointing": false,
338
+ "fsdp_fsdp_limit_all_gathers": true,
339
+ "fsdp_fsdp_sync_module_states": true,
340
+ "use_git": true,
341
+ "random_seed": 0
342
+ }
343
+ },
344
+ "commonsense_qa": {
345
+ "acc": 0.7248157248157249,
346
+ "acc_stderr": 0.012786323696375931,
347
+ "args": {
348
+ "exp_name": "results/v2/commonsense_qa/Qwen/Qwen1.5-1.8B",
349
+ "task_name": "commonsense_qa",
350
+ "model_name": "Qwen/Qwen1.5-1.8B",
351
+ "dataset_param": {
352
+ "max_num_train": 50000,
353
+ "completion_only": false,
354
+ "apply_chat_template": false
355
+ },
356
+ "resume_from_checkpoint": false,
357
+ "train_param": {
358
+ "max_length": 1024,
359
+ "per_device_train_batch_size": 8,
360
+ "auto_find_batch_size": true,
361
+ "num_train_epochs": 5,
362
+ "max_steps": -1,
363
+ "logging_strategy": "epoch",
364
+ "save_strategy": "epoch",
365
+ "logging_steps": 0.2,
366
+ "save_steps": 0.2,
367
+ "gradient_accumulation_steps": 1,
368
+ "gradient_checkpointing": false,
369
+ "optim": "adamw_torch",
370
+ "learning_rate": 5e-05,
371
+ "weight_decay": 0.01,
372
+ "warmup_steps": 0,
373
+ "report_to": "tensorboard",
374
+ "save_safetensors": true,
375
+ "packing": false
376
+ },
377
+ "use_peft": true,
378
+ "peft_task_type": "CAUSAL_LM",
379
+ "peft_inference_mode": false,
380
+ "peft_r": 8,
381
+ "peft_lora_alpha": 32,
382
+ "peft_lora_dropout": 0.1,
383
+ "fsdp": "none",
384
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
385
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
386
+ "fsdp_fsdp_min_num_params": 100000000.0,
387
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
388
+ "fsdp_fsdp_forward_prefetch": false,
389
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
390
+ "fsdp_fsdp_offload_params": false,
391
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
392
+ "fsdp_fsdp_use_orig_params": true,
393
+ "fsdp_fsdp_activation_checkpointing": false,
394
+ "fsdp_fsdp_limit_all_gathers": true,
395
+ "fsdp_fsdp_sync_module_states": true,
396
+ "use_git": true,
397
+ "random_seed": 0
398
+ }
399
+ },
400
+ "gsm8k": {
401
+ "exact_match,flexible-extract": 0.3730098559514784,
402
+ "exact_match,flexible-extract_stderr": 0.013320876609777217,
403
+ "args": {
404
+ "exp_name": "results/v2/gsm8k/Qwen/Qwen1.5-1.8B",
405
+ "task_name": "gsm8k",
406
+ "model_name": "Qwen/Qwen1.5-1.8B",
407
+ "dataset_param": {
408
+ "max_num_train": 50000,
409
+ "completion_only": false,
410
+ "apply_chat_template": false
411
+ },
412
+ "resume_from_checkpoint": false,
413
+ "train_param": {
414
+ "max_length": 1024,
415
+ "per_device_train_batch_size": 8,
416
+ "auto_find_batch_size": true,
417
+ "num_train_epochs": 5,
418
+ "max_steps": -1,
419
+ "logging_strategy": "epoch",
420
+ "save_strategy": "epoch",
421
+ "logging_steps": 0.2,
422
+ "save_steps": 0.2,
423
+ "gradient_accumulation_steps": 1,
424
+ "gradient_checkpointing": false,
425
+ "optim": "adamw_torch",
426
+ "learning_rate": 5e-05,
427
+ "weight_decay": 0.01,
428
+ "warmup_steps": 0,
429
+ "report_to": "tensorboard",
430
+ "save_safetensors": true,
431
+ "packing": false
432
+ },
433
+ "use_peft": true,
434
+ "peft_task_type": "CAUSAL_LM",
435
+ "peft_inference_mode": false,
436
+ "peft_r": 8,
437
+ "peft_lora_alpha": 32,
438
+ "peft_lora_dropout": 0.1,
439
+ "fsdp": "none",
440
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
441
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
442
+ "fsdp_fsdp_min_num_params": 100000000.0,
443
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
444
+ "fsdp_fsdp_forward_prefetch": false,
445
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
446
+ "fsdp_fsdp_offload_params": false,
447
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
448
+ "fsdp_fsdp_use_orig_params": true,
449
+ "fsdp_fsdp_activation_checkpointing": false,
450
+ "fsdp_fsdp_limit_all_gathers": true,
451
+ "fsdp_fsdp_sync_module_states": true,
452
+ "use_git": true,
453
+ "random_seed": 0
454
+ }
455
+ },
456
+ "arc_challenge": {
457
+ "acc_norm": 0.41467576791808874,
458
+ "acc_norm_stderr": 0.014397070564409174,
459
+ "args": {
460
+ "exp_name": "results/v2/arc_challenge/Qwen/Qwen1.5-1.8B",
461
+ "task_name": "arc_challenge",
462
+ "model_name": "Qwen/Qwen1.5-1.8B",
463
+ "dataset_param": {
464
+ "max_num_train": 50000,
465
+ "completion_only": false,
466
+ "apply_chat_template": false
467
+ },
468
+ "resume_from_checkpoint": false,
469
+ "train_param": {
470
+ "max_length": 1024,
471
+ "per_device_train_batch_size": 8,
472
+ "auto_find_batch_size": true,
473
+ "num_train_epochs": 5,
474
+ "max_steps": -1,
475
+ "logging_strategy": "epoch",
476
+ "save_strategy": "epoch",
477
+ "logging_steps": 0.2,
478
+ "save_steps": 0.2,
479
+ "gradient_accumulation_steps": 1,
480
+ "gradient_checkpointing": false,
481
+ "optim": "adamw_torch",
482
+ "learning_rate": 5e-05,
483
+ "weight_decay": 0.01,
484
+ "warmup_steps": 0,
485
+ "report_to": "tensorboard",
486
+ "save_safetensors": true,
487
+ "packing": false
488
+ },
489
+ "use_peft": true,
490
+ "peft_task_type": "CAUSAL_LM",
491
+ "peft_inference_mode": false,
492
+ "peft_r": 8,
493
+ "peft_lora_alpha": 32,
494
+ "peft_lora_dropout": 0.1,
495
+ "fsdp": "none",
496
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
497
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
498
+ "fsdp_fsdp_min_num_params": 100000000.0,
499
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
500
+ "fsdp_fsdp_forward_prefetch": false,
501
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
502
+ "fsdp_fsdp_offload_params": false,
503
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
504
+ "fsdp_fsdp_use_orig_params": true,
505
+ "fsdp_fsdp_activation_checkpointing": false,
506
+ "fsdp_fsdp_limit_all_gathers": true,
507
+ "fsdp_fsdp_sync_module_states": true,
508
+ "use_git": true,
509
+ "random_seed": 0
510
+ }
511
+ }
512
+ }
513
+ }
Qwen/Qwen1.5-14B-Chat/results_2026-01-05-04-27-51.json ADDED
@@ -0,0 +1,513 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "model_name": "Qwen/Qwen1.5-14B-Chat",
4
+ "model_type": "instruction-tuned",
5
+ "params": 14.0
6
+ },
7
+ "results": {
8
+ "medmcqa": {
9
+ "acc_norm": 0.5436289744202726,
10
+ "acc_norm_stderr": 0.007702262497652573,
11
+ "args": {
12
+ "exp_name": "results/v1/medmcqa/Qwen/Qwen1.5-14B-Chat",
13
+ "task_name": "medmcqa",
14
+ "model_name": "Qwen/Qwen1.5-14B-Chat",
15
+ "dataset_param": {
16
+ "max_num_train": 50000,
17
+ "completion_only": false,
18
+ "apply_chat_template": true
19
+ },
20
+ "resume_from_checkpoint": false,
21
+ "train_param": {
22
+ "max_length": 1024,
23
+ "per_device_train_batch_size": 8,
24
+ "auto_find_batch_size": true,
25
+ "num_train_epochs": 1,
26
+ "max_steps": -1,
27
+ "logging_strategy": "steps",
28
+ "save_strategy": "steps",
29
+ "logging_steps": 0.2,
30
+ "save_steps": 0.2,
31
+ "gradient_accumulation_steps": 1,
32
+ "gradient_checkpointing": false,
33
+ "optim": "adamw_torch",
34
+ "learning_rate": 5e-05,
35
+ "weight_decay": 0.01,
36
+ "warmup_steps": 0,
37
+ "report_to": "tensorboard",
38
+ "save_safetensors": true,
39
+ "packing": false
40
+ },
41
+ "use_peft": true,
42
+ "peft_task_type": "CAUSAL_LM",
43
+ "peft_inference_mode": false,
44
+ "peft_r": 8,
45
+ "peft_lora_alpha": 32,
46
+ "peft_lora_dropout": 0.1,
47
+ "fsdp": "none",
48
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
49
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
50
+ "fsdp_fsdp_min_num_params": 100000000.0,
51
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
52
+ "fsdp_fsdp_forward_prefetch": false,
53
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
54
+ "fsdp_fsdp_offload_params": false,
55
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
56
+ "fsdp_fsdp_use_orig_params": true,
57
+ "fsdp_fsdp_activation_checkpointing": false,
58
+ "fsdp_fsdp_limit_all_gathers": true,
59
+ "fsdp_fsdp_sync_module_states": true,
60
+ "use_git": true,
61
+ "random_seed": 0
62
+ }
63
+ },
64
+ "nq_open": {
65
+ "exact_match,remove_whitespace": 0.26565096952908585,
66
+ "exact_match,remove_whitespace_stderr": 0.00735213587258742,
67
+ "args": {
68
+ "exp_name": "results/v1/nq_open/Qwen/Qwen1.5-14B-Chat",
69
+ "task_name": "nq_open",
70
+ "model_name": "Qwen/Qwen1.5-14B-Chat",
71
+ "dataset_param": {
72
+ "max_num_train": 50000,
73
+ "completion_only": false,
74
+ "apply_chat_template": true
75
+ },
76
+ "resume_from_checkpoint": false,
77
+ "train_param": {
78
+ "max_length": 1024,
79
+ "per_device_train_batch_size": 8,
80
+ "auto_find_batch_size": true,
81
+ "num_train_epochs": 1,
82
+ "max_steps": -1,
83
+ "logging_strategy": "steps",
84
+ "save_strategy": "steps",
85
+ "logging_steps": 0.2,
86
+ "save_steps": 0.2,
87
+ "gradient_accumulation_steps": 1,
88
+ "gradient_checkpointing": false,
89
+ "optim": "adamw_torch",
90
+ "learning_rate": 5e-05,
91
+ "weight_decay": 0.01,
92
+ "warmup_steps": 0,
93
+ "report_to": "tensorboard",
94
+ "save_safetensors": true,
95
+ "packing": false
96
+ },
97
+ "use_peft": true,
98
+ "peft_task_type": "CAUSAL_LM",
99
+ "peft_inference_mode": false,
100
+ "peft_r": 8,
101
+ "peft_lora_alpha": 32,
102
+ "peft_lora_dropout": 0.1,
103
+ "fsdp": "none",
104
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
105
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
106
+ "fsdp_fsdp_min_num_params": 100000000.0,
107
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
108
+ "fsdp_fsdp_forward_prefetch": false,
109
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
110
+ "fsdp_fsdp_offload_params": false,
111
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
112
+ "fsdp_fsdp_use_orig_params": true,
113
+ "fsdp_fsdp_activation_checkpointing": false,
114
+ "fsdp_fsdp_limit_all_gathers": true,
115
+ "fsdp_fsdp_sync_module_states": true,
116
+ "use_git": true,
117
+ "random_seed": 0
118
+ }
119
+ },
120
+ "winogrande": {
121
+ "acc": 0.8271507498026835,
122
+ "acc_stderr": 0.010626964529971859,
123
+ "args": {
124
+ "exp_name": "results/v1/winogrande/Qwen/Qwen1.5-14B-Chat",
125
+ "task_name": "winogrande",
126
+ "model_name": "Qwen/Qwen1.5-14B-Chat",
127
+ "dataset_param": {
128
+ "max_num_train": 50000,
129
+ "completion_only": false,
130
+ "apply_chat_template": true
131
+ },
132
+ "resume_from_checkpoint": false,
133
+ "train_param": {
134
+ "max_length": 1024,
135
+ "per_device_train_batch_size": 8,
136
+ "auto_find_batch_size": true,
137
+ "num_train_epochs": 1,
138
+ "max_steps": -1,
139
+ "logging_strategy": "steps",
140
+ "save_strategy": "steps",
141
+ "logging_steps": 0.2,
142
+ "save_steps": 0.2,
143
+ "gradient_accumulation_steps": 1,
144
+ "gradient_checkpointing": false,
145
+ "optim": "adamw_torch",
146
+ "learning_rate": 5e-05,
147
+ "weight_decay": 0.01,
148
+ "warmup_steps": 0,
149
+ "report_to": "tensorboard",
150
+ "save_safetensors": true,
151
+ "packing": false
152
+ },
153
+ "use_peft": true,
154
+ "peft_task_type": "CAUSAL_LM",
155
+ "peft_inference_mode": false,
156
+ "peft_r": 8,
157
+ "peft_lora_alpha": 32,
158
+ "peft_lora_dropout": 0.1,
159
+ "fsdp": "none",
160
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
161
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
162
+ "fsdp_fsdp_min_num_params": 100000000.0,
163
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
164
+ "fsdp_fsdp_forward_prefetch": false,
165
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
166
+ "fsdp_fsdp_offload_params": false,
167
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
168
+ "fsdp_fsdp_use_orig_params": true,
169
+ "fsdp_fsdp_activation_checkpointing": false,
170
+ "fsdp_fsdp_limit_all_gathers": true,
171
+ "fsdp_fsdp_sync_module_states": true,
172
+ "use_git": true,
173
+ "random_seed": 0
174
+ }
175
+ },
176
+ "hellaswag": {
177
+ "acc_norm": 0.815,
178
+ "acc_norm_stderr": 0.003883169826416608,
179
+ "args": {
180
+ "exp_name": "results/v1/hellaswag/Qwen/Qwen1.5-14B-Chat",
181
+ "task_name": "hellaswag",
182
+ "model_name": "Qwen/Qwen1.5-14B-Chat",
183
+ "dataset_param": {
184
+ "max_num_train": 50000,
185
+ "completion_only": false,
186
+ "apply_chat_template": true
187
+ },
188
+ "resume_from_checkpoint": false,
189
+ "train_param": {
190
+ "max_length": 1024,
191
+ "per_device_train_batch_size": 8,
192
+ "auto_find_batch_size": true,
193
+ "num_train_epochs": 1,
194
+ "max_steps": -1,
195
+ "logging_strategy": "steps",
196
+ "save_strategy": "steps",
197
+ "logging_steps": 0.2,
198
+ "save_steps": 0.2,
199
+ "gradient_accumulation_steps": 1,
200
+ "gradient_checkpointing": false,
201
+ "optim": "adamw_torch",
202
+ "learning_rate": 5e-05,
203
+ "weight_decay": 0.01,
204
+ "warmup_steps": 0,
205
+ "report_to": "tensorboard",
206
+ "save_safetensors": true,
207
+ "packing": false
208
+ },
209
+ "use_peft": true,
210
+ "peft_task_type": "CAUSAL_LM",
211
+ "peft_inference_mode": false,
212
+ "peft_r": 8,
213
+ "peft_lora_alpha": 32,
214
+ "peft_lora_dropout": 0.1,
215
+ "fsdp": "none",
216
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
217
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
218
+ "fsdp_fsdp_min_num_params": 100000000.0,
219
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
220
+ "fsdp_fsdp_forward_prefetch": false,
221
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
222
+ "fsdp_fsdp_offload_params": false,
223
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
224
+ "fsdp_fsdp_use_orig_params": true,
225
+ "fsdp_fsdp_activation_checkpointing": false,
226
+ "fsdp_fsdp_limit_all_gathers": true,
227
+ "fsdp_fsdp_sync_module_states": true,
228
+ "use_git": true,
229
+ "random_seed": 0
230
+ }
231
+ },
232
+ "social_iqa": {
233
+ "acc": 0.5475946775844421,
234
+ "acc_stderr": 0.011262695440459563,
235
+ "args": {
236
+ "exp_name": "results/v1/social_iqa/Qwen/Qwen1.5-14B-Chat",
237
+ "task_name": "social_iqa",
238
+ "model_name": "Qwen/Qwen1.5-14B-Chat",
239
+ "dataset_param": {
240
+ "max_num_train": 50000,
241
+ "completion_only": false,
242
+ "apply_chat_template": true
243
+ },
244
+ "resume_from_checkpoint": false,
245
+ "train_param": {
246
+ "max_length": 1024,
247
+ "per_device_train_batch_size": 8,
248
+ "auto_find_batch_size": true,
249
+ "num_train_epochs": 1,
250
+ "max_steps": -1,
251
+ "logging_strategy": "steps",
252
+ "save_strategy": "steps",
253
+ "logging_steps": 0.2,
254
+ "save_steps": 0.2,
255
+ "gradient_accumulation_steps": 1,
256
+ "gradient_checkpointing": false,
257
+ "optim": "adamw_torch",
258
+ "learning_rate": 5e-05,
259
+ "weight_decay": 0.01,
260
+ "warmup_steps": 0,
261
+ "report_to": "tensorboard",
262
+ "save_safetensors": true,
263
+ "packing": false
264
+ },
265
+ "use_peft": true,
266
+ "peft_task_type": "CAUSAL_LM",
267
+ "peft_inference_mode": false,
268
+ "peft_r": 8,
269
+ "peft_lora_alpha": 32,
270
+ "peft_lora_dropout": 0.1,
271
+ "fsdp": "none",
272
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
273
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
274
+ "fsdp_fsdp_min_num_params": 100000000.0,
275
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
276
+ "fsdp_fsdp_forward_prefetch": false,
277
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
278
+ "fsdp_fsdp_offload_params": false,
279
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
280
+ "fsdp_fsdp_use_orig_params": true,
281
+ "fsdp_fsdp_activation_checkpointing": false,
282
+ "fsdp_fsdp_limit_all_gathers": true,
283
+ "fsdp_fsdp_sync_module_states": true,
284
+ "use_git": true,
285
+ "random_seed": 0
286
+ }
287
+ },
288
+ "piqa": {
289
+ "acc_norm": 0.8248095756256801,
290
+ "acc_norm_stderr": 0.008869057107124918,
291
+ "args": {
292
+ "exp_name": "results/v2/piqa/Qwen/Qwen1.5-14B-Chat",
293
+ "task_name": "piqa",
294
+ "model_name": "Qwen/Qwen1.5-14B-Chat",
295
+ "dataset_param": {
296
+ "max_num_train": 50000,
297
+ "completion_only": false,
298
+ "apply_chat_template": true
299
+ },
300
+ "resume_from_checkpoint": false,
301
+ "train_param": {
302
+ "max_length": 1024,
303
+ "per_device_train_batch_size": 8,
304
+ "auto_find_batch_size": true,
305
+ "num_train_epochs": 5,
306
+ "max_steps": -1,
307
+ "logging_strategy": "epoch",
308
+ "save_strategy": "epoch",
309
+ "logging_steps": 0.2,
310
+ "save_steps": 0.2,
311
+ "gradient_accumulation_steps": 1,
312
+ "gradient_checkpointing": false,
313
+ "optim": "adamw_torch",
314
+ "learning_rate": 5e-05,
315
+ "weight_decay": 0.01,
316
+ "warmup_steps": 0,
317
+ "report_to": "tensorboard",
318
+ "save_safetensors": true,
319
+ "packing": false
320
+ },
321
+ "use_peft": true,
322
+ "peft_task_type": "CAUSAL_LM",
323
+ "peft_inference_mode": false,
324
+ "peft_r": 8,
325
+ "peft_lora_alpha": 32,
326
+ "peft_lora_dropout": 0.1,
327
+ "fsdp": "none",
328
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
329
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
330
+ "fsdp_fsdp_min_num_params": 100000000.0,
331
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
332
+ "fsdp_fsdp_forward_prefetch": false,
333
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
334
+ "fsdp_fsdp_offload_params": false,
335
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
336
+ "fsdp_fsdp_use_orig_params": true,
337
+ "fsdp_fsdp_activation_checkpointing": false,
338
+ "fsdp_fsdp_limit_all_gathers": true,
339
+ "fsdp_fsdp_sync_module_states": true,
340
+ "use_git": true,
341
+ "random_seed": 0
342
+ }
343
+ },
344
+ "commonsense_qa": {
345
+ "acc": 0.8648648648648649,
346
+ "acc_stderr": 0.009787648220812867,
347
+ "args": {
348
+ "exp_name": "results/v2/commonsense_qa/Qwen/Qwen1.5-14B-Chat",
349
+ "task_name": "commonsense_qa",
350
+ "model_name": "Qwen/Qwen1.5-14B-Chat",
351
+ "dataset_param": {
352
+ "max_num_train": 50000,
353
+ "completion_only": false,
354
+ "apply_chat_template": true
355
+ },
356
+ "resume_from_checkpoint": false,
357
+ "train_param": {
358
+ "max_length": 1024,
359
+ "per_device_train_batch_size": 8,
360
+ "auto_find_batch_size": true,
361
+ "num_train_epochs": 5,
362
+ "max_steps": -1,
363
+ "logging_strategy": "epoch",
364
+ "save_strategy": "epoch",
365
+ "logging_steps": 0.2,
366
+ "save_steps": 0.2,
367
+ "gradient_accumulation_steps": 1,
368
+ "gradient_checkpointing": false,
369
+ "optim": "adamw_torch",
370
+ "learning_rate": 5e-05,
371
+ "weight_decay": 0.01,
372
+ "warmup_steps": 0,
373
+ "report_to": "tensorboard",
374
+ "save_safetensors": true,
375
+ "packing": false
376
+ },
377
+ "use_peft": true,
378
+ "peft_task_type": "CAUSAL_LM",
379
+ "peft_inference_mode": false,
380
+ "peft_r": 8,
381
+ "peft_lora_alpha": 32,
382
+ "peft_lora_dropout": 0.1,
383
+ "fsdp": "none",
384
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
385
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
386
+ "fsdp_fsdp_min_num_params": 100000000.0,
387
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
388
+ "fsdp_fsdp_forward_prefetch": false,
389
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
390
+ "fsdp_fsdp_offload_params": false,
391
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
392
+ "fsdp_fsdp_use_orig_params": true,
393
+ "fsdp_fsdp_activation_checkpointing": false,
394
+ "fsdp_fsdp_limit_all_gathers": true,
395
+ "fsdp_fsdp_sync_module_states": true,
396
+ "use_git": true,
397
+ "random_seed": 0
398
+ }
399
+ },
400
+ "gsm8k": {
401
+ "exact_match,flexible-extract": 0.6770280515542078,
402
+ "exact_match,flexible-extract_stderr": 0.012880360794851812,
403
+ "args": {
404
+ "exp_name": "results/v2/gsm8k/Qwen/Qwen1.5-14B-Chat",
405
+ "task_name": "gsm8k",
406
+ "model_name": "Qwen/Qwen1.5-14B-Chat",
407
+ "dataset_param": {
408
+ "max_num_train": 50000,
409
+ "completion_only": false,
410
+ "apply_chat_template": true
411
+ },
412
+ "resume_from_checkpoint": false,
413
+ "train_param": {
414
+ "max_length": 1024,
415
+ "per_device_train_batch_size": 8,
416
+ "auto_find_batch_size": true,
417
+ "num_train_epochs": 5,
418
+ "max_steps": -1,
419
+ "logging_strategy": "epoch",
420
+ "save_strategy": "epoch",
421
+ "logging_steps": 0.2,
422
+ "save_steps": 0.2,
423
+ "gradient_accumulation_steps": 1,
424
+ "gradient_checkpointing": false,
425
+ "optim": "adamw_torch",
426
+ "learning_rate": 5e-05,
427
+ "weight_decay": 0.01,
428
+ "warmup_steps": 0,
429
+ "report_to": "tensorboard",
430
+ "save_safetensors": true,
431
+ "packing": false
432
+ },
433
+ "use_peft": true,
434
+ "peft_task_type": "CAUSAL_LM",
435
+ "peft_inference_mode": false,
436
+ "peft_r": 8,
437
+ "peft_lora_alpha": 32,
438
+ "peft_lora_dropout": 0.1,
439
+ "fsdp": "none",
440
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
441
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
442
+ "fsdp_fsdp_min_num_params": 100000000.0,
443
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
444
+ "fsdp_fsdp_forward_prefetch": false,
445
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
446
+ "fsdp_fsdp_offload_params": false,
447
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
448
+ "fsdp_fsdp_use_orig_params": true,
449
+ "fsdp_fsdp_activation_checkpointing": false,
450
+ "fsdp_fsdp_limit_all_gathers": true,
451
+ "fsdp_fsdp_sync_module_states": true,
452
+ "use_git": true,
453
+ "random_seed": 0
454
+ }
455
+ },
456
+ "arc_challenge": {
457
+ "acc_norm": 0.5870307167235495,
458
+ "acc_norm_stderr": 0.014388344935398326,
459
+ "args": {
460
+ "exp_name": "results/v2/arc_challenge/Qwen/Qwen1.5-14B-Chat",
461
+ "task_name": "arc_challenge",
462
+ "model_name": "Qwen/Qwen1.5-14B-Chat",
463
+ "dataset_param": {
464
+ "max_num_train": 50000,
465
+ "completion_only": false,
466
+ "apply_chat_template": true
467
+ },
468
+ "resume_from_checkpoint": false,
469
+ "train_param": {
470
+ "max_length": 1024,
471
+ "per_device_train_batch_size": 8,
472
+ "auto_find_batch_size": true,
473
+ "num_train_epochs": 5,
474
+ "max_steps": -1,
475
+ "logging_strategy": "epoch",
476
+ "save_strategy": "epoch",
477
+ "logging_steps": 0.2,
478
+ "save_steps": 0.2,
479
+ "gradient_accumulation_steps": 1,
480
+ "gradient_checkpointing": false,
481
+ "optim": "adamw_torch",
482
+ "learning_rate": 5e-05,
483
+ "weight_decay": 0.01,
484
+ "warmup_steps": 0,
485
+ "report_to": "tensorboard",
486
+ "save_safetensors": true,
487
+ "packing": false
488
+ },
489
+ "use_peft": true,
490
+ "peft_task_type": "CAUSAL_LM",
491
+ "peft_inference_mode": false,
492
+ "peft_r": 8,
493
+ "peft_lora_alpha": 32,
494
+ "peft_lora_dropout": 0.1,
495
+ "fsdp": "none",
496
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
497
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
498
+ "fsdp_fsdp_min_num_params": 100000000.0,
499
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
500
+ "fsdp_fsdp_forward_prefetch": false,
501
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
502
+ "fsdp_fsdp_offload_params": false,
503
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
504
+ "fsdp_fsdp_use_orig_params": true,
505
+ "fsdp_fsdp_activation_checkpointing": false,
506
+ "fsdp_fsdp_limit_all_gathers": true,
507
+ "fsdp_fsdp_sync_module_states": true,
508
+ "use_git": true,
509
+ "random_seed": 0
510
+ }
511
+ }
512
+ }
513
+ }
Qwen/Qwen1.5-14B/results_2026-01-05-04-27-50.json ADDED
@@ -0,0 +1,513 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "model_name": "Qwen/Qwen1.5-14B",
4
+ "model_type": "pretrained",
5
+ "params": 14.0
6
+ },
7
+ "results": {
8
+ "medmcqa": {
9
+ "acc_norm": 0.5543868037293809,
10
+ "acc_norm_stderr": 0.007685877345127753,
11
+ "args": {
12
+ "exp_name": "results/v1/medmcqa/Qwen/Qwen1.5-14B",
13
+ "task_name": "medmcqa",
14
+ "model_name": "Qwen/Qwen1.5-14B",
15
+ "dataset_param": {
16
+ "max_num_train": 50000,
17
+ "completion_only": false,
18
+ "apply_chat_template": false
19
+ },
20
+ "resume_from_checkpoint": false,
21
+ "train_param": {
22
+ "max_length": 1024,
23
+ "per_device_train_batch_size": 8,
24
+ "auto_find_batch_size": true,
25
+ "num_train_epochs": 1,
26
+ "max_steps": -1,
27
+ "logging_strategy": "steps",
28
+ "save_strategy": "steps",
29
+ "logging_steps": 0.2,
30
+ "save_steps": 0.2,
31
+ "gradient_accumulation_steps": 1,
32
+ "gradient_checkpointing": false,
33
+ "optim": "adamw_torch",
34
+ "learning_rate": 5e-05,
35
+ "weight_decay": 0.01,
36
+ "warmup_steps": 0,
37
+ "report_to": "tensorboard",
38
+ "save_safetensors": true,
39
+ "packing": false
40
+ },
41
+ "use_peft": true,
42
+ "peft_task_type": "CAUSAL_LM",
43
+ "peft_inference_mode": false,
44
+ "peft_r": 8,
45
+ "peft_lora_alpha": 32,
46
+ "peft_lora_dropout": 0.1,
47
+ "fsdp": "none",
48
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
49
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
50
+ "fsdp_fsdp_min_num_params": 100000000.0,
51
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
52
+ "fsdp_fsdp_forward_prefetch": false,
53
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
54
+ "fsdp_fsdp_offload_params": false,
55
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
56
+ "fsdp_fsdp_use_orig_params": true,
57
+ "fsdp_fsdp_activation_checkpointing": false,
58
+ "fsdp_fsdp_limit_all_gathers": true,
59
+ "fsdp_fsdp_sync_module_states": true,
60
+ "use_git": true,
61
+ "random_seed": 0
62
+ }
63
+ },
64
+ "nq_open": {
65
+ "exact_match,remove_whitespace": 0.2756232686980609,
66
+ "exact_match,remove_whitespace_stderr": 0.007437838497928537,
67
+ "args": {
68
+ "exp_name": "results/v1/nq_open/Qwen/Qwen1.5-14B",
69
+ "task_name": "nq_open",
70
+ "model_name": "Qwen/Qwen1.5-14B",
71
+ "dataset_param": {
72
+ "max_num_train": 50000,
73
+ "completion_only": false,
74
+ "apply_chat_template": false
75
+ },
76
+ "resume_from_checkpoint": false,
77
+ "train_param": {
78
+ "max_length": 1024,
79
+ "per_device_train_batch_size": 8,
80
+ "auto_find_batch_size": true,
81
+ "num_train_epochs": 1,
82
+ "max_steps": -1,
83
+ "logging_strategy": "steps",
84
+ "save_strategy": "steps",
85
+ "logging_steps": 0.2,
86
+ "save_steps": 0.2,
87
+ "gradient_accumulation_steps": 1,
88
+ "gradient_checkpointing": false,
89
+ "optim": "adamw_torch",
90
+ "learning_rate": 5e-05,
91
+ "weight_decay": 0.01,
92
+ "warmup_steps": 0,
93
+ "report_to": "tensorboard",
94
+ "save_safetensors": true,
95
+ "packing": false
96
+ },
97
+ "use_peft": true,
98
+ "peft_task_type": "CAUSAL_LM",
99
+ "peft_inference_mode": false,
100
+ "peft_r": 8,
101
+ "peft_lora_alpha": 32,
102
+ "peft_lora_dropout": 0.1,
103
+ "fsdp": "none",
104
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
105
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
106
+ "fsdp_fsdp_min_num_params": 100000000.0,
107
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
108
+ "fsdp_fsdp_forward_prefetch": false,
109
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
110
+ "fsdp_fsdp_offload_params": false,
111
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
112
+ "fsdp_fsdp_use_orig_params": true,
113
+ "fsdp_fsdp_activation_checkpointing": false,
114
+ "fsdp_fsdp_limit_all_gathers": true,
115
+ "fsdp_fsdp_sync_module_states": true,
116
+ "use_git": true,
117
+ "random_seed": 0
118
+ }
119
+ },
120
+ "winogrande": {
121
+ "acc": 0.8208366219415943,
122
+ "acc_stderr": 0.010777949156047984,
123
+ "args": {
124
+ "exp_name": "results/v1/winogrande/Qwen/Qwen1.5-14B",
125
+ "task_name": "winogrande",
126
+ "model_name": "Qwen/Qwen1.5-14B",
127
+ "dataset_param": {
128
+ "max_num_train": 50000,
129
+ "completion_only": false,
130
+ "apply_chat_template": false
131
+ },
132
+ "resume_from_checkpoint": false,
133
+ "train_param": {
134
+ "max_length": 1024,
135
+ "per_device_train_batch_size": 8,
136
+ "auto_find_batch_size": true,
137
+ "num_train_epochs": 1,
138
+ "max_steps": -1,
139
+ "logging_strategy": "steps",
140
+ "save_strategy": "steps",
141
+ "logging_steps": 0.2,
142
+ "save_steps": 0.2,
143
+ "gradient_accumulation_steps": 1,
144
+ "gradient_checkpointing": false,
145
+ "optim": "adamw_torch",
146
+ "learning_rate": 5e-05,
147
+ "weight_decay": 0.01,
148
+ "warmup_steps": 0,
149
+ "report_to": "tensorboard",
150
+ "save_safetensors": true,
151
+ "packing": false
152
+ },
153
+ "use_peft": true,
154
+ "peft_task_type": "CAUSAL_LM",
155
+ "peft_inference_mode": false,
156
+ "peft_r": 8,
157
+ "peft_lora_alpha": 32,
158
+ "peft_lora_dropout": 0.1,
159
+ "fsdp": "none",
160
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
161
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
162
+ "fsdp_fsdp_min_num_params": 100000000.0,
163
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
164
+ "fsdp_fsdp_forward_prefetch": false,
165
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
166
+ "fsdp_fsdp_offload_params": false,
167
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
168
+ "fsdp_fsdp_use_orig_params": true,
169
+ "fsdp_fsdp_activation_checkpointing": false,
170
+ "fsdp_fsdp_limit_all_gathers": true,
171
+ "fsdp_fsdp_sync_module_states": true,
172
+ "use_git": true,
173
+ "random_seed": 0
174
+ }
175
+ },
176
+ "hellaswag": {
177
+ "acc_norm": 0.8176,
178
+ "acc_norm_stderr": 0.0038619315700236558,
179
+ "args": {
180
+ "exp_name": "results/v1/hellaswag/Qwen/Qwen1.5-14B",
181
+ "task_name": "hellaswag",
182
+ "model_name": "Qwen/Qwen1.5-14B",
183
+ "dataset_param": {
184
+ "max_num_train": 50000,
185
+ "completion_only": false,
186
+ "apply_chat_template": false
187
+ },
188
+ "resume_from_checkpoint": false,
189
+ "train_param": {
190
+ "max_length": 1024,
191
+ "per_device_train_batch_size": 8,
192
+ "auto_find_batch_size": true,
193
+ "num_train_epochs": 1,
194
+ "max_steps": -1,
195
+ "logging_strategy": "steps",
196
+ "save_strategy": "steps",
197
+ "logging_steps": 0.2,
198
+ "save_steps": 0.2,
199
+ "gradient_accumulation_steps": 1,
200
+ "gradient_checkpointing": false,
201
+ "optim": "adamw_torch",
202
+ "learning_rate": 5e-05,
203
+ "weight_decay": 0.01,
204
+ "warmup_steps": 0,
205
+ "report_to": "tensorboard",
206
+ "save_safetensors": true,
207
+ "packing": false
208
+ },
209
+ "use_peft": true,
210
+ "peft_task_type": "CAUSAL_LM",
211
+ "peft_inference_mode": false,
212
+ "peft_r": 8,
213
+ "peft_lora_alpha": 32,
214
+ "peft_lora_dropout": 0.1,
215
+ "fsdp": "none",
216
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
217
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
218
+ "fsdp_fsdp_min_num_params": 100000000.0,
219
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
220
+ "fsdp_fsdp_forward_prefetch": false,
221
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
222
+ "fsdp_fsdp_offload_params": false,
223
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
224
+ "fsdp_fsdp_use_orig_params": true,
225
+ "fsdp_fsdp_activation_checkpointing": false,
226
+ "fsdp_fsdp_limit_all_gathers": true,
227
+ "fsdp_fsdp_sync_module_states": true,
228
+ "use_git": true,
229
+ "random_seed": 0
230
+ }
231
+ },
232
+ "social_iqa": {
233
+ "acc": 0.5547594677584442,
234
+ "acc_stderr": 0.011246013134156761,
235
+ "args": {
236
+ "exp_name": "results/v1/social_iqa/Qwen/Qwen1.5-14B",
237
+ "task_name": "social_iqa",
238
+ "model_name": "Qwen/Qwen1.5-14B",
239
+ "dataset_param": {
240
+ "max_num_train": 50000,
241
+ "completion_only": false,
242
+ "apply_chat_template": false
243
+ },
244
+ "resume_from_checkpoint": false,
245
+ "train_param": {
246
+ "max_length": 1024,
247
+ "per_device_train_batch_size": 8,
248
+ "auto_find_batch_size": true,
249
+ "num_train_epochs": 1,
250
+ "max_steps": -1,
251
+ "logging_strategy": "steps",
252
+ "save_strategy": "steps",
253
+ "logging_steps": 0.2,
254
+ "save_steps": 0.2,
255
+ "gradient_accumulation_steps": 1,
256
+ "gradient_checkpointing": false,
257
+ "optim": "adamw_torch",
258
+ "learning_rate": 5e-05,
259
+ "weight_decay": 0.01,
260
+ "warmup_steps": 0,
261
+ "report_to": "tensorboard",
262
+ "save_safetensors": true,
263
+ "packing": false
264
+ },
265
+ "use_peft": true,
266
+ "peft_task_type": "CAUSAL_LM",
267
+ "peft_inference_mode": false,
268
+ "peft_r": 8,
269
+ "peft_lora_alpha": 32,
270
+ "peft_lora_dropout": 0.1,
271
+ "fsdp": "none",
272
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
273
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
274
+ "fsdp_fsdp_min_num_params": 100000000.0,
275
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
276
+ "fsdp_fsdp_forward_prefetch": false,
277
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
278
+ "fsdp_fsdp_offload_params": false,
279
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
280
+ "fsdp_fsdp_use_orig_params": true,
281
+ "fsdp_fsdp_activation_checkpointing": false,
282
+ "fsdp_fsdp_limit_all_gathers": true,
283
+ "fsdp_fsdp_sync_module_states": true,
284
+ "use_git": true,
285
+ "random_seed": 0
286
+ }
287
+ },
288
+ "piqa": {
289
+ "acc_norm": 0.8193688792165397,
290
+ "acc_norm_stderr": 0.008975971037338033,
291
+ "args": {
292
+ "exp_name": "results/v2/piqa/Qwen/Qwen1.5-14B",
293
+ "task_name": "piqa",
294
+ "model_name": "Qwen/Qwen1.5-14B",
295
+ "dataset_param": {
296
+ "max_num_train": 50000,
297
+ "completion_only": false,
298
+ "apply_chat_template": false
299
+ },
300
+ "resume_from_checkpoint": false,
301
+ "train_param": {
302
+ "max_length": 1024,
303
+ "per_device_train_batch_size": 8,
304
+ "auto_find_batch_size": true,
305
+ "num_train_epochs": 5,
306
+ "max_steps": -1,
307
+ "logging_strategy": "epoch",
308
+ "save_strategy": "epoch",
309
+ "logging_steps": 0.2,
310
+ "save_steps": 0.2,
311
+ "gradient_accumulation_steps": 1,
312
+ "gradient_checkpointing": false,
313
+ "optim": "adamw_torch",
314
+ "learning_rate": 5e-05,
315
+ "weight_decay": 0.01,
316
+ "warmup_steps": 0,
317
+ "report_to": "tensorboard",
318
+ "save_safetensors": true,
319
+ "packing": false
320
+ },
321
+ "use_peft": true,
322
+ "peft_task_type": "CAUSAL_LM",
323
+ "peft_inference_mode": false,
324
+ "peft_r": 8,
325
+ "peft_lora_alpha": 32,
326
+ "peft_lora_dropout": 0.1,
327
+ "fsdp": "none",
328
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
329
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
330
+ "fsdp_fsdp_min_num_params": 100000000.0,
331
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
332
+ "fsdp_fsdp_forward_prefetch": false,
333
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
334
+ "fsdp_fsdp_offload_params": false,
335
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
336
+ "fsdp_fsdp_use_orig_params": true,
337
+ "fsdp_fsdp_activation_checkpointing": false,
338
+ "fsdp_fsdp_limit_all_gathers": true,
339
+ "fsdp_fsdp_sync_module_states": true,
340
+ "use_git": true,
341
+ "random_seed": 0
342
+ }
343
+ },
344
+ "commonsense_qa": {
345
+ "acc": 0.8763308763308764,
346
+ "acc_stderr": 0.00942507431461914,
347
+ "args": {
348
+ "exp_name": "results/v2/commonsense_qa/Qwen/Qwen1.5-14B",
349
+ "task_name": "commonsense_qa",
350
+ "model_name": "Qwen/Qwen1.5-14B",
351
+ "dataset_param": {
352
+ "max_num_train": 50000,
353
+ "completion_only": false,
354
+ "apply_chat_template": false
355
+ },
356
+ "resume_from_checkpoint": false,
357
+ "train_param": {
358
+ "max_length": 1024,
359
+ "per_device_train_batch_size": 8,
360
+ "auto_find_batch_size": true,
361
+ "num_train_epochs": 5,
362
+ "max_steps": -1,
363
+ "logging_strategy": "epoch",
364
+ "save_strategy": "epoch",
365
+ "logging_steps": 0.2,
366
+ "save_steps": 0.2,
367
+ "gradient_accumulation_steps": 1,
368
+ "gradient_checkpointing": false,
369
+ "optim": "adamw_torch",
370
+ "learning_rate": 5e-05,
371
+ "weight_decay": 0.01,
372
+ "warmup_steps": 0,
373
+ "report_to": "tensorboard",
374
+ "save_safetensors": true,
375
+ "packing": false
376
+ },
377
+ "use_peft": true,
378
+ "peft_task_type": "CAUSAL_LM",
379
+ "peft_inference_mode": false,
380
+ "peft_r": 8,
381
+ "peft_lora_alpha": 32,
382
+ "peft_lora_dropout": 0.1,
383
+ "fsdp": "none",
384
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
385
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
386
+ "fsdp_fsdp_min_num_params": 100000000.0,
387
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
388
+ "fsdp_fsdp_forward_prefetch": false,
389
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
390
+ "fsdp_fsdp_offload_params": false,
391
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
392
+ "fsdp_fsdp_use_orig_params": true,
393
+ "fsdp_fsdp_activation_checkpointing": false,
394
+ "fsdp_fsdp_limit_all_gathers": true,
395
+ "fsdp_fsdp_sync_module_states": true,
396
+ "use_git": true,
397
+ "random_seed": 0
398
+ }
399
+ },
400
+ "gsm8k": {
401
+ "exact_match,flexible-extract": 0.6959818043972706,
402
+ "exact_match,flexible-extract_stderr": 0.012670420440198667,
403
+ "args": {
404
+ "exp_name": "results/v2/gsm8k/Qwen/Qwen1.5-14B",
405
+ "task_name": "gsm8k",
406
+ "model_name": "Qwen/Qwen1.5-14B",
407
+ "dataset_param": {
408
+ "max_num_train": 50000,
409
+ "completion_only": false,
410
+ "apply_chat_template": false
411
+ },
412
+ "resume_from_checkpoint": false,
413
+ "train_param": {
414
+ "max_length": 1024,
415
+ "per_device_train_batch_size": 8,
416
+ "auto_find_batch_size": true,
417
+ "num_train_epochs": 5,
418
+ "max_steps": -1,
419
+ "logging_strategy": "epoch",
420
+ "save_strategy": "epoch",
421
+ "logging_steps": 0.2,
422
+ "save_steps": 0.2,
423
+ "gradient_accumulation_steps": 1,
424
+ "gradient_checkpointing": false,
425
+ "optim": "adamw_torch",
426
+ "learning_rate": 5e-05,
427
+ "weight_decay": 0.01,
428
+ "warmup_steps": 0,
429
+ "report_to": "tensorboard",
430
+ "save_safetensors": true,
431
+ "packing": false
432
+ },
433
+ "use_peft": true,
434
+ "peft_task_type": "CAUSAL_LM",
435
+ "peft_inference_mode": false,
436
+ "peft_r": 8,
437
+ "peft_lora_alpha": 32,
438
+ "peft_lora_dropout": 0.1,
439
+ "fsdp": "none",
440
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
441
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
442
+ "fsdp_fsdp_min_num_params": 100000000.0,
443
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
444
+ "fsdp_fsdp_forward_prefetch": false,
445
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
446
+ "fsdp_fsdp_offload_params": false,
447
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
448
+ "fsdp_fsdp_use_orig_params": true,
449
+ "fsdp_fsdp_activation_checkpointing": false,
450
+ "fsdp_fsdp_limit_all_gathers": true,
451
+ "fsdp_fsdp_sync_module_states": true,
452
+ "use_git": true,
453
+ "random_seed": 0
454
+ }
455
+ },
456
+ "arc_challenge": {
457
+ "acc_norm": 0.6006825938566553,
458
+ "acc_norm_stderr": 0.014312094557946705,
459
+ "args": {
460
+ "exp_name": "results/v2/arc_challenge/Qwen/Qwen1.5-14B",
461
+ "task_name": "arc_challenge",
462
+ "model_name": "Qwen/Qwen1.5-14B",
463
+ "dataset_param": {
464
+ "max_num_train": 50000,
465
+ "completion_only": false,
466
+ "apply_chat_template": false
467
+ },
468
+ "resume_from_checkpoint": false,
469
+ "train_param": {
470
+ "max_length": 1024,
471
+ "per_device_train_batch_size": 8,
472
+ "auto_find_batch_size": true,
473
+ "num_train_epochs": 5,
474
+ "max_steps": -1,
475
+ "logging_strategy": "epoch",
476
+ "save_strategy": "epoch",
477
+ "logging_steps": 0.2,
478
+ "save_steps": 0.2,
479
+ "gradient_accumulation_steps": 1,
480
+ "gradient_checkpointing": false,
481
+ "optim": "adamw_torch",
482
+ "learning_rate": 5e-05,
483
+ "weight_decay": 0.01,
484
+ "warmup_steps": 0,
485
+ "report_to": "tensorboard",
486
+ "save_safetensors": true,
487
+ "packing": false
488
+ },
489
+ "use_peft": true,
490
+ "peft_task_type": "CAUSAL_LM",
491
+ "peft_inference_mode": false,
492
+ "peft_r": 8,
493
+ "peft_lora_alpha": 32,
494
+ "peft_lora_dropout": 0.1,
495
+ "fsdp": "none",
496
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
497
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
498
+ "fsdp_fsdp_min_num_params": 100000000.0,
499
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
500
+ "fsdp_fsdp_forward_prefetch": false,
501
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
502
+ "fsdp_fsdp_offload_params": false,
503
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
504
+ "fsdp_fsdp_use_orig_params": true,
505
+ "fsdp_fsdp_activation_checkpointing": false,
506
+ "fsdp_fsdp_limit_all_gathers": true,
507
+ "fsdp_fsdp_sync_module_states": true,
508
+ "use_git": true,
509
+ "random_seed": 0
510
+ }
511
+ }
512
+ }
513
+ }
Qwen/Qwen1.5-4B-Chat/results_2026-01-05-04-27-51.json ADDED
@@ -0,0 +1,513 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "model_name": "Qwen/Qwen1.5-4B-Chat",
4
+ "model_type": "instruction-tuned",
5
+ "params": 4.0
6
+ },
7
+ "results": {
8
+ "medmcqa": {
9
+ "acc_norm": 0.4630647860387282,
10
+ "acc_norm_stderr": 0.007710629002237254,
11
+ "args": {
12
+ "exp_name": "results/v1/medmcqa/Qwen/Qwen1.5-4B-Chat",
13
+ "task_name": "medmcqa",
14
+ "model_name": "Qwen/Qwen1.5-4B-Chat",
15
+ "dataset_param": {
16
+ "max_num_train": 50000,
17
+ "completion_only": false,
18
+ "apply_chat_template": true
19
+ },
20
+ "resume_from_checkpoint": false,
21
+ "train_param": {
22
+ "max_length": 1024,
23
+ "per_device_train_batch_size": 8,
24
+ "auto_find_batch_size": true,
25
+ "num_train_epochs": 1,
26
+ "max_steps": -1,
27
+ "logging_strategy": "steps",
28
+ "save_strategy": "steps",
29
+ "logging_steps": 0.2,
30
+ "save_steps": 0.2,
31
+ "gradient_accumulation_steps": 1,
32
+ "gradient_checkpointing": false,
33
+ "optim": "adamw_torch",
34
+ "learning_rate": 5e-05,
35
+ "weight_decay": 0.01,
36
+ "warmup_steps": 0,
37
+ "report_to": "tensorboard",
38
+ "save_safetensors": true,
39
+ "packing": false
40
+ },
41
+ "use_peft": true,
42
+ "peft_task_type": "CAUSAL_LM",
43
+ "peft_inference_mode": false,
44
+ "peft_r": 8,
45
+ "peft_lora_alpha": 32,
46
+ "peft_lora_dropout": 0.1,
47
+ "fsdp": "none",
48
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
49
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
50
+ "fsdp_fsdp_min_num_params": 100000000.0,
51
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
52
+ "fsdp_fsdp_forward_prefetch": false,
53
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
54
+ "fsdp_fsdp_offload_params": false,
55
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
56
+ "fsdp_fsdp_use_orig_params": true,
57
+ "fsdp_fsdp_activation_checkpointing": false,
58
+ "fsdp_fsdp_limit_all_gathers": true,
59
+ "fsdp_fsdp_sync_module_states": true,
60
+ "use_git": true,
61
+ "random_seed": 0
62
+ }
63
+ },
64
+ "nq_open": {
65
+ "exact_match,remove_whitespace": 0.17146814404432134,
66
+ "exact_match,remove_whitespace_stderr": 0.0062741209715567355,
67
+ "args": {
68
+ "exp_name": "results/v1/nq_open/Qwen/Qwen1.5-4B-Chat",
69
+ "task_name": "nq_open",
70
+ "model_name": "Qwen/Qwen1.5-4B-Chat",
71
+ "dataset_param": {
72
+ "max_num_train": 50000,
73
+ "completion_only": false,
74
+ "apply_chat_template": true
75
+ },
76
+ "resume_from_checkpoint": false,
77
+ "train_param": {
78
+ "max_length": 1024,
79
+ "per_device_train_batch_size": 8,
80
+ "auto_find_batch_size": true,
81
+ "num_train_epochs": 1,
82
+ "max_steps": -1,
83
+ "logging_strategy": "steps",
84
+ "save_strategy": "steps",
85
+ "logging_steps": 0.2,
86
+ "save_steps": 0.2,
87
+ "gradient_accumulation_steps": 1,
88
+ "gradient_checkpointing": false,
89
+ "optim": "adamw_torch",
90
+ "learning_rate": 5e-05,
91
+ "weight_decay": 0.01,
92
+ "warmup_steps": 0,
93
+ "report_to": "tensorboard",
94
+ "save_safetensors": true,
95
+ "packing": false
96
+ },
97
+ "use_peft": true,
98
+ "peft_task_type": "CAUSAL_LM",
99
+ "peft_inference_mode": false,
100
+ "peft_r": 8,
101
+ "peft_lora_alpha": 32,
102
+ "peft_lora_dropout": 0.1,
103
+ "fsdp": "none",
104
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
105
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
106
+ "fsdp_fsdp_min_num_params": 100000000.0,
107
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
108
+ "fsdp_fsdp_forward_prefetch": false,
109
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
110
+ "fsdp_fsdp_offload_params": false,
111
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
112
+ "fsdp_fsdp_use_orig_params": true,
113
+ "fsdp_fsdp_activation_checkpointing": false,
114
+ "fsdp_fsdp_limit_all_gathers": true,
115
+ "fsdp_fsdp_sync_module_states": true,
116
+ "use_git": true,
117
+ "random_seed": 0
118
+ }
119
+ },
120
+ "winogrande": {
121
+ "acc": 0.7458563535911602,
122
+ "acc_stderr": 0.012236307219708267,
123
+ "args": {
124
+ "exp_name": "results/v1/winogrande/Qwen/Qwen1.5-4B-Chat",
125
+ "task_name": "winogrande",
126
+ "model_name": "Qwen/Qwen1.5-4B-Chat",
127
+ "dataset_param": {
128
+ "max_num_train": 50000,
129
+ "completion_only": false,
130
+ "apply_chat_template": true
131
+ },
132
+ "resume_from_checkpoint": false,
133
+ "train_param": {
134
+ "max_length": 1024,
135
+ "per_device_train_batch_size": 8,
136
+ "auto_find_batch_size": true,
137
+ "num_train_epochs": 1,
138
+ "max_steps": -1,
139
+ "logging_strategy": "steps",
140
+ "save_strategy": "steps",
141
+ "logging_steps": 0.2,
142
+ "save_steps": 0.2,
143
+ "gradient_accumulation_steps": 1,
144
+ "gradient_checkpointing": false,
145
+ "optim": "adamw_torch",
146
+ "learning_rate": 5e-05,
147
+ "weight_decay": 0.01,
148
+ "warmup_steps": 0,
149
+ "report_to": "tensorboard",
150
+ "save_safetensors": true,
151
+ "packing": false
152
+ },
153
+ "use_peft": true,
154
+ "peft_task_type": "CAUSAL_LM",
155
+ "peft_inference_mode": false,
156
+ "peft_r": 8,
157
+ "peft_lora_alpha": 32,
158
+ "peft_lora_dropout": 0.1,
159
+ "fsdp": "none",
160
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
161
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
162
+ "fsdp_fsdp_min_num_params": 100000000.0,
163
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
164
+ "fsdp_fsdp_forward_prefetch": false,
165
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
166
+ "fsdp_fsdp_offload_params": false,
167
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
168
+ "fsdp_fsdp_use_orig_params": true,
169
+ "fsdp_fsdp_activation_checkpointing": false,
170
+ "fsdp_fsdp_limit_all_gathers": true,
171
+ "fsdp_fsdp_sync_module_states": true,
172
+ "use_git": true,
173
+ "random_seed": 0
174
+ }
175
+ },
176
+ "hellaswag": {
177
+ "acc_norm": 0.7261,
178
+ "acc_norm_stderr": 0.004459805823889447,
179
+ "args": {
180
+ "exp_name": "results/v1/hellaswag/Qwen/Qwen1.5-4B-Chat",
181
+ "task_name": "hellaswag",
182
+ "model_name": "Qwen/Qwen1.5-4B-Chat",
183
+ "dataset_param": {
184
+ "max_num_train": 50000,
185
+ "completion_only": false,
186
+ "apply_chat_template": true
187
+ },
188
+ "resume_from_checkpoint": false,
189
+ "train_param": {
190
+ "max_length": 1024,
191
+ "per_device_train_batch_size": 8,
192
+ "auto_find_batch_size": true,
193
+ "num_train_epochs": 1,
194
+ "max_steps": -1,
195
+ "logging_strategy": "steps",
196
+ "save_strategy": "steps",
197
+ "logging_steps": 0.2,
198
+ "save_steps": 0.2,
199
+ "gradient_accumulation_steps": 1,
200
+ "gradient_checkpointing": false,
201
+ "optim": "adamw_torch",
202
+ "learning_rate": 5e-05,
203
+ "weight_decay": 0.01,
204
+ "warmup_steps": 0,
205
+ "report_to": "tensorboard",
206
+ "save_safetensors": true,
207
+ "packing": false
208
+ },
209
+ "use_peft": true,
210
+ "peft_task_type": "CAUSAL_LM",
211
+ "peft_inference_mode": false,
212
+ "peft_r": 8,
213
+ "peft_lora_alpha": 32,
214
+ "peft_lora_dropout": 0.1,
215
+ "fsdp": "none",
216
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
217
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
218
+ "fsdp_fsdp_min_num_params": 100000000.0,
219
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
220
+ "fsdp_fsdp_forward_prefetch": false,
221
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
222
+ "fsdp_fsdp_offload_params": false,
223
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
224
+ "fsdp_fsdp_use_orig_params": true,
225
+ "fsdp_fsdp_activation_checkpointing": false,
226
+ "fsdp_fsdp_limit_all_gathers": true,
227
+ "fsdp_fsdp_sync_module_states": true,
228
+ "use_git": true,
229
+ "random_seed": 0
230
+ }
231
+ },
232
+ "social_iqa": {
233
+ "acc": 0.537871033776868,
234
+ "acc_stderr": 0.011281570251648942,
235
+ "args": {
236
+ "exp_name": "results/v1/social_iqa/Qwen/Qwen1.5-4B-Chat",
237
+ "task_name": "social_iqa",
238
+ "model_name": "Qwen/Qwen1.5-4B-Chat",
239
+ "dataset_param": {
240
+ "max_num_train": 50000,
241
+ "completion_only": false,
242
+ "apply_chat_template": true
243
+ },
244
+ "resume_from_checkpoint": false,
245
+ "train_param": {
246
+ "max_length": 1024,
247
+ "per_device_train_batch_size": 8,
248
+ "auto_find_batch_size": true,
249
+ "num_train_epochs": 1,
250
+ "max_steps": -1,
251
+ "logging_strategy": "steps",
252
+ "save_strategy": "steps",
253
+ "logging_steps": 0.2,
254
+ "save_steps": 0.2,
255
+ "gradient_accumulation_steps": 1,
256
+ "gradient_checkpointing": false,
257
+ "optim": "adamw_torch",
258
+ "learning_rate": 5e-05,
259
+ "weight_decay": 0.01,
260
+ "warmup_steps": 0,
261
+ "report_to": "tensorboard",
262
+ "save_safetensors": true,
263
+ "packing": false
264
+ },
265
+ "use_peft": true,
266
+ "peft_task_type": "CAUSAL_LM",
267
+ "peft_inference_mode": false,
268
+ "peft_r": 8,
269
+ "peft_lora_alpha": 32,
270
+ "peft_lora_dropout": 0.1,
271
+ "fsdp": "none",
272
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
273
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
274
+ "fsdp_fsdp_min_num_params": 100000000.0,
275
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
276
+ "fsdp_fsdp_forward_prefetch": false,
277
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
278
+ "fsdp_fsdp_offload_params": false,
279
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
280
+ "fsdp_fsdp_use_orig_params": true,
281
+ "fsdp_fsdp_activation_checkpointing": false,
282
+ "fsdp_fsdp_limit_all_gathers": true,
283
+ "fsdp_fsdp_sync_module_states": true,
284
+ "use_git": true,
285
+ "random_seed": 0
286
+ }
287
+ },
288
+ "piqa": {
289
+ "acc_norm": 0.7850924918389554,
290
+ "acc_norm_stderr": 0.0095836650826533,
291
+ "args": {
292
+ "exp_name": "results/v2/piqa/Qwen/Qwen1.5-4B-Chat",
293
+ "task_name": "piqa",
294
+ "model_name": "Qwen/Qwen1.5-4B-Chat",
295
+ "dataset_param": {
296
+ "max_num_train": 50000,
297
+ "completion_only": false,
298
+ "apply_chat_template": true
299
+ },
300
+ "resume_from_checkpoint": false,
301
+ "train_param": {
302
+ "max_length": 1024,
303
+ "per_device_train_batch_size": 8,
304
+ "auto_find_batch_size": true,
305
+ "num_train_epochs": 5,
306
+ "max_steps": -1,
307
+ "logging_strategy": "epoch",
308
+ "save_strategy": "epoch",
309
+ "logging_steps": 0.2,
310
+ "save_steps": 0.2,
311
+ "gradient_accumulation_steps": 1,
312
+ "gradient_checkpointing": false,
313
+ "optim": "adamw_torch",
314
+ "learning_rate": 5e-05,
315
+ "weight_decay": 0.01,
316
+ "warmup_steps": 0,
317
+ "report_to": "tensorboard",
318
+ "save_safetensors": true,
319
+ "packing": false
320
+ },
321
+ "use_peft": true,
322
+ "peft_task_type": "CAUSAL_LM",
323
+ "peft_inference_mode": false,
324
+ "peft_r": 8,
325
+ "peft_lora_alpha": 32,
326
+ "peft_lora_dropout": 0.1,
327
+ "fsdp": "none",
328
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
329
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
330
+ "fsdp_fsdp_min_num_params": 100000000.0,
331
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
332
+ "fsdp_fsdp_forward_prefetch": false,
333
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
334
+ "fsdp_fsdp_offload_params": false,
335
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
336
+ "fsdp_fsdp_use_orig_params": true,
337
+ "fsdp_fsdp_activation_checkpointing": false,
338
+ "fsdp_fsdp_limit_all_gathers": true,
339
+ "fsdp_fsdp_sync_module_states": true,
340
+ "use_git": true,
341
+ "random_seed": 0
342
+ }
343
+ },
344
+ "commonsense_qa": {
345
+ "acc": 0.8124488124488124,
346
+ "acc_stderr": 0.011175783964114762,
347
+ "args": {
348
+ "exp_name": "results/v2/commonsense_qa/Qwen/Qwen1.5-4B-Chat",
349
+ "task_name": "commonsense_qa",
350
+ "model_name": "Qwen/Qwen1.5-4B-Chat",
351
+ "dataset_param": {
352
+ "max_num_train": 50000,
353
+ "completion_only": false,
354
+ "apply_chat_template": true
355
+ },
356
+ "resume_from_checkpoint": false,
357
+ "train_param": {
358
+ "max_length": 1024,
359
+ "per_device_train_batch_size": 8,
360
+ "auto_find_batch_size": true,
361
+ "num_train_epochs": 5,
362
+ "max_steps": -1,
363
+ "logging_strategy": "epoch",
364
+ "save_strategy": "epoch",
365
+ "logging_steps": 0.2,
366
+ "save_steps": 0.2,
367
+ "gradient_accumulation_steps": 1,
368
+ "gradient_checkpointing": false,
369
+ "optim": "adamw_torch",
370
+ "learning_rate": 5e-05,
371
+ "weight_decay": 0.01,
372
+ "warmup_steps": 0,
373
+ "report_to": "tensorboard",
374
+ "save_safetensors": true,
375
+ "packing": false
376
+ },
377
+ "use_peft": true,
378
+ "peft_task_type": "CAUSAL_LM",
379
+ "peft_inference_mode": false,
380
+ "peft_r": 8,
381
+ "peft_lora_alpha": 32,
382
+ "peft_lora_dropout": 0.1,
383
+ "fsdp": "none",
384
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
385
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
386
+ "fsdp_fsdp_min_num_params": 100000000.0,
387
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
388
+ "fsdp_fsdp_forward_prefetch": false,
389
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
390
+ "fsdp_fsdp_offload_params": false,
391
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
392
+ "fsdp_fsdp_use_orig_params": true,
393
+ "fsdp_fsdp_activation_checkpointing": false,
394
+ "fsdp_fsdp_limit_all_gathers": true,
395
+ "fsdp_fsdp_sync_module_states": true,
396
+ "use_git": true,
397
+ "random_seed": 0
398
+ }
399
+ },
400
+ "gsm8k": {
401
+ "exact_match,flexible-extract": 0.4564063684609553,
402
+ "exact_match,flexible-extract_stderr": 0.013720038270485329,
403
+ "args": {
404
+ "exp_name": "results/v2/gsm8k/Qwen/Qwen1.5-4B-Chat",
405
+ "task_name": "gsm8k",
406
+ "model_name": "Qwen/Qwen1.5-4B-Chat",
407
+ "dataset_param": {
408
+ "max_num_train": 50000,
409
+ "completion_only": false,
410
+ "apply_chat_template": true
411
+ },
412
+ "resume_from_checkpoint": false,
413
+ "train_param": {
414
+ "max_length": 1024,
415
+ "per_device_train_batch_size": 8,
416
+ "auto_find_batch_size": true,
417
+ "num_train_epochs": 5,
418
+ "max_steps": -1,
419
+ "logging_strategy": "epoch",
420
+ "save_strategy": "epoch",
421
+ "logging_steps": 0.2,
422
+ "save_steps": 0.2,
423
+ "gradient_accumulation_steps": 1,
424
+ "gradient_checkpointing": false,
425
+ "optim": "adamw_torch",
426
+ "learning_rate": 5e-05,
427
+ "weight_decay": 0.01,
428
+ "warmup_steps": 0,
429
+ "report_to": "tensorboard",
430
+ "save_safetensors": true,
431
+ "packing": false
432
+ },
433
+ "use_peft": true,
434
+ "peft_task_type": "CAUSAL_LM",
435
+ "peft_inference_mode": false,
436
+ "peft_r": 8,
437
+ "peft_lora_alpha": 32,
438
+ "peft_lora_dropout": 0.1,
439
+ "fsdp": "none",
440
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
441
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
442
+ "fsdp_fsdp_min_num_params": 100000000.0,
443
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
444
+ "fsdp_fsdp_forward_prefetch": false,
445
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
446
+ "fsdp_fsdp_offload_params": false,
447
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
448
+ "fsdp_fsdp_use_orig_params": true,
449
+ "fsdp_fsdp_activation_checkpointing": false,
450
+ "fsdp_fsdp_limit_all_gathers": true,
451
+ "fsdp_fsdp_sync_module_states": true,
452
+ "use_git": true,
453
+ "random_seed": 0
454
+ }
455
+ },
456
+ "arc_challenge": {
457
+ "acc_norm": 0.45819112627986347,
458
+ "acc_norm_stderr": 0.014560220308714698,
459
+ "args": {
460
+ "exp_name": "results/v2/arc_challenge/Qwen/Qwen1.5-4B-Chat",
461
+ "task_name": "arc_challenge",
462
+ "model_name": "Qwen/Qwen1.5-4B-Chat",
463
+ "dataset_param": {
464
+ "max_num_train": 50000,
465
+ "completion_only": false,
466
+ "apply_chat_template": true
467
+ },
468
+ "resume_from_checkpoint": false,
469
+ "train_param": {
470
+ "max_length": 1024,
471
+ "per_device_train_batch_size": 8,
472
+ "auto_find_batch_size": true,
473
+ "num_train_epochs": 5,
474
+ "max_steps": -1,
475
+ "logging_strategy": "epoch",
476
+ "save_strategy": "epoch",
477
+ "logging_steps": 0.2,
478
+ "save_steps": 0.2,
479
+ "gradient_accumulation_steps": 1,
480
+ "gradient_checkpointing": false,
481
+ "optim": "adamw_torch",
482
+ "learning_rate": 5e-05,
483
+ "weight_decay": 0.01,
484
+ "warmup_steps": 0,
485
+ "report_to": "tensorboard",
486
+ "save_safetensors": true,
487
+ "packing": false
488
+ },
489
+ "use_peft": true,
490
+ "peft_task_type": "CAUSAL_LM",
491
+ "peft_inference_mode": false,
492
+ "peft_r": 8,
493
+ "peft_lora_alpha": 32,
494
+ "peft_lora_dropout": 0.1,
495
+ "fsdp": "none",
496
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
497
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
498
+ "fsdp_fsdp_min_num_params": 100000000.0,
499
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
500
+ "fsdp_fsdp_forward_prefetch": false,
501
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
502
+ "fsdp_fsdp_offload_params": false,
503
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
504
+ "fsdp_fsdp_use_orig_params": true,
505
+ "fsdp_fsdp_activation_checkpointing": false,
506
+ "fsdp_fsdp_limit_all_gathers": true,
507
+ "fsdp_fsdp_sync_module_states": true,
508
+ "use_git": true,
509
+ "random_seed": 0
510
+ }
511
+ }
512
+ }
513
+ }
Qwen/Qwen1.5-4B/results_2026-01-05-04-27-50.json ADDED
@@ -0,0 +1,513 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "model_name": "Qwen/Qwen1.5-4B",
4
+ "model_type": "pretrained",
5
+ "params": 4.0
6
+ },
7
+ "results": {
8
+ "medmcqa": {
9
+ "acc_norm": 0.47788668419794406,
10
+ "acc_norm_stderr": 0.00772418799411289,
11
+ "args": {
12
+ "exp_name": "results/v1/medmcqa/Qwen/Qwen1.5-4B",
13
+ "task_name": "medmcqa",
14
+ "model_name": "Qwen/Qwen1.5-4B",
15
+ "dataset_param": {
16
+ "max_num_train": 50000,
17
+ "completion_only": false,
18
+ "apply_chat_template": false
19
+ },
20
+ "resume_from_checkpoint": false,
21
+ "train_param": {
22
+ "max_length": 1024,
23
+ "per_device_train_batch_size": 8,
24
+ "auto_find_batch_size": true,
25
+ "num_train_epochs": 1,
26
+ "max_steps": -1,
27
+ "logging_strategy": "steps",
28
+ "save_strategy": "steps",
29
+ "logging_steps": 0.2,
30
+ "save_steps": 0.2,
31
+ "gradient_accumulation_steps": 1,
32
+ "gradient_checkpointing": false,
33
+ "optim": "adamw_torch",
34
+ "learning_rate": 5e-05,
35
+ "weight_decay": 0.01,
36
+ "warmup_steps": 0,
37
+ "report_to": "tensorboard",
38
+ "save_safetensors": true,
39
+ "packing": false
40
+ },
41
+ "use_peft": true,
42
+ "peft_task_type": "CAUSAL_LM",
43
+ "peft_inference_mode": false,
44
+ "peft_r": 8,
45
+ "peft_lora_alpha": 32,
46
+ "peft_lora_dropout": 0.1,
47
+ "fsdp": "none",
48
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
49
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
50
+ "fsdp_fsdp_min_num_params": 100000000.0,
51
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
52
+ "fsdp_fsdp_forward_prefetch": false,
53
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
54
+ "fsdp_fsdp_offload_params": false,
55
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
56
+ "fsdp_fsdp_use_orig_params": true,
57
+ "fsdp_fsdp_activation_checkpointing": false,
58
+ "fsdp_fsdp_limit_all_gathers": true,
59
+ "fsdp_fsdp_sync_module_states": true,
60
+ "use_git": true,
61
+ "random_seed": 0
62
+ }
63
+ },
64
+ "nq_open": {
65
+ "exact_match,remove_whitespace": 0.19750692520775623,
66
+ "exact_match,remove_whitespace_stderr": 0.006627021347718897,
67
+ "args": {
68
+ "exp_name": "results/v1/nq_open/Qwen/Qwen1.5-4B",
69
+ "task_name": "nq_open",
70
+ "model_name": "Qwen/Qwen1.5-4B",
71
+ "dataset_param": {
72
+ "max_num_train": 50000,
73
+ "completion_only": false,
74
+ "apply_chat_template": false
75
+ },
76
+ "resume_from_checkpoint": false,
77
+ "train_param": {
78
+ "max_length": 1024,
79
+ "per_device_train_batch_size": 8,
80
+ "auto_find_batch_size": true,
81
+ "num_train_epochs": 1,
82
+ "max_steps": -1,
83
+ "logging_strategy": "steps",
84
+ "save_strategy": "steps",
85
+ "logging_steps": 0.2,
86
+ "save_steps": 0.2,
87
+ "gradient_accumulation_steps": 1,
88
+ "gradient_checkpointing": false,
89
+ "optim": "adamw_torch",
90
+ "learning_rate": 5e-05,
91
+ "weight_decay": 0.01,
92
+ "warmup_steps": 0,
93
+ "report_to": "tensorboard",
94
+ "save_safetensors": true,
95
+ "packing": false
96
+ },
97
+ "use_peft": true,
98
+ "peft_task_type": "CAUSAL_LM",
99
+ "peft_inference_mode": false,
100
+ "peft_r": 8,
101
+ "peft_lora_alpha": 32,
102
+ "peft_lora_dropout": 0.1,
103
+ "fsdp": "none",
104
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
105
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
106
+ "fsdp_fsdp_min_num_params": 100000000.0,
107
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
108
+ "fsdp_fsdp_forward_prefetch": false,
109
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
110
+ "fsdp_fsdp_offload_params": false,
111
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
112
+ "fsdp_fsdp_use_orig_params": true,
113
+ "fsdp_fsdp_activation_checkpointing": false,
114
+ "fsdp_fsdp_limit_all_gathers": true,
115
+ "fsdp_fsdp_sync_module_states": true,
116
+ "use_git": true,
117
+ "random_seed": 0
118
+ }
119
+ },
120
+ "winogrande": {
121
+ "acc": 0.7521704814522494,
122
+ "acc_stderr": 0.012134386019865344,
123
+ "args": {
124
+ "exp_name": "results/v1/winogrande/Qwen/Qwen1.5-4B",
125
+ "task_name": "winogrande",
126
+ "model_name": "Qwen/Qwen1.5-4B",
127
+ "dataset_param": {
128
+ "max_num_train": 50000,
129
+ "completion_only": false,
130
+ "apply_chat_template": false
131
+ },
132
+ "resume_from_checkpoint": false,
133
+ "train_param": {
134
+ "max_length": 1024,
135
+ "per_device_train_batch_size": 8,
136
+ "auto_find_batch_size": true,
137
+ "num_train_epochs": 1,
138
+ "max_steps": -1,
139
+ "logging_strategy": "steps",
140
+ "save_strategy": "steps",
141
+ "logging_steps": 0.2,
142
+ "save_steps": 0.2,
143
+ "gradient_accumulation_steps": 1,
144
+ "gradient_checkpointing": false,
145
+ "optim": "adamw_torch",
146
+ "learning_rate": 5e-05,
147
+ "weight_decay": 0.01,
148
+ "warmup_steps": 0,
149
+ "report_to": "tensorboard",
150
+ "save_safetensors": true,
151
+ "packing": false
152
+ },
153
+ "use_peft": true,
154
+ "peft_task_type": "CAUSAL_LM",
155
+ "peft_inference_mode": false,
156
+ "peft_r": 8,
157
+ "peft_lora_alpha": 32,
158
+ "peft_lora_dropout": 0.1,
159
+ "fsdp": "none",
160
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
161
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
162
+ "fsdp_fsdp_min_num_params": 100000000.0,
163
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
164
+ "fsdp_fsdp_forward_prefetch": false,
165
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
166
+ "fsdp_fsdp_offload_params": false,
167
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
168
+ "fsdp_fsdp_use_orig_params": true,
169
+ "fsdp_fsdp_activation_checkpointing": false,
170
+ "fsdp_fsdp_limit_all_gathers": true,
171
+ "fsdp_fsdp_sync_module_states": true,
172
+ "use_git": true,
173
+ "random_seed": 0
174
+ }
175
+ },
176
+ "hellaswag": {
177
+ "acc_norm": 0.738,
178
+ "acc_norm_stderr": 0.004397446276348706,
179
+ "args": {
180
+ "exp_name": "results/v1/hellaswag/Qwen/Qwen1.5-4B",
181
+ "task_name": "hellaswag",
182
+ "model_name": "Qwen/Qwen1.5-4B",
183
+ "dataset_param": {
184
+ "max_num_train": 50000,
185
+ "completion_only": false,
186
+ "apply_chat_template": false
187
+ },
188
+ "resume_from_checkpoint": false,
189
+ "train_param": {
190
+ "max_length": 1024,
191
+ "per_device_train_batch_size": 8,
192
+ "auto_find_batch_size": true,
193
+ "num_train_epochs": 1,
194
+ "max_steps": -1,
195
+ "logging_strategy": "steps",
196
+ "save_strategy": "steps",
197
+ "logging_steps": 0.2,
198
+ "save_steps": 0.2,
199
+ "gradient_accumulation_steps": 1,
200
+ "gradient_checkpointing": false,
201
+ "optim": "adamw_torch",
202
+ "learning_rate": 5e-05,
203
+ "weight_decay": 0.01,
204
+ "warmup_steps": 0,
205
+ "report_to": "tensorboard",
206
+ "save_safetensors": true,
207
+ "packing": false
208
+ },
209
+ "use_peft": true,
210
+ "peft_task_type": "CAUSAL_LM",
211
+ "peft_inference_mode": false,
212
+ "peft_r": 8,
213
+ "peft_lora_alpha": 32,
214
+ "peft_lora_dropout": 0.1,
215
+ "fsdp": "none",
216
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
217
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
218
+ "fsdp_fsdp_min_num_params": 100000000.0,
219
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
220
+ "fsdp_fsdp_forward_prefetch": false,
221
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
222
+ "fsdp_fsdp_offload_params": false,
223
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
224
+ "fsdp_fsdp_use_orig_params": true,
225
+ "fsdp_fsdp_activation_checkpointing": false,
226
+ "fsdp_fsdp_limit_all_gathers": true,
227
+ "fsdp_fsdp_sync_module_states": true,
228
+ "use_git": true,
229
+ "random_seed": 0
230
+ }
231
+ },
232
+ "social_iqa": {
233
+ "acc": 0.5547594677584442,
234
+ "acc_stderr": 0.011246013134156763,
235
+ "args": {
236
+ "exp_name": "results/v1/social_iqa/Qwen/Qwen1.5-4B",
237
+ "task_name": "social_iqa",
238
+ "model_name": "Qwen/Qwen1.5-4B",
239
+ "dataset_param": {
240
+ "max_num_train": 50000,
241
+ "completion_only": false,
242
+ "apply_chat_template": false
243
+ },
244
+ "resume_from_checkpoint": false,
245
+ "train_param": {
246
+ "max_length": 1024,
247
+ "per_device_train_batch_size": 8,
248
+ "auto_find_batch_size": true,
249
+ "num_train_epochs": 1,
250
+ "max_steps": -1,
251
+ "logging_strategy": "steps",
252
+ "save_strategy": "steps",
253
+ "logging_steps": 0.2,
254
+ "save_steps": 0.2,
255
+ "gradient_accumulation_steps": 1,
256
+ "gradient_checkpointing": false,
257
+ "optim": "adamw_torch",
258
+ "learning_rate": 5e-05,
259
+ "weight_decay": 0.01,
260
+ "warmup_steps": 0,
261
+ "report_to": "tensorboard",
262
+ "save_safetensors": true,
263
+ "packing": false
264
+ },
265
+ "use_peft": true,
266
+ "peft_task_type": "CAUSAL_LM",
267
+ "peft_inference_mode": false,
268
+ "peft_r": 8,
269
+ "peft_lora_alpha": 32,
270
+ "peft_lora_dropout": 0.1,
271
+ "fsdp": "none",
272
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
273
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
274
+ "fsdp_fsdp_min_num_params": 100000000.0,
275
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
276
+ "fsdp_fsdp_forward_prefetch": false,
277
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
278
+ "fsdp_fsdp_offload_params": false,
279
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
280
+ "fsdp_fsdp_use_orig_params": true,
281
+ "fsdp_fsdp_activation_checkpointing": false,
282
+ "fsdp_fsdp_limit_all_gathers": true,
283
+ "fsdp_fsdp_sync_module_states": true,
284
+ "use_git": true,
285
+ "random_seed": 0
286
+ }
287
+ },
288
+ "piqa": {
289
+ "acc_norm": 0.7878128400435256,
290
+ "acc_norm_stderr": 0.009539299828174088,
291
+ "args": {
292
+ "exp_name": "results/v2/piqa/Qwen/Qwen1.5-4B",
293
+ "task_name": "piqa",
294
+ "model_name": "Qwen/Qwen1.5-4B",
295
+ "dataset_param": {
296
+ "max_num_train": 50000,
297
+ "completion_only": false,
298
+ "apply_chat_template": false
299
+ },
300
+ "resume_from_checkpoint": false,
301
+ "train_param": {
302
+ "max_length": 1024,
303
+ "per_device_train_batch_size": 8,
304
+ "auto_find_batch_size": true,
305
+ "num_train_epochs": 5,
306
+ "max_steps": -1,
307
+ "logging_strategy": "epoch",
308
+ "save_strategy": "epoch",
309
+ "logging_steps": 0.2,
310
+ "save_steps": 0.2,
311
+ "gradient_accumulation_steps": 1,
312
+ "gradient_checkpointing": false,
313
+ "optim": "adamw_torch",
314
+ "learning_rate": 5e-05,
315
+ "weight_decay": 0.01,
316
+ "warmup_steps": 0,
317
+ "report_to": "tensorboard",
318
+ "save_safetensors": true,
319
+ "packing": false
320
+ },
321
+ "use_peft": true,
322
+ "peft_task_type": "CAUSAL_LM",
323
+ "peft_inference_mode": false,
324
+ "peft_r": 8,
325
+ "peft_lora_alpha": 32,
326
+ "peft_lora_dropout": 0.1,
327
+ "fsdp": "none",
328
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
329
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
330
+ "fsdp_fsdp_min_num_params": 100000000.0,
331
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
332
+ "fsdp_fsdp_forward_prefetch": false,
333
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
334
+ "fsdp_fsdp_offload_params": false,
335
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
336
+ "fsdp_fsdp_use_orig_params": true,
337
+ "fsdp_fsdp_activation_checkpointing": false,
338
+ "fsdp_fsdp_limit_all_gathers": true,
339
+ "fsdp_fsdp_sync_module_states": true,
340
+ "use_git": true,
341
+ "random_seed": 0
342
+ }
343
+ },
344
+ "commonsense_qa": {
345
+ "acc": 0.8124488124488124,
346
+ "acc_stderr": 0.011175783964114751,
347
+ "args": {
348
+ "exp_name": "results/v2/commonsense_qa/Qwen/Qwen1.5-4B",
349
+ "task_name": "commonsense_qa",
350
+ "model_name": "Qwen/Qwen1.5-4B",
351
+ "dataset_param": {
352
+ "max_num_train": 50000,
353
+ "completion_only": false,
354
+ "apply_chat_template": false
355
+ },
356
+ "resume_from_checkpoint": false,
357
+ "train_param": {
358
+ "max_length": 1024,
359
+ "per_device_train_batch_size": 8,
360
+ "auto_find_batch_size": true,
361
+ "num_train_epochs": 5,
362
+ "max_steps": -1,
363
+ "logging_strategy": "epoch",
364
+ "save_strategy": "epoch",
365
+ "logging_steps": 0.2,
366
+ "save_steps": 0.2,
367
+ "gradient_accumulation_steps": 1,
368
+ "gradient_checkpointing": false,
369
+ "optim": "adamw_torch",
370
+ "learning_rate": 5e-05,
371
+ "weight_decay": 0.01,
372
+ "warmup_steps": 0,
373
+ "report_to": "tensorboard",
374
+ "save_safetensors": true,
375
+ "packing": false
376
+ },
377
+ "use_peft": true,
378
+ "peft_task_type": "CAUSAL_LM",
379
+ "peft_inference_mode": false,
380
+ "peft_r": 8,
381
+ "peft_lora_alpha": 32,
382
+ "peft_lora_dropout": 0.1,
383
+ "fsdp": "none",
384
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
385
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
386
+ "fsdp_fsdp_min_num_params": 100000000.0,
387
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
388
+ "fsdp_fsdp_forward_prefetch": false,
389
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
390
+ "fsdp_fsdp_offload_params": false,
391
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
392
+ "fsdp_fsdp_use_orig_params": true,
393
+ "fsdp_fsdp_activation_checkpointing": false,
394
+ "fsdp_fsdp_limit_all_gathers": true,
395
+ "fsdp_fsdp_sync_module_states": true,
396
+ "use_git": true,
397
+ "random_seed": 0
398
+ }
399
+ },
400
+ "gsm8k": {
401
+ "exact_match,flexible-extract": 0.5261561789234268,
402
+ "exact_match,flexible-extract_stderr": 0.013753627037255045,
403
+ "args": {
404
+ "exp_name": "results/v2/gsm8k/Qwen/Qwen1.5-4B",
405
+ "task_name": "gsm8k",
406
+ "model_name": "Qwen/Qwen1.5-4B",
407
+ "dataset_param": {
408
+ "max_num_train": 50000,
409
+ "completion_only": false,
410
+ "apply_chat_template": false
411
+ },
412
+ "resume_from_checkpoint": false,
413
+ "train_param": {
414
+ "max_length": 1024,
415
+ "per_device_train_batch_size": 8,
416
+ "auto_find_batch_size": true,
417
+ "num_train_epochs": 5,
418
+ "max_steps": -1,
419
+ "logging_strategy": "epoch",
420
+ "save_strategy": "epoch",
421
+ "logging_steps": 0.2,
422
+ "save_steps": 0.2,
423
+ "gradient_accumulation_steps": 1,
424
+ "gradient_checkpointing": false,
425
+ "optim": "adamw_torch",
426
+ "learning_rate": 5e-05,
427
+ "weight_decay": 0.01,
428
+ "warmup_steps": 0,
429
+ "report_to": "tensorboard",
430
+ "save_safetensors": true,
431
+ "packing": false
432
+ },
433
+ "use_peft": true,
434
+ "peft_task_type": "CAUSAL_LM",
435
+ "peft_inference_mode": false,
436
+ "peft_r": 8,
437
+ "peft_lora_alpha": 32,
438
+ "peft_lora_dropout": 0.1,
439
+ "fsdp": "none",
440
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
441
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
442
+ "fsdp_fsdp_min_num_params": 100000000.0,
443
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
444
+ "fsdp_fsdp_forward_prefetch": false,
445
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
446
+ "fsdp_fsdp_offload_params": false,
447
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
448
+ "fsdp_fsdp_use_orig_params": true,
449
+ "fsdp_fsdp_activation_checkpointing": false,
450
+ "fsdp_fsdp_limit_all_gathers": true,
451
+ "fsdp_fsdp_sync_module_states": true,
452
+ "use_git": true,
453
+ "random_seed": 0
454
+ }
455
+ },
456
+ "arc_challenge": {
457
+ "acc_norm": 0.5008532423208191,
458
+ "acc_norm_stderr": 0.014611369529813279,
459
+ "args": {
460
+ "exp_name": "results/v2/arc_challenge/Qwen/Qwen1.5-4B",
461
+ "task_name": "arc_challenge",
462
+ "model_name": "Qwen/Qwen1.5-4B",
463
+ "dataset_param": {
464
+ "max_num_train": 50000,
465
+ "completion_only": false,
466
+ "apply_chat_template": false
467
+ },
468
+ "resume_from_checkpoint": false,
469
+ "train_param": {
470
+ "max_length": 1024,
471
+ "per_device_train_batch_size": 8,
472
+ "auto_find_batch_size": true,
473
+ "num_train_epochs": 5,
474
+ "max_steps": -1,
475
+ "logging_strategy": "epoch",
476
+ "save_strategy": "epoch",
477
+ "logging_steps": 0.2,
478
+ "save_steps": 0.2,
479
+ "gradient_accumulation_steps": 1,
480
+ "gradient_checkpointing": false,
481
+ "optim": "adamw_torch",
482
+ "learning_rate": 5e-05,
483
+ "weight_decay": 0.01,
484
+ "warmup_steps": 0,
485
+ "report_to": "tensorboard",
486
+ "save_safetensors": true,
487
+ "packing": false
488
+ },
489
+ "use_peft": true,
490
+ "peft_task_type": "CAUSAL_LM",
491
+ "peft_inference_mode": false,
492
+ "peft_r": 8,
493
+ "peft_lora_alpha": 32,
494
+ "peft_lora_dropout": 0.1,
495
+ "fsdp": "none",
496
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
497
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
498
+ "fsdp_fsdp_min_num_params": 100000000.0,
499
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
500
+ "fsdp_fsdp_forward_prefetch": false,
501
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
502
+ "fsdp_fsdp_offload_params": false,
503
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
504
+ "fsdp_fsdp_use_orig_params": true,
505
+ "fsdp_fsdp_activation_checkpointing": false,
506
+ "fsdp_fsdp_limit_all_gathers": true,
507
+ "fsdp_fsdp_sync_module_states": true,
508
+ "use_git": true,
509
+ "random_seed": 0
510
+ }
511
+ }
512
+ }
513
+ }
Qwen/Qwen1.5-7B-Chat/results_2026-01-05-04-27-51.json ADDED
@@ -0,0 +1,513 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "model_name": "Qwen/Qwen1.5-7B-Chat",
4
+ "model_type": "instruction-tuned",
5
+ "params": 7.0
6
+ },
7
+ "results": {
8
+ "medmcqa": {
9
+ "acc_norm": 0.5010757829309108,
10
+ "acc_norm_stderr": 0.007731735434496672,
11
+ "args": {
12
+ "exp_name": "results/v1/medmcqa/Qwen/Qwen1.5-7B-Chat",
13
+ "task_name": "medmcqa",
14
+ "model_name": "Qwen/Qwen1.5-7B-Chat",
15
+ "dataset_param": {
16
+ "max_num_train": 50000,
17
+ "completion_only": false,
18
+ "apply_chat_template": true
19
+ },
20
+ "resume_from_checkpoint": false,
21
+ "train_param": {
22
+ "max_length": 1024,
23
+ "per_device_train_batch_size": 8,
24
+ "auto_find_batch_size": true,
25
+ "num_train_epochs": 1,
26
+ "max_steps": -1,
27
+ "logging_strategy": "steps",
28
+ "save_strategy": "steps",
29
+ "logging_steps": 0.2,
30
+ "save_steps": 0.2,
31
+ "gradient_accumulation_steps": 1,
32
+ "gradient_checkpointing": false,
33
+ "optim": "adamw_torch",
34
+ "learning_rate": 5e-05,
35
+ "weight_decay": 0.01,
36
+ "warmup_steps": 0,
37
+ "report_to": "tensorboard",
38
+ "save_safetensors": true,
39
+ "packing": false
40
+ },
41
+ "use_peft": true,
42
+ "peft_task_type": "CAUSAL_LM",
43
+ "peft_inference_mode": false,
44
+ "peft_r": 8,
45
+ "peft_lora_alpha": 32,
46
+ "peft_lora_dropout": 0.1,
47
+ "fsdp": "none",
48
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
49
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
50
+ "fsdp_fsdp_min_num_params": 100000000.0,
51
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
52
+ "fsdp_fsdp_forward_prefetch": false,
53
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
54
+ "fsdp_fsdp_offload_params": false,
55
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
56
+ "fsdp_fsdp_use_orig_params": true,
57
+ "fsdp_fsdp_activation_checkpointing": false,
58
+ "fsdp_fsdp_limit_all_gathers": true,
59
+ "fsdp_fsdp_sync_module_states": true,
60
+ "use_git": true,
61
+ "random_seed": 0
62
+ }
63
+ },
64
+ "nq_open": {
65
+ "exact_match,remove_whitespace": 0.2329639889196676,
66
+ "exact_match,remove_whitespace_stderr": 0.007036534686169712,
67
+ "args": {
68
+ "exp_name": "results/v1/nq_open/Qwen/Qwen1.5-7B-Chat",
69
+ "task_name": "nq_open",
70
+ "model_name": "Qwen/Qwen1.5-7B-Chat",
71
+ "dataset_param": {
72
+ "max_num_train": 50000,
73
+ "completion_only": false,
74
+ "apply_chat_template": true
75
+ },
76
+ "resume_from_checkpoint": false,
77
+ "train_param": {
78
+ "max_length": 1024,
79
+ "per_device_train_batch_size": 8,
80
+ "auto_find_batch_size": true,
81
+ "num_train_epochs": 1,
82
+ "max_steps": -1,
83
+ "logging_strategy": "steps",
84
+ "save_strategy": "steps",
85
+ "logging_steps": 0.2,
86
+ "save_steps": 0.2,
87
+ "gradient_accumulation_steps": 1,
88
+ "gradient_checkpointing": false,
89
+ "optim": "adamw_torch",
90
+ "learning_rate": 5e-05,
91
+ "weight_decay": 0.01,
92
+ "warmup_steps": 0,
93
+ "report_to": "tensorboard",
94
+ "save_safetensors": true,
95
+ "packing": false
96
+ },
97
+ "use_peft": true,
98
+ "peft_task_type": "CAUSAL_LM",
99
+ "peft_inference_mode": false,
100
+ "peft_r": 8,
101
+ "peft_lora_alpha": 32,
102
+ "peft_lora_dropout": 0.1,
103
+ "fsdp": "none",
104
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
105
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
106
+ "fsdp_fsdp_min_num_params": 100000000.0,
107
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
108
+ "fsdp_fsdp_forward_prefetch": false,
109
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
110
+ "fsdp_fsdp_offload_params": false,
111
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
112
+ "fsdp_fsdp_use_orig_params": true,
113
+ "fsdp_fsdp_activation_checkpointing": false,
114
+ "fsdp_fsdp_limit_all_gathers": true,
115
+ "fsdp_fsdp_sync_module_states": true,
116
+ "use_git": true,
117
+ "random_seed": 0
118
+ }
119
+ },
120
+ "winogrande": {
121
+ "acc": 0.7797947908445146,
122
+ "acc_stderr": 0.011646276755089679,
123
+ "args": {
124
+ "exp_name": "results/v1/winogrande/Qwen/Qwen1.5-7B-Chat",
125
+ "task_name": "winogrande",
126
+ "model_name": "Qwen/Qwen1.5-7B-Chat",
127
+ "dataset_param": {
128
+ "max_num_train": 50000,
129
+ "completion_only": false,
130
+ "apply_chat_template": true
131
+ },
132
+ "resume_from_checkpoint": false,
133
+ "train_param": {
134
+ "max_length": 1024,
135
+ "per_device_train_batch_size": 8,
136
+ "auto_find_batch_size": true,
137
+ "num_train_epochs": 1,
138
+ "max_steps": -1,
139
+ "logging_strategy": "steps",
140
+ "save_strategy": "steps",
141
+ "logging_steps": 0.2,
142
+ "save_steps": 0.2,
143
+ "gradient_accumulation_steps": 1,
144
+ "gradient_checkpointing": false,
145
+ "optim": "adamw_torch",
146
+ "learning_rate": 5e-05,
147
+ "weight_decay": 0.01,
148
+ "warmup_steps": 0,
149
+ "report_to": "tensorboard",
150
+ "save_safetensors": true,
151
+ "packing": false
152
+ },
153
+ "use_peft": true,
154
+ "peft_task_type": "CAUSAL_LM",
155
+ "peft_inference_mode": false,
156
+ "peft_r": 8,
157
+ "peft_lora_alpha": 32,
158
+ "peft_lora_dropout": 0.1,
159
+ "fsdp": "none",
160
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
161
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
162
+ "fsdp_fsdp_min_num_params": 100000000.0,
163
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
164
+ "fsdp_fsdp_forward_prefetch": false,
165
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
166
+ "fsdp_fsdp_offload_params": false,
167
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
168
+ "fsdp_fsdp_use_orig_params": true,
169
+ "fsdp_fsdp_activation_checkpointing": false,
170
+ "fsdp_fsdp_limit_all_gathers": true,
171
+ "fsdp_fsdp_sync_module_states": true,
172
+ "use_git": true,
173
+ "random_seed": 0
174
+ }
175
+ },
176
+ "hellaswag": {
177
+ "acc_norm": 0.7801,
178
+ "acc_norm_stderr": 0.004141994038076248,
179
+ "args": {
180
+ "exp_name": "results/v1/hellaswag/Qwen/Qwen1.5-7B-Chat",
181
+ "task_name": "hellaswag",
182
+ "model_name": "Qwen/Qwen1.5-7B-Chat",
183
+ "dataset_param": {
184
+ "max_num_train": 50000,
185
+ "completion_only": false,
186
+ "apply_chat_template": true
187
+ },
188
+ "resume_from_checkpoint": false,
189
+ "train_param": {
190
+ "max_length": 1024,
191
+ "per_device_train_batch_size": 8,
192
+ "auto_find_batch_size": true,
193
+ "num_train_epochs": 1,
194
+ "max_steps": -1,
195
+ "logging_strategy": "steps",
196
+ "save_strategy": "steps",
197
+ "logging_steps": 0.2,
198
+ "save_steps": 0.2,
199
+ "gradient_accumulation_steps": 1,
200
+ "gradient_checkpointing": false,
201
+ "optim": "adamw_torch",
202
+ "learning_rate": 5e-05,
203
+ "weight_decay": 0.01,
204
+ "warmup_steps": 0,
205
+ "report_to": "tensorboard",
206
+ "save_safetensors": true,
207
+ "packing": false
208
+ },
209
+ "use_peft": true,
210
+ "peft_task_type": "CAUSAL_LM",
211
+ "peft_inference_mode": false,
212
+ "peft_r": 8,
213
+ "peft_lora_alpha": 32,
214
+ "peft_lora_dropout": 0.1,
215
+ "fsdp": "none",
216
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
217
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
218
+ "fsdp_fsdp_min_num_params": 100000000.0,
219
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
220
+ "fsdp_fsdp_forward_prefetch": false,
221
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
222
+ "fsdp_fsdp_offload_params": false,
223
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
224
+ "fsdp_fsdp_use_orig_params": true,
225
+ "fsdp_fsdp_activation_checkpointing": false,
226
+ "fsdp_fsdp_limit_all_gathers": true,
227
+ "fsdp_fsdp_sync_module_states": true,
228
+ "use_git": true,
229
+ "random_seed": 0
230
+ }
231
+ },
232
+ "social_iqa": {
233
+ "acc": 0.5353121801432958,
234
+ "acc_stderr": 0.011285819120808803,
235
+ "args": {
236
+ "exp_name": "results/v1/social_iqa/Qwen/Qwen1.5-7B-Chat",
237
+ "task_name": "social_iqa",
238
+ "model_name": "Qwen/Qwen1.5-7B-Chat",
239
+ "dataset_param": {
240
+ "max_num_train": 50000,
241
+ "completion_only": false,
242
+ "apply_chat_template": true
243
+ },
244
+ "resume_from_checkpoint": false,
245
+ "train_param": {
246
+ "max_length": 1024,
247
+ "per_device_train_batch_size": 8,
248
+ "auto_find_batch_size": true,
249
+ "num_train_epochs": 1,
250
+ "max_steps": -1,
251
+ "logging_strategy": "steps",
252
+ "save_strategy": "steps",
253
+ "logging_steps": 0.2,
254
+ "save_steps": 0.2,
255
+ "gradient_accumulation_steps": 1,
256
+ "gradient_checkpointing": false,
257
+ "optim": "adamw_torch",
258
+ "learning_rate": 5e-05,
259
+ "weight_decay": 0.01,
260
+ "warmup_steps": 0,
261
+ "report_to": "tensorboard",
262
+ "save_safetensors": true,
263
+ "packing": false
264
+ },
265
+ "use_peft": true,
266
+ "peft_task_type": "CAUSAL_LM",
267
+ "peft_inference_mode": false,
268
+ "peft_r": 8,
269
+ "peft_lora_alpha": 32,
270
+ "peft_lora_dropout": 0.1,
271
+ "fsdp": "none",
272
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
273
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
274
+ "fsdp_fsdp_min_num_params": 100000000.0,
275
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
276
+ "fsdp_fsdp_forward_prefetch": false,
277
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
278
+ "fsdp_fsdp_offload_params": false,
279
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
280
+ "fsdp_fsdp_use_orig_params": true,
281
+ "fsdp_fsdp_activation_checkpointing": false,
282
+ "fsdp_fsdp_limit_all_gathers": true,
283
+ "fsdp_fsdp_sync_module_states": true,
284
+ "use_git": true,
285
+ "random_seed": 0
286
+ }
287
+ },
288
+ "piqa": {
289
+ "acc_norm": 0.8003264417845484,
290
+ "acc_norm_stderr": 0.009326942154519176,
291
+ "args": {
292
+ "exp_name": "results/v2/piqa/Qwen/Qwen1.5-7B-Chat",
293
+ "task_name": "piqa",
294
+ "model_name": "Qwen/Qwen1.5-7B-Chat",
295
+ "dataset_param": {
296
+ "max_num_train": 50000,
297
+ "completion_only": false,
298
+ "apply_chat_template": true
299
+ },
300
+ "resume_from_checkpoint": false,
301
+ "train_param": {
302
+ "max_length": 1024,
303
+ "per_device_train_batch_size": 8,
304
+ "auto_find_batch_size": true,
305
+ "num_train_epochs": 5,
306
+ "max_steps": -1,
307
+ "logging_strategy": "epoch",
308
+ "save_strategy": "epoch",
309
+ "logging_steps": 0.2,
310
+ "save_steps": 0.2,
311
+ "gradient_accumulation_steps": 1,
312
+ "gradient_checkpointing": false,
313
+ "optim": "adamw_torch",
314
+ "learning_rate": 5e-05,
315
+ "weight_decay": 0.01,
316
+ "warmup_steps": 0,
317
+ "report_to": "tensorboard",
318
+ "save_safetensors": true,
319
+ "packing": false
320
+ },
321
+ "use_peft": true,
322
+ "peft_task_type": "CAUSAL_LM",
323
+ "peft_inference_mode": false,
324
+ "peft_r": 8,
325
+ "peft_lora_alpha": 32,
326
+ "peft_lora_dropout": 0.1,
327
+ "fsdp": "none",
328
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
329
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
330
+ "fsdp_fsdp_min_num_params": 100000000.0,
331
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
332
+ "fsdp_fsdp_forward_prefetch": false,
333
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
334
+ "fsdp_fsdp_offload_params": false,
335
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
336
+ "fsdp_fsdp_use_orig_params": true,
337
+ "fsdp_fsdp_activation_checkpointing": false,
338
+ "fsdp_fsdp_limit_all_gathers": true,
339
+ "fsdp_fsdp_sync_module_states": true,
340
+ "use_git": true,
341
+ "random_seed": 0
342
+ }
343
+ },
344
+ "commonsense_qa": {
345
+ "acc": 0.8411138411138411,
346
+ "acc_stderr": 0.010466235619039333,
347
+ "args": {
348
+ "exp_name": "results/v2/commonsense_qa/Qwen/Qwen1.5-7B-Chat",
349
+ "task_name": "commonsense_qa",
350
+ "model_name": "Qwen/Qwen1.5-7B-Chat",
351
+ "dataset_param": {
352
+ "max_num_train": 50000,
353
+ "completion_only": false,
354
+ "apply_chat_template": true
355
+ },
356
+ "resume_from_checkpoint": false,
357
+ "train_param": {
358
+ "max_length": 1024,
359
+ "per_device_train_batch_size": 8,
360
+ "auto_find_batch_size": true,
361
+ "num_train_epochs": 5,
362
+ "max_steps": -1,
363
+ "logging_strategy": "epoch",
364
+ "save_strategy": "epoch",
365
+ "logging_steps": 0.2,
366
+ "save_steps": 0.2,
367
+ "gradient_accumulation_steps": 1,
368
+ "gradient_checkpointing": false,
369
+ "optim": "adamw_torch",
370
+ "learning_rate": 5e-05,
371
+ "weight_decay": 0.01,
372
+ "warmup_steps": 0,
373
+ "report_to": "tensorboard",
374
+ "save_safetensors": true,
375
+ "packing": false
376
+ },
377
+ "use_peft": true,
378
+ "peft_task_type": "CAUSAL_LM",
379
+ "peft_inference_mode": false,
380
+ "peft_r": 8,
381
+ "peft_lora_alpha": 32,
382
+ "peft_lora_dropout": 0.1,
383
+ "fsdp": "none",
384
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
385
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
386
+ "fsdp_fsdp_min_num_params": 100000000.0,
387
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
388
+ "fsdp_fsdp_forward_prefetch": false,
389
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
390
+ "fsdp_fsdp_offload_params": false,
391
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
392
+ "fsdp_fsdp_use_orig_params": true,
393
+ "fsdp_fsdp_activation_checkpointing": false,
394
+ "fsdp_fsdp_limit_all_gathers": true,
395
+ "fsdp_fsdp_sync_module_states": true,
396
+ "use_git": true,
397
+ "random_seed": 0
398
+ }
399
+ },
400
+ "gsm8k": {
401
+ "exact_match,flexible-extract": 0.5398028809704322,
402
+ "exact_match,flexible-extract_stderr": 0.013728776714099363,
403
+ "args": {
404
+ "exp_name": "results/v2/gsm8k/Qwen/Qwen1.5-7B-Chat",
405
+ "task_name": "gsm8k",
406
+ "model_name": "Qwen/Qwen1.5-7B-Chat",
407
+ "dataset_param": {
408
+ "max_num_train": 50000,
409
+ "completion_only": false,
410
+ "apply_chat_template": true
411
+ },
412
+ "resume_from_checkpoint": false,
413
+ "train_param": {
414
+ "max_length": 1024,
415
+ "per_device_train_batch_size": 8,
416
+ "auto_find_batch_size": true,
417
+ "num_train_epochs": 5,
418
+ "max_steps": -1,
419
+ "logging_strategy": "epoch",
420
+ "save_strategy": "epoch",
421
+ "logging_steps": 0.2,
422
+ "save_steps": 0.2,
423
+ "gradient_accumulation_steps": 1,
424
+ "gradient_checkpointing": false,
425
+ "optim": "adamw_torch",
426
+ "learning_rate": 5e-05,
427
+ "weight_decay": 0.01,
428
+ "warmup_steps": 0,
429
+ "report_to": "tensorboard",
430
+ "save_safetensors": true,
431
+ "packing": false
432
+ },
433
+ "use_peft": true,
434
+ "peft_task_type": "CAUSAL_LM",
435
+ "peft_inference_mode": false,
436
+ "peft_r": 8,
437
+ "peft_lora_alpha": 32,
438
+ "peft_lora_dropout": 0.1,
439
+ "fsdp": "none",
440
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
441
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
442
+ "fsdp_fsdp_min_num_params": 100000000.0,
443
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
444
+ "fsdp_fsdp_forward_prefetch": false,
445
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
446
+ "fsdp_fsdp_offload_params": false,
447
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
448
+ "fsdp_fsdp_use_orig_params": true,
449
+ "fsdp_fsdp_activation_checkpointing": false,
450
+ "fsdp_fsdp_limit_all_gathers": true,
451
+ "fsdp_fsdp_sync_module_states": true,
452
+ "use_git": true,
453
+ "random_seed": 0
454
+ }
455
+ },
456
+ "arc_challenge": {
457
+ "acc_norm": 0.5247440273037542,
458
+ "acc_norm_stderr": 0.01459348769493774,
459
+ "args": {
460
+ "exp_name": "results/v2/arc_challenge/Qwen/Qwen1.5-7B-Chat",
461
+ "task_name": "arc_challenge",
462
+ "model_name": "Qwen/Qwen1.5-7B-Chat",
463
+ "dataset_param": {
464
+ "max_num_train": 50000,
465
+ "completion_only": false,
466
+ "apply_chat_template": true
467
+ },
468
+ "resume_from_checkpoint": false,
469
+ "train_param": {
470
+ "max_length": 1024,
471
+ "per_device_train_batch_size": 8,
472
+ "auto_find_batch_size": true,
473
+ "num_train_epochs": 5,
474
+ "max_steps": -1,
475
+ "logging_strategy": "epoch",
476
+ "save_strategy": "epoch",
477
+ "logging_steps": 0.2,
478
+ "save_steps": 0.2,
479
+ "gradient_accumulation_steps": 1,
480
+ "gradient_checkpointing": false,
481
+ "optim": "adamw_torch",
482
+ "learning_rate": 5e-05,
483
+ "weight_decay": 0.01,
484
+ "warmup_steps": 0,
485
+ "report_to": "tensorboard",
486
+ "save_safetensors": true,
487
+ "packing": false
488
+ },
489
+ "use_peft": true,
490
+ "peft_task_type": "CAUSAL_LM",
491
+ "peft_inference_mode": false,
492
+ "peft_r": 8,
493
+ "peft_lora_alpha": 32,
494
+ "peft_lora_dropout": 0.1,
495
+ "fsdp": "none",
496
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
497
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
498
+ "fsdp_fsdp_min_num_params": 100000000.0,
499
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
500
+ "fsdp_fsdp_forward_prefetch": false,
501
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
502
+ "fsdp_fsdp_offload_params": false,
503
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
504
+ "fsdp_fsdp_use_orig_params": true,
505
+ "fsdp_fsdp_activation_checkpointing": false,
506
+ "fsdp_fsdp_limit_all_gathers": true,
507
+ "fsdp_fsdp_sync_module_states": true,
508
+ "use_git": true,
509
+ "random_seed": 0
510
+ }
511
+ }
512
+ }
513
+ }
Qwen/Qwen1.5-7B/results_2026-01-05-04-27-50.json ADDED
@@ -0,0 +1,513 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "model_name": "Qwen/Qwen1.5-7B",
4
+ "model_type": "pretrained",
5
+ "params": 7.0
6
+ },
7
+ "results": {
8
+ "medmcqa": {
9
+ "acc_norm": 0.508247669136983,
10
+ "acc_norm_stderr": 0.007730701368716393,
11
+ "args": {
12
+ "exp_name": "results/v1/medmcqa/Qwen/Qwen1.5-7B",
13
+ "task_name": "medmcqa",
14
+ "model_name": "Qwen/Qwen1.5-7B",
15
+ "dataset_param": {
16
+ "max_num_train": 50000,
17
+ "completion_only": false,
18
+ "apply_chat_template": false
19
+ },
20
+ "resume_from_checkpoint": false,
21
+ "train_param": {
22
+ "max_length": 1024,
23
+ "per_device_train_batch_size": 8,
24
+ "auto_find_batch_size": true,
25
+ "num_train_epochs": 1,
26
+ "max_steps": -1,
27
+ "logging_strategy": "steps",
28
+ "save_strategy": "steps",
29
+ "logging_steps": 0.2,
30
+ "save_steps": 0.2,
31
+ "gradient_accumulation_steps": 1,
32
+ "gradient_checkpointing": false,
33
+ "optim": "adamw_torch",
34
+ "learning_rate": 5e-05,
35
+ "weight_decay": 0.01,
36
+ "warmup_steps": 0,
37
+ "report_to": "tensorboard",
38
+ "save_safetensors": true,
39
+ "packing": false
40
+ },
41
+ "use_peft": true,
42
+ "peft_task_type": "CAUSAL_LM",
43
+ "peft_inference_mode": false,
44
+ "peft_r": 8,
45
+ "peft_lora_alpha": 32,
46
+ "peft_lora_dropout": 0.1,
47
+ "fsdp": "none",
48
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
49
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
50
+ "fsdp_fsdp_min_num_params": 100000000.0,
51
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
52
+ "fsdp_fsdp_forward_prefetch": false,
53
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
54
+ "fsdp_fsdp_offload_params": false,
55
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
56
+ "fsdp_fsdp_use_orig_params": true,
57
+ "fsdp_fsdp_activation_checkpointing": false,
58
+ "fsdp_fsdp_limit_all_gathers": true,
59
+ "fsdp_fsdp_sync_module_states": true,
60
+ "use_git": true,
61
+ "random_seed": 0
62
+ }
63
+ },
64
+ "nq_open": {
65
+ "exact_match,remove_whitespace": 0.2554016620498615,
66
+ "exact_match,remove_whitespace_stderr": 0.007259044349946153,
67
+ "args": {
68
+ "exp_name": "results/v1/nq_open/Qwen/Qwen1.5-7B",
69
+ "task_name": "nq_open",
70
+ "model_name": "Qwen/Qwen1.5-7B",
71
+ "dataset_param": {
72
+ "max_num_train": 50000,
73
+ "completion_only": false,
74
+ "apply_chat_template": false
75
+ },
76
+ "resume_from_checkpoint": false,
77
+ "train_param": {
78
+ "max_length": 1024,
79
+ "per_device_train_batch_size": 8,
80
+ "auto_find_batch_size": true,
81
+ "num_train_epochs": 1,
82
+ "max_steps": -1,
83
+ "logging_strategy": "steps",
84
+ "save_strategy": "steps",
85
+ "logging_steps": 0.2,
86
+ "save_steps": 0.2,
87
+ "gradient_accumulation_steps": 1,
88
+ "gradient_checkpointing": false,
89
+ "optim": "adamw_torch",
90
+ "learning_rate": 5e-05,
91
+ "weight_decay": 0.01,
92
+ "warmup_steps": 0,
93
+ "report_to": "tensorboard",
94
+ "save_safetensors": true,
95
+ "packing": false
96
+ },
97
+ "use_peft": true,
98
+ "peft_task_type": "CAUSAL_LM",
99
+ "peft_inference_mode": false,
100
+ "peft_r": 8,
101
+ "peft_lora_alpha": 32,
102
+ "peft_lora_dropout": 0.1,
103
+ "fsdp": "none",
104
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
105
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
106
+ "fsdp_fsdp_min_num_params": 100000000.0,
107
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
108
+ "fsdp_fsdp_forward_prefetch": false,
109
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
110
+ "fsdp_fsdp_offload_params": false,
111
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
112
+ "fsdp_fsdp_use_orig_params": true,
113
+ "fsdp_fsdp_activation_checkpointing": false,
114
+ "fsdp_fsdp_limit_all_gathers": true,
115
+ "fsdp_fsdp_sync_module_states": true,
116
+ "use_git": true,
117
+ "random_seed": 0
118
+ }
119
+ },
120
+ "winogrande": {
121
+ "acc": 0.7971586424625099,
122
+ "acc_stderr": 0.011301439925936657,
123
+ "args": {
124
+ "exp_name": "results/v1/winogrande/Qwen/Qwen1.5-7B",
125
+ "task_name": "winogrande",
126
+ "model_name": "Qwen/Qwen1.5-7B",
127
+ "dataset_param": {
128
+ "max_num_train": 50000,
129
+ "completion_only": false,
130
+ "apply_chat_template": false
131
+ },
132
+ "resume_from_checkpoint": false,
133
+ "train_param": {
134
+ "max_length": 1024,
135
+ "per_device_train_batch_size": 8,
136
+ "auto_find_batch_size": true,
137
+ "num_train_epochs": 1,
138
+ "max_steps": -1,
139
+ "logging_strategy": "steps",
140
+ "save_strategy": "steps",
141
+ "logging_steps": 0.2,
142
+ "save_steps": 0.2,
143
+ "gradient_accumulation_steps": 1,
144
+ "gradient_checkpointing": false,
145
+ "optim": "adamw_torch",
146
+ "learning_rate": 5e-05,
147
+ "weight_decay": 0.01,
148
+ "warmup_steps": 0,
149
+ "report_to": "tensorboard",
150
+ "save_safetensors": true,
151
+ "packing": false
152
+ },
153
+ "use_peft": true,
154
+ "peft_task_type": "CAUSAL_LM",
155
+ "peft_inference_mode": false,
156
+ "peft_r": 8,
157
+ "peft_lora_alpha": 32,
158
+ "peft_lora_dropout": 0.1,
159
+ "fsdp": "none",
160
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
161
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
162
+ "fsdp_fsdp_min_num_params": 100000000.0,
163
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
164
+ "fsdp_fsdp_forward_prefetch": false,
165
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
166
+ "fsdp_fsdp_offload_params": false,
167
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
168
+ "fsdp_fsdp_use_orig_params": true,
169
+ "fsdp_fsdp_activation_checkpointing": false,
170
+ "fsdp_fsdp_limit_all_gathers": true,
171
+ "fsdp_fsdp_sync_module_states": true,
172
+ "use_git": true,
173
+ "random_seed": 0
174
+ }
175
+ },
176
+ "hellaswag": {
177
+ "acc_norm": 0.7918,
178
+ "acc_norm_stderr": 0.0040604094242415074,
179
+ "args": {
180
+ "exp_name": "results/v1/hellaswag/Qwen/Qwen1.5-7B",
181
+ "task_name": "hellaswag",
182
+ "model_name": "Qwen/Qwen1.5-7B",
183
+ "dataset_param": {
184
+ "max_num_train": 50000,
185
+ "completion_only": false,
186
+ "apply_chat_template": false
187
+ },
188
+ "resume_from_checkpoint": false,
189
+ "train_param": {
190
+ "max_length": 1024,
191
+ "per_device_train_batch_size": 8,
192
+ "auto_find_batch_size": true,
193
+ "num_train_epochs": 1,
194
+ "max_steps": -1,
195
+ "logging_strategy": "steps",
196
+ "save_strategy": "steps",
197
+ "logging_steps": 0.2,
198
+ "save_steps": 0.2,
199
+ "gradient_accumulation_steps": 1,
200
+ "gradient_checkpointing": false,
201
+ "optim": "adamw_torch",
202
+ "learning_rate": 5e-05,
203
+ "weight_decay": 0.01,
204
+ "warmup_steps": 0,
205
+ "report_to": "tensorboard",
206
+ "save_safetensors": true,
207
+ "packing": false
208
+ },
209
+ "use_peft": true,
210
+ "peft_task_type": "CAUSAL_LM",
211
+ "peft_inference_mode": false,
212
+ "peft_r": 8,
213
+ "peft_lora_alpha": 32,
214
+ "peft_lora_dropout": 0.1,
215
+ "fsdp": "none",
216
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
217
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
218
+ "fsdp_fsdp_min_num_params": 100000000.0,
219
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
220
+ "fsdp_fsdp_forward_prefetch": false,
221
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
222
+ "fsdp_fsdp_offload_params": false,
223
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
224
+ "fsdp_fsdp_use_orig_params": true,
225
+ "fsdp_fsdp_activation_checkpointing": false,
226
+ "fsdp_fsdp_limit_all_gathers": true,
227
+ "fsdp_fsdp_sync_module_states": true,
228
+ "use_git": true,
229
+ "random_seed": 0
230
+ }
231
+ },
232
+ "social_iqa": {
233
+ "acc": 0.5414534288638689,
234
+ "acc_stderr": 0.011275119608640891,
235
+ "args": {
236
+ "exp_name": "results/v1/social_iqa/Qwen/Qwen1.5-7B",
237
+ "task_name": "social_iqa",
238
+ "model_name": "Qwen/Qwen1.5-7B",
239
+ "dataset_param": {
240
+ "max_num_train": 50000,
241
+ "completion_only": false,
242
+ "apply_chat_template": false
243
+ },
244
+ "resume_from_checkpoint": false,
245
+ "train_param": {
246
+ "max_length": 1024,
247
+ "per_device_train_batch_size": 8,
248
+ "auto_find_batch_size": true,
249
+ "num_train_epochs": 1,
250
+ "max_steps": -1,
251
+ "logging_strategy": "steps",
252
+ "save_strategy": "steps",
253
+ "logging_steps": 0.2,
254
+ "save_steps": 0.2,
255
+ "gradient_accumulation_steps": 1,
256
+ "gradient_checkpointing": false,
257
+ "optim": "adamw_torch",
258
+ "learning_rate": 5e-05,
259
+ "weight_decay": 0.01,
260
+ "warmup_steps": 0,
261
+ "report_to": "tensorboard",
262
+ "save_safetensors": true,
263
+ "packing": false
264
+ },
265
+ "use_peft": true,
266
+ "peft_task_type": "CAUSAL_LM",
267
+ "peft_inference_mode": false,
268
+ "peft_r": 8,
269
+ "peft_lora_alpha": 32,
270
+ "peft_lora_dropout": 0.1,
271
+ "fsdp": "none",
272
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
273
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
274
+ "fsdp_fsdp_min_num_params": 100000000.0,
275
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
276
+ "fsdp_fsdp_forward_prefetch": false,
277
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
278
+ "fsdp_fsdp_offload_params": false,
279
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
280
+ "fsdp_fsdp_use_orig_params": true,
281
+ "fsdp_fsdp_activation_checkpointing": false,
282
+ "fsdp_fsdp_limit_all_gathers": true,
283
+ "fsdp_fsdp_sync_module_states": true,
284
+ "use_git": true,
285
+ "random_seed": 0
286
+ }
287
+ },
288
+ "piqa": {
289
+ "acc_norm": 0.8133841131664853,
290
+ "acc_norm_stderr": 0.009090077190470808,
291
+ "args": {
292
+ "exp_name": "results/v2/piqa/Qwen/Qwen1.5-7B",
293
+ "task_name": "piqa",
294
+ "model_name": "Qwen/Qwen1.5-7B",
295
+ "dataset_param": {
296
+ "max_num_train": 50000,
297
+ "completion_only": false,
298
+ "apply_chat_template": false
299
+ },
300
+ "resume_from_checkpoint": false,
301
+ "train_param": {
302
+ "max_length": 1024,
303
+ "per_device_train_batch_size": 8,
304
+ "auto_find_batch_size": true,
305
+ "num_train_epochs": 5,
306
+ "max_steps": -1,
307
+ "logging_strategy": "epoch",
308
+ "save_strategy": "epoch",
309
+ "logging_steps": 0.2,
310
+ "save_steps": 0.2,
311
+ "gradient_accumulation_steps": 1,
312
+ "gradient_checkpointing": false,
313
+ "optim": "adamw_torch",
314
+ "learning_rate": 5e-05,
315
+ "weight_decay": 0.01,
316
+ "warmup_steps": 0,
317
+ "report_to": "tensorboard",
318
+ "save_safetensors": true,
319
+ "packing": false
320
+ },
321
+ "use_peft": true,
322
+ "peft_task_type": "CAUSAL_LM",
323
+ "peft_inference_mode": false,
324
+ "peft_r": 8,
325
+ "peft_lora_alpha": 32,
326
+ "peft_lora_dropout": 0.1,
327
+ "fsdp": "none",
328
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
329
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
330
+ "fsdp_fsdp_min_num_params": 100000000.0,
331
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
332
+ "fsdp_fsdp_forward_prefetch": false,
333
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
334
+ "fsdp_fsdp_offload_params": false,
335
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
336
+ "fsdp_fsdp_use_orig_params": true,
337
+ "fsdp_fsdp_activation_checkpointing": false,
338
+ "fsdp_fsdp_limit_all_gathers": true,
339
+ "fsdp_fsdp_sync_module_states": true,
340
+ "use_git": true,
341
+ "random_seed": 0
342
+ }
343
+ },
344
+ "commonsense_qa": {
345
+ "acc": 0.8533988533988534,
346
+ "acc_stderr": 0.010126626581199826,
347
+ "args": {
348
+ "exp_name": "results/v2/commonsense_qa/Qwen/Qwen1.5-7B",
349
+ "task_name": "commonsense_qa",
350
+ "model_name": "Qwen/Qwen1.5-7B",
351
+ "dataset_param": {
352
+ "max_num_train": 50000,
353
+ "completion_only": false,
354
+ "apply_chat_template": false
355
+ },
356
+ "resume_from_checkpoint": false,
357
+ "train_param": {
358
+ "max_length": 1024,
359
+ "per_device_train_batch_size": 8,
360
+ "auto_find_batch_size": true,
361
+ "num_train_epochs": 5,
362
+ "max_steps": -1,
363
+ "logging_strategy": "epoch",
364
+ "save_strategy": "epoch",
365
+ "logging_steps": 0.2,
366
+ "save_steps": 0.2,
367
+ "gradient_accumulation_steps": 1,
368
+ "gradient_checkpointing": false,
369
+ "optim": "adamw_torch",
370
+ "learning_rate": 5e-05,
371
+ "weight_decay": 0.01,
372
+ "warmup_steps": 0,
373
+ "report_to": "tensorboard",
374
+ "save_safetensors": true,
375
+ "packing": false
376
+ },
377
+ "use_peft": true,
378
+ "peft_task_type": "CAUSAL_LM",
379
+ "peft_inference_mode": false,
380
+ "peft_r": 8,
381
+ "peft_lora_alpha": 32,
382
+ "peft_lora_dropout": 0.1,
383
+ "fsdp": "none",
384
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
385
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
386
+ "fsdp_fsdp_min_num_params": 100000000.0,
387
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
388
+ "fsdp_fsdp_forward_prefetch": false,
389
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
390
+ "fsdp_fsdp_offload_params": false,
391
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
392
+ "fsdp_fsdp_use_orig_params": true,
393
+ "fsdp_fsdp_activation_checkpointing": false,
394
+ "fsdp_fsdp_limit_all_gathers": true,
395
+ "fsdp_fsdp_sync_module_states": true,
396
+ "use_git": true,
397
+ "random_seed": 0
398
+ }
399
+ },
400
+ "gsm8k": {
401
+ "exact_match,flexible-extract": 0.5830174374526156,
402
+ "exact_match,flexible-extract_stderr": 0.013581320997216591,
403
+ "args": {
404
+ "exp_name": "results/v2/gsm8k/Qwen/Qwen1.5-7B",
405
+ "task_name": "gsm8k",
406
+ "model_name": "Qwen/Qwen1.5-7B",
407
+ "dataset_param": {
408
+ "max_num_train": 50000,
409
+ "completion_only": false,
410
+ "apply_chat_template": false
411
+ },
412
+ "resume_from_checkpoint": false,
413
+ "train_param": {
414
+ "max_length": 1024,
415
+ "per_device_train_batch_size": 8,
416
+ "auto_find_batch_size": true,
417
+ "num_train_epochs": 5,
418
+ "max_steps": -1,
419
+ "logging_strategy": "epoch",
420
+ "save_strategy": "epoch",
421
+ "logging_steps": 0.2,
422
+ "save_steps": 0.2,
423
+ "gradient_accumulation_steps": 1,
424
+ "gradient_checkpointing": false,
425
+ "optim": "adamw_torch",
426
+ "learning_rate": 5e-05,
427
+ "weight_decay": 0.01,
428
+ "warmup_steps": 0,
429
+ "report_to": "tensorboard",
430
+ "save_safetensors": true,
431
+ "packing": false
432
+ },
433
+ "use_peft": true,
434
+ "peft_task_type": "CAUSAL_LM",
435
+ "peft_inference_mode": false,
436
+ "peft_r": 8,
437
+ "peft_lora_alpha": 32,
438
+ "peft_lora_dropout": 0.1,
439
+ "fsdp": "none",
440
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
441
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
442
+ "fsdp_fsdp_min_num_params": 100000000.0,
443
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
444
+ "fsdp_fsdp_forward_prefetch": false,
445
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
446
+ "fsdp_fsdp_offload_params": false,
447
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
448
+ "fsdp_fsdp_use_orig_params": true,
449
+ "fsdp_fsdp_activation_checkpointing": false,
450
+ "fsdp_fsdp_limit_all_gathers": true,
451
+ "fsdp_fsdp_sync_module_states": true,
452
+ "use_git": true,
453
+ "random_seed": 0
454
+ }
455
+ },
456
+ "arc_challenge": {
457
+ "acc_norm": 0.5622866894197952,
458
+ "acc_norm_stderr": 0.014497573881108282,
459
+ "args": {
460
+ "exp_name": "results/v2/arc_challenge/Qwen/Qwen1.5-7B",
461
+ "task_name": "arc_challenge",
462
+ "model_name": "Qwen/Qwen1.5-7B",
463
+ "dataset_param": {
464
+ "max_num_train": 50000,
465
+ "completion_only": false,
466
+ "apply_chat_template": false
467
+ },
468
+ "resume_from_checkpoint": false,
469
+ "train_param": {
470
+ "max_length": 1024,
471
+ "per_device_train_batch_size": 8,
472
+ "auto_find_batch_size": true,
473
+ "num_train_epochs": 5,
474
+ "max_steps": -1,
475
+ "logging_strategy": "epoch",
476
+ "save_strategy": "epoch",
477
+ "logging_steps": 0.2,
478
+ "save_steps": 0.2,
479
+ "gradient_accumulation_steps": 1,
480
+ "gradient_checkpointing": false,
481
+ "optim": "adamw_torch",
482
+ "learning_rate": 5e-05,
483
+ "weight_decay": 0.01,
484
+ "warmup_steps": 0,
485
+ "report_to": "tensorboard",
486
+ "save_safetensors": true,
487
+ "packing": false
488
+ },
489
+ "use_peft": true,
490
+ "peft_task_type": "CAUSAL_LM",
491
+ "peft_inference_mode": false,
492
+ "peft_r": 8,
493
+ "peft_lora_alpha": 32,
494
+ "peft_lora_dropout": 0.1,
495
+ "fsdp": "none",
496
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
497
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
498
+ "fsdp_fsdp_min_num_params": 100000000.0,
499
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
500
+ "fsdp_fsdp_forward_prefetch": false,
501
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
502
+ "fsdp_fsdp_offload_params": false,
503
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
504
+ "fsdp_fsdp_use_orig_params": true,
505
+ "fsdp_fsdp_activation_checkpointing": false,
506
+ "fsdp_fsdp_limit_all_gathers": true,
507
+ "fsdp_fsdp_sync_module_states": true,
508
+ "use_git": true,
509
+ "random_seed": 0
510
+ }
511
+ }
512
+ }
513
+ }
Qwen/Qwen2-0.5B-Instruct/results_2026-01-05-04-27-51.json ADDED
@@ -0,0 +1,513 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "model_name": "Qwen/Qwen2-0.5B-Instruct",
4
+ "model_type": "instruction-tuned",
5
+ "params": 0.5
6
+ },
7
+ "results": {
8
+ "medmcqa": {
9
+ "acc_norm": 0.38250059765718386,
10
+ "acc_norm_stderr": 0.0075152306764844365,
11
+ "args": {
12
+ "exp_name": "results/v1/medmcqa/Qwen/Qwen2-0.5B-Instruct",
13
+ "task_name": "medmcqa",
14
+ "model_name": "Qwen/Qwen2-0.5B-Instruct",
15
+ "dataset_param": {
16
+ "max_num_train": 50000,
17
+ "completion_only": false,
18
+ "apply_chat_template": true
19
+ },
20
+ "resume_from_checkpoint": false,
21
+ "train_param": {
22
+ "max_length": 1024,
23
+ "per_device_train_batch_size": 8,
24
+ "auto_find_batch_size": true,
25
+ "num_train_epochs": 1,
26
+ "max_steps": -1,
27
+ "logging_strategy": "steps",
28
+ "save_strategy": "steps",
29
+ "logging_steps": 0.2,
30
+ "save_steps": 0.2,
31
+ "gradient_accumulation_steps": 1,
32
+ "gradient_checkpointing": false,
33
+ "optim": "adamw_torch",
34
+ "learning_rate": 5e-05,
35
+ "weight_decay": 0.01,
36
+ "warmup_steps": 0,
37
+ "report_to": "tensorboard",
38
+ "save_safetensors": true,
39
+ "packing": false
40
+ },
41
+ "use_peft": true,
42
+ "peft_task_type": "CAUSAL_LM",
43
+ "peft_inference_mode": false,
44
+ "peft_r": 8,
45
+ "peft_lora_alpha": 32,
46
+ "peft_lora_dropout": 0.1,
47
+ "fsdp": "none",
48
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
49
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
50
+ "fsdp_fsdp_min_num_params": 100000000.0,
51
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
52
+ "fsdp_fsdp_forward_prefetch": false,
53
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
54
+ "fsdp_fsdp_offload_params": false,
55
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
56
+ "fsdp_fsdp_use_orig_params": true,
57
+ "fsdp_fsdp_activation_checkpointing": false,
58
+ "fsdp_fsdp_limit_all_gathers": true,
59
+ "fsdp_fsdp_sync_module_states": true,
60
+ "use_git": true,
61
+ "random_seed": 0
62
+ }
63
+ },
64
+ "nq_open": {
65
+ "exact_match,remove_whitespace": 0.0667590027700831,
66
+ "exact_match,remove_whitespace_stderr": 0.004154879030909522,
67
+ "args": {
68
+ "exp_name": "results/v1/nq_open/Qwen/Qwen2-0.5B-Instruct",
69
+ "task_name": "nq_open",
70
+ "model_name": "Qwen/Qwen2-0.5B-Instruct",
71
+ "dataset_param": {
72
+ "max_num_train": 50000,
73
+ "completion_only": false,
74
+ "apply_chat_template": true
75
+ },
76
+ "resume_from_checkpoint": false,
77
+ "train_param": {
78
+ "max_length": 1024,
79
+ "per_device_train_batch_size": 8,
80
+ "auto_find_batch_size": true,
81
+ "num_train_epochs": 1,
82
+ "max_steps": -1,
83
+ "logging_strategy": "steps",
84
+ "save_strategy": "steps",
85
+ "logging_steps": 0.2,
86
+ "save_steps": 0.2,
87
+ "gradient_accumulation_steps": 1,
88
+ "gradient_checkpointing": false,
89
+ "optim": "adamw_torch",
90
+ "learning_rate": 5e-05,
91
+ "weight_decay": 0.01,
92
+ "warmup_steps": 0,
93
+ "report_to": "tensorboard",
94
+ "save_safetensors": true,
95
+ "packing": false
96
+ },
97
+ "use_peft": true,
98
+ "peft_task_type": "CAUSAL_LM",
99
+ "peft_inference_mode": false,
100
+ "peft_r": 8,
101
+ "peft_lora_alpha": 32,
102
+ "peft_lora_dropout": 0.1,
103
+ "fsdp": "none",
104
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
105
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
106
+ "fsdp_fsdp_min_num_params": 100000000.0,
107
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
108
+ "fsdp_fsdp_forward_prefetch": false,
109
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
110
+ "fsdp_fsdp_offload_params": false,
111
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
112
+ "fsdp_fsdp_use_orig_params": true,
113
+ "fsdp_fsdp_activation_checkpointing": false,
114
+ "fsdp_fsdp_limit_all_gathers": true,
115
+ "fsdp_fsdp_sync_module_states": true,
116
+ "use_git": true,
117
+ "random_seed": 0
118
+ }
119
+ },
120
+ "winogrande": {
121
+ "acc": 0.595895816890292,
122
+ "acc_stderr": 0.013791610664670856,
123
+ "args": {
124
+ "exp_name": "results/v1/winogrande/Qwen/Qwen2-0.5B-Instruct",
125
+ "task_name": "winogrande",
126
+ "model_name": "Qwen/Qwen2-0.5B-Instruct",
127
+ "dataset_param": {
128
+ "max_num_train": 50000,
129
+ "completion_only": false,
130
+ "apply_chat_template": true
131
+ },
132
+ "resume_from_checkpoint": false,
133
+ "train_param": {
134
+ "max_length": 1024,
135
+ "per_device_train_batch_size": 8,
136
+ "auto_find_batch_size": true,
137
+ "num_train_epochs": 1,
138
+ "max_steps": -1,
139
+ "logging_strategy": "steps",
140
+ "save_strategy": "steps",
141
+ "logging_steps": 0.2,
142
+ "save_steps": 0.2,
143
+ "gradient_accumulation_steps": 1,
144
+ "gradient_checkpointing": false,
145
+ "optim": "adamw_torch",
146
+ "learning_rate": 5e-05,
147
+ "weight_decay": 0.01,
148
+ "warmup_steps": 0,
149
+ "report_to": "tensorboard",
150
+ "save_safetensors": true,
151
+ "packing": false
152
+ },
153
+ "use_peft": true,
154
+ "peft_task_type": "CAUSAL_LM",
155
+ "peft_inference_mode": false,
156
+ "peft_r": 8,
157
+ "peft_lora_alpha": 32,
158
+ "peft_lora_dropout": 0.1,
159
+ "fsdp": "none",
160
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
161
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
162
+ "fsdp_fsdp_min_num_params": 100000000.0,
163
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
164
+ "fsdp_fsdp_forward_prefetch": false,
165
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
166
+ "fsdp_fsdp_offload_params": false,
167
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
168
+ "fsdp_fsdp_use_orig_params": true,
169
+ "fsdp_fsdp_activation_checkpointing": false,
170
+ "fsdp_fsdp_limit_all_gathers": true,
171
+ "fsdp_fsdp_sync_module_states": true,
172
+ "use_git": true,
173
+ "random_seed": 0
174
+ }
175
+ },
176
+ "hellaswag": {
177
+ "acc_norm": 0.5065,
178
+ "acc_norm_stderr": 0.004999827479771912,
179
+ "args": {
180
+ "exp_name": "results/v1/hellaswag/Qwen/Qwen2-0.5B-Instruct",
181
+ "task_name": "hellaswag",
182
+ "model_name": "Qwen/Qwen2-0.5B-Instruct",
183
+ "dataset_param": {
184
+ "max_num_train": 50000,
185
+ "completion_only": false,
186
+ "apply_chat_template": true
187
+ },
188
+ "resume_from_checkpoint": false,
189
+ "train_param": {
190
+ "max_length": 1024,
191
+ "per_device_train_batch_size": 8,
192
+ "auto_find_batch_size": true,
193
+ "num_train_epochs": 1,
194
+ "max_steps": -1,
195
+ "logging_strategy": "steps",
196
+ "save_strategy": "steps",
197
+ "logging_steps": 0.2,
198
+ "save_steps": 0.2,
199
+ "gradient_accumulation_steps": 1,
200
+ "gradient_checkpointing": false,
201
+ "optim": "adamw_torch",
202
+ "learning_rate": 5e-05,
203
+ "weight_decay": 0.01,
204
+ "warmup_steps": 0,
205
+ "report_to": "tensorboard",
206
+ "save_safetensors": true,
207
+ "packing": false
208
+ },
209
+ "use_peft": true,
210
+ "peft_task_type": "CAUSAL_LM",
211
+ "peft_inference_mode": false,
212
+ "peft_r": 8,
213
+ "peft_lora_alpha": 32,
214
+ "peft_lora_dropout": 0.1,
215
+ "fsdp": "none",
216
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
217
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
218
+ "fsdp_fsdp_min_num_params": 100000000.0,
219
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
220
+ "fsdp_fsdp_forward_prefetch": false,
221
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
222
+ "fsdp_fsdp_offload_params": false,
223
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
224
+ "fsdp_fsdp_use_orig_params": true,
225
+ "fsdp_fsdp_activation_checkpointing": false,
226
+ "fsdp_fsdp_limit_all_gathers": true,
227
+ "fsdp_fsdp_sync_module_states": true,
228
+ "use_git": true,
229
+ "random_seed": 0
230
+ }
231
+ },
232
+ "social_iqa": {
233
+ "acc": 0.45496417604913,
234
+ "acc_stderr": 0.011268082118729844,
235
+ "args": {
236
+ "exp_name": "results/v1/social_iqa/Qwen/Qwen2-0.5B-Instruct",
237
+ "task_name": "social_iqa",
238
+ "model_name": "Qwen/Qwen2-0.5B-Instruct",
239
+ "dataset_param": {
240
+ "max_num_train": 50000,
241
+ "completion_only": false,
242
+ "apply_chat_template": true
243
+ },
244
+ "resume_from_checkpoint": false,
245
+ "train_param": {
246
+ "max_length": 1024,
247
+ "per_device_train_batch_size": 8,
248
+ "auto_find_batch_size": true,
249
+ "num_train_epochs": 1,
250
+ "max_steps": -1,
251
+ "logging_strategy": "steps",
252
+ "save_strategy": "steps",
253
+ "logging_steps": 0.2,
254
+ "save_steps": 0.2,
255
+ "gradient_accumulation_steps": 1,
256
+ "gradient_checkpointing": false,
257
+ "optim": "adamw_torch",
258
+ "learning_rate": 5e-05,
259
+ "weight_decay": 0.01,
260
+ "warmup_steps": 0,
261
+ "report_to": "tensorboard",
262
+ "save_safetensors": true,
263
+ "packing": false
264
+ },
265
+ "use_peft": true,
266
+ "peft_task_type": "CAUSAL_LM",
267
+ "peft_inference_mode": false,
268
+ "peft_r": 8,
269
+ "peft_lora_alpha": 32,
270
+ "peft_lora_dropout": 0.1,
271
+ "fsdp": "none",
272
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
273
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
274
+ "fsdp_fsdp_min_num_params": 100000000.0,
275
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
276
+ "fsdp_fsdp_forward_prefetch": false,
277
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
278
+ "fsdp_fsdp_offload_params": false,
279
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
280
+ "fsdp_fsdp_use_orig_params": true,
281
+ "fsdp_fsdp_activation_checkpointing": false,
282
+ "fsdp_fsdp_limit_all_gathers": true,
283
+ "fsdp_fsdp_sync_module_states": true,
284
+ "use_git": true,
285
+ "random_seed": 0
286
+ }
287
+ },
288
+ "piqa": {
289
+ "acc_norm": 0.6974972796517954,
290
+ "acc_norm_stderr": 0.010717199698083895,
291
+ "args": {
292
+ "exp_name": "results/v2/piqa/Qwen/Qwen2-0.5B-Instruct",
293
+ "task_name": "piqa",
294
+ "model_name": "Qwen/Qwen2-0.5B-Instruct",
295
+ "dataset_param": {
296
+ "max_num_train": 50000,
297
+ "completion_only": false,
298
+ "apply_chat_template": true
299
+ },
300
+ "resume_from_checkpoint": false,
301
+ "train_param": {
302
+ "max_length": 1024,
303
+ "per_device_train_batch_size": 8,
304
+ "auto_find_batch_size": true,
305
+ "num_train_epochs": 5,
306
+ "max_steps": -1,
307
+ "logging_strategy": "epoch",
308
+ "save_strategy": "epoch",
309
+ "logging_steps": 0.2,
310
+ "save_steps": 0.2,
311
+ "gradient_accumulation_steps": 1,
312
+ "gradient_checkpointing": false,
313
+ "optim": "adamw_torch",
314
+ "learning_rate": 5e-05,
315
+ "weight_decay": 0.01,
316
+ "warmup_steps": 0,
317
+ "report_to": "tensorboard",
318
+ "save_safetensors": true,
319
+ "packing": false
320
+ },
321
+ "use_peft": true,
322
+ "peft_task_type": "CAUSAL_LM",
323
+ "peft_inference_mode": false,
324
+ "peft_r": 8,
325
+ "peft_lora_alpha": 32,
326
+ "peft_lora_dropout": 0.1,
327
+ "fsdp": "none",
328
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
329
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
330
+ "fsdp_fsdp_min_num_params": 100000000.0,
331
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
332
+ "fsdp_fsdp_forward_prefetch": false,
333
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
334
+ "fsdp_fsdp_offload_params": false,
335
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
336
+ "fsdp_fsdp_use_orig_params": true,
337
+ "fsdp_fsdp_activation_checkpointing": false,
338
+ "fsdp_fsdp_limit_all_gathers": true,
339
+ "fsdp_fsdp_sync_module_states": true,
340
+ "use_git": true,
341
+ "random_seed": 0
342
+ }
343
+ },
344
+ "commonsense_qa": {
345
+ "acc": 0.579033579033579,
346
+ "acc_stderr": 0.014134995908456714,
347
+ "args": {
348
+ "exp_name": "results/v2/commonsense_qa/Qwen/Qwen2-0.5B-Instruct",
349
+ "task_name": "commonsense_qa",
350
+ "model_name": "Qwen/Qwen2-0.5B-Instruct",
351
+ "dataset_param": {
352
+ "max_num_train": 50000,
353
+ "completion_only": false,
354
+ "apply_chat_template": true
355
+ },
356
+ "resume_from_checkpoint": false,
357
+ "train_param": {
358
+ "max_length": 1024,
359
+ "per_device_train_batch_size": 8,
360
+ "auto_find_batch_size": true,
361
+ "num_train_epochs": 5,
362
+ "max_steps": -1,
363
+ "logging_strategy": "epoch",
364
+ "save_strategy": "epoch",
365
+ "logging_steps": 0.2,
366
+ "save_steps": 0.2,
367
+ "gradient_accumulation_steps": 1,
368
+ "gradient_checkpointing": false,
369
+ "optim": "adamw_torch",
370
+ "learning_rate": 5e-05,
371
+ "weight_decay": 0.01,
372
+ "warmup_steps": 0,
373
+ "report_to": "tensorboard",
374
+ "save_safetensors": true,
375
+ "packing": false
376
+ },
377
+ "use_peft": true,
378
+ "peft_task_type": "CAUSAL_LM",
379
+ "peft_inference_mode": false,
380
+ "peft_r": 8,
381
+ "peft_lora_alpha": 32,
382
+ "peft_lora_dropout": 0.1,
383
+ "fsdp": "none",
384
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
385
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
386
+ "fsdp_fsdp_min_num_params": 100000000.0,
387
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
388
+ "fsdp_fsdp_forward_prefetch": false,
389
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
390
+ "fsdp_fsdp_offload_params": false,
391
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
392
+ "fsdp_fsdp_use_orig_params": true,
393
+ "fsdp_fsdp_activation_checkpointing": false,
394
+ "fsdp_fsdp_limit_all_gathers": true,
395
+ "fsdp_fsdp_sync_module_states": true,
396
+ "use_git": true,
397
+ "random_seed": 0
398
+ }
399
+ },
400
+ "gsm8k": {
401
+ "exact_match,flexible-extract": 0.2744503411675512,
402
+ "exact_match,flexible-extract_stderr": 0.012291581170814898,
403
+ "args": {
404
+ "exp_name": "results/v2/gsm8k/Qwen/Qwen2-0.5B-Instruct",
405
+ "task_name": "gsm8k",
406
+ "model_name": "Qwen/Qwen2-0.5B-Instruct",
407
+ "dataset_param": {
408
+ "max_num_train": 50000,
409
+ "completion_only": false,
410
+ "apply_chat_template": true
411
+ },
412
+ "resume_from_checkpoint": false,
413
+ "train_param": {
414
+ "max_length": 1024,
415
+ "per_device_train_batch_size": 8,
416
+ "auto_find_batch_size": true,
417
+ "num_train_epochs": 5,
418
+ "max_steps": -1,
419
+ "logging_strategy": "epoch",
420
+ "save_strategy": "epoch",
421
+ "logging_steps": 0.2,
422
+ "save_steps": 0.2,
423
+ "gradient_accumulation_steps": 1,
424
+ "gradient_checkpointing": false,
425
+ "optim": "adamw_torch",
426
+ "learning_rate": 5e-05,
427
+ "weight_decay": 0.01,
428
+ "warmup_steps": 0,
429
+ "report_to": "tensorboard",
430
+ "save_safetensors": true,
431
+ "packing": false
432
+ },
433
+ "use_peft": true,
434
+ "peft_task_type": "CAUSAL_LM",
435
+ "peft_inference_mode": false,
436
+ "peft_r": 8,
437
+ "peft_lora_alpha": 32,
438
+ "peft_lora_dropout": 0.1,
439
+ "fsdp": "none",
440
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
441
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
442
+ "fsdp_fsdp_min_num_params": 100000000.0,
443
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
444
+ "fsdp_fsdp_forward_prefetch": false,
445
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
446
+ "fsdp_fsdp_offload_params": false,
447
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
448
+ "fsdp_fsdp_use_orig_params": true,
449
+ "fsdp_fsdp_activation_checkpointing": false,
450
+ "fsdp_fsdp_limit_all_gathers": true,
451
+ "fsdp_fsdp_sync_module_states": true,
452
+ "use_git": true,
453
+ "random_seed": 0
454
+ }
455
+ },
456
+ "arc_challenge": {
457
+ "acc_norm": 0.29948805460750855,
458
+ "acc_norm_stderr": 0.013385021637313565,
459
+ "args": {
460
+ "exp_name": "results/v2/arc_challenge/Qwen/Qwen2-0.5B-Instruct",
461
+ "task_name": "arc_challenge",
462
+ "model_name": "Qwen/Qwen2-0.5B-Instruct",
463
+ "dataset_param": {
464
+ "max_num_train": 50000,
465
+ "completion_only": false,
466
+ "apply_chat_template": true
467
+ },
468
+ "resume_from_checkpoint": false,
469
+ "train_param": {
470
+ "max_length": 1024,
471
+ "per_device_train_batch_size": 8,
472
+ "auto_find_batch_size": true,
473
+ "num_train_epochs": 5,
474
+ "max_steps": -1,
475
+ "logging_strategy": "epoch",
476
+ "save_strategy": "epoch",
477
+ "logging_steps": 0.2,
478
+ "save_steps": 0.2,
479
+ "gradient_accumulation_steps": 1,
480
+ "gradient_checkpointing": false,
481
+ "optim": "adamw_torch",
482
+ "learning_rate": 5e-05,
483
+ "weight_decay": 0.01,
484
+ "warmup_steps": 0,
485
+ "report_to": "tensorboard",
486
+ "save_safetensors": true,
487
+ "packing": false
488
+ },
489
+ "use_peft": true,
490
+ "peft_task_type": "CAUSAL_LM",
491
+ "peft_inference_mode": false,
492
+ "peft_r": 8,
493
+ "peft_lora_alpha": 32,
494
+ "peft_lora_dropout": 0.1,
495
+ "fsdp": "none",
496
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
497
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
498
+ "fsdp_fsdp_min_num_params": 100000000.0,
499
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
500
+ "fsdp_fsdp_forward_prefetch": false,
501
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
502
+ "fsdp_fsdp_offload_params": false,
503
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
504
+ "fsdp_fsdp_use_orig_params": true,
505
+ "fsdp_fsdp_activation_checkpointing": false,
506
+ "fsdp_fsdp_limit_all_gathers": true,
507
+ "fsdp_fsdp_sync_module_states": true,
508
+ "use_git": true,
509
+ "random_seed": 0
510
+ }
511
+ }
512
+ }
513
+ }
Qwen/Qwen2-0.5B/results_2026-01-05-04-27-50.json ADDED
@@ -0,0 +1,513 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "model_name": "Qwen/Qwen2-0.5B",
4
+ "model_type": "pretrained",
5
+ "params": 0.5
6
+ },
7
+ "results": {
8
+ "medmcqa": {
9
+ "acc_norm": 0.38680372938082713,
10
+ "acc_norm_stderr": 0.007531007230050308,
11
+ "args": {
12
+ "exp_name": "results/v1/medmcqa/Qwen/Qwen2-0.5B",
13
+ "task_name": "medmcqa",
14
+ "model_name": "Qwen/Qwen2-0.5B",
15
+ "dataset_param": {
16
+ "max_num_train": 50000,
17
+ "completion_only": false,
18
+ "apply_chat_template": false
19
+ },
20
+ "resume_from_checkpoint": false,
21
+ "train_param": {
22
+ "max_length": 1024,
23
+ "per_device_train_batch_size": 8,
24
+ "auto_find_batch_size": true,
25
+ "num_train_epochs": 1,
26
+ "max_steps": -1,
27
+ "logging_strategy": "steps",
28
+ "save_strategy": "steps",
29
+ "logging_steps": 0.2,
30
+ "save_steps": 0.2,
31
+ "gradient_accumulation_steps": 1,
32
+ "gradient_checkpointing": false,
33
+ "optim": "adamw_torch",
34
+ "learning_rate": 5e-05,
35
+ "weight_decay": 0.01,
36
+ "warmup_steps": 0,
37
+ "report_to": "tensorboard",
38
+ "save_safetensors": true,
39
+ "packing": false
40
+ },
41
+ "use_peft": true,
42
+ "peft_task_type": "CAUSAL_LM",
43
+ "peft_inference_mode": false,
44
+ "peft_r": 8,
45
+ "peft_lora_alpha": 32,
46
+ "peft_lora_dropout": 0.1,
47
+ "fsdp": "none",
48
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
49
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
50
+ "fsdp_fsdp_min_num_params": 100000000.0,
51
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
52
+ "fsdp_fsdp_forward_prefetch": false,
53
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
54
+ "fsdp_fsdp_offload_params": false,
55
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
56
+ "fsdp_fsdp_use_orig_params": true,
57
+ "fsdp_fsdp_activation_checkpointing": false,
58
+ "fsdp_fsdp_limit_all_gathers": true,
59
+ "fsdp_fsdp_sync_module_states": true,
60
+ "use_git": true,
61
+ "random_seed": 0
62
+ }
63
+ },
64
+ "nq_open": {
65
+ "exact_match,remove_whitespace": 0.0745152354570637,
66
+ "exact_match,remove_whitespace_stderr": 0.004371331423672333,
67
+ "args": {
68
+ "exp_name": "results/v1/nq_open/Qwen/Qwen2-0.5B",
69
+ "task_name": "nq_open",
70
+ "model_name": "Qwen/Qwen2-0.5B",
71
+ "dataset_param": {
72
+ "max_num_train": 50000,
73
+ "completion_only": false,
74
+ "apply_chat_template": false
75
+ },
76
+ "resume_from_checkpoint": false,
77
+ "train_param": {
78
+ "max_length": 1024,
79
+ "per_device_train_batch_size": 8,
80
+ "auto_find_batch_size": true,
81
+ "num_train_epochs": 1,
82
+ "max_steps": -1,
83
+ "logging_strategy": "steps",
84
+ "save_strategy": "steps",
85
+ "logging_steps": 0.2,
86
+ "save_steps": 0.2,
87
+ "gradient_accumulation_steps": 1,
88
+ "gradient_checkpointing": false,
89
+ "optim": "adamw_torch",
90
+ "learning_rate": 5e-05,
91
+ "weight_decay": 0.01,
92
+ "warmup_steps": 0,
93
+ "report_to": "tensorboard",
94
+ "save_safetensors": true,
95
+ "packing": false
96
+ },
97
+ "use_peft": true,
98
+ "peft_task_type": "CAUSAL_LM",
99
+ "peft_inference_mode": false,
100
+ "peft_r": 8,
101
+ "peft_lora_alpha": 32,
102
+ "peft_lora_dropout": 0.1,
103
+ "fsdp": "none",
104
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
105
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
106
+ "fsdp_fsdp_min_num_params": 100000000.0,
107
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
108
+ "fsdp_fsdp_forward_prefetch": false,
109
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
110
+ "fsdp_fsdp_offload_params": false,
111
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
112
+ "fsdp_fsdp_use_orig_params": true,
113
+ "fsdp_fsdp_activation_checkpointing": false,
114
+ "fsdp_fsdp_limit_all_gathers": true,
115
+ "fsdp_fsdp_sync_module_states": true,
116
+ "use_git": true,
117
+ "random_seed": 0
118
+ }
119
+ },
120
+ "winogrande": {
121
+ "acc": 0.5895816890292028,
122
+ "acc_stderr": 0.01382510712003586,
123
+ "args": {
124
+ "exp_name": "results/v1/winogrande/Qwen/Qwen2-0.5B",
125
+ "task_name": "winogrande",
126
+ "model_name": "Qwen/Qwen2-0.5B",
127
+ "dataset_param": {
128
+ "max_num_train": 50000,
129
+ "completion_only": false,
130
+ "apply_chat_template": false
131
+ },
132
+ "resume_from_checkpoint": false,
133
+ "train_param": {
134
+ "max_length": 1024,
135
+ "per_device_train_batch_size": 8,
136
+ "auto_find_batch_size": true,
137
+ "num_train_epochs": 1,
138
+ "max_steps": -1,
139
+ "logging_strategy": "steps",
140
+ "save_strategy": "steps",
141
+ "logging_steps": 0.2,
142
+ "save_steps": 0.2,
143
+ "gradient_accumulation_steps": 1,
144
+ "gradient_checkpointing": false,
145
+ "optim": "adamw_torch",
146
+ "learning_rate": 5e-05,
147
+ "weight_decay": 0.01,
148
+ "warmup_steps": 0,
149
+ "report_to": "tensorboard",
150
+ "save_safetensors": true,
151
+ "packing": false
152
+ },
153
+ "use_peft": true,
154
+ "peft_task_type": "CAUSAL_LM",
155
+ "peft_inference_mode": false,
156
+ "peft_r": 8,
157
+ "peft_lora_alpha": 32,
158
+ "peft_lora_dropout": 0.1,
159
+ "fsdp": "none",
160
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
161
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
162
+ "fsdp_fsdp_min_num_params": 100000000.0,
163
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
164
+ "fsdp_fsdp_forward_prefetch": false,
165
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
166
+ "fsdp_fsdp_offload_params": false,
167
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
168
+ "fsdp_fsdp_use_orig_params": true,
169
+ "fsdp_fsdp_activation_checkpointing": false,
170
+ "fsdp_fsdp_limit_all_gathers": true,
171
+ "fsdp_fsdp_sync_module_states": true,
172
+ "use_git": true,
173
+ "random_seed": 0
174
+ }
175
+ },
176
+ "hellaswag": {
177
+ "acc_norm": 0.506,
178
+ "acc_norm_stderr": 0.004999889987789013,
179
+ "args": {
180
+ "exp_name": "results/v1/hellaswag/Qwen/Qwen2-0.5B",
181
+ "task_name": "hellaswag",
182
+ "model_name": "Qwen/Qwen2-0.5B",
183
+ "dataset_param": {
184
+ "max_num_train": 50000,
185
+ "completion_only": false,
186
+ "apply_chat_template": false
187
+ },
188
+ "resume_from_checkpoint": false,
189
+ "train_param": {
190
+ "max_length": 1024,
191
+ "per_device_train_batch_size": 8,
192
+ "auto_find_batch_size": true,
193
+ "num_train_epochs": 1,
194
+ "max_steps": -1,
195
+ "logging_strategy": "steps",
196
+ "save_strategy": "steps",
197
+ "logging_steps": 0.2,
198
+ "save_steps": 0.2,
199
+ "gradient_accumulation_steps": 1,
200
+ "gradient_checkpointing": false,
201
+ "optim": "adamw_torch",
202
+ "learning_rate": 5e-05,
203
+ "weight_decay": 0.01,
204
+ "warmup_steps": 0,
205
+ "report_to": "tensorboard",
206
+ "save_safetensors": true,
207
+ "packing": false
208
+ },
209
+ "use_peft": true,
210
+ "peft_task_type": "CAUSAL_LM",
211
+ "peft_inference_mode": false,
212
+ "peft_r": 8,
213
+ "peft_lora_alpha": 32,
214
+ "peft_lora_dropout": 0.1,
215
+ "fsdp": "none",
216
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
217
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
218
+ "fsdp_fsdp_min_num_params": 100000000.0,
219
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
220
+ "fsdp_fsdp_forward_prefetch": false,
221
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
222
+ "fsdp_fsdp_offload_params": false,
223
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
224
+ "fsdp_fsdp_use_orig_params": true,
225
+ "fsdp_fsdp_activation_checkpointing": false,
226
+ "fsdp_fsdp_limit_all_gathers": true,
227
+ "fsdp_fsdp_sync_module_states": true,
228
+ "use_git": true,
229
+ "random_seed": 0
230
+ }
231
+ },
232
+ "social_iqa": {
233
+ "acc": 0.47850562947799385,
234
+ "acc_stderr": 0.011303611339153372,
235
+ "args": {
236
+ "exp_name": "results/v1/social_iqa/Qwen/Qwen2-0.5B",
237
+ "task_name": "social_iqa",
238
+ "model_name": "Qwen/Qwen2-0.5B",
239
+ "dataset_param": {
240
+ "max_num_train": 50000,
241
+ "completion_only": false,
242
+ "apply_chat_template": false
243
+ },
244
+ "resume_from_checkpoint": false,
245
+ "train_param": {
246
+ "max_length": 1024,
247
+ "per_device_train_batch_size": 8,
248
+ "auto_find_batch_size": true,
249
+ "num_train_epochs": 1,
250
+ "max_steps": -1,
251
+ "logging_strategy": "steps",
252
+ "save_strategy": "steps",
253
+ "logging_steps": 0.2,
254
+ "save_steps": 0.2,
255
+ "gradient_accumulation_steps": 1,
256
+ "gradient_checkpointing": false,
257
+ "optim": "adamw_torch",
258
+ "learning_rate": 5e-05,
259
+ "weight_decay": 0.01,
260
+ "warmup_steps": 0,
261
+ "report_to": "tensorboard",
262
+ "save_safetensors": true,
263
+ "packing": false
264
+ },
265
+ "use_peft": true,
266
+ "peft_task_type": "CAUSAL_LM",
267
+ "peft_inference_mode": false,
268
+ "peft_r": 8,
269
+ "peft_lora_alpha": 32,
270
+ "peft_lora_dropout": 0.1,
271
+ "fsdp": "none",
272
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
273
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
274
+ "fsdp_fsdp_min_num_params": 100000000.0,
275
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
276
+ "fsdp_fsdp_forward_prefetch": false,
277
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
278
+ "fsdp_fsdp_offload_params": false,
279
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
280
+ "fsdp_fsdp_use_orig_params": true,
281
+ "fsdp_fsdp_activation_checkpointing": false,
282
+ "fsdp_fsdp_limit_all_gathers": true,
283
+ "fsdp_fsdp_sync_module_states": true,
284
+ "use_git": true,
285
+ "random_seed": 0
286
+ }
287
+ },
288
+ "piqa": {
289
+ "acc_norm": 0.6931447225244831,
290
+ "acc_norm_stderr": 0.010760295070580374,
291
+ "args": {
292
+ "exp_name": "results/v2/piqa/Qwen/Qwen2-0.5B",
293
+ "task_name": "piqa",
294
+ "model_name": "Qwen/Qwen2-0.5B",
295
+ "dataset_param": {
296
+ "max_num_train": 50000,
297
+ "completion_only": false,
298
+ "apply_chat_template": false
299
+ },
300
+ "resume_from_checkpoint": false,
301
+ "train_param": {
302
+ "max_length": 1024,
303
+ "per_device_train_batch_size": 8,
304
+ "auto_find_batch_size": true,
305
+ "num_train_epochs": 5,
306
+ "max_steps": -1,
307
+ "logging_strategy": "epoch",
308
+ "save_strategy": "epoch",
309
+ "logging_steps": 0.2,
310
+ "save_steps": 0.2,
311
+ "gradient_accumulation_steps": 1,
312
+ "gradient_checkpointing": false,
313
+ "optim": "adamw_torch",
314
+ "learning_rate": 5e-05,
315
+ "weight_decay": 0.01,
316
+ "warmup_steps": 0,
317
+ "report_to": "tensorboard",
318
+ "save_safetensors": true,
319
+ "packing": false
320
+ },
321
+ "use_peft": true,
322
+ "peft_task_type": "CAUSAL_LM",
323
+ "peft_inference_mode": false,
324
+ "peft_r": 8,
325
+ "peft_lora_alpha": 32,
326
+ "peft_lora_dropout": 0.1,
327
+ "fsdp": "none",
328
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
329
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
330
+ "fsdp_fsdp_min_num_params": 100000000.0,
331
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
332
+ "fsdp_fsdp_forward_prefetch": false,
333
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
334
+ "fsdp_fsdp_offload_params": false,
335
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
336
+ "fsdp_fsdp_use_orig_params": true,
337
+ "fsdp_fsdp_activation_checkpointing": false,
338
+ "fsdp_fsdp_limit_all_gathers": true,
339
+ "fsdp_fsdp_sync_module_states": true,
340
+ "use_git": true,
341
+ "random_seed": 0
342
+ }
343
+ },
344
+ "commonsense_qa": {
345
+ "acc": 0.601965601965602,
346
+ "acc_stderr": 0.014014132679800204,
347
+ "args": {
348
+ "exp_name": "results/v2/commonsense_qa/Qwen/Qwen2-0.5B",
349
+ "task_name": "commonsense_qa",
350
+ "model_name": "Qwen/Qwen2-0.5B",
351
+ "dataset_param": {
352
+ "max_num_train": 50000,
353
+ "completion_only": false,
354
+ "apply_chat_template": false
355
+ },
356
+ "resume_from_checkpoint": false,
357
+ "train_param": {
358
+ "max_length": 1024,
359
+ "per_device_train_batch_size": 8,
360
+ "auto_find_batch_size": true,
361
+ "num_train_epochs": 5,
362
+ "max_steps": -1,
363
+ "logging_strategy": "epoch",
364
+ "save_strategy": "epoch",
365
+ "logging_steps": 0.2,
366
+ "save_steps": 0.2,
367
+ "gradient_accumulation_steps": 1,
368
+ "gradient_checkpointing": false,
369
+ "optim": "adamw_torch",
370
+ "learning_rate": 5e-05,
371
+ "weight_decay": 0.01,
372
+ "warmup_steps": 0,
373
+ "report_to": "tensorboard",
374
+ "save_safetensors": true,
375
+ "packing": false
376
+ },
377
+ "use_peft": true,
378
+ "peft_task_type": "CAUSAL_LM",
379
+ "peft_inference_mode": false,
380
+ "peft_r": 8,
381
+ "peft_lora_alpha": 32,
382
+ "peft_lora_dropout": 0.1,
383
+ "fsdp": "none",
384
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
385
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
386
+ "fsdp_fsdp_min_num_params": 100000000.0,
387
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
388
+ "fsdp_fsdp_forward_prefetch": false,
389
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
390
+ "fsdp_fsdp_offload_params": false,
391
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
392
+ "fsdp_fsdp_use_orig_params": true,
393
+ "fsdp_fsdp_activation_checkpointing": false,
394
+ "fsdp_fsdp_limit_all_gathers": true,
395
+ "fsdp_fsdp_sync_module_states": true,
396
+ "use_git": true,
397
+ "random_seed": 0
398
+ }
399
+ },
400
+ "gsm8k": {
401
+ "exact_match,flexible-extract": 0.3206974981046247,
402
+ "exact_match,flexible-extract_stderr": 0.012856468433722281,
403
+ "args": {
404
+ "exp_name": "results/v2/gsm8k/Qwen/Qwen2-0.5B",
405
+ "task_name": "gsm8k",
406
+ "model_name": "Qwen/Qwen2-0.5B",
407
+ "dataset_param": {
408
+ "max_num_train": 50000,
409
+ "completion_only": false,
410
+ "apply_chat_template": false
411
+ },
412
+ "resume_from_checkpoint": false,
413
+ "train_param": {
414
+ "max_length": 1024,
415
+ "per_device_train_batch_size": 8,
416
+ "auto_find_batch_size": true,
417
+ "num_train_epochs": 5,
418
+ "max_steps": -1,
419
+ "logging_strategy": "epoch",
420
+ "save_strategy": "epoch",
421
+ "logging_steps": 0.2,
422
+ "save_steps": 0.2,
423
+ "gradient_accumulation_steps": 1,
424
+ "gradient_checkpointing": false,
425
+ "optim": "adamw_torch",
426
+ "learning_rate": 5e-05,
427
+ "weight_decay": 0.01,
428
+ "warmup_steps": 0,
429
+ "report_to": "tensorboard",
430
+ "save_safetensors": true,
431
+ "packing": false
432
+ },
433
+ "use_peft": true,
434
+ "peft_task_type": "CAUSAL_LM",
435
+ "peft_inference_mode": false,
436
+ "peft_r": 8,
437
+ "peft_lora_alpha": 32,
438
+ "peft_lora_dropout": 0.1,
439
+ "fsdp": "none",
440
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
441
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
442
+ "fsdp_fsdp_min_num_params": 100000000.0,
443
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
444
+ "fsdp_fsdp_forward_prefetch": false,
445
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
446
+ "fsdp_fsdp_offload_params": false,
447
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
448
+ "fsdp_fsdp_use_orig_params": true,
449
+ "fsdp_fsdp_activation_checkpointing": false,
450
+ "fsdp_fsdp_limit_all_gathers": true,
451
+ "fsdp_fsdp_sync_module_states": true,
452
+ "use_git": true,
453
+ "random_seed": 0
454
+ }
455
+ },
456
+ "arc_challenge": {
457
+ "acc_norm": 0.30631399317406144,
458
+ "acc_norm_stderr": 0.013470584417276513,
459
+ "args": {
460
+ "exp_name": "results/v2/arc_challenge/Qwen/Qwen2-0.5B",
461
+ "task_name": "arc_challenge",
462
+ "model_name": "Qwen/Qwen2-0.5B",
463
+ "dataset_param": {
464
+ "max_num_train": 50000,
465
+ "completion_only": false,
466
+ "apply_chat_template": false
467
+ },
468
+ "resume_from_checkpoint": false,
469
+ "train_param": {
470
+ "max_length": 1024,
471
+ "per_device_train_batch_size": 8,
472
+ "auto_find_batch_size": true,
473
+ "num_train_epochs": 5,
474
+ "max_steps": -1,
475
+ "logging_strategy": "epoch",
476
+ "save_strategy": "epoch",
477
+ "logging_steps": 0.2,
478
+ "save_steps": 0.2,
479
+ "gradient_accumulation_steps": 1,
480
+ "gradient_checkpointing": false,
481
+ "optim": "adamw_torch",
482
+ "learning_rate": 5e-05,
483
+ "weight_decay": 0.01,
484
+ "warmup_steps": 0,
485
+ "report_to": "tensorboard",
486
+ "save_safetensors": true,
487
+ "packing": false
488
+ },
489
+ "use_peft": true,
490
+ "peft_task_type": "CAUSAL_LM",
491
+ "peft_inference_mode": false,
492
+ "peft_r": 8,
493
+ "peft_lora_alpha": 32,
494
+ "peft_lora_dropout": 0.1,
495
+ "fsdp": "none",
496
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
497
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
498
+ "fsdp_fsdp_min_num_params": 100000000.0,
499
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
500
+ "fsdp_fsdp_forward_prefetch": false,
501
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
502
+ "fsdp_fsdp_offload_params": false,
503
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
504
+ "fsdp_fsdp_use_orig_params": true,
505
+ "fsdp_fsdp_activation_checkpointing": false,
506
+ "fsdp_fsdp_limit_all_gathers": true,
507
+ "fsdp_fsdp_sync_module_states": true,
508
+ "use_git": true,
509
+ "random_seed": 0
510
+ }
511
+ }
512
+ }
513
+ }
Qwen/Qwen2-1.5B-Instruct/results_2026-01-05-04-27-51.json ADDED
@@ -0,0 +1,513 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "model_name": "Qwen/Qwen2-1.5B-Instruct",
4
+ "model_type": "instruction-tuned",
5
+ "params": 1.5
6
+ },
7
+ "results": {
8
+ "medmcqa": {
9
+ "acc_norm": 0.4499163279942625,
10
+ "acc_norm_stderr": 0.007692867282733471,
11
+ "args": {
12
+ "exp_name": "results/v1/medmcqa/Qwen/Qwen2-1.5B-Instruct",
13
+ "task_name": "medmcqa",
14
+ "model_name": "Qwen/Qwen2-1.5B-Instruct",
15
+ "dataset_param": {
16
+ "max_num_train": 50000,
17
+ "completion_only": false,
18
+ "apply_chat_template": true
19
+ },
20
+ "resume_from_checkpoint": false,
21
+ "train_param": {
22
+ "max_length": 1024,
23
+ "per_device_train_batch_size": 8,
24
+ "auto_find_batch_size": true,
25
+ "num_train_epochs": 1,
26
+ "max_steps": -1,
27
+ "logging_strategy": "steps",
28
+ "save_strategy": "steps",
29
+ "logging_steps": 0.2,
30
+ "save_steps": 0.2,
31
+ "gradient_accumulation_steps": 1,
32
+ "gradient_checkpointing": false,
33
+ "optim": "adamw_torch",
34
+ "learning_rate": 5e-05,
35
+ "weight_decay": 0.01,
36
+ "warmup_steps": 0,
37
+ "report_to": "tensorboard",
38
+ "save_safetensors": true,
39
+ "packing": false
40
+ },
41
+ "use_peft": true,
42
+ "peft_task_type": "CAUSAL_LM",
43
+ "peft_inference_mode": false,
44
+ "peft_r": 8,
45
+ "peft_lora_alpha": 32,
46
+ "peft_lora_dropout": 0.1,
47
+ "fsdp": "none",
48
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
49
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
50
+ "fsdp_fsdp_min_num_params": 100000000.0,
51
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
52
+ "fsdp_fsdp_forward_prefetch": false,
53
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
54
+ "fsdp_fsdp_offload_params": false,
55
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
56
+ "fsdp_fsdp_use_orig_params": true,
57
+ "fsdp_fsdp_activation_checkpointing": false,
58
+ "fsdp_fsdp_limit_all_gathers": true,
59
+ "fsdp_fsdp_sync_module_states": true,
60
+ "use_git": true,
61
+ "random_seed": 0
62
+ }
63
+ },
64
+ "nq_open": {
65
+ "exact_match,remove_whitespace": 0.13933518005540166,
66
+ "exact_match,remove_whitespace_stderr": 0.005764398230895091,
67
+ "args": {
68
+ "exp_name": "results/v1/nq_open/Qwen/Qwen2-1.5B-Instruct",
69
+ "task_name": "nq_open",
70
+ "model_name": "Qwen/Qwen2-1.5B-Instruct",
71
+ "dataset_param": {
72
+ "max_num_train": 50000,
73
+ "completion_only": false,
74
+ "apply_chat_template": true
75
+ },
76
+ "resume_from_checkpoint": false,
77
+ "train_param": {
78
+ "max_length": 1024,
79
+ "per_device_train_batch_size": 8,
80
+ "auto_find_batch_size": true,
81
+ "num_train_epochs": 1,
82
+ "max_steps": -1,
83
+ "logging_strategy": "steps",
84
+ "save_strategy": "steps",
85
+ "logging_steps": 0.2,
86
+ "save_steps": 0.2,
87
+ "gradient_accumulation_steps": 1,
88
+ "gradient_checkpointing": false,
89
+ "optim": "adamw_torch",
90
+ "learning_rate": 5e-05,
91
+ "weight_decay": 0.01,
92
+ "warmup_steps": 0,
93
+ "report_to": "tensorboard",
94
+ "save_safetensors": true,
95
+ "packing": false
96
+ },
97
+ "use_peft": true,
98
+ "peft_task_type": "CAUSAL_LM",
99
+ "peft_inference_mode": false,
100
+ "peft_r": 8,
101
+ "peft_lora_alpha": 32,
102
+ "peft_lora_dropout": 0.1,
103
+ "fsdp": "none",
104
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
105
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
106
+ "fsdp_fsdp_min_num_params": 100000000.0,
107
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
108
+ "fsdp_fsdp_forward_prefetch": false,
109
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
110
+ "fsdp_fsdp_offload_params": false,
111
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
112
+ "fsdp_fsdp_use_orig_params": true,
113
+ "fsdp_fsdp_activation_checkpointing": false,
114
+ "fsdp_fsdp_limit_all_gathers": true,
115
+ "fsdp_fsdp_sync_module_states": true,
116
+ "use_git": true,
117
+ "random_seed": 0
118
+ }
119
+ },
120
+ "winogrande": {
121
+ "acc": 0.696921862667719,
122
+ "acc_stderr": 0.012916727462634472,
123
+ "args": {
124
+ "exp_name": "results/v1/winogrande/Qwen/Qwen2-1.5B-Instruct",
125
+ "task_name": "winogrande",
126
+ "model_name": "Qwen/Qwen2-1.5B-Instruct",
127
+ "dataset_param": {
128
+ "max_num_train": 50000,
129
+ "completion_only": false,
130
+ "apply_chat_template": true
131
+ },
132
+ "resume_from_checkpoint": false,
133
+ "train_param": {
134
+ "max_length": 1024,
135
+ "per_device_train_batch_size": 8,
136
+ "auto_find_batch_size": true,
137
+ "num_train_epochs": 1,
138
+ "max_steps": -1,
139
+ "logging_strategy": "steps",
140
+ "save_strategy": "steps",
141
+ "logging_steps": 0.2,
142
+ "save_steps": 0.2,
143
+ "gradient_accumulation_steps": 1,
144
+ "gradient_checkpointing": false,
145
+ "optim": "adamw_torch",
146
+ "learning_rate": 5e-05,
147
+ "weight_decay": 0.01,
148
+ "warmup_steps": 0,
149
+ "report_to": "tensorboard",
150
+ "save_safetensors": true,
151
+ "packing": false
152
+ },
153
+ "use_peft": true,
154
+ "peft_task_type": "CAUSAL_LM",
155
+ "peft_inference_mode": false,
156
+ "peft_r": 8,
157
+ "peft_lora_alpha": 32,
158
+ "peft_lora_dropout": 0.1,
159
+ "fsdp": "none",
160
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
161
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
162
+ "fsdp_fsdp_min_num_params": 100000000.0,
163
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
164
+ "fsdp_fsdp_forward_prefetch": false,
165
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
166
+ "fsdp_fsdp_offload_params": false,
167
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
168
+ "fsdp_fsdp_use_orig_params": true,
169
+ "fsdp_fsdp_activation_checkpointing": false,
170
+ "fsdp_fsdp_limit_all_gathers": true,
171
+ "fsdp_fsdp_sync_module_states": true,
172
+ "use_git": true,
173
+ "random_seed": 0
174
+ }
175
+ },
176
+ "hellaswag": {
177
+ "acc_norm": 0.6761,
178
+ "acc_norm_stderr": 0.0046798578084070825,
179
+ "args": {
180
+ "exp_name": "results/v1/hellaswag/Qwen/Qwen2-1.5B-Instruct",
181
+ "task_name": "hellaswag",
182
+ "model_name": "Qwen/Qwen2-1.5B-Instruct",
183
+ "dataset_param": {
184
+ "max_num_train": 50000,
185
+ "completion_only": false,
186
+ "apply_chat_template": true
187
+ },
188
+ "resume_from_checkpoint": false,
189
+ "train_param": {
190
+ "max_length": 1024,
191
+ "per_device_train_batch_size": 8,
192
+ "auto_find_batch_size": true,
193
+ "num_train_epochs": 1,
194
+ "max_steps": -1,
195
+ "logging_strategy": "steps",
196
+ "save_strategy": "steps",
197
+ "logging_steps": 0.2,
198
+ "save_steps": 0.2,
199
+ "gradient_accumulation_steps": 1,
200
+ "gradient_checkpointing": false,
201
+ "optim": "adamw_torch",
202
+ "learning_rate": 5e-05,
203
+ "weight_decay": 0.01,
204
+ "warmup_steps": 0,
205
+ "report_to": "tensorboard",
206
+ "save_safetensors": true,
207
+ "packing": false
208
+ },
209
+ "use_peft": true,
210
+ "peft_task_type": "CAUSAL_LM",
211
+ "peft_inference_mode": false,
212
+ "peft_r": 8,
213
+ "peft_lora_alpha": 32,
214
+ "peft_lora_dropout": 0.1,
215
+ "fsdp": "none",
216
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
217
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
218
+ "fsdp_fsdp_min_num_params": 100000000.0,
219
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
220
+ "fsdp_fsdp_forward_prefetch": false,
221
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
222
+ "fsdp_fsdp_offload_params": false,
223
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
224
+ "fsdp_fsdp_use_orig_params": true,
225
+ "fsdp_fsdp_activation_checkpointing": false,
226
+ "fsdp_fsdp_limit_all_gathers": true,
227
+ "fsdp_fsdp_sync_module_states": true,
228
+ "use_git": true,
229
+ "random_seed": 0
230
+ }
231
+ },
232
+ "social_iqa": {
233
+ "acc": 0.5174002047082907,
234
+ "acc_stderr": 0.011307217422300708,
235
+ "args": {
236
+ "exp_name": "results/v1/social_iqa/Qwen/Qwen2-1.5B-Instruct",
237
+ "task_name": "social_iqa",
238
+ "model_name": "Qwen/Qwen2-1.5B-Instruct",
239
+ "dataset_param": {
240
+ "max_num_train": 50000,
241
+ "completion_only": false,
242
+ "apply_chat_template": true
243
+ },
244
+ "resume_from_checkpoint": false,
245
+ "train_param": {
246
+ "max_length": 1024,
247
+ "per_device_train_batch_size": 8,
248
+ "auto_find_batch_size": true,
249
+ "num_train_epochs": 1,
250
+ "max_steps": -1,
251
+ "logging_strategy": "steps",
252
+ "save_strategy": "steps",
253
+ "logging_steps": 0.2,
254
+ "save_steps": 0.2,
255
+ "gradient_accumulation_steps": 1,
256
+ "gradient_checkpointing": false,
257
+ "optim": "adamw_torch",
258
+ "learning_rate": 5e-05,
259
+ "weight_decay": 0.01,
260
+ "warmup_steps": 0,
261
+ "report_to": "tensorboard",
262
+ "save_safetensors": true,
263
+ "packing": false
264
+ },
265
+ "use_peft": true,
266
+ "peft_task_type": "CAUSAL_LM",
267
+ "peft_inference_mode": false,
268
+ "peft_r": 8,
269
+ "peft_lora_alpha": 32,
270
+ "peft_lora_dropout": 0.1,
271
+ "fsdp": "none",
272
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
273
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
274
+ "fsdp_fsdp_min_num_params": 100000000.0,
275
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
276
+ "fsdp_fsdp_forward_prefetch": false,
277
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
278
+ "fsdp_fsdp_offload_params": false,
279
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
280
+ "fsdp_fsdp_use_orig_params": true,
281
+ "fsdp_fsdp_activation_checkpointing": false,
282
+ "fsdp_fsdp_limit_all_gathers": true,
283
+ "fsdp_fsdp_sync_module_states": true,
284
+ "use_git": true,
285
+ "random_seed": 0
286
+ }
287
+ },
288
+ "piqa": {
289
+ "acc_norm": 0.7687704026115343,
290
+ "acc_norm_stderr": 0.009837063180625327,
291
+ "args": {
292
+ "exp_name": "results/v2/piqa/Qwen/Qwen2-1.5B-Instruct",
293
+ "task_name": "piqa",
294
+ "model_name": "Qwen/Qwen2-1.5B-Instruct",
295
+ "dataset_param": {
296
+ "max_num_train": 50000,
297
+ "completion_only": false,
298
+ "apply_chat_template": true
299
+ },
300
+ "resume_from_checkpoint": false,
301
+ "train_param": {
302
+ "max_length": 1024,
303
+ "per_device_train_batch_size": 8,
304
+ "auto_find_batch_size": true,
305
+ "num_train_epochs": 5,
306
+ "max_steps": -1,
307
+ "logging_strategy": "epoch",
308
+ "save_strategy": "epoch",
309
+ "logging_steps": 0.2,
310
+ "save_steps": 0.2,
311
+ "gradient_accumulation_steps": 1,
312
+ "gradient_checkpointing": false,
313
+ "optim": "adamw_torch",
314
+ "learning_rate": 5e-05,
315
+ "weight_decay": 0.01,
316
+ "warmup_steps": 0,
317
+ "report_to": "tensorboard",
318
+ "save_safetensors": true,
319
+ "packing": false
320
+ },
321
+ "use_peft": true,
322
+ "peft_task_type": "CAUSAL_LM",
323
+ "peft_inference_mode": false,
324
+ "peft_r": 8,
325
+ "peft_lora_alpha": 32,
326
+ "peft_lora_dropout": 0.1,
327
+ "fsdp": "none",
328
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
329
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
330
+ "fsdp_fsdp_min_num_params": 100000000.0,
331
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
332
+ "fsdp_fsdp_forward_prefetch": false,
333
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
334
+ "fsdp_fsdp_offload_params": false,
335
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
336
+ "fsdp_fsdp_use_orig_params": true,
337
+ "fsdp_fsdp_activation_checkpointing": false,
338
+ "fsdp_fsdp_limit_all_gathers": true,
339
+ "fsdp_fsdp_sync_module_states": true,
340
+ "use_git": true,
341
+ "random_seed": 0
342
+ }
343
+ },
344
+ "commonsense_qa": {
345
+ "acc": 0.7403767403767404,
346
+ "acc_stderr": 0.012552154236305988,
347
+ "args": {
348
+ "exp_name": "results/v2/commonsense_qa/Qwen/Qwen2-1.5B-Instruct",
349
+ "task_name": "commonsense_qa",
350
+ "model_name": "Qwen/Qwen2-1.5B-Instruct",
351
+ "dataset_param": {
352
+ "max_num_train": 50000,
353
+ "completion_only": false,
354
+ "apply_chat_template": true
355
+ },
356
+ "resume_from_checkpoint": false,
357
+ "train_param": {
358
+ "max_length": 1024,
359
+ "per_device_train_batch_size": 8,
360
+ "auto_find_batch_size": true,
361
+ "num_train_epochs": 5,
362
+ "max_steps": -1,
363
+ "logging_strategy": "epoch",
364
+ "save_strategy": "epoch",
365
+ "logging_steps": 0.2,
366
+ "save_steps": 0.2,
367
+ "gradient_accumulation_steps": 1,
368
+ "gradient_checkpointing": false,
369
+ "optim": "adamw_torch",
370
+ "learning_rate": 5e-05,
371
+ "weight_decay": 0.01,
372
+ "warmup_steps": 0,
373
+ "report_to": "tensorboard",
374
+ "save_safetensors": true,
375
+ "packing": false
376
+ },
377
+ "use_peft": true,
378
+ "peft_task_type": "CAUSAL_LM",
379
+ "peft_inference_mode": false,
380
+ "peft_r": 8,
381
+ "peft_lora_alpha": 32,
382
+ "peft_lora_dropout": 0.1,
383
+ "fsdp": "none",
384
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
385
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
386
+ "fsdp_fsdp_min_num_params": 100000000.0,
387
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
388
+ "fsdp_fsdp_forward_prefetch": false,
389
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
390
+ "fsdp_fsdp_offload_params": false,
391
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
392
+ "fsdp_fsdp_use_orig_params": true,
393
+ "fsdp_fsdp_activation_checkpointing": false,
394
+ "fsdp_fsdp_limit_all_gathers": true,
395
+ "fsdp_fsdp_sync_module_states": true,
396
+ "use_git": true,
397
+ "random_seed": 0
398
+ }
399
+ },
400
+ "gsm8k": {
401
+ "exact_match,flexible-extract": 0.47384382107657314,
402
+ "exact_match,flexible-extract_stderr": 0.013753627037255045,
403
+ "args": {
404
+ "exp_name": "results/v2/gsm8k/Qwen/Qwen2-1.5B-Instruct",
405
+ "task_name": "gsm8k",
406
+ "model_name": "Qwen/Qwen2-1.5B-Instruct",
407
+ "dataset_param": {
408
+ "max_num_train": 50000,
409
+ "completion_only": false,
410
+ "apply_chat_template": true
411
+ },
412
+ "resume_from_checkpoint": false,
413
+ "train_param": {
414
+ "max_length": 1024,
415
+ "per_device_train_batch_size": 8,
416
+ "auto_find_batch_size": true,
417
+ "num_train_epochs": 5,
418
+ "max_steps": -1,
419
+ "logging_strategy": "epoch",
420
+ "save_strategy": "epoch",
421
+ "logging_steps": 0.2,
422
+ "save_steps": 0.2,
423
+ "gradient_accumulation_steps": 1,
424
+ "gradient_checkpointing": false,
425
+ "optim": "adamw_torch",
426
+ "learning_rate": 5e-05,
427
+ "weight_decay": 0.01,
428
+ "warmup_steps": 0,
429
+ "report_to": "tensorboard",
430
+ "save_safetensors": true,
431
+ "packing": false
432
+ },
433
+ "use_peft": true,
434
+ "peft_task_type": "CAUSAL_LM",
435
+ "peft_inference_mode": false,
436
+ "peft_r": 8,
437
+ "peft_lora_alpha": 32,
438
+ "peft_lora_dropout": 0.1,
439
+ "fsdp": "none",
440
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
441
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
442
+ "fsdp_fsdp_min_num_params": 100000000.0,
443
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
444
+ "fsdp_fsdp_forward_prefetch": false,
445
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
446
+ "fsdp_fsdp_offload_params": false,
447
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
448
+ "fsdp_fsdp_use_orig_params": true,
449
+ "fsdp_fsdp_activation_checkpointing": false,
450
+ "fsdp_fsdp_limit_all_gathers": true,
451
+ "fsdp_fsdp_sync_module_states": true,
452
+ "use_git": true,
453
+ "random_seed": 0
454
+ }
455
+ },
456
+ "arc_challenge": {
457
+ "acc_norm": 0.4052901023890785,
458
+ "acc_norm_stderr": 0.014346869060229328,
459
+ "args": {
460
+ "exp_name": "results/v2/arc_challenge/Qwen/Qwen2-1.5B-Instruct",
461
+ "task_name": "arc_challenge",
462
+ "model_name": "Qwen/Qwen2-1.5B-Instruct",
463
+ "dataset_param": {
464
+ "max_num_train": 50000,
465
+ "completion_only": false,
466
+ "apply_chat_template": true
467
+ },
468
+ "resume_from_checkpoint": false,
469
+ "train_param": {
470
+ "max_length": 1024,
471
+ "per_device_train_batch_size": 8,
472
+ "auto_find_batch_size": true,
473
+ "num_train_epochs": 5,
474
+ "max_steps": -1,
475
+ "logging_strategy": "epoch",
476
+ "save_strategy": "epoch",
477
+ "logging_steps": 0.2,
478
+ "save_steps": 0.2,
479
+ "gradient_accumulation_steps": 1,
480
+ "gradient_checkpointing": false,
481
+ "optim": "adamw_torch",
482
+ "learning_rate": 5e-05,
483
+ "weight_decay": 0.01,
484
+ "warmup_steps": 0,
485
+ "report_to": "tensorboard",
486
+ "save_safetensors": true,
487
+ "packing": false
488
+ },
489
+ "use_peft": true,
490
+ "peft_task_type": "CAUSAL_LM",
491
+ "peft_inference_mode": false,
492
+ "peft_r": 8,
493
+ "peft_lora_alpha": 32,
494
+ "peft_lora_dropout": 0.1,
495
+ "fsdp": "none",
496
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
497
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
498
+ "fsdp_fsdp_min_num_params": 100000000.0,
499
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
500
+ "fsdp_fsdp_forward_prefetch": false,
501
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
502
+ "fsdp_fsdp_offload_params": false,
503
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
504
+ "fsdp_fsdp_use_orig_params": true,
505
+ "fsdp_fsdp_activation_checkpointing": false,
506
+ "fsdp_fsdp_limit_all_gathers": true,
507
+ "fsdp_fsdp_sync_module_states": true,
508
+ "use_git": true,
509
+ "random_seed": 0
510
+ }
511
+ }
512
+ }
513
+ }
Qwen/Qwen2-1.5B/results_2026-01-05-04-27-50.json ADDED
@@ -0,0 +1,513 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "model_name": "Qwen/Qwen2-1.5B",
4
+ "model_type": "pretrained",
5
+ "params": 1.5
6
+ },
7
+ "results": {
8
+ "medmcqa": {
9
+ "acc_norm": 0.4594788429356921,
10
+ "acc_norm_stderr": 0.007706320978890783,
11
+ "args": {
12
+ "exp_name": "results/v1/medmcqa/Qwen/Qwen2-1.5B",
13
+ "task_name": "medmcqa",
14
+ "model_name": "Qwen/Qwen2-1.5B",
15
+ "dataset_param": {
16
+ "max_num_train": 50000,
17
+ "completion_only": false,
18
+ "apply_chat_template": false
19
+ },
20
+ "resume_from_checkpoint": false,
21
+ "train_param": {
22
+ "max_length": 1024,
23
+ "per_device_train_batch_size": 8,
24
+ "auto_find_batch_size": true,
25
+ "num_train_epochs": 1,
26
+ "max_steps": -1,
27
+ "logging_strategy": "steps",
28
+ "save_strategy": "steps",
29
+ "logging_steps": 0.2,
30
+ "save_steps": 0.2,
31
+ "gradient_accumulation_steps": 1,
32
+ "gradient_checkpointing": false,
33
+ "optim": "adamw_torch",
34
+ "learning_rate": 5e-05,
35
+ "weight_decay": 0.01,
36
+ "warmup_steps": 0,
37
+ "report_to": "tensorboard",
38
+ "save_safetensors": true,
39
+ "packing": false
40
+ },
41
+ "use_peft": true,
42
+ "peft_task_type": "CAUSAL_LM",
43
+ "peft_inference_mode": false,
44
+ "peft_r": 8,
45
+ "peft_lora_alpha": 32,
46
+ "peft_lora_dropout": 0.1,
47
+ "fsdp": "none",
48
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
49
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
50
+ "fsdp_fsdp_min_num_params": 100000000.0,
51
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
52
+ "fsdp_fsdp_forward_prefetch": false,
53
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
54
+ "fsdp_fsdp_offload_params": false,
55
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
56
+ "fsdp_fsdp_use_orig_params": true,
57
+ "fsdp_fsdp_activation_checkpointing": false,
58
+ "fsdp_fsdp_limit_all_gathers": true,
59
+ "fsdp_fsdp_sync_module_states": true,
60
+ "use_git": true,
61
+ "random_seed": 0
62
+ }
63
+ },
64
+ "nq_open": {
65
+ "exact_match,remove_whitespace": 0.15318559556786704,
66
+ "exact_match,remove_whitespace_stderr": 0.005995282362521945,
67
+ "args": {
68
+ "exp_name": "results/v1/nq_open/Qwen/Qwen2-1.5B",
69
+ "task_name": "nq_open",
70
+ "model_name": "Qwen/Qwen2-1.5B",
71
+ "dataset_param": {
72
+ "max_num_train": 50000,
73
+ "completion_only": false,
74
+ "apply_chat_template": false
75
+ },
76
+ "resume_from_checkpoint": false,
77
+ "train_param": {
78
+ "max_length": 1024,
79
+ "per_device_train_batch_size": 8,
80
+ "auto_find_batch_size": true,
81
+ "num_train_epochs": 1,
82
+ "max_steps": -1,
83
+ "logging_strategy": "steps",
84
+ "save_strategy": "steps",
85
+ "logging_steps": 0.2,
86
+ "save_steps": 0.2,
87
+ "gradient_accumulation_steps": 1,
88
+ "gradient_checkpointing": false,
89
+ "optim": "adamw_torch",
90
+ "learning_rate": 5e-05,
91
+ "weight_decay": 0.01,
92
+ "warmup_steps": 0,
93
+ "report_to": "tensorboard",
94
+ "save_safetensors": true,
95
+ "packing": false
96
+ },
97
+ "use_peft": true,
98
+ "peft_task_type": "CAUSAL_LM",
99
+ "peft_inference_mode": false,
100
+ "peft_r": 8,
101
+ "peft_lora_alpha": 32,
102
+ "peft_lora_dropout": 0.1,
103
+ "fsdp": "none",
104
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
105
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
106
+ "fsdp_fsdp_min_num_params": 100000000.0,
107
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
108
+ "fsdp_fsdp_forward_prefetch": false,
109
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
110
+ "fsdp_fsdp_offload_params": false,
111
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
112
+ "fsdp_fsdp_use_orig_params": true,
113
+ "fsdp_fsdp_activation_checkpointing": false,
114
+ "fsdp_fsdp_limit_all_gathers": true,
115
+ "fsdp_fsdp_sync_module_states": true,
116
+ "use_git": true,
117
+ "random_seed": 0
118
+ }
119
+ },
120
+ "winogrande": {
121
+ "acc": 0.7024467245461721,
122
+ "acc_stderr": 0.012849085254614647,
123
+ "args": {
124
+ "exp_name": "results/v1/winogrande/Qwen/Qwen2-1.5B",
125
+ "task_name": "winogrande",
126
+ "model_name": "Qwen/Qwen2-1.5B",
127
+ "dataset_param": {
128
+ "max_num_train": 50000,
129
+ "completion_only": false,
130
+ "apply_chat_template": false
131
+ },
132
+ "resume_from_checkpoint": false,
133
+ "train_param": {
134
+ "max_length": 1024,
135
+ "per_device_train_batch_size": 8,
136
+ "auto_find_batch_size": true,
137
+ "num_train_epochs": 1,
138
+ "max_steps": -1,
139
+ "logging_strategy": "steps",
140
+ "save_strategy": "steps",
141
+ "logging_steps": 0.2,
142
+ "save_steps": 0.2,
143
+ "gradient_accumulation_steps": 1,
144
+ "gradient_checkpointing": false,
145
+ "optim": "adamw_torch",
146
+ "learning_rate": 5e-05,
147
+ "weight_decay": 0.01,
148
+ "warmup_steps": 0,
149
+ "report_to": "tensorboard",
150
+ "save_safetensors": true,
151
+ "packing": false
152
+ },
153
+ "use_peft": true,
154
+ "peft_task_type": "CAUSAL_LM",
155
+ "peft_inference_mode": false,
156
+ "peft_r": 8,
157
+ "peft_lora_alpha": 32,
158
+ "peft_lora_dropout": 0.1,
159
+ "fsdp": "none",
160
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
161
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
162
+ "fsdp_fsdp_min_num_params": 100000000.0,
163
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
164
+ "fsdp_fsdp_forward_prefetch": false,
165
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
166
+ "fsdp_fsdp_offload_params": false,
167
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
168
+ "fsdp_fsdp_use_orig_params": true,
169
+ "fsdp_fsdp_activation_checkpointing": false,
170
+ "fsdp_fsdp_limit_all_gathers": true,
171
+ "fsdp_fsdp_sync_module_states": true,
172
+ "use_git": true,
173
+ "random_seed": 0
174
+ }
175
+ },
176
+ "hellaswag": {
177
+ "acc_norm": 0.6845,
178
+ "acc_norm_stderr": 0.004647379348996679,
179
+ "args": {
180
+ "exp_name": "results/v1/hellaswag/Qwen/Qwen2-1.5B",
181
+ "task_name": "hellaswag",
182
+ "model_name": "Qwen/Qwen2-1.5B",
183
+ "dataset_param": {
184
+ "max_num_train": 50000,
185
+ "completion_only": false,
186
+ "apply_chat_template": false
187
+ },
188
+ "resume_from_checkpoint": false,
189
+ "train_param": {
190
+ "max_length": 1024,
191
+ "per_device_train_batch_size": 8,
192
+ "auto_find_batch_size": true,
193
+ "num_train_epochs": 1,
194
+ "max_steps": -1,
195
+ "logging_strategy": "steps",
196
+ "save_strategy": "steps",
197
+ "logging_steps": 0.2,
198
+ "save_steps": 0.2,
199
+ "gradient_accumulation_steps": 1,
200
+ "gradient_checkpointing": false,
201
+ "optim": "adamw_torch",
202
+ "learning_rate": 5e-05,
203
+ "weight_decay": 0.01,
204
+ "warmup_steps": 0,
205
+ "report_to": "tensorboard",
206
+ "save_safetensors": true,
207
+ "packing": false
208
+ },
209
+ "use_peft": true,
210
+ "peft_task_type": "CAUSAL_LM",
211
+ "peft_inference_mode": false,
212
+ "peft_r": 8,
213
+ "peft_lora_alpha": 32,
214
+ "peft_lora_dropout": 0.1,
215
+ "fsdp": "none",
216
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
217
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
218
+ "fsdp_fsdp_min_num_params": 100000000.0,
219
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
220
+ "fsdp_fsdp_forward_prefetch": false,
221
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
222
+ "fsdp_fsdp_offload_params": false,
223
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
224
+ "fsdp_fsdp_use_orig_params": true,
225
+ "fsdp_fsdp_activation_checkpointing": false,
226
+ "fsdp_fsdp_limit_all_gathers": true,
227
+ "fsdp_fsdp_sync_module_states": true,
228
+ "use_git": true,
229
+ "random_seed": 0
230
+ }
231
+ },
232
+ "social_iqa": {
233
+ "acc": 0.5250767656090072,
234
+ "acc_stderr": 0.011299832020886457,
235
+ "args": {
236
+ "exp_name": "results/v1/social_iqa/Qwen/Qwen2-1.5B",
237
+ "task_name": "social_iqa",
238
+ "model_name": "Qwen/Qwen2-1.5B",
239
+ "dataset_param": {
240
+ "max_num_train": 50000,
241
+ "completion_only": false,
242
+ "apply_chat_template": false
243
+ },
244
+ "resume_from_checkpoint": false,
245
+ "train_param": {
246
+ "max_length": 1024,
247
+ "per_device_train_batch_size": 8,
248
+ "auto_find_batch_size": true,
249
+ "num_train_epochs": 1,
250
+ "max_steps": -1,
251
+ "logging_strategy": "steps",
252
+ "save_strategy": "steps",
253
+ "logging_steps": 0.2,
254
+ "save_steps": 0.2,
255
+ "gradient_accumulation_steps": 1,
256
+ "gradient_checkpointing": false,
257
+ "optim": "adamw_torch",
258
+ "learning_rate": 5e-05,
259
+ "weight_decay": 0.01,
260
+ "warmup_steps": 0,
261
+ "report_to": "tensorboard",
262
+ "save_safetensors": true,
263
+ "packing": false
264
+ },
265
+ "use_peft": true,
266
+ "peft_task_type": "CAUSAL_LM",
267
+ "peft_inference_mode": false,
268
+ "peft_r": 8,
269
+ "peft_lora_alpha": 32,
270
+ "peft_lora_dropout": 0.1,
271
+ "fsdp": "none",
272
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
273
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
274
+ "fsdp_fsdp_min_num_params": 100000000.0,
275
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
276
+ "fsdp_fsdp_forward_prefetch": false,
277
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
278
+ "fsdp_fsdp_offload_params": false,
279
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
280
+ "fsdp_fsdp_use_orig_params": true,
281
+ "fsdp_fsdp_activation_checkpointing": false,
282
+ "fsdp_fsdp_limit_all_gathers": true,
283
+ "fsdp_fsdp_sync_module_states": true,
284
+ "use_git": true,
285
+ "random_seed": 0
286
+ }
287
+ },
288
+ "piqa": {
289
+ "acc_norm": 0.7676822633297062,
290
+ "acc_norm_stderr": 0.009853201384168241,
291
+ "args": {
292
+ "exp_name": "results/v2/piqa/Qwen/Qwen2-1.5B",
293
+ "task_name": "piqa",
294
+ "model_name": "Qwen/Qwen2-1.5B",
295
+ "dataset_param": {
296
+ "max_num_train": 50000,
297
+ "completion_only": false,
298
+ "apply_chat_template": false
299
+ },
300
+ "resume_from_checkpoint": false,
301
+ "train_param": {
302
+ "max_length": 1024,
303
+ "per_device_train_batch_size": 8,
304
+ "auto_find_batch_size": true,
305
+ "num_train_epochs": 5,
306
+ "max_steps": -1,
307
+ "logging_strategy": "epoch",
308
+ "save_strategy": "epoch",
309
+ "logging_steps": 0.2,
310
+ "save_steps": 0.2,
311
+ "gradient_accumulation_steps": 1,
312
+ "gradient_checkpointing": false,
313
+ "optim": "adamw_torch",
314
+ "learning_rate": 5e-05,
315
+ "weight_decay": 0.01,
316
+ "warmup_steps": 0,
317
+ "report_to": "tensorboard",
318
+ "save_safetensors": true,
319
+ "packing": false
320
+ },
321
+ "use_peft": true,
322
+ "peft_task_type": "CAUSAL_LM",
323
+ "peft_inference_mode": false,
324
+ "peft_r": 8,
325
+ "peft_lora_alpha": 32,
326
+ "peft_lora_dropout": 0.1,
327
+ "fsdp": "none",
328
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
329
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
330
+ "fsdp_fsdp_min_num_params": 100000000.0,
331
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
332
+ "fsdp_fsdp_forward_prefetch": false,
333
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
334
+ "fsdp_fsdp_offload_params": false,
335
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
336
+ "fsdp_fsdp_use_orig_params": true,
337
+ "fsdp_fsdp_activation_checkpointing": false,
338
+ "fsdp_fsdp_limit_all_gathers": true,
339
+ "fsdp_fsdp_sync_module_states": true,
340
+ "use_git": true,
341
+ "random_seed": 0
342
+ }
343
+ },
344
+ "commonsense_qa": {
345
+ "acc": 0.7346437346437347,
346
+ "acc_stderr": 0.012640758880532864,
347
+ "args": {
348
+ "exp_name": "results/v2/commonsense_qa/Qwen/Qwen2-1.5B",
349
+ "task_name": "commonsense_qa",
350
+ "model_name": "Qwen/Qwen2-1.5B",
351
+ "dataset_param": {
352
+ "max_num_train": 50000,
353
+ "completion_only": false,
354
+ "apply_chat_template": false
355
+ },
356
+ "resume_from_checkpoint": false,
357
+ "train_param": {
358
+ "max_length": 1024,
359
+ "per_device_train_batch_size": 8,
360
+ "auto_find_batch_size": true,
361
+ "num_train_epochs": 5,
362
+ "max_steps": -1,
363
+ "logging_strategy": "epoch",
364
+ "save_strategy": "epoch",
365
+ "logging_steps": 0.2,
366
+ "save_steps": 0.2,
367
+ "gradient_accumulation_steps": 1,
368
+ "gradient_checkpointing": false,
369
+ "optim": "adamw_torch",
370
+ "learning_rate": 5e-05,
371
+ "weight_decay": 0.01,
372
+ "warmup_steps": 0,
373
+ "report_to": "tensorboard",
374
+ "save_safetensors": true,
375
+ "packing": false
376
+ },
377
+ "use_peft": true,
378
+ "peft_task_type": "CAUSAL_LM",
379
+ "peft_inference_mode": false,
380
+ "peft_r": 8,
381
+ "peft_lora_alpha": 32,
382
+ "peft_lora_dropout": 0.1,
383
+ "fsdp": "none",
384
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
385
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
386
+ "fsdp_fsdp_min_num_params": 100000000.0,
387
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
388
+ "fsdp_fsdp_forward_prefetch": false,
389
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
390
+ "fsdp_fsdp_offload_params": false,
391
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
392
+ "fsdp_fsdp_use_orig_params": true,
393
+ "fsdp_fsdp_activation_checkpointing": false,
394
+ "fsdp_fsdp_limit_all_gathers": true,
395
+ "fsdp_fsdp_sync_module_states": true,
396
+ "use_git": true,
397
+ "random_seed": 0
398
+ }
399
+ },
400
+ "gsm8k": {
401
+ "exact_match,flexible-extract": 0.5473843821076573,
402
+ "exact_match,flexible-extract_stderr": 0.013710499070935137,
403
+ "args": {
404
+ "exp_name": "results/v2/gsm8k/Qwen/Qwen2-1.5B",
405
+ "task_name": "gsm8k",
406
+ "model_name": "Qwen/Qwen2-1.5B",
407
+ "dataset_param": {
408
+ "max_num_train": 50000,
409
+ "completion_only": false,
410
+ "apply_chat_template": false
411
+ },
412
+ "resume_from_checkpoint": false,
413
+ "train_param": {
414
+ "max_length": 1024,
415
+ "per_device_train_batch_size": 8,
416
+ "auto_find_batch_size": true,
417
+ "num_train_epochs": 5,
418
+ "max_steps": -1,
419
+ "logging_strategy": "epoch",
420
+ "save_strategy": "epoch",
421
+ "logging_steps": 0.2,
422
+ "save_steps": 0.2,
423
+ "gradient_accumulation_steps": 1,
424
+ "gradient_checkpointing": false,
425
+ "optim": "adamw_torch",
426
+ "learning_rate": 5e-05,
427
+ "weight_decay": 0.01,
428
+ "warmup_steps": 0,
429
+ "report_to": "tensorboard",
430
+ "save_safetensors": true,
431
+ "packing": false
432
+ },
433
+ "use_peft": true,
434
+ "peft_task_type": "CAUSAL_LM",
435
+ "peft_inference_mode": false,
436
+ "peft_r": 8,
437
+ "peft_lora_alpha": 32,
438
+ "peft_lora_dropout": 0.1,
439
+ "fsdp": "none",
440
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
441
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
442
+ "fsdp_fsdp_min_num_params": 100000000.0,
443
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
444
+ "fsdp_fsdp_forward_prefetch": false,
445
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
446
+ "fsdp_fsdp_offload_params": false,
447
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
448
+ "fsdp_fsdp_use_orig_params": true,
449
+ "fsdp_fsdp_activation_checkpointing": false,
450
+ "fsdp_fsdp_limit_all_gathers": true,
451
+ "fsdp_fsdp_sync_module_states": true,
452
+ "use_git": true,
453
+ "random_seed": 0
454
+ }
455
+ },
456
+ "arc_challenge": {
457
+ "acc_norm": 0.447098976109215,
458
+ "acc_norm_stderr": 0.014529380160526848,
459
+ "args": {
460
+ "exp_name": "results/v2/arc_challenge/Qwen/Qwen2-1.5B",
461
+ "task_name": "arc_challenge",
462
+ "model_name": "Qwen/Qwen2-1.5B",
463
+ "dataset_param": {
464
+ "max_num_train": 50000,
465
+ "completion_only": false,
466
+ "apply_chat_template": false
467
+ },
468
+ "resume_from_checkpoint": false,
469
+ "train_param": {
470
+ "max_length": 1024,
471
+ "per_device_train_batch_size": 8,
472
+ "auto_find_batch_size": true,
473
+ "num_train_epochs": 5,
474
+ "max_steps": -1,
475
+ "logging_strategy": "epoch",
476
+ "save_strategy": "epoch",
477
+ "logging_steps": 0.2,
478
+ "save_steps": 0.2,
479
+ "gradient_accumulation_steps": 1,
480
+ "gradient_checkpointing": false,
481
+ "optim": "adamw_torch",
482
+ "learning_rate": 5e-05,
483
+ "weight_decay": 0.01,
484
+ "warmup_steps": 0,
485
+ "report_to": "tensorboard",
486
+ "save_safetensors": true,
487
+ "packing": false
488
+ },
489
+ "use_peft": true,
490
+ "peft_task_type": "CAUSAL_LM",
491
+ "peft_inference_mode": false,
492
+ "peft_r": 8,
493
+ "peft_lora_alpha": 32,
494
+ "peft_lora_dropout": 0.1,
495
+ "fsdp": "none",
496
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
497
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
498
+ "fsdp_fsdp_min_num_params": 100000000.0,
499
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
500
+ "fsdp_fsdp_forward_prefetch": false,
501
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
502
+ "fsdp_fsdp_offload_params": false,
503
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
504
+ "fsdp_fsdp_use_orig_params": true,
505
+ "fsdp_fsdp_activation_checkpointing": false,
506
+ "fsdp_fsdp_limit_all_gathers": true,
507
+ "fsdp_fsdp_sync_module_states": true,
508
+ "use_git": true,
509
+ "random_seed": 0
510
+ }
511
+ }
512
+ }
513
+ }
Qwen/Qwen2-7B-Instruct/results_2026-01-05-04-27-51.json ADDED
@@ -0,0 +1,513 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "model_name": "Qwen/Qwen2-7B-Instruct",
4
+ "model_type": "instruction-tuned",
5
+ "params": 7.0
6
+ },
7
+ "results": {
8
+ "medmcqa": {
9
+ "acc_norm": 0.5723165192445613,
10
+ "acc_norm_stderr": 0.007650456753217209,
11
+ "args": {
12
+ "exp_name": "results/v1/medmcqa/Qwen/Qwen2-7B-Instruct",
13
+ "task_name": "medmcqa",
14
+ "model_name": "Qwen/Qwen2-7B-Instruct",
15
+ "dataset_param": {
16
+ "max_num_train": 50000,
17
+ "completion_only": false,
18
+ "apply_chat_template": true
19
+ },
20
+ "resume_from_checkpoint": false,
21
+ "train_param": {
22
+ "max_length": 1024,
23
+ "per_device_train_batch_size": 8,
24
+ "auto_find_batch_size": true,
25
+ "num_train_epochs": 1,
26
+ "max_steps": -1,
27
+ "logging_strategy": "steps",
28
+ "save_strategy": "steps",
29
+ "logging_steps": 0.2,
30
+ "save_steps": 0.2,
31
+ "gradient_accumulation_steps": 1,
32
+ "gradient_checkpointing": false,
33
+ "optim": "adamw_torch",
34
+ "learning_rate": 5e-05,
35
+ "weight_decay": 0.01,
36
+ "warmup_steps": 0,
37
+ "report_to": "tensorboard",
38
+ "save_safetensors": true,
39
+ "packing": false
40
+ },
41
+ "use_peft": true,
42
+ "peft_task_type": "CAUSAL_LM",
43
+ "peft_inference_mode": false,
44
+ "peft_r": 8,
45
+ "peft_lora_alpha": 32,
46
+ "peft_lora_dropout": 0.1,
47
+ "fsdp": "none",
48
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
49
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
50
+ "fsdp_fsdp_min_num_params": 100000000.0,
51
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
52
+ "fsdp_fsdp_forward_prefetch": false,
53
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
54
+ "fsdp_fsdp_offload_params": false,
55
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
56
+ "fsdp_fsdp_use_orig_params": true,
57
+ "fsdp_fsdp_activation_checkpointing": false,
58
+ "fsdp_fsdp_limit_all_gathers": true,
59
+ "fsdp_fsdp_sync_module_states": true,
60
+ "use_git": true,
61
+ "random_seed": 0
62
+ }
63
+ },
64
+ "nq_open": {
65
+ "exact_match,remove_whitespace": 0.26481994459833796,
66
+ "exact_match,remove_whitespace_stderr": 0.0073447795040573385,
67
+ "args": {
68
+ "exp_name": "results/v1/nq_open/Qwen/Qwen2-7B-Instruct",
69
+ "task_name": "nq_open",
70
+ "model_name": "Qwen/Qwen2-7B-Instruct",
71
+ "dataset_param": {
72
+ "max_num_train": 50000,
73
+ "completion_only": false,
74
+ "apply_chat_template": true
75
+ },
76
+ "resume_from_checkpoint": false,
77
+ "train_param": {
78
+ "max_length": 1024,
79
+ "per_device_train_batch_size": 8,
80
+ "auto_find_batch_size": true,
81
+ "num_train_epochs": 1,
82
+ "max_steps": -1,
83
+ "logging_strategy": "steps",
84
+ "save_strategy": "steps",
85
+ "logging_steps": 0.2,
86
+ "save_steps": 0.2,
87
+ "gradient_accumulation_steps": 1,
88
+ "gradient_checkpointing": false,
89
+ "optim": "adamw_torch",
90
+ "learning_rate": 5e-05,
91
+ "weight_decay": 0.01,
92
+ "warmup_steps": 0,
93
+ "report_to": "tensorboard",
94
+ "save_safetensors": true,
95
+ "packing": false
96
+ },
97
+ "use_peft": true,
98
+ "peft_task_type": "CAUSAL_LM",
99
+ "peft_inference_mode": false,
100
+ "peft_r": 8,
101
+ "peft_lora_alpha": 32,
102
+ "peft_lora_dropout": 0.1,
103
+ "fsdp": "none",
104
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
105
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
106
+ "fsdp_fsdp_min_num_params": 100000000.0,
107
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
108
+ "fsdp_fsdp_forward_prefetch": false,
109
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
110
+ "fsdp_fsdp_offload_params": false,
111
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
112
+ "fsdp_fsdp_use_orig_params": true,
113
+ "fsdp_fsdp_activation_checkpointing": false,
114
+ "fsdp_fsdp_limit_all_gathers": true,
115
+ "fsdp_fsdp_sync_module_states": true,
116
+ "use_git": true,
117
+ "random_seed": 0
118
+ }
119
+ },
120
+ "winogrande": {
121
+ "acc": 0.813733228097869,
122
+ "acc_stderr": 0.010941877955676211,
123
+ "args": {
124
+ "exp_name": "results/v1/winogrande/Qwen/Qwen2-7B-Instruct",
125
+ "task_name": "winogrande",
126
+ "model_name": "Qwen/Qwen2-7B-Instruct",
127
+ "dataset_param": {
128
+ "max_num_train": 50000,
129
+ "completion_only": false,
130
+ "apply_chat_template": true
131
+ },
132
+ "resume_from_checkpoint": false,
133
+ "train_param": {
134
+ "max_length": 1024,
135
+ "per_device_train_batch_size": 8,
136
+ "auto_find_batch_size": true,
137
+ "num_train_epochs": 1,
138
+ "max_steps": -1,
139
+ "logging_strategy": "steps",
140
+ "save_strategy": "steps",
141
+ "logging_steps": 0.2,
142
+ "save_steps": 0.2,
143
+ "gradient_accumulation_steps": 1,
144
+ "gradient_checkpointing": false,
145
+ "optim": "adamw_torch",
146
+ "learning_rate": 5e-05,
147
+ "weight_decay": 0.01,
148
+ "warmup_steps": 0,
149
+ "report_to": "tensorboard",
150
+ "save_safetensors": true,
151
+ "packing": false
152
+ },
153
+ "use_peft": true,
154
+ "peft_task_type": "CAUSAL_LM",
155
+ "peft_inference_mode": false,
156
+ "peft_r": 8,
157
+ "peft_lora_alpha": 32,
158
+ "peft_lora_dropout": 0.1,
159
+ "fsdp": "none",
160
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
161
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
162
+ "fsdp_fsdp_min_num_params": 100000000.0,
163
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
164
+ "fsdp_fsdp_forward_prefetch": false,
165
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
166
+ "fsdp_fsdp_offload_params": false,
167
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
168
+ "fsdp_fsdp_use_orig_params": true,
169
+ "fsdp_fsdp_activation_checkpointing": false,
170
+ "fsdp_fsdp_limit_all_gathers": true,
171
+ "fsdp_fsdp_sync_module_states": true,
172
+ "use_git": true,
173
+ "random_seed": 0
174
+ }
175
+ },
176
+ "hellaswag": {
177
+ "acc_norm": 0.8129,
178
+ "acc_norm_stderr": 0.0039001128301642362,
179
+ "args": {
180
+ "exp_name": "results/v1/hellaswag/Qwen/Qwen2-7B-Instruct",
181
+ "task_name": "hellaswag",
182
+ "model_name": "Qwen/Qwen2-7B-Instruct",
183
+ "dataset_param": {
184
+ "max_num_train": 50000,
185
+ "completion_only": false,
186
+ "apply_chat_template": true
187
+ },
188
+ "resume_from_checkpoint": false,
189
+ "train_param": {
190
+ "max_length": 1024,
191
+ "per_device_train_batch_size": 8,
192
+ "auto_find_batch_size": true,
193
+ "num_train_epochs": 1,
194
+ "max_steps": -1,
195
+ "logging_strategy": "steps",
196
+ "save_strategy": "steps",
197
+ "logging_steps": 0.2,
198
+ "save_steps": 0.2,
199
+ "gradient_accumulation_steps": 1,
200
+ "gradient_checkpointing": false,
201
+ "optim": "adamw_torch",
202
+ "learning_rate": 5e-05,
203
+ "weight_decay": 0.01,
204
+ "warmup_steps": 0,
205
+ "report_to": "tensorboard",
206
+ "save_safetensors": true,
207
+ "packing": false
208
+ },
209
+ "use_peft": true,
210
+ "peft_task_type": "CAUSAL_LM",
211
+ "peft_inference_mode": false,
212
+ "peft_r": 8,
213
+ "peft_lora_alpha": 32,
214
+ "peft_lora_dropout": 0.1,
215
+ "fsdp": "none",
216
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
217
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
218
+ "fsdp_fsdp_min_num_params": 100000000.0,
219
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
220
+ "fsdp_fsdp_forward_prefetch": false,
221
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
222
+ "fsdp_fsdp_offload_params": false,
223
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
224
+ "fsdp_fsdp_use_orig_params": true,
225
+ "fsdp_fsdp_activation_checkpointing": false,
226
+ "fsdp_fsdp_limit_all_gathers": true,
227
+ "fsdp_fsdp_sync_module_states": true,
228
+ "use_git": true,
229
+ "random_seed": 0
230
+ }
231
+ },
232
+ "social_iqa": {
233
+ "acc": 0.5619242579324463,
234
+ "acc_stderr": 0.01122696506802993,
235
+ "args": {
236
+ "exp_name": "results/v1/social_iqa/Qwen/Qwen2-7B-Instruct",
237
+ "task_name": "social_iqa",
238
+ "model_name": "Qwen/Qwen2-7B-Instruct",
239
+ "dataset_param": {
240
+ "max_num_train": 50000,
241
+ "completion_only": false,
242
+ "apply_chat_template": true
243
+ },
244
+ "resume_from_checkpoint": false,
245
+ "train_param": {
246
+ "max_length": 1024,
247
+ "per_device_train_batch_size": 8,
248
+ "auto_find_batch_size": true,
249
+ "num_train_epochs": 1,
250
+ "max_steps": -1,
251
+ "logging_strategy": "steps",
252
+ "save_strategy": "steps",
253
+ "logging_steps": 0.2,
254
+ "save_steps": 0.2,
255
+ "gradient_accumulation_steps": 1,
256
+ "gradient_checkpointing": false,
257
+ "optim": "adamw_torch",
258
+ "learning_rate": 5e-05,
259
+ "weight_decay": 0.01,
260
+ "warmup_steps": 0,
261
+ "report_to": "tensorboard",
262
+ "save_safetensors": true,
263
+ "packing": false
264
+ },
265
+ "use_peft": true,
266
+ "peft_task_type": "CAUSAL_LM",
267
+ "peft_inference_mode": false,
268
+ "peft_r": 8,
269
+ "peft_lora_alpha": 32,
270
+ "peft_lora_dropout": 0.1,
271
+ "fsdp": "none",
272
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
273
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
274
+ "fsdp_fsdp_min_num_params": 100000000.0,
275
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
276
+ "fsdp_fsdp_forward_prefetch": false,
277
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
278
+ "fsdp_fsdp_offload_params": false,
279
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
280
+ "fsdp_fsdp_use_orig_params": true,
281
+ "fsdp_fsdp_activation_checkpointing": false,
282
+ "fsdp_fsdp_limit_all_gathers": true,
283
+ "fsdp_fsdp_sync_module_states": true,
284
+ "use_git": true,
285
+ "random_seed": 0
286
+ }
287
+ },
288
+ "piqa": {
289
+ "acc_norm": 0.8199129488574538,
290
+ "acc_norm_stderr": 0.008965417898903191,
291
+ "args": {
292
+ "exp_name": "results/v2/piqa/Qwen/Qwen2-7B-Instruct",
293
+ "task_name": "piqa",
294
+ "model_name": "Qwen/Qwen2-7B-Instruct",
295
+ "dataset_param": {
296
+ "max_num_train": 50000,
297
+ "completion_only": false,
298
+ "apply_chat_template": true
299
+ },
300
+ "resume_from_checkpoint": false,
301
+ "train_param": {
302
+ "max_length": 1024,
303
+ "per_device_train_batch_size": 8,
304
+ "auto_find_batch_size": true,
305
+ "num_train_epochs": 5,
306
+ "max_steps": -1,
307
+ "logging_strategy": "epoch",
308
+ "save_strategy": "epoch",
309
+ "logging_steps": 0.2,
310
+ "save_steps": 0.2,
311
+ "gradient_accumulation_steps": 1,
312
+ "gradient_checkpointing": false,
313
+ "optim": "adamw_torch",
314
+ "learning_rate": 5e-05,
315
+ "weight_decay": 0.01,
316
+ "warmup_steps": 0,
317
+ "report_to": "tensorboard",
318
+ "save_safetensors": true,
319
+ "packing": false
320
+ },
321
+ "use_peft": true,
322
+ "peft_task_type": "CAUSAL_LM",
323
+ "peft_inference_mode": false,
324
+ "peft_r": 8,
325
+ "peft_lora_alpha": 32,
326
+ "peft_lora_dropout": 0.1,
327
+ "fsdp": "none",
328
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
329
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
330
+ "fsdp_fsdp_min_num_params": 100000000.0,
331
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
332
+ "fsdp_fsdp_forward_prefetch": false,
333
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
334
+ "fsdp_fsdp_offload_params": false,
335
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
336
+ "fsdp_fsdp_use_orig_params": true,
337
+ "fsdp_fsdp_activation_checkpointing": false,
338
+ "fsdp_fsdp_limit_all_gathers": true,
339
+ "fsdp_fsdp_sync_module_states": true,
340
+ "use_git": true,
341
+ "random_seed": 0
342
+ }
343
+ },
344
+ "commonsense_qa": {
345
+ "acc": 0.8419328419328419,
346
+ "acc_stderr": 0.010444307085591483,
347
+ "args": {
348
+ "exp_name": "results/v2/commonsense_qa/Qwen/Qwen2-7B-Instruct",
349
+ "task_name": "commonsense_qa",
350
+ "model_name": "Qwen/Qwen2-7B-Instruct",
351
+ "dataset_param": {
352
+ "max_num_train": 50000,
353
+ "completion_only": false,
354
+ "apply_chat_template": true
355
+ },
356
+ "resume_from_checkpoint": false,
357
+ "train_param": {
358
+ "max_length": 1024,
359
+ "per_device_train_batch_size": 8,
360
+ "auto_find_batch_size": true,
361
+ "num_train_epochs": 5,
362
+ "max_steps": -1,
363
+ "logging_strategy": "epoch",
364
+ "save_strategy": "epoch",
365
+ "logging_steps": 0.2,
366
+ "save_steps": 0.2,
367
+ "gradient_accumulation_steps": 1,
368
+ "gradient_checkpointing": false,
369
+ "optim": "adamw_torch",
370
+ "learning_rate": 5e-05,
371
+ "weight_decay": 0.01,
372
+ "warmup_steps": 0,
373
+ "report_to": "tensorboard",
374
+ "save_safetensors": true,
375
+ "packing": false
376
+ },
377
+ "use_peft": true,
378
+ "peft_task_type": "CAUSAL_LM",
379
+ "peft_inference_mode": false,
380
+ "peft_r": 8,
381
+ "peft_lora_alpha": 32,
382
+ "peft_lora_dropout": 0.1,
383
+ "fsdp": "none",
384
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
385
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
386
+ "fsdp_fsdp_min_num_params": 100000000.0,
387
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
388
+ "fsdp_fsdp_forward_prefetch": false,
389
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
390
+ "fsdp_fsdp_offload_params": false,
391
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
392
+ "fsdp_fsdp_use_orig_params": true,
393
+ "fsdp_fsdp_activation_checkpointing": false,
394
+ "fsdp_fsdp_limit_all_gathers": true,
395
+ "fsdp_fsdp_sync_module_states": true,
396
+ "use_git": true,
397
+ "random_seed": 0
398
+ }
399
+ },
400
+ "gsm8k": {
401
+ "exact_match,flexible-extract": 0.7498104624715694,
402
+ "exact_match,flexible-extract_stderr": 0.011930334350873354,
403
+ "args": {
404
+ "exp_name": "results/v2/gsm8k/Qwen/Qwen2-7B-Instruct",
405
+ "task_name": "gsm8k",
406
+ "model_name": "Qwen/Qwen2-7B-Instruct",
407
+ "dataset_param": {
408
+ "max_num_train": 50000,
409
+ "completion_only": false,
410
+ "apply_chat_template": true
411
+ },
412
+ "resume_from_checkpoint": false,
413
+ "train_param": {
414
+ "max_length": 1024,
415
+ "per_device_train_batch_size": 8,
416
+ "auto_find_batch_size": true,
417
+ "num_train_epochs": 5,
418
+ "max_steps": -1,
419
+ "logging_strategy": "epoch",
420
+ "save_strategy": "epoch",
421
+ "logging_steps": 0.2,
422
+ "save_steps": 0.2,
423
+ "gradient_accumulation_steps": 1,
424
+ "gradient_checkpointing": false,
425
+ "optim": "adamw_torch",
426
+ "learning_rate": 5e-05,
427
+ "weight_decay": 0.01,
428
+ "warmup_steps": 0,
429
+ "report_to": "tensorboard",
430
+ "save_safetensors": true,
431
+ "packing": false
432
+ },
433
+ "use_peft": true,
434
+ "peft_task_type": "CAUSAL_LM",
435
+ "peft_inference_mode": false,
436
+ "peft_r": 8,
437
+ "peft_lora_alpha": 32,
438
+ "peft_lora_dropout": 0.1,
439
+ "fsdp": "none",
440
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
441
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
442
+ "fsdp_fsdp_min_num_params": 100000000.0,
443
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
444
+ "fsdp_fsdp_forward_prefetch": false,
445
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
446
+ "fsdp_fsdp_offload_params": false,
447
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
448
+ "fsdp_fsdp_use_orig_params": true,
449
+ "fsdp_fsdp_activation_checkpointing": false,
450
+ "fsdp_fsdp_limit_all_gathers": true,
451
+ "fsdp_fsdp_sync_module_states": true,
452
+ "use_git": true,
453
+ "random_seed": 0
454
+ }
455
+ },
456
+ "arc_challenge": {
457
+ "acc_norm": 0.568259385665529,
458
+ "acc_norm_stderr": 0.014474591427196204,
459
+ "args": {
460
+ "exp_name": "results/v2/arc_challenge/Qwen/Qwen2-7B-Instruct",
461
+ "task_name": "arc_challenge",
462
+ "model_name": "Qwen/Qwen2-7B-Instruct",
463
+ "dataset_param": {
464
+ "max_num_train": 50000,
465
+ "completion_only": false,
466
+ "apply_chat_template": true
467
+ },
468
+ "resume_from_checkpoint": false,
469
+ "train_param": {
470
+ "max_length": 1024,
471
+ "per_device_train_batch_size": 8,
472
+ "auto_find_batch_size": true,
473
+ "num_train_epochs": 5,
474
+ "max_steps": -1,
475
+ "logging_strategy": "epoch",
476
+ "save_strategy": "epoch",
477
+ "logging_steps": 0.2,
478
+ "save_steps": 0.2,
479
+ "gradient_accumulation_steps": 1,
480
+ "gradient_checkpointing": false,
481
+ "optim": "adamw_torch",
482
+ "learning_rate": 5e-05,
483
+ "weight_decay": 0.01,
484
+ "warmup_steps": 0,
485
+ "report_to": "tensorboard",
486
+ "save_safetensors": true,
487
+ "packing": false
488
+ },
489
+ "use_peft": true,
490
+ "peft_task_type": "CAUSAL_LM",
491
+ "peft_inference_mode": false,
492
+ "peft_r": 8,
493
+ "peft_lora_alpha": 32,
494
+ "peft_lora_dropout": 0.1,
495
+ "fsdp": "none",
496
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
497
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
498
+ "fsdp_fsdp_min_num_params": 100000000.0,
499
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
500
+ "fsdp_fsdp_forward_prefetch": false,
501
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
502
+ "fsdp_fsdp_offload_params": false,
503
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
504
+ "fsdp_fsdp_use_orig_params": true,
505
+ "fsdp_fsdp_activation_checkpointing": false,
506
+ "fsdp_fsdp_limit_all_gathers": true,
507
+ "fsdp_fsdp_sync_module_states": true,
508
+ "use_git": true,
509
+ "random_seed": 0
510
+ }
511
+ }
512
+ }
513
+ }
Qwen/Qwen2-7B/results_2026-01-05-04-27-50.json ADDED
@@ -0,0 +1,513 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "model_name": "Qwen/Qwen2-7B",
4
+ "model_type": "pretrained",
5
+ "params": 7.0
6
+ },
7
+ "results": {
8
+ "medmcqa": {
9
+ "acc_norm": 0.5794884054506335,
10
+ "acc_norm_stderr": 0.007633423343206934,
11
+ "args": {
12
+ "exp_name": "results/v1/medmcqa/Qwen/Qwen2-7B",
13
+ "task_name": "medmcqa",
14
+ "model_name": "Qwen/Qwen2-7B",
15
+ "dataset_param": {
16
+ "max_num_train": 50000,
17
+ "completion_only": false,
18
+ "apply_chat_template": false
19
+ },
20
+ "resume_from_checkpoint": false,
21
+ "train_param": {
22
+ "max_length": 1024,
23
+ "per_device_train_batch_size": 8,
24
+ "auto_find_batch_size": true,
25
+ "num_train_epochs": 1,
26
+ "max_steps": -1,
27
+ "logging_strategy": "steps",
28
+ "save_strategy": "steps",
29
+ "logging_steps": 0.2,
30
+ "save_steps": 0.2,
31
+ "gradient_accumulation_steps": 1,
32
+ "gradient_checkpointing": false,
33
+ "optim": "adamw_torch",
34
+ "learning_rate": 5e-05,
35
+ "weight_decay": 0.01,
36
+ "warmup_steps": 0,
37
+ "report_to": "tensorboard",
38
+ "save_safetensors": true,
39
+ "packing": false
40
+ },
41
+ "use_peft": true,
42
+ "peft_task_type": "CAUSAL_LM",
43
+ "peft_inference_mode": false,
44
+ "peft_r": 8,
45
+ "peft_lora_alpha": 32,
46
+ "peft_lora_dropout": 0.1,
47
+ "fsdp": "none",
48
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
49
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
50
+ "fsdp_fsdp_min_num_params": 100000000.0,
51
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
52
+ "fsdp_fsdp_forward_prefetch": false,
53
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
54
+ "fsdp_fsdp_offload_params": false,
55
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
56
+ "fsdp_fsdp_use_orig_params": true,
57
+ "fsdp_fsdp_activation_checkpointing": false,
58
+ "fsdp_fsdp_limit_all_gathers": true,
59
+ "fsdp_fsdp_sync_module_states": true,
60
+ "use_git": true,
61
+ "random_seed": 0
62
+ }
63
+ },
64
+ "nq_open": {
65
+ "exact_match,remove_whitespace": 0.2889196675900277,
66
+ "exact_match,remove_whitespace_stderr": 0.007544916636132609,
67
+ "args": {
68
+ "exp_name": "results/v1/nq_open/Qwen/Qwen2-7B",
69
+ "task_name": "nq_open",
70
+ "model_name": "Qwen/Qwen2-7B",
71
+ "dataset_param": {
72
+ "max_num_train": 50000,
73
+ "completion_only": false,
74
+ "apply_chat_template": false
75
+ },
76
+ "resume_from_checkpoint": false,
77
+ "train_param": {
78
+ "max_length": 1024,
79
+ "per_device_train_batch_size": 8,
80
+ "auto_find_batch_size": true,
81
+ "num_train_epochs": 1,
82
+ "max_steps": -1,
83
+ "logging_strategy": "steps",
84
+ "save_strategy": "steps",
85
+ "logging_steps": 0.2,
86
+ "save_steps": 0.2,
87
+ "gradient_accumulation_steps": 1,
88
+ "gradient_checkpointing": false,
89
+ "optim": "adamw_torch",
90
+ "learning_rate": 5e-05,
91
+ "weight_decay": 0.01,
92
+ "warmup_steps": 0,
93
+ "report_to": "tensorboard",
94
+ "save_safetensors": true,
95
+ "packing": false
96
+ },
97
+ "use_peft": true,
98
+ "peft_task_type": "CAUSAL_LM",
99
+ "peft_inference_mode": false,
100
+ "peft_r": 8,
101
+ "peft_lora_alpha": 32,
102
+ "peft_lora_dropout": 0.1,
103
+ "fsdp": "none",
104
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
105
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
106
+ "fsdp_fsdp_min_num_params": 100000000.0,
107
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
108
+ "fsdp_fsdp_forward_prefetch": false,
109
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
110
+ "fsdp_fsdp_offload_params": false,
111
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
112
+ "fsdp_fsdp_use_orig_params": true,
113
+ "fsdp_fsdp_activation_checkpointing": false,
114
+ "fsdp_fsdp_limit_all_gathers": true,
115
+ "fsdp_fsdp_sync_module_states": true,
116
+ "use_git": true,
117
+ "random_seed": 0
118
+ }
119
+ },
120
+ "winogrande": {
121
+ "acc": 0.8342541436464088,
122
+ "acc_stderr": 0.010450899545370644,
123
+ "args": {
124
+ "exp_name": "results/v1/winogrande/Qwen/Qwen2-7B",
125
+ "task_name": "winogrande",
126
+ "model_name": "Qwen/Qwen2-7B",
127
+ "dataset_param": {
128
+ "max_num_train": 50000,
129
+ "completion_only": false,
130
+ "apply_chat_template": false
131
+ },
132
+ "resume_from_checkpoint": false,
133
+ "train_param": {
134
+ "max_length": 1024,
135
+ "per_device_train_batch_size": 8,
136
+ "auto_find_batch_size": true,
137
+ "num_train_epochs": 1,
138
+ "max_steps": -1,
139
+ "logging_strategy": "steps",
140
+ "save_strategy": "steps",
141
+ "logging_steps": 0.2,
142
+ "save_steps": 0.2,
143
+ "gradient_accumulation_steps": 1,
144
+ "gradient_checkpointing": false,
145
+ "optim": "adamw_torch",
146
+ "learning_rate": 5e-05,
147
+ "weight_decay": 0.01,
148
+ "warmup_steps": 0,
149
+ "report_to": "tensorboard",
150
+ "save_safetensors": true,
151
+ "packing": false
152
+ },
153
+ "use_peft": true,
154
+ "peft_task_type": "CAUSAL_LM",
155
+ "peft_inference_mode": false,
156
+ "peft_r": 8,
157
+ "peft_lora_alpha": 32,
158
+ "peft_lora_dropout": 0.1,
159
+ "fsdp": "none",
160
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
161
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
162
+ "fsdp_fsdp_min_num_params": 100000000.0,
163
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
164
+ "fsdp_fsdp_forward_prefetch": false,
165
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
166
+ "fsdp_fsdp_offload_params": false,
167
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
168
+ "fsdp_fsdp_use_orig_params": true,
169
+ "fsdp_fsdp_activation_checkpointing": false,
170
+ "fsdp_fsdp_limit_all_gathers": true,
171
+ "fsdp_fsdp_sync_module_states": true,
172
+ "use_git": true,
173
+ "random_seed": 0
174
+ }
175
+ },
176
+ "hellaswag": {
177
+ "acc_norm": 0.8164,
178
+ "acc_norm_stderr": 0.003871770016453381,
179
+ "args": {
180
+ "exp_name": "results/v1/hellaswag/Qwen/Qwen2-7B",
181
+ "task_name": "hellaswag",
182
+ "model_name": "Qwen/Qwen2-7B",
183
+ "dataset_param": {
184
+ "max_num_train": 50000,
185
+ "completion_only": false,
186
+ "apply_chat_template": false
187
+ },
188
+ "resume_from_checkpoint": false,
189
+ "train_param": {
190
+ "max_length": 1024,
191
+ "per_device_train_batch_size": 8,
192
+ "auto_find_batch_size": true,
193
+ "num_train_epochs": 1,
194
+ "max_steps": -1,
195
+ "logging_strategy": "steps",
196
+ "save_strategy": "steps",
197
+ "logging_steps": 0.2,
198
+ "save_steps": 0.2,
199
+ "gradient_accumulation_steps": 1,
200
+ "gradient_checkpointing": false,
201
+ "optim": "adamw_torch",
202
+ "learning_rate": 5e-05,
203
+ "weight_decay": 0.01,
204
+ "warmup_steps": 0,
205
+ "report_to": "tensorboard",
206
+ "save_safetensors": true,
207
+ "packing": false
208
+ },
209
+ "use_peft": true,
210
+ "peft_task_type": "CAUSAL_LM",
211
+ "peft_inference_mode": false,
212
+ "peft_r": 8,
213
+ "peft_lora_alpha": 32,
214
+ "peft_lora_dropout": 0.1,
215
+ "fsdp": "none",
216
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
217
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
218
+ "fsdp_fsdp_min_num_params": 100000000.0,
219
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
220
+ "fsdp_fsdp_forward_prefetch": false,
221
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
222
+ "fsdp_fsdp_offload_params": false,
223
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
224
+ "fsdp_fsdp_use_orig_params": true,
225
+ "fsdp_fsdp_activation_checkpointing": false,
226
+ "fsdp_fsdp_limit_all_gathers": true,
227
+ "fsdp_fsdp_sync_module_states": true,
228
+ "use_git": true,
229
+ "random_seed": 0
230
+ }
231
+ },
232
+ "social_iqa": {
233
+ "acc": 0.5736949846468782,
234
+ "acc_stderr": 0.011190503463264756,
235
+ "args": {
236
+ "exp_name": "results/v1/social_iqa/Qwen/Qwen2-7B",
237
+ "task_name": "social_iqa",
238
+ "model_name": "Qwen/Qwen2-7B",
239
+ "dataset_param": {
240
+ "max_num_train": 50000,
241
+ "completion_only": false,
242
+ "apply_chat_template": false
243
+ },
244
+ "resume_from_checkpoint": false,
245
+ "train_param": {
246
+ "max_length": 1024,
247
+ "per_device_train_batch_size": 8,
248
+ "auto_find_batch_size": true,
249
+ "num_train_epochs": 1,
250
+ "max_steps": -1,
251
+ "logging_strategy": "steps",
252
+ "save_strategy": "steps",
253
+ "logging_steps": 0.2,
254
+ "save_steps": 0.2,
255
+ "gradient_accumulation_steps": 1,
256
+ "gradient_checkpointing": false,
257
+ "optim": "adamw_torch",
258
+ "learning_rate": 5e-05,
259
+ "weight_decay": 0.01,
260
+ "warmup_steps": 0,
261
+ "report_to": "tensorboard",
262
+ "save_safetensors": true,
263
+ "packing": false
264
+ },
265
+ "use_peft": true,
266
+ "peft_task_type": "CAUSAL_LM",
267
+ "peft_inference_mode": false,
268
+ "peft_r": 8,
269
+ "peft_lora_alpha": 32,
270
+ "peft_lora_dropout": 0.1,
271
+ "fsdp": "none",
272
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
273
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
274
+ "fsdp_fsdp_min_num_params": 100000000.0,
275
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
276
+ "fsdp_fsdp_forward_prefetch": false,
277
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
278
+ "fsdp_fsdp_offload_params": false,
279
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
280
+ "fsdp_fsdp_use_orig_params": true,
281
+ "fsdp_fsdp_activation_checkpointing": false,
282
+ "fsdp_fsdp_limit_all_gathers": true,
283
+ "fsdp_fsdp_sync_module_states": true,
284
+ "use_git": true,
285
+ "random_seed": 0
286
+ }
287
+ },
288
+ "piqa": {
289
+ "acc_norm": 0.8286180631120783,
290
+ "acc_norm_stderr": 0.008792353552929524,
291
+ "args": {
292
+ "exp_name": "results/v2/piqa/Qwen/Qwen2-7B",
293
+ "task_name": "piqa",
294
+ "model_name": "Qwen/Qwen2-7B",
295
+ "dataset_param": {
296
+ "max_num_train": 50000,
297
+ "completion_only": false,
298
+ "apply_chat_template": false
299
+ },
300
+ "resume_from_checkpoint": false,
301
+ "train_param": {
302
+ "max_length": 1024,
303
+ "per_device_train_batch_size": 8,
304
+ "auto_find_batch_size": true,
305
+ "num_train_epochs": 5,
306
+ "max_steps": -1,
307
+ "logging_strategy": "epoch",
308
+ "save_strategy": "epoch",
309
+ "logging_steps": 0.2,
310
+ "save_steps": 0.2,
311
+ "gradient_accumulation_steps": 1,
312
+ "gradient_checkpointing": false,
313
+ "optim": "adamw_torch",
314
+ "learning_rate": 5e-05,
315
+ "weight_decay": 0.01,
316
+ "warmup_steps": 0,
317
+ "report_to": "tensorboard",
318
+ "save_safetensors": true,
319
+ "packing": false
320
+ },
321
+ "use_peft": true,
322
+ "peft_task_type": "CAUSAL_LM",
323
+ "peft_inference_mode": false,
324
+ "peft_r": 8,
325
+ "peft_lora_alpha": 32,
326
+ "peft_lora_dropout": 0.1,
327
+ "fsdp": "none",
328
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
329
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
330
+ "fsdp_fsdp_min_num_params": 100000000.0,
331
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
332
+ "fsdp_fsdp_forward_prefetch": false,
333
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
334
+ "fsdp_fsdp_offload_params": false,
335
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
336
+ "fsdp_fsdp_use_orig_params": true,
337
+ "fsdp_fsdp_activation_checkpointing": false,
338
+ "fsdp_fsdp_limit_all_gathers": true,
339
+ "fsdp_fsdp_sync_module_states": true,
340
+ "use_git": true,
341
+ "random_seed": 0
342
+ }
343
+ },
344
+ "commonsense_qa": {
345
+ "acc": 0.8574938574938575,
346
+ "acc_stderr": 0.01000811727328727,
347
+ "args": {
348
+ "exp_name": "results/v2/commonsense_qa/Qwen/Qwen2-7B",
349
+ "task_name": "commonsense_qa",
350
+ "model_name": "Qwen/Qwen2-7B",
351
+ "dataset_param": {
352
+ "max_num_train": 50000,
353
+ "completion_only": false,
354
+ "apply_chat_template": false
355
+ },
356
+ "resume_from_checkpoint": false,
357
+ "train_param": {
358
+ "max_length": 1024,
359
+ "per_device_train_batch_size": 8,
360
+ "auto_find_batch_size": true,
361
+ "num_train_epochs": 5,
362
+ "max_steps": -1,
363
+ "logging_strategy": "epoch",
364
+ "save_strategy": "epoch",
365
+ "logging_steps": 0.2,
366
+ "save_steps": 0.2,
367
+ "gradient_accumulation_steps": 1,
368
+ "gradient_checkpointing": false,
369
+ "optim": "adamw_torch",
370
+ "learning_rate": 5e-05,
371
+ "weight_decay": 0.01,
372
+ "warmup_steps": 0,
373
+ "report_to": "tensorboard",
374
+ "save_safetensors": true,
375
+ "packing": false
376
+ },
377
+ "use_peft": true,
378
+ "peft_task_type": "CAUSAL_LM",
379
+ "peft_inference_mode": false,
380
+ "peft_r": 8,
381
+ "peft_lora_alpha": 32,
382
+ "peft_lora_dropout": 0.1,
383
+ "fsdp": "none",
384
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
385
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
386
+ "fsdp_fsdp_min_num_params": 100000000.0,
387
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
388
+ "fsdp_fsdp_forward_prefetch": false,
389
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
390
+ "fsdp_fsdp_offload_params": false,
391
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
392
+ "fsdp_fsdp_use_orig_params": true,
393
+ "fsdp_fsdp_activation_checkpointing": false,
394
+ "fsdp_fsdp_limit_all_gathers": true,
395
+ "fsdp_fsdp_sync_module_states": true,
396
+ "use_git": true,
397
+ "random_seed": 0
398
+ }
399
+ },
400
+ "gsm8k": {
401
+ "exact_match,flexible-extract": 0.7687642153146323,
402
+ "exact_match,flexible-extract_stderr": 0.011613587503166605,
403
+ "args": {
404
+ "exp_name": "results/v2/gsm8k/Qwen/Qwen2-7B",
405
+ "task_name": "gsm8k",
406
+ "model_name": "Qwen/Qwen2-7B",
407
+ "dataset_param": {
408
+ "max_num_train": 50000,
409
+ "completion_only": false,
410
+ "apply_chat_template": false
411
+ },
412
+ "resume_from_checkpoint": false,
413
+ "train_param": {
414
+ "max_length": 1024,
415
+ "per_device_train_batch_size": 8,
416
+ "auto_find_batch_size": true,
417
+ "num_train_epochs": 5,
418
+ "max_steps": -1,
419
+ "logging_strategy": "epoch",
420
+ "save_strategy": "epoch",
421
+ "logging_steps": 0.2,
422
+ "save_steps": 0.2,
423
+ "gradient_accumulation_steps": 1,
424
+ "gradient_checkpointing": false,
425
+ "optim": "adamw_torch",
426
+ "learning_rate": 5e-05,
427
+ "weight_decay": 0.01,
428
+ "warmup_steps": 0,
429
+ "report_to": "tensorboard",
430
+ "save_safetensors": true,
431
+ "packing": false
432
+ },
433
+ "use_peft": true,
434
+ "peft_task_type": "CAUSAL_LM",
435
+ "peft_inference_mode": false,
436
+ "peft_r": 8,
437
+ "peft_lora_alpha": 32,
438
+ "peft_lora_dropout": 0.1,
439
+ "fsdp": "none",
440
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
441
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
442
+ "fsdp_fsdp_min_num_params": 100000000.0,
443
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
444
+ "fsdp_fsdp_forward_prefetch": false,
445
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
446
+ "fsdp_fsdp_offload_params": false,
447
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
448
+ "fsdp_fsdp_use_orig_params": true,
449
+ "fsdp_fsdp_activation_checkpointing": false,
450
+ "fsdp_fsdp_limit_all_gathers": true,
451
+ "fsdp_fsdp_sync_module_states": true,
452
+ "use_git": true,
453
+ "random_seed": 0
454
+ }
455
+ },
456
+ "arc_challenge": {
457
+ "acc_norm": 0.5964163822525598,
458
+ "acc_norm_stderr": 0.01433715891426844,
459
+ "args": {
460
+ "exp_name": "results/v2/arc_challenge/Qwen/Qwen2-7B",
461
+ "task_name": "arc_challenge",
462
+ "model_name": "Qwen/Qwen2-7B",
463
+ "dataset_param": {
464
+ "max_num_train": 50000,
465
+ "completion_only": false,
466
+ "apply_chat_template": false
467
+ },
468
+ "resume_from_checkpoint": false,
469
+ "train_param": {
470
+ "max_length": 1024,
471
+ "per_device_train_batch_size": 8,
472
+ "auto_find_batch_size": true,
473
+ "num_train_epochs": 5,
474
+ "max_steps": -1,
475
+ "logging_strategy": "epoch",
476
+ "save_strategy": "epoch",
477
+ "logging_steps": 0.2,
478
+ "save_steps": 0.2,
479
+ "gradient_accumulation_steps": 1,
480
+ "gradient_checkpointing": false,
481
+ "optim": "adamw_torch",
482
+ "learning_rate": 5e-05,
483
+ "weight_decay": 0.01,
484
+ "warmup_steps": 0,
485
+ "report_to": "tensorboard",
486
+ "save_safetensors": true,
487
+ "packing": false
488
+ },
489
+ "use_peft": true,
490
+ "peft_task_type": "CAUSAL_LM",
491
+ "peft_inference_mode": false,
492
+ "peft_r": 8,
493
+ "peft_lora_alpha": 32,
494
+ "peft_lora_dropout": 0.1,
495
+ "fsdp": "none",
496
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
497
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
498
+ "fsdp_fsdp_min_num_params": 100000000.0,
499
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
500
+ "fsdp_fsdp_forward_prefetch": false,
501
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
502
+ "fsdp_fsdp_offload_params": false,
503
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
504
+ "fsdp_fsdp_use_orig_params": true,
505
+ "fsdp_fsdp_activation_checkpointing": false,
506
+ "fsdp_fsdp_limit_all_gathers": true,
507
+ "fsdp_fsdp_sync_module_states": true,
508
+ "use_git": true,
509
+ "random_seed": 0
510
+ }
511
+ }
512
+ }
513
+ }
Qwen/Qwen2.5-0.5B-Instruct/results_2026-01-05-04-27-51.json ADDED
@@ -0,0 +1,513 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "model_name": "Qwen/Qwen2.5-0.5B-Instruct",
4
+ "model_type": "instruction-tuned",
5
+ "params": 0.5
6
+ },
7
+ "results": {
8
+ "medmcqa": {
9
+ "acc_norm": 0.40186469041357875,
10
+ "acc_norm_stderr": 0.0075813689208232326,
11
+ "args": {
12
+ "exp_name": "results/v1/medmcqa/Qwen/Qwen2.5-0.5B-Instruct",
13
+ "task_name": "medmcqa",
14
+ "model_name": "Qwen/Qwen2.5-0.5B-Instruct",
15
+ "dataset_param": {
16
+ "max_num_train": 50000,
17
+ "completion_only": false,
18
+ "apply_chat_template": true
19
+ },
20
+ "resume_from_checkpoint": false,
21
+ "train_param": {
22
+ "max_length": 1024,
23
+ "per_device_train_batch_size": 8,
24
+ "auto_find_batch_size": true,
25
+ "num_train_epochs": 1,
26
+ "max_steps": -1,
27
+ "logging_strategy": "steps",
28
+ "save_strategy": "steps",
29
+ "logging_steps": 0.2,
30
+ "save_steps": 0.2,
31
+ "gradient_accumulation_steps": 1,
32
+ "gradient_checkpointing": false,
33
+ "optim": "adamw_torch",
34
+ "learning_rate": 5e-05,
35
+ "weight_decay": 0.01,
36
+ "warmup_steps": 0,
37
+ "report_to": "tensorboard",
38
+ "save_safetensors": true,
39
+ "packing": false
40
+ },
41
+ "use_peft": true,
42
+ "peft_task_type": "CAUSAL_LM",
43
+ "peft_inference_mode": false,
44
+ "peft_r": 8,
45
+ "peft_lora_alpha": 32,
46
+ "peft_lora_dropout": 0.1,
47
+ "fsdp": "none",
48
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
49
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
50
+ "fsdp_fsdp_min_num_params": 100000000.0,
51
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
52
+ "fsdp_fsdp_forward_prefetch": false,
53
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
54
+ "fsdp_fsdp_offload_params": false,
55
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
56
+ "fsdp_fsdp_use_orig_params": true,
57
+ "fsdp_fsdp_activation_checkpointing": false,
58
+ "fsdp_fsdp_limit_all_gathers": true,
59
+ "fsdp_fsdp_sync_module_states": true,
60
+ "use_git": true,
61
+ "random_seed": 0
62
+ }
63
+ },
64
+ "nq_open": {
65
+ "exact_match,remove_whitespace": 0.08005540166204986,
66
+ "exact_match,remove_whitespace_stderr": 0.004517339345170745,
67
+ "args": {
68
+ "exp_name": "results/v1/nq_open/Qwen/Qwen2.5-0.5B-Instruct",
69
+ "task_name": "nq_open",
70
+ "model_name": "Qwen/Qwen2.5-0.5B-Instruct",
71
+ "dataset_param": {
72
+ "max_num_train": 50000,
73
+ "completion_only": false,
74
+ "apply_chat_template": true
75
+ },
76
+ "resume_from_checkpoint": false,
77
+ "train_param": {
78
+ "max_length": 1024,
79
+ "per_device_train_batch_size": 8,
80
+ "auto_find_batch_size": true,
81
+ "num_train_epochs": 1,
82
+ "max_steps": -1,
83
+ "logging_strategy": "steps",
84
+ "save_strategy": "steps",
85
+ "logging_steps": 0.2,
86
+ "save_steps": 0.2,
87
+ "gradient_accumulation_steps": 1,
88
+ "gradient_checkpointing": false,
89
+ "optim": "adamw_torch",
90
+ "learning_rate": 5e-05,
91
+ "weight_decay": 0.01,
92
+ "warmup_steps": 0,
93
+ "report_to": "tensorboard",
94
+ "save_safetensors": true,
95
+ "packing": false
96
+ },
97
+ "use_peft": true,
98
+ "peft_task_type": "CAUSAL_LM",
99
+ "peft_inference_mode": false,
100
+ "peft_r": 8,
101
+ "peft_lora_alpha": 32,
102
+ "peft_lora_dropout": 0.1,
103
+ "fsdp": "none",
104
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
105
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
106
+ "fsdp_fsdp_min_num_params": 100000000.0,
107
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
108
+ "fsdp_fsdp_forward_prefetch": false,
109
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
110
+ "fsdp_fsdp_offload_params": false,
111
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
112
+ "fsdp_fsdp_use_orig_params": true,
113
+ "fsdp_fsdp_activation_checkpointing": false,
114
+ "fsdp_fsdp_limit_all_gathers": true,
115
+ "fsdp_fsdp_sync_module_states": true,
116
+ "use_git": true,
117
+ "random_seed": 0
118
+ }
119
+ },
120
+ "winogrande": {
121
+ "acc": 0.5919494869771112,
122
+ "acc_stderr": 0.013812822643745015,
123
+ "args": {
124
+ "exp_name": "results/v1/winogrande/Qwen/Qwen2.5-0.5B-Instruct",
125
+ "task_name": "winogrande",
126
+ "model_name": "Qwen/Qwen2.5-0.5B-Instruct",
127
+ "dataset_param": {
128
+ "max_num_train": 50000,
129
+ "completion_only": false,
130
+ "apply_chat_template": true
131
+ },
132
+ "resume_from_checkpoint": false,
133
+ "train_param": {
134
+ "max_length": 1024,
135
+ "per_device_train_batch_size": 8,
136
+ "auto_find_batch_size": true,
137
+ "num_train_epochs": 1,
138
+ "max_steps": -1,
139
+ "logging_strategy": "steps",
140
+ "save_strategy": "steps",
141
+ "logging_steps": 0.2,
142
+ "save_steps": 0.2,
143
+ "gradient_accumulation_steps": 1,
144
+ "gradient_checkpointing": false,
145
+ "optim": "adamw_torch",
146
+ "learning_rate": 5e-05,
147
+ "weight_decay": 0.01,
148
+ "warmup_steps": 0,
149
+ "report_to": "tensorboard",
150
+ "save_safetensors": true,
151
+ "packing": false
152
+ },
153
+ "use_peft": true,
154
+ "peft_task_type": "CAUSAL_LM",
155
+ "peft_inference_mode": false,
156
+ "peft_r": 8,
157
+ "peft_lora_alpha": 32,
158
+ "peft_lora_dropout": 0.1,
159
+ "fsdp": "none",
160
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
161
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
162
+ "fsdp_fsdp_min_num_params": 100000000.0,
163
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
164
+ "fsdp_fsdp_forward_prefetch": false,
165
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
166
+ "fsdp_fsdp_offload_params": false,
167
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
168
+ "fsdp_fsdp_use_orig_params": true,
169
+ "fsdp_fsdp_activation_checkpointing": false,
170
+ "fsdp_fsdp_limit_all_gathers": true,
171
+ "fsdp_fsdp_sync_module_states": true,
172
+ "use_git": true,
173
+ "random_seed": 0
174
+ }
175
+ },
176
+ "hellaswag": {
177
+ "acc_norm": 0.5292,
178
+ "acc_norm_stderr": 0.004991715909661663,
179
+ "args": {
180
+ "exp_name": "results/v1/hellaswag/Qwen/Qwen2.5-0.5B-Instruct",
181
+ "task_name": "hellaswag",
182
+ "model_name": "Qwen/Qwen2.5-0.5B-Instruct",
183
+ "dataset_param": {
184
+ "max_num_train": 50000,
185
+ "completion_only": false,
186
+ "apply_chat_template": true
187
+ },
188
+ "resume_from_checkpoint": false,
189
+ "train_param": {
190
+ "max_length": 1024,
191
+ "per_device_train_batch_size": 8,
192
+ "auto_find_batch_size": true,
193
+ "num_train_epochs": 1,
194
+ "max_steps": -1,
195
+ "logging_strategy": "steps",
196
+ "save_strategy": "steps",
197
+ "logging_steps": 0.2,
198
+ "save_steps": 0.2,
199
+ "gradient_accumulation_steps": 1,
200
+ "gradient_checkpointing": false,
201
+ "optim": "adamw_torch",
202
+ "learning_rate": 5e-05,
203
+ "weight_decay": 0.01,
204
+ "warmup_steps": 0,
205
+ "report_to": "tensorboard",
206
+ "save_safetensors": true,
207
+ "packing": false
208
+ },
209
+ "use_peft": true,
210
+ "peft_task_type": "CAUSAL_LM",
211
+ "peft_inference_mode": false,
212
+ "peft_r": 8,
213
+ "peft_lora_alpha": 32,
214
+ "peft_lora_dropout": 0.1,
215
+ "fsdp": "none",
216
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
217
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
218
+ "fsdp_fsdp_min_num_params": 100000000.0,
219
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
220
+ "fsdp_fsdp_forward_prefetch": false,
221
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
222
+ "fsdp_fsdp_offload_params": false,
223
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
224
+ "fsdp_fsdp_use_orig_params": true,
225
+ "fsdp_fsdp_activation_checkpointing": false,
226
+ "fsdp_fsdp_limit_all_gathers": true,
227
+ "fsdp_fsdp_sync_module_states": true,
228
+ "use_git": true,
229
+ "random_seed": 0
230
+ }
231
+ },
232
+ "social_iqa": {
233
+ "acc": 0.4631525076765609,
234
+ "acc_stderr": 0.011283305647440943,
235
+ "args": {
236
+ "exp_name": "results/v1/social_iqa/Qwen/Qwen2.5-0.5B-Instruct",
237
+ "task_name": "social_iqa",
238
+ "model_name": "Qwen/Qwen2.5-0.5B-Instruct",
239
+ "dataset_param": {
240
+ "max_num_train": 50000,
241
+ "completion_only": false,
242
+ "apply_chat_template": true
243
+ },
244
+ "resume_from_checkpoint": false,
245
+ "train_param": {
246
+ "max_length": 1024,
247
+ "per_device_train_batch_size": 8,
248
+ "auto_find_batch_size": true,
249
+ "num_train_epochs": 1,
250
+ "max_steps": -1,
251
+ "logging_strategy": "steps",
252
+ "save_strategy": "steps",
253
+ "logging_steps": 0.2,
254
+ "save_steps": 0.2,
255
+ "gradient_accumulation_steps": 1,
256
+ "gradient_checkpointing": false,
257
+ "optim": "adamw_torch",
258
+ "learning_rate": 5e-05,
259
+ "weight_decay": 0.01,
260
+ "warmup_steps": 0,
261
+ "report_to": "tensorboard",
262
+ "save_safetensors": true,
263
+ "packing": false
264
+ },
265
+ "use_peft": true,
266
+ "peft_task_type": "CAUSAL_LM",
267
+ "peft_inference_mode": false,
268
+ "peft_r": 8,
269
+ "peft_lora_alpha": 32,
270
+ "peft_lora_dropout": 0.1,
271
+ "fsdp": "none",
272
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
273
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
274
+ "fsdp_fsdp_min_num_params": 100000000.0,
275
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
276
+ "fsdp_fsdp_forward_prefetch": false,
277
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
278
+ "fsdp_fsdp_offload_params": false,
279
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
280
+ "fsdp_fsdp_use_orig_params": true,
281
+ "fsdp_fsdp_activation_checkpointing": false,
282
+ "fsdp_fsdp_limit_all_gathers": true,
283
+ "fsdp_fsdp_sync_module_states": true,
284
+ "use_git": true,
285
+ "random_seed": 0
286
+ }
287
+ },
288
+ "piqa": {
289
+ "acc_norm": 0.7072905331882481,
290
+ "acc_norm_stderr": 0.010616044462393094,
291
+ "args": {
292
+ "exp_name": "results/v2/piqa/Qwen/Qwen2.5-0.5B-Instruct",
293
+ "task_name": "piqa",
294
+ "model_name": "Qwen/Qwen2.5-0.5B-Instruct",
295
+ "dataset_param": {
296
+ "max_num_train": 50000,
297
+ "completion_only": false,
298
+ "apply_chat_template": true
299
+ },
300
+ "resume_from_checkpoint": false,
301
+ "train_param": {
302
+ "max_length": 1024,
303
+ "per_device_train_batch_size": 8,
304
+ "auto_find_batch_size": true,
305
+ "num_train_epochs": 5,
306
+ "max_steps": -1,
307
+ "logging_strategy": "epoch",
308
+ "save_strategy": "epoch",
309
+ "logging_steps": 0.2,
310
+ "save_steps": 0.2,
311
+ "gradient_accumulation_steps": 1,
312
+ "gradient_checkpointing": false,
313
+ "optim": "adamw_torch",
314
+ "learning_rate": 5e-05,
315
+ "weight_decay": 0.01,
316
+ "warmup_steps": 0,
317
+ "report_to": "tensorboard",
318
+ "save_safetensors": true,
319
+ "packing": false
320
+ },
321
+ "use_peft": true,
322
+ "peft_task_type": "CAUSAL_LM",
323
+ "peft_inference_mode": false,
324
+ "peft_r": 8,
325
+ "peft_lora_alpha": 32,
326
+ "peft_lora_dropout": 0.1,
327
+ "fsdp": "none",
328
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
329
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
330
+ "fsdp_fsdp_min_num_params": 100000000.0,
331
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
332
+ "fsdp_fsdp_forward_prefetch": false,
333
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
334
+ "fsdp_fsdp_offload_params": false,
335
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
336
+ "fsdp_fsdp_use_orig_params": true,
337
+ "fsdp_fsdp_activation_checkpointing": false,
338
+ "fsdp_fsdp_limit_all_gathers": true,
339
+ "fsdp_fsdp_sync_module_states": true,
340
+ "use_git": true,
341
+ "random_seed": 0
342
+ }
343
+ },
344
+ "commonsense_qa": {
345
+ "acc": 0.6347256347256347,
346
+ "acc_stderr": 0.013785505673659645,
347
+ "args": {
348
+ "exp_name": "results/v2/commonsense_qa/Qwen/Qwen2.5-0.5B-Instruct",
349
+ "task_name": "commonsense_qa",
350
+ "model_name": "Qwen/Qwen2.5-0.5B-Instruct",
351
+ "dataset_param": {
352
+ "max_num_train": 50000,
353
+ "completion_only": false,
354
+ "apply_chat_template": true
355
+ },
356
+ "resume_from_checkpoint": false,
357
+ "train_param": {
358
+ "max_length": 1024,
359
+ "per_device_train_batch_size": 8,
360
+ "auto_find_batch_size": true,
361
+ "num_train_epochs": 5,
362
+ "max_steps": -1,
363
+ "logging_strategy": "epoch",
364
+ "save_strategy": "epoch",
365
+ "logging_steps": 0.2,
366
+ "save_steps": 0.2,
367
+ "gradient_accumulation_steps": 1,
368
+ "gradient_checkpointing": false,
369
+ "optim": "adamw_torch",
370
+ "learning_rate": 5e-05,
371
+ "weight_decay": 0.01,
372
+ "warmup_steps": 0,
373
+ "report_to": "tensorboard",
374
+ "save_safetensors": true,
375
+ "packing": false
376
+ },
377
+ "use_peft": true,
378
+ "peft_task_type": "CAUSAL_LM",
379
+ "peft_inference_mode": false,
380
+ "peft_r": 8,
381
+ "peft_lora_alpha": 32,
382
+ "peft_lora_dropout": 0.1,
383
+ "fsdp": "none",
384
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
385
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
386
+ "fsdp_fsdp_min_num_params": 100000000.0,
387
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
388
+ "fsdp_fsdp_forward_prefetch": false,
389
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
390
+ "fsdp_fsdp_offload_params": false,
391
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
392
+ "fsdp_fsdp_use_orig_params": true,
393
+ "fsdp_fsdp_activation_checkpointing": false,
394
+ "fsdp_fsdp_limit_all_gathers": true,
395
+ "fsdp_fsdp_sync_module_states": true,
396
+ "use_git": true,
397
+ "random_seed": 0
398
+ }
399
+ },
400
+ "gsm8k": {
401
+ "exact_match,flexible-extract": 0.30856709628506446,
402
+ "exact_match,flexible-extract_stderr": 0.012723076049815908,
403
+ "args": {
404
+ "exp_name": "results/v2/gsm8k/Qwen/Qwen2.5-0.5B-Instruct",
405
+ "task_name": "gsm8k",
406
+ "model_name": "Qwen/Qwen2.5-0.5B-Instruct",
407
+ "dataset_param": {
408
+ "max_num_train": 50000,
409
+ "completion_only": false,
410
+ "apply_chat_template": true
411
+ },
412
+ "resume_from_checkpoint": false,
413
+ "train_param": {
414
+ "max_length": 1024,
415
+ "per_device_train_batch_size": 8,
416
+ "auto_find_batch_size": true,
417
+ "num_train_epochs": 5,
418
+ "max_steps": -1,
419
+ "logging_strategy": "epoch",
420
+ "save_strategy": "epoch",
421
+ "logging_steps": 0.2,
422
+ "save_steps": 0.2,
423
+ "gradient_accumulation_steps": 1,
424
+ "gradient_checkpointing": false,
425
+ "optim": "adamw_torch",
426
+ "learning_rate": 5e-05,
427
+ "weight_decay": 0.01,
428
+ "warmup_steps": 0,
429
+ "report_to": "tensorboard",
430
+ "save_safetensors": true,
431
+ "packing": false
432
+ },
433
+ "use_peft": true,
434
+ "peft_task_type": "CAUSAL_LM",
435
+ "peft_inference_mode": false,
436
+ "peft_r": 8,
437
+ "peft_lora_alpha": 32,
438
+ "peft_lora_dropout": 0.1,
439
+ "fsdp": "none",
440
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
441
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
442
+ "fsdp_fsdp_min_num_params": 100000000.0,
443
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
444
+ "fsdp_fsdp_forward_prefetch": false,
445
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
446
+ "fsdp_fsdp_offload_params": false,
447
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
448
+ "fsdp_fsdp_use_orig_params": true,
449
+ "fsdp_fsdp_activation_checkpointing": false,
450
+ "fsdp_fsdp_limit_all_gathers": true,
451
+ "fsdp_fsdp_sync_module_states": true,
452
+ "use_git": true,
453
+ "random_seed": 0
454
+ }
455
+ },
456
+ "arc_challenge": {
457
+ "acc_norm": 0.3532423208191126,
458
+ "acc_norm_stderr": 0.013967822714840053,
459
+ "args": {
460
+ "exp_name": "results/v2/arc_challenge/Qwen/Qwen2.5-0.5B-Instruct",
461
+ "task_name": "arc_challenge",
462
+ "model_name": "Qwen/Qwen2.5-0.5B-Instruct",
463
+ "dataset_param": {
464
+ "max_num_train": 50000,
465
+ "completion_only": false,
466
+ "apply_chat_template": true
467
+ },
468
+ "resume_from_checkpoint": false,
469
+ "train_param": {
470
+ "max_length": 1024,
471
+ "per_device_train_batch_size": 8,
472
+ "auto_find_batch_size": true,
473
+ "num_train_epochs": 5,
474
+ "max_steps": -1,
475
+ "logging_strategy": "epoch",
476
+ "save_strategy": "epoch",
477
+ "logging_steps": 0.2,
478
+ "save_steps": 0.2,
479
+ "gradient_accumulation_steps": 1,
480
+ "gradient_checkpointing": false,
481
+ "optim": "adamw_torch",
482
+ "learning_rate": 5e-05,
483
+ "weight_decay": 0.01,
484
+ "warmup_steps": 0,
485
+ "report_to": "tensorboard",
486
+ "save_safetensors": true,
487
+ "packing": false
488
+ },
489
+ "use_peft": true,
490
+ "peft_task_type": "CAUSAL_LM",
491
+ "peft_inference_mode": false,
492
+ "peft_r": 8,
493
+ "peft_lora_alpha": 32,
494
+ "peft_lora_dropout": 0.1,
495
+ "fsdp": "none",
496
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
497
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
498
+ "fsdp_fsdp_min_num_params": 100000000.0,
499
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
500
+ "fsdp_fsdp_forward_prefetch": false,
501
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
502
+ "fsdp_fsdp_offload_params": false,
503
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
504
+ "fsdp_fsdp_use_orig_params": true,
505
+ "fsdp_fsdp_activation_checkpointing": false,
506
+ "fsdp_fsdp_limit_all_gathers": true,
507
+ "fsdp_fsdp_sync_module_states": true,
508
+ "use_git": true,
509
+ "random_seed": 0
510
+ }
511
+ }
512
+ }
513
+ }
Qwen/Qwen2.5-0.5B/results_2026-01-05-04-27-50.json ADDED
@@ -0,0 +1,513 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "model_name": "Qwen/Qwen2.5-0.5B",
4
+ "model_type": "pretrained",
5
+ "params": 0.5
6
+ },
7
+ "results": {
8
+ "medmcqa": {
9
+ "acc_norm": 0.4066459478842936,
10
+ "acc_norm_stderr": 0.0075957937993202645,
11
+ "args": {
12
+ "exp_name": "results/v1/medmcqa/Qwen/Qwen2.5-0.5B",
13
+ "task_name": "medmcqa",
14
+ "model_name": "Qwen/Qwen2.5-0.5B",
15
+ "dataset_param": {
16
+ "max_num_train": 50000,
17
+ "completion_only": false,
18
+ "apply_chat_template": false
19
+ },
20
+ "resume_from_checkpoint": false,
21
+ "train_param": {
22
+ "max_length": 1024,
23
+ "per_device_train_batch_size": 8,
24
+ "auto_find_batch_size": true,
25
+ "num_train_epochs": 1,
26
+ "max_steps": -1,
27
+ "logging_strategy": "steps",
28
+ "save_strategy": "steps",
29
+ "logging_steps": 0.2,
30
+ "save_steps": 0.2,
31
+ "gradient_accumulation_steps": 1,
32
+ "gradient_checkpointing": false,
33
+ "optim": "adamw_torch",
34
+ "learning_rate": 5e-05,
35
+ "weight_decay": 0.01,
36
+ "warmup_steps": 0,
37
+ "report_to": "tensorboard",
38
+ "save_safetensors": true,
39
+ "packing": false
40
+ },
41
+ "use_peft": true,
42
+ "peft_task_type": "CAUSAL_LM",
43
+ "peft_inference_mode": false,
44
+ "peft_r": 8,
45
+ "peft_lora_alpha": 32,
46
+ "peft_lora_dropout": 0.1,
47
+ "fsdp": "none",
48
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
49
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
50
+ "fsdp_fsdp_min_num_params": 100000000.0,
51
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
52
+ "fsdp_fsdp_forward_prefetch": false,
53
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
54
+ "fsdp_fsdp_offload_params": false,
55
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
56
+ "fsdp_fsdp_use_orig_params": true,
57
+ "fsdp_fsdp_activation_checkpointing": false,
58
+ "fsdp_fsdp_limit_all_gathers": true,
59
+ "fsdp_fsdp_sync_module_states": true,
60
+ "use_git": true,
61
+ "random_seed": 0
62
+ }
63
+ },
64
+ "nq_open": {
65
+ "exact_match,remove_whitespace": 0.08642659279778393,
66
+ "exact_match,remove_whitespace_stderr": 0.004677372734591098,
67
+ "args": {
68
+ "exp_name": "results/v1/nq_open/Qwen/Qwen2.5-0.5B",
69
+ "task_name": "nq_open",
70
+ "model_name": "Qwen/Qwen2.5-0.5B",
71
+ "dataset_param": {
72
+ "max_num_train": 50000,
73
+ "completion_only": false,
74
+ "apply_chat_template": false
75
+ },
76
+ "resume_from_checkpoint": false,
77
+ "train_param": {
78
+ "max_length": 1024,
79
+ "per_device_train_batch_size": 8,
80
+ "auto_find_batch_size": true,
81
+ "num_train_epochs": 1,
82
+ "max_steps": -1,
83
+ "logging_strategy": "steps",
84
+ "save_strategy": "steps",
85
+ "logging_steps": 0.2,
86
+ "save_steps": 0.2,
87
+ "gradient_accumulation_steps": 1,
88
+ "gradient_checkpointing": false,
89
+ "optim": "adamw_torch",
90
+ "learning_rate": 5e-05,
91
+ "weight_decay": 0.01,
92
+ "warmup_steps": 0,
93
+ "report_to": "tensorboard",
94
+ "save_safetensors": true,
95
+ "packing": false
96
+ },
97
+ "use_peft": true,
98
+ "peft_task_type": "CAUSAL_LM",
99
+ "peft_inference_mode": false,
100
+ "peft_r": 8,
101
+ "peft_lora_alpha": 32,
102
+ "peft_lora_dropout": 0.1,
103
+ "fsdp": "none",
104
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
105
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
106
+ "fsdp_fsdp_min_num_params": 100000000.0,
107
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
108
+ "fsdp_fsdp_forward_prefetch": false,
109
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
110
+ "fsdp_fsdp_offload_params": false,
111
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
112
+ "fsdp_fsdp_use_orig_params": true,
113
+ "fsdp_fsdp_activation_checkpointing": false,
114
+ "fsdp_fsdp_limit_all_gathers": true,
115
+ "fsdp_fsdp_sync_module_states": true,
116
+ "use_git": true,
117
+ "random_seed": 0
118
+ }
119
+ },
120
+ "winogrande": {
121
+ "acc": 0.579321231254933,
122
+ "acc_stderr": 0.013874526372008327,
123
+ "args": {
124
+ "exp_name": "results/v1/winogrande/Qwen/Qwen2.5-0.5B",
125
+ "task_name": "winogrande",
126
+ "model_name": "Qwen/Qwen2.5-0.5B",
127
+ "dataset_param": {
128
+ "max_num_train": 50000,
129
+ "completion_only": false,
130
+ "apply_chat_template": false
131
+ },
132
+ "resume_from_checkpoint": false,
133
+ "train_param": {
134
+ "max_length": 1024,
135
+ "per_device_train_batch_size": 8,
136
+ "auto_find_batch_size": true,
137
+ "num_train_epochs": 1,
138
+ "max_steps": -1,
139
+ "logging_strategy": "steps",
140
+ "save_strategy": "steps",
141
+ "logging_steps": 0.2,
142
+ "save_steps": 0.2,
143
+ "gradient_accumulation_steps": 1,
144
+ "gradient_checkpointing": false,
145
+ "optim": "adamw_torch",
146
+ "learning_rate": 5e-05,
147
+ "weight_decay": 0.01,
148
+ "warmup_steps": 0,
149
+ "report_to": "tensorboard",
150
+ "save_safetensors": true,
151
+ "packing": false
152
+ },
153
+ "use_peft": true,
154
+ "peft_task_type": "CAUSAL_LM",
155
+ "peft_inference_mode": false,
156
+ "peft_r": 8,
157
+ "peft_lora_alpha": 32,
158
+ "peft_lora_dropout": 0.1,
159
+ "fsdp": "none",
160
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
161
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
162
+ "fsdp_fsdp_min_num_params": 100000000.0,
163
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
164
+ "fsdp_fsdp_forward_prefetch": false,
165
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
166
+ "fsdp_fsdp_offload_params": false,
167
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
168
+ "fsdp_fsdp_use_orig_params": true,
169
+ "fsdp_fsdp_activation_checkpointing": false,
170
+ "fsdp_fsdp_limit_all_gathers": true,
171
+ "fsdp_fsdp_sync_module_states": true,
172
+ "use_git": true,
173
+ "random_seed": 0
174
+ }
175
+ },
176
+ "hellaswag": {
177
+ "acc_norm": 0.5352,
178
+ "acc_norm_stderr": 0.004987843608051959,
179
+ "args": {
180
+ "exp_name": "results/v1/hellaswag/Qwen/Qwen2.5-0.5B",
181
+ "task_name": "hellaswag",
182
+ "model_name": "Qwen/Qwen2.5-0.5B",
183
+ "dataset_param": {
184
+ "max_num_train": 50000,
185
+ "completion_only": false,
186
+ "apply_chat_template": false
187
+ },
188
+ "resume_from_checkpoint": false,
189
+ "train_param": {
190
+ "max_length": 1024,
191
+ "per_device_train_batch_size": 8,
192
+ "auto_find_batch_size": true,
193
+ "num_train_epochs": 1,
194
+ "max_steps": -1,
195
+ "logging_strategy": "steps",
196
+ "save_strategy": "steps",
197
+ "logging_steps": 0.2,
198
+ "save_steps": 0.2,
199
+ "gradient_accumulation_steps": 1,
200
+ "gradient_checkpointing": false,
201
+ "optim": "adamw_torch",
202
+ "learning_rate": 5e-05,
203
+ "weight_decay": 0.01,
204
+ "warmup_steps": 0,
205
+ "report_to": "tensorboard",
206
+ "save_safetensors": true,
207
+ "packing": false
208
+ },
209
+ "use_peft": true,
210
+ "peft_task_type": "CAUSAL_LM",
211
+ "peft_inference_mode": false,
212
+ "peft_r": 8,
213
+ "peft_lora_alpha": 32,
214
+ "peft_lora_dropout": 0.1,
215
+ "fsdp": "none",
216
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
217
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
218
+ "fsdp_fsdp_min_num_params": 100000000.0,
219
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
220
+ "fsdp_fsdp_forward_prefetch": false,
221
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
222
+ "fsdp_fsdp_offload_params": false,
223
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
224
+ "fsdp_fsdp_use_orig_params": true,
225
+ "fsdp_fsdp_activation_checkpointing": false,
226
+ "fsdp_fsdp_limit_all_gathers": true,
227
+ "fsdp_fsdp_sync_module_states": true,
228
+ "use_git": true,
229
+ "random_seed": 0
230
+ }
231
+ },
232
+ "social_iqa": {
233
+ "acc": 0.4779938587512794,
234
+ "acc_stderr": 0.011303107107577816,
235
+ "args": {
236
+ "exp_name": "results/v1/social_iqa/Qwen/Qwen2.5-0.5B",
237
+ "task_name": "social_iqa",
238
+ "model_name": "Qwen/Qwen2.5-0.5B",
239
+ "dataset_param": {
240
+ "max_num_train": 50000,
241
+ "completion_only": false,
242
+ "apply_chat_template": false
243
+ },
244
+ "resume_from_checkpoint": false,
245
+ "train_param": {
246
+ "max_length": 1024,
247
+ "per_device_train_batch_size": 8,
248
+ "auto_find_batch_size": true,
249
+ "num_train_epochs": 1,
250
+ "max_steps": -1,
251
+ "logging_strategy": "steps",
252
+ "save_strategy": "steps",
253
+ "logging_steps": 0.2,
254
+ "save_steps": 0.2,
255
+ "gradient_accumulation_steps": 1,
256
+ "gradient_checkpointing": false,
257
+ "optim": "adamw_torch",
258
+ "learning_rate": 5e-05,
259
+ "weight_decay": 0.01,
260
+ "warmup_steps": 0,
261
+ "report_to": "tensorboard",
262
+ "save_safetensors": true,
263
+ "packing": false
264
+ },
265
+ "use_peft": true,
266
+ "peft_task_type": "CAUSAL_LM",
267
+ "peft_inference_mode": false,
268
+ "peft_r": 8,
269
+ "peft_lora_alpha": 32,
270
+ "peft_lora_dropout": 0.1,
271
+ "fsdp": "none",
272
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
273
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
274
+ "fsdp_fsdp_min_num_params": 100000000.0,
275
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
276
+ "fsdp_fsdp_forward_prefetch": false,
277
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
278
+ "fsdp_fsdp_offload_params": false,
279
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
280
+ "fsdp_fsdp_use_orig_params": true,
281
+ "fsdp_fsdp_activation_checkpointing": false,
282
+ "fsdp_fsdp_limit_all_gathers": true,
283
+ "fsdp_fsdp_sync_module_states": true,
284
+ "use_git": true,
285
+ "random_seed": 0
286
+ }
287
+ },
288
+ "piqa": {
289
+ "acc_norm": 0.7078346028291621,
290
+ "acc_norm_stderr": 0.01061025217451366,
291
+ "args": {
292
+ "exp_name": "results/v2/piqa/Qwen/Qwen2.5-0.5B",
293
+ "task_name": "piqa",
294
+ "model_name": "Qwen/Qwen2.5-0.5B",
295
+ "dataset_param": {
296
+ "max_num_train": 50000,
297
+ "completion_only": false,
298
+ "apply_chat_template": false
299
+ },
300
+ "resume_from_checkpoint": false,
301
+ "train_param": {
302
+ "max_length": 1024,
303
+ "per_device_train_batch_size": 8,
304
+ "auto_find_batch_size": true,
305
+ "num_train_epochs": 5,
306
+ "max_steps": -1,
307
+ "logging_strategy": "epoch",
308
+ "save_strategy": "epoch",
309
+ "logging_steps": 0.2,
310
+ "save_steps": 0.2,
311
+ "gradient_accumulation_steps": 1,
312
+ "gradient_checkpointing": false,
313
+ "optim": "adamw_torch",
314
+ "learning_rate": 5e-05,
315
+ "weight_decay": 0.01,
316
+ "warmup_steps": 0,
317
+ "report_to": "tensorboard",
318
+ "save_safetensors": true,
319
+ "packing": false
320
+ },
321
+ "use_peft": true,
322
+ "peft_task_type": "CAUSAL_LM",
323
+ "peft_inference_mode": false,
324
+ "peft_r": 8,
325
+ "peft_lora_alpha": 32,
326
+ "peft_lora_dropout": 0.1,
327
+ "fsdp": "none",
328
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
329
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
330
+ "fsdp_fsdp_min_num_params": 100000000.0,
331
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
332
+ "fsdp_fsdp_forward_prefetch": false,
333
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
334
+ "fsdp_fsdp_offload_params": false,
335
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
336
+ "fsdp_fsdp_use_orig_params": true,
337
+ "fsdp_fsdp_activation_checkpointing": false,
338
+ "fsdp_fsdp_limit_all_gathers": true,
339
+ "fsdp_fsdp_sync_module_states": true,
340
+ "use_git": true,
341
+ "random_seed": 0
342
+ }
343
+ },
344
+ "commonsense_qa": {
345
+ "acc": 0.6306306306306306,
346
+ "acc_stderr": 0.01381777301518641,
347
+ "args": {
348
+ "exp_name": "results/v2/commonsense_qa/Qwen/Qwen2.5-0.5B",
349
+ "task_name": "commonsense_qa",
350
+ "model_name": "Qwen/Qwen2.5-0.5B",
351
+ "dataset_param": {
352
+ "max_num_train": 50000,
353
+ "completion_only": false,
354
+ "apply_chat_template": false
355
+ },
356
+ "resume_from_checkpoint": false,
357
+ "train_param": {
358
+ "max_length": 1024,
359
+ "per_device_train_batch_size": 8,
360
+ "auto_find_batch_size": true,
361
+ "num_train_epochs": 5,
362
+ "max_steps": -1,
363
+ "logging_strategy": "epoch",
364
+ "save_strategy": "epoch",
365
+ "logging_steps": 0.2,
366
+ "save_steps": 0.2,
367
+ "gradient_accumulation_steps": 1,
368
+ "gradient_checkpointing": false,
369
+ "optim": "adamw_torch",
370
+ "learning_rate": 5e-05,
371
+ "weight_decay": 0.01,
372
+ "warmup_steps": 0,
373
+ "report_to": "tensorboard",
374
+ "save_safetensors": true,
375
+ "packing": false
376
+ },
377
+ "use_peft": true,
378
+ "peft_task_type": "CAUSAL_LM",
379
+ "peft_inference_mode": false,
380
+ "peft_r": 8,
381
+ "peft_lora_alpha": 32,
382
+ "peft_lora_dropout": 0.1,
383
+ "fsdp": "none",
384
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
385
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
386
+ "fsdp_fsdp_min_num_params": 100000000.0,
387
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
388
+ "fsdp_fsdp_forward_prefetch": false,
389
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
390
+ "fsdp_fsdp_offload_params": false,
391
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
392
+ "fsdp_fsdp_use_orig_params": true,
393
+ "fsdp_fsdp_activation_checkpointing": false,
394
+ "fsdp_fsdp_limit_all_gathers": true,
395
+ "fsdp_fsdp_sync_module_states": true,
396
+ "use_git": true,
397
+ "random_seed": 0
398
+ }
399
+ },
400
+ "gsm8k": {
401
+ "exact_match,flexible-extract": 0.35860500379075055,
402
+ "exact_match,flexible-extract_stderr": 0.013210317364134033,
403
+ "args": {
404
+ "exp_name": "results/v2/gsm8k/Qwen/Qwen2.5-0.5B",
405
+ "task_name": "gsm8k",
406
+ "model_name": "Qwen/Qwen2.5-0.5B",
407
+ "dataset_param": {
408
+ "max_num_train": 50000,
409
+ "completion_only": false,
410
+ "apply_chat_template": false
411
+ },
412
+ "resume_from_checkpoint": false,
413
+ "train_param": {
414
+ "max_length": 1024,
415
+ "per_device_train_batch_size": 8,
416
+ "auto_find_batch_size": true,
417
+ "num_train_epochs": 5,
418
+ "max_steps": -1,
419
+ "logging_strategy": "epoch",
420
+ "save_strategy": "epoch",
421
+ "logging_steps": 0.2,
422
+ "save_steps": 0.2,
423
+ "gradient_accumulation_steps": 1,
424
+ "gradient_checkpointing": false,
425
+ "optim": "adamw_torch",
426
+ "learning_rate": 5e-05,
427
+ "weight_decay": 0.01,
428
+ "warmup_steps": 0,
429
+ "report_to": "tensorboard",
430
+ "save_safetensors": true,
431
+ "packing": false
432
+ },
433
+ "use_peft": true,
434
+ "peft_task_type": "CAUSAL_LM",
435
+ "peft_inference_mode": false,
436
+ "peft_r": 8,
437
+ "peft_lora_alpha": 32,
438
+ "peft_lora_dropout": 0.1,
439
+ "fsdp": "none",
440
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
441
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
442
+ "fsdp_fsdp_min_num_params": 100000000.0,
443
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
444
+ "fsdp_fsdp_forward_prefetch": false,
445
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
446
+ "fsdp_fsdp_offload_params": false,
447
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
448
+ "fsdp_fsdp_use_orig_params": true,
449
+ "fsdp_fsdp_activation_checkpointing": false,
450
+ "fsdp_fsdp_limit_all_gathers": true,
451
+ "fsdp_fsdp_sync_module_states": true,
452
+ "use_git": true,
453
+ "random_seed": 0
454
+ }
455
+ },
456
+ "arc_challenge": {
457
+ "acc_norm": 0.35238907849829354,
458
+ "acc_norm_stderr": 0.01396014260059868,
459
+ "args": {
460
+ "exp_name": "results/v2/arc_challenge/Qwen/Qwen2.5-0.5B",
461
+ "task_name": "arc_challenge",
462
+ "model_name": "Qwen/Qwen2.5-0.5B",
463
+ "dataset_param": {
464
+ "max_num_train": 50000,
465
+ "completion_only": false,
466
+ "apply_chat_template": false
467
+ },
468
+ "resume_from_checkpoint": false,
469
+ "train_param": {
470
+ "max_length": 1024,
471
+ "per_device_train_batch_size": 8,
472
+ "auto_find_batch_size": true,
473
+ "num_train_epochs": 5,
474
+ "max_steps": -1,
475
+ "logging_strategy": "epoch",
476
+ "save_strategy": "epoch",
477
+ "logging_steps": 0.2,
478
+ "save_steps": 0.2,
479
+ "gradient_accumulation_steps": 1,
480
+ "gradient_checkpointing": false,
481
+ "optim": "adamw_torch",
482
+ "learning_rate": 5e-05,
483
+ "weight_decay": 0.01,
484
+ "warmup_steps": 0,
485
+ "report_to": "tensorboard",
486
+ "save_safetensors": true,
487
+ "packing": false
488
+ },
489
+ "use_peft": true,
490
+ "peft_task_type": "CAUSAL_LM",
491
+ "peft_inference_mode": false,
492
+ "peft_r": 8,
493
+ "peft_lora_alpha": 32,
494
+ "peft_lora_dropout": 0.1,
495
+ "fsdp": "none",
496
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
497
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
498
+ "fsdp_fsdp_min_num_params": 100000000.0,
499
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
500
+ "fsdp_fsdp_forward_prefetch": false,
501
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
502
+ "fsdp_fsdp_offload_params": false,
503
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
504
+ "fsdp_fsdp_use_orig_params": true,
505
+ "fsdp_fsdp_activation_checkpointing": false,
506
+ "fsdp_fsdp_limit_all_gathers": true,
507
+ "fsdp_fsdp_sync_module_states": true,
508
+ "use_git": true,
509
+ "random_seed": 0
510
+ }
511
+ }
512
+ }
513
+ }
Qwen/Qwen2.5-1.5B-Instruct/results_2026-01-05-04-27-51.json ADDED
@@ -0,0 +1,513 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "model_name": "Qwen/Qwen2.5-1.5B-Instruct",
4
+ "model_type": "instruction-tuned",
5
+ "params": 1.5
6
+ },
7
+ "results": {
8
+ "medmcqa": {
9
+ "acc_norm": 0.4936648338513029,
10
+ "acc_norm_stderr": 0.007731132688179789,
11
+ "args": {
12
+ "exp_name": "results/v1/medmcqa/Qwen/Qwen2.5-1.5B-Instruct",
13
+ "task_name": "medmcqa",
14
+ "model_name": "Qwen/Qwen2.5-1.5B-Instruct",
15
+ "dataset_param": {
16
+ "max_num_train": 50000,
17
+ "completion_only": false,
18
+ "apply_chat_template": true
19
+ },
20
+ "resume_from_checkpoint": false,
21
+ "train_param": {
22
+ "max_length": 1024,
23
+ "per_device_train_batch_size": 8,
24
+ "auto_find_batch_size": true,
25
+ "num_train_epochs": 1,
26
+ "max_steps": -1,
27
+ "logging_strategy": "steps",
28
+ "save_strategy": "steps",
29
+ "logging_steps": 0.2,
30
+ "save_steps": 0.2,
31
+ "gradient_accumulation_steps": 1,
32
+ "gradient_checkpointing": false,
33
+ "optim": "adamw_torch",
34
+ "learning_rate": 5e-05,
35
+ "weight_decay": 0.01,
36
+ "warmup_steps": 0,
37
+ "report_to": "tensorboard",
38
+ "save_safetensors": true,
39
+ "packing": false
40
+ },
41
+ "use_peft": true,
42
+ "peft_task_type": "CAUSAL_LM",
43
+ "peft_inference_mode": false,
44
+ "peft_r": 8,
45
+ "peft_lora_alpha": 32,
46
+ "peft_lora_dropout": 0.1,
47
+ "fsdp": "none",
48
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
49
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
50
+ "fsdp_fsdp_min_num_params": 100000000.0,
51
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
52
+ "fsdp_fsdp_forward_prefetch": false,
53
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
54
+ "fsdp_fsdp_offload_params": false,
55
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
56
+ "fsdp_fsdp_use_orig_params": true,
57
+ "fsdp_fsdp_activation_checkpointing": false,
58
+ "fsdp_fsdp_limit_all_gathers": true,
59
+ "fsdp_fsdp_sync_module_states": true,
60
+ "use_git": true,
61
+ "random_seed": 0
62
+ }
63
+ },
64
+ "nq_open": {
65
+ "exact_match,remove_whitespace": 0.14570637119113575,
66
+ "exact_match,remove_whitespace_stderr": 0.005872856832710493,
67
+ "args": {
68
+ "exp_name": "results/v1/nq_open/Qwen/Qwen2.5-1.5B-Instruct",
69
+ "task_name": "nq_open",
70
+ "model_name": "Qwen/Qwen2.5-1.5B-Instruct",
71
+ "dataset_param": {
72
+ "max_num_train": 50000,
73
+ "completion_only": false,
74
+ "apply_chat_template": true
75
+ },
76
+ "resume_from_checkpoint": false,
77
+ "train_param": {
78
+ "max_length": 1024,
79
+ "per_device_train_batch_size": 8,
80
+ "auto_find_batch_size": true,
81
+ "num_train_epochs": 1,
82
+ "max_steps": -1,
83
+ "logging_strategy": "steps",
84
+ "save_strategy": "steps",
85
+ "logging_steps": 0.2,
86
+ "save_steps": 0.2,
87
+ "gradient_accumulation_steps": 1,
88
+ "gradient_checkpointing": false,
89
+ "optim": "adamw_torch",
90
+ "learning_rate": 5e-05,
91
+ "weight_decay": 0.01,
92
+ "warmup_steps": 0,
93
+ "report_to": "tensorboard",
94
+ "save_safetensors": true,
95
+ "packing": false
96
+ },
97
+ "use_peft": true,
98
+ "peft_task_type": "CAUSAL_LM",
99
+ "peft_inference_mode": false,
100
+ "peft_r": 8,
101
+ "peft_lora_alpha": 32,
102
+ "peft_lora_dropout": 0.1,
103
+ "fsdp": "none",
104
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
105
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
106
+ "fsdp_fsdp_min_num_params": 100000000.0,
107
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
108
+ "fsdp_fsdp_forward_prefetch": false,
109
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
110
+ "fsdp_fsdp_offload_params": false,
111
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
112
+ "fsdp_fsdp_use_orig_params": true,
113
+ "fsdp_fsdp_activation_checkpointing": false,
114
+ "fsdp_fsdp_limit_all_gathers": true,
115
+ "fsdp_fsdp_sync_module_states": true,
116
+ "use_git": true,
117
+ "random_seed": 0
118
+ }
119
+ },
120
+ "winogrande": {
121
+ "acc": 0.6985003946329913,
122
+ "acc_stderr": 0.012897628072546689,
123
+ "args": {
124
+ "exp_name": "results/v1/winogrande/Qwen/Qwen2.5-1.5B-Instruct",
125
+ "task_name": "winogrande",
126
+ "model_name": "Qwen/Qwen2.5-1.5B-Instruct",
127
+ "dataset_param": {
128
+ "max_num_train": 50000,
129
+ "completion_only": false,
130
+ "apply_chat_template": true
131
+ },
132
+ "resume_from_checkpoint": false,
133
+ "train_param": {
134
+ "max_length": 1024,
135
+ "per_device_train_batch_size": 8,
136
+ "auto_find_batch_size": true,
137
+ "num_train_epochs": 1,
138
+ "max_steps": -1,
139
+ "logging_strategy": "steps",
140
+ "save_strategy": "steps",
141
+ "logging_steps": 0.2,
142
+ "save_steps": 0.2,
143
+ "gradient_accumulation_steps": 1,
144
+ "gradient_checkpointing": false,
145
+ "optim": "adamw_torch",
146
+ "learning_rate": 5e-05,
147
+ "weight_decay": 0.01,
148
+ "warmup_steps": 0,
149
+ "report_to": "tensorboard",
150
+ "save_safetensors": true,
151
+ "packing": false
152
+ },
153
+ "use_peft": true,
154
+ "peft_task_type": "CAUSAL_LM",
155
+ "peft_inference_mode": false,
156
+ "peft_r": 8,
157
+ "peft_lora_alpha": 32,
158
+ "peft_lora_dropout": 0.1,
159
+ "fsdp": "none",
160
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
161
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
162
+ "fsdp_fsdp_min_num_params": 100000000.0,
163
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
164
+ "fsdp_fsdp_forward_prefetch": false,
165
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
166
+ "fsdp_fsdp_offload_params": false,
167
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
168
+ "fsdp_fsdp_use_orig_params": true,
169
+ "fsdp_fsdp_activation_checkpointing": false,
170
+ "fsdp_fsdp_limit_all_gathers": true,
171
+ "fsdp_fsdp_sync_module_states": true,
172
+ "use_git": true,
173
+ "random_seed": 0
174
+ }
175
+ },
176
+ "hellaswag": {
177
+ "acc_norm": 0.6971,
178
+ "acc_norm_stderr": 0.0045953531667407765,
179
+ "args": {
180
+ "exp_name": "results/v1/hellaswag/Qwen/Qwen2.5-1.5B-Instruct",
181
+ "task_name": "hellaswag",
182
+ "model_name": "Qwen/Qwen2.5-1.5B-Instruct",
183
+ "dataset_param": {
184
+ "max_num_train": 50000,
185
+ "completion_only": false,
186
+ "apply_chat_template": true
187
+ },
188
+ "resume_from_checkpoint": false,
189
+ "train_param": {
190
+ "max_length": 1024,
191
+ "per_device_train_batch_size": 8,
192
+ "auto_find_batch_size": true,
193
+ "num_train_epochs": 1,
194
+ "max_steps": -1,
195
+ "logging_strategy": "steps",
196
+ "save_strategy": "steps",
197
+ "logging_steps": 0.2,
198
+ "save_steps": 0.2,
199
+ "gradient_accumulation_steps": 1,
200
+ "gradient_checkpointing": false,
201
+ "optim": "adamw_torch",
202
+ "learning_rate": 5e-05,
203
+ "weight_decay": 0.01,
204
+ "warmup_steps": 0,
205
+ "report_to": "tensorboard",
206
+ "save_safetensors": true,
207
+ "packing": false
208
+ },
209
+ "use_peft": true,
210
+ "peft_task_type": "CAUSAL_LM",
211
+ "peft_inference_mode": false,
212
+ "peft_r": 8,
213
+ "peft_lora_alpha": 32,
214
+ "peft_lora_dropout": 0.1,
215
+ "fsdp": "none",
216
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
217
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
218
+ "fsdp_fsdp_min_num_params": 100000000.0,
219
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
220
+ "fsdp_fsdp_forward_prefetch": false,
221
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
222
+ "fsdp_fsdp_offload_params": false,
223
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
224
+ "fsdp_fsdp_use_orig_params": true,
225
+ "fsdp_fsdp_activation_checkpointing": false,
226
+ "fsdp_fsdp_limit_all_gathers": true,
227
+ "fsdp_fsdp_sync_module_states": true,
228
+ "use_git": true,
229
+ "random_seed": 0
230
+ }
231
+ },
232
+ "social_iqa": {
233
+ "acc": 0.5133060388945753,
234
+ "acc_stderr": 0.011310063517892635,
235
+ "args": {
236
+ "exp_name": "results/v1/social_iqa/Qwen/Qwen2.5-1.5B-Instruct",
237
+ "task_name": "social_iqa",
238
+ "model_name": "Qwen/Qwen2.5-1.5B-Instruct",
239
+ "dataset_param": {
240
+ "max_num_train": 50000,
241
+ "completion_only": false,
242
+ "apply_chat_template": true
243
+ },
244
+ "resume_from_checkpoint": false,
245
+ "train_param": {
246
+ "max_length": 1024,
247
+ "per_device_train_batch_size": 8,
248
+ "auto_find_batch_size": true,
249
+ "num_train_epochs": 1,
250
+ "max_steps": -1,
251
+ "logging_strategy": "steps",
252
+ "save_strategy": "steps",
253
+ "logging_steps": 0.2,
254
+ "save_steps": 0.2,
255
+ "gradient_accumulation_steps": 1,
256
+ "gradient_checkpointing": false,
257
+ "optim": "adamw_torch",
258
+ "learning_rate": 5e-05,
259
+ "weight_decay": 0.01,
260
+ "warmup_steps": 0,
261
+ "report_to": "tensorboard",
262
+ "save_safetensors": true,
263
+ "packing": false
264
+ },
265
+ "use_peft": true,
266
+ "peft_task_type": "CAUSAL_LM",
267
+ "peft_inference_mode": false,
268
+ "peft_r": 8,
269
+ "peft_lora_alpha": 32,
270
+ "peft_lora_dropout": 0.1,
271
+ "fsdp": "none",
272
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
273
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
274
+ "fsdp_fsdp_min_num_params": 100000000.0,
275
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
276
+ "fsdp_fsdp_forward_prefetch": false,
277
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
278
+ "fsdp_fsdp_offload_params": false,
279
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
280
+ "fsdp_fsdp_use_orig_params": true,
281
+ "fsdp_fsdp_activation_checkpointing": false,
282
+ "fsdp_fsdp_limit_all_gathers": true,
283
+ "fsdp_fsdp_sync_module_states": true,
284
+ "use_git": true,
285
+ "random_seed": 0
286
+ }
287
+ },
288
+ "piqa": {
289
+ "acc_norm": 0.7693144722524483,
290
+ "acc_norm_stderr": 0.009828959550983086,
291
+ "args": {
292
+ "exp_name": "results/v2/piqa/Qwen/Qwen2.5-1.5B-Instruct",
293
+ "task_name": "piqa",
294
+ "model_name": "Qwen/Qwen2.5-1.5B-Instruct",
295
+ "dataset_param": {
296
+ "max_num_train": 50000,
297
+ "completion_only": false,
298
+ "apply_chat_template": true
299
+ },
300
+ "resume_from_checkpoint": false,
301
+ "train_param": {
302
+ "max_length": 1024,
303
+ "per_device_train_batch_size": 8,
304
+ "auto_find_batch_size": true,
305
+ "num_train_epochs": 5,
306
+ "max_steps": -1,
307
+ "logging_strategy": "epoch",
308
+ "save_strategy": "epoch",
309
+ "logging_steps": 0.2,
310
+ "save_steps": 0.2,
311
+ "gradient_accumulation_steps": 1,
312
+ "gradient_checkpointing": false,
313
+ "optim": "adamw_torch",
314
+ "learning_rate": 5e-05,
315
+ "weight_decay": 0.01,
316
+ "warmup_steps": 0,
317
+ "report_to": "tensorboard",
318
+ "save_safetensors": true,
319
+ "packing": false
320
+ },
321
+ "use_peft": true,
322
+ "peft_task_type": "CAUSAL_LM",
323
+ "peft_inference_mode": false,
324
+ "peft_r": 8,
325
+ "peft_lora_alpha": 32,
326
+ "peft_lora_dropout": 0.1,
327
+ "fsdp": "none",
328
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
329
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
330
+ "fsdp_fsdp_min_num_params": 100000000.0,
331
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
332
+ "fsdp_fsdp_forward_prefetch": false,
333
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
334
+ "fsdp_fsdp_offload_params": false,
335
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
336
+ "fsdp_fsdp_use_orig_params": true,
337
+ "fsdp_fsdp_activation_checkpointing": false,
338
+ "fsdp_fsdp_limit_all_gathers": true,
339
+ "fsdp_fsdp_sync_module_states": true,
340
+ "use_git": true,
341
+ "random_seed": 0
342
+ }
343
+ },
344
+ "commonsense_qa": {
345
+ "acc": 0.773955773955774,
346
+ "acc_stderr": 0.011974981909575622,
347
+ "args": {
348
+ "exp_name": "results/v2/commonsense_qa/Qwen/Qwen2.5-1.5B-Instruct",
349
+ "task_name": "commonsense_qa",
350
+ "model_name": "Qwen/Qwen2.5-1.5B-Instruct",
351
+ "dataset_param": {
352
+ "max_num_train": 50000,
353
+ "completion_only": false,
354
+ "apply_chat_template": true
355
+ },
356
+ "resume_from_checkpoint": false,
357
+ "train_param": {
358
+ "max_length": 1024,
359
+ "per_device_train_batch_size": 8,
360
+ "auto_find_batch_size": true,
361
+ "num_train_epochs": 5,
362
+ "max_steps": -1,
363
+ "logging_strategy": "epoch",
364
+ "save_strategy": "epoch",
365
+ "logging_steps": 0.2,
366
+ "save_steps": 0.2,
367
+ "gradient_accumulation_steps": 1,
368
+ "gradient_checkpointing": false,
369
+ "optim": "adamw_torch",
370
+ "learning_rate": 5e-05,
371
+ "weight_decay": 0.01,
372
+ "warmup_steps": 0,
373
+ "report_to": "tensorboard",
374
+ "save_safetensors": true,
375
+ "packing": false
376
+ },
377
+ "use_peft": true,
378
+ "peft_task_type": "CAUSAL_LM",
379
+ "peft_inference_mode": false,
380
+ "peft_r": 8,
381
+ "peft_lora_alpha": 32,
382
+ "peft_lora_dropout": 0.1,
383
+ "fsdp": "none",
384
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
385
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
386
+ "fsdp_fsdp_min_num_params": 100000000.0,
387
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
388
+ "fsdp_fsdp_forward_prefetch": false,
389
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
390
+ "fsdp_fsdp_offload_params": false,
391
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
392
+ "fsdp_fsdp_use_orig_params": true,
393
+ "fsdp_fsdp_activation_checkpointing": false,
394
+ "fsdp_fsdp_limit_all_gathers": true,
395
+ "fsdp_fsdp_sync_module_states": true,
396
+ "use_git": true,
397
+ "random_seed": 0
398
+ }
399
+ },
400
+ "gsm8k": {
401
+ "exact_match,flexible-extract": 0.5034116755117514,
402
+ "exact_match,flexible-extract_stderr": 0.013772164105556744,
403
+ "args": {
404
+ "exp_name": "results/v2/gsm8k/Qwen/Qwen2.5-1.5B-Instruct",
405
+ "task_name": "gsm8k",
406
+ "model_name": "Qwen/Qwen2.5-1.5B-Instruct",
407
+ "dataset_param": {
408
+ "max_num_train": 50000,
409
+ "completion_only": false,
410
+ "apply_chat_template": true
411
+ },
412
+ "resume_from_checkpoint": false,
413
+ "train_param": {
414
+ "max_length": 1024,
415
+ "per_device_train_batch_size": 8,
416
+ "auto_find_batch_size": true,
417
+ "num_train_epochs": 5,
418
+ "max_steps": -1,
419
+ "logging_strategy": "epoch",
420
+ "save_strategy": "epoch",
421
+ "logging_steps": 0.2,
422
+ "save_steps": 0.2,
423
+ "gradient_accumulation_steps": 1,
424
+ "gradient_checkpointing": false,
425
+ "optim": "adamw_torch",
426
+ "learning_rate": 5e-05,
427
+ "weight_decay": 0.01,
428
+ "warmup_steps": 0,
429
+ "report_to": "tensorboard",
430
+ "save_safetensors": true,
431
+ "packing": false
432
+ },
433
+ "use_peft": true,
434
+ "peft_task_type": "CAUSAL_LM",
435
+ "peft_inference_mode": false,
436
+ "peft_r": 8,
437
+ "peft_lora_alpha": 32,
438
+ "peft_lora_dropout": 0.1,
439
+ "fsdp": "none",
440
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
441
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
442
+ "fsdp_fsdp_min_num_params": 100000000.0,
443
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
444
+ "fsdp_fsdp_forward_prefetch": false,
445
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
446
+ "fsdp_fsdp_offload_params": false,
447
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
448
+ "fsdp_fsdp_use_orig_params": true,
449
+ "fsdp_fsdp_activation_checkpointing": false,
450
+ "fsdp_fsdp_limit_all_gathers": true,
451
+ "fsdp_fsdp_sync_module_states": true,
452
+ "use_git": true,
453
+ "random_seed": 0
454
+ }
455
+ },
456
+ "arc_challenge": {
457
+ "acc_norm": 0.4709897610921502,
458
+ "acc_norm_stderr": 0.014586776355294314,
459
+ "args": {
460
+ "exp_name": "results/v2/arc_challenge/Qwen/Qwen2.5-1.5B-Instruct",
461
+ "task_name": "arc_challenge",
462
+ "model_name": "Qwen/Qwen2.5-1.5B-Instruct",
463
+ "dataset_param": {
464
+ "max_num_train": 50000,
465
+ "completion_only": false,
466
+ "apply_chat_template": true
467
+ },
468
+ "resume_from_checkpoint": false,
469
+ "train_param": {
470
+ "max_length": 1024,
471
+ "per_device_train_batch_size": 8,
472
+ "auto_find_batch_size": true,
473
+ "num_train_epochs": 5,
474
+ "max_steps": -1,
475
+ "logging_strategy": "epoch",
476
+ "save_strategy": "epoch",
477
+ "logging_steps": 0.2,
478
+ "save_steps": 0.2,
479
+ "gradient_accumulation_steps": 1,
480
+ "gradient_checkpointing": false,
481
+ "optim": "adamw_torch",
482
+ "learning_rate": 5e-05,
483
+ "weight_decay": 0.01,
484
+ "warmup_steps": 0,
485
+ "report_to": "tensorboard",
486
+ "save_safetensors": true,
487
+ "packing": false
488
+ },
489
+ "use_peft": true,
490
+ "peft_task_type": "CAUSAL_LM",
491
+ "peft_inference_mode": false,
492
+ "peft_r": 8,
493
+ "peft_lora_alpha": 32,
494
+ "peft_lora_dropout": 0.1,
495
+ "fsdp": "none",
496
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
497
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
498
+ "fsdp_fsdp_min_num_params": 100000000.0,
499
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
500
+ "fsdp_fsdp_forward_prefetch": false,
501
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
502
+ "fsdp_fsdp_offload_params": false,
503
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
504
+ "fsdp_fsdp_use_orig_params": true,
505
+ "fsdp_fsdp_activation_checkpointing": false,
506
+ "fsdp_fsdp_limit_all_gathers": true,
507
+ "fsdp_fsdp_sync_module_states": true,
508
+ "use_git": true,
509
+ "random_seed": 0
510
+ }
511
+ }
512
+ }
513
+ }
Qwen/Qwen2.5-1.5B/results_2026-01-05-04-27-50.json ADDED
@@ -0,0 +1,513 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "model_name": "Qwen/Qwen2.5-1.5B",
4
+ "model_type": "pretrained",
5
+ "params": 1.5
6
+ },
7
+ "results": {
8
+ "medmcqa": {
9
+ "acc_norm": 0.4989242170690892,
10
+ "acc_norm_stderr": 0.007731735434496672,
11
+ "args": {
12
+ "exp_name": "results/v1/medmcqa/Qwen/Qwen2.5-1.5B",
13
+ "task_name": "medmcqa",
14
+ "model_name": "Qwen/Qwen2.5-1.5B",
15
+ "dataset_param": {
16
+ "max_num_train": 50000,
17
+ "completion_only": false,
18
+ "apply_chat_template": false
19
+ },
20
+ "resume_from_checkpoint": false,
21
+ "train_param": {
22
+ "max_length": 1024,
23
+ "per_device_train_batch_size": 8,
24
+ "auto_find_batch_size": true,
25
+ "num_train_epochs": 1,
26
+ "max_steps": -1,
27
+ "logging_strategy": "steps",
28
+ "save_strategy": "steps",
29
+ "logging_steps": 0.2,
30
+ "save_steps": 0.2,
31
+ "gradient_accumulation_steps": 1,
32
+ "gradient_checkpointing": false,
33
+ "optim": "adamw_torch",
34
+ "learning_rate": 5e-05,
35
+ "weight_decay": 0.01,
36
+ "warmup_steps": 0,
37
+ "report_to": "tensorboard",
38
+ "save_safetensors": true,
39
+ "packing": false
40
+ },
41
+ "use_peft": true,
42
+ "peft_task_type": "CAUSAL_LM",
43
+ "peft_inference_mode": false,
44
+ "peft_r": 8,
45
+ "peft_lora_alpha": 32,
46
+ "peft_lora_dropout": 0.1,
47
+ "fsdp": "none",
48
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
49
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
50
+ "fsdp_fsdp_min_num_params": 100000000.0,
51
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
52
+ "fsdp_fsdp_forward_prefetch": false,
53
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
54
+ "fsdp_fsdp_offload_params": false,
55
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
56
+ "fsdp_fsdp_use_orig_params": true,
57
+ "fsdp_fsdp_activation_checkpointing": false,
58
+ "fsdp_fsdp_limit_all_gathers": true,
59
+ "fsdp_fsdp_sync_module_states": true,
60
+ "use_git": true,
61
+ "random_seed": 0
62
+ }
63
+ },
64
+ "nq_open": {
65
+ "exact_match,remove_whitespace": 0.15595567867036012,
66
+ "exact_match,remove_whitespace_stderr": 0.00603934419398217,
67
+ "args": {
68
+ "exp_name": "results/v1/nq_open/Qwen/Qwen2.5-1.5B",
69
+ "task_name": "nq_open",
70
+ "model_name": "Qwen/Qwen2.5-1.5B",
71
+ "dataset_param": {
72
+ "max_num_train": 50000,
73
+ "completion_only": false,
74
+ "apply_chat_template": false
75
+ },
76
+ "resume_from_checkpoint": false,
77
+ "train_param": {
78
+ "max_length": 1024,
79
+ "per_device_train_batch_size": 8,
80
+ "auto_find_batch_size": true,
81
+ "num_train_epochs": 1,
82
+ "max_steps": -1,
83
+ "logging_strategy": "steps",
84
+ "save_strategy": "steps",
85
+ "logging_steps": 0.2,
86
+ "save_steps": 0.2,
87
+ "gradient_accumulation_steps": 1,
88
+ "gradient_checkpointing": false,
89
+ "optim": "adamw_torch",
90
+ "learning_rate": 5e-05,
91
+ "weight_decay": 0.01,
92
+ "warmup_steps": 0,
93
+ "report_to": "tensorboard",
94
+ "save_safetensors": true,
95
+ "packing": false
96
+ },
97
+ "use_peft": true,
98
+ "peft_task_type": "CAUSAL_LM",
99
+ "peft_inference_mode": false,
100
+ "peft_r": 8,
101
+ "peft_lora_alpha": 32,
102
+ "peft_lora_dropout": 0.1,
103
+ "fsdp": "none",
104
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
105
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
106
+ "fsdp_fsdp_min_num_params": 100000000.0,
107
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
108
+ "fsdp_fsdp_forward_prefetch": false,
109
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
110
+ "fsdp_fsdp_offload_params": false,
111
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
112
+ "fsdp_fsdp_use_orig_params": true,
113
+ "fsdp_fsdp_activation_checkpointing": false,
114
+ "fsdp_fsdp_limit_all_gathers": true,
115
+ "fsdp_fsdp_sync_module_states": true,
116
+ "use_git": true,
117
+ "random_seed": 0
118
+ }
119
+ },
120
+ "winogrande": {
121
+ "acc": 0.7040252565114443,
122
+ "acc_stderr": 0.01282934822633901,
123
+ "args": {
124
+ "exp_name": "results/v1/winogrande/Qwen/Qwen2.5-1.5B",
125
+ "task_name": "winogrande",
126
+ "model_name": "Qwen/Qwen2.5-1.5B",
127
+ "dataset_param": {
128
+ "max_num_train": 50000,
129
+ "completion_only": false,
130
+ "apply_chat_template": false
131
+ },
132
+ "resume_from_checkpoint": false,
133
+ "train_param": {
134
+ "max_length": 1024,
135
+ "per_device_train_batch_size": 8,
136
+ "auto_find_batch_size": true,
137
+ "num_train_epochs": 1,
138
+ "max_steps": -1,
139
+ "logging_strategy": "steps",
140
+ "save_strategy": "steps",
141
+ "logging_steps": 0.2,
142
+ "save_steps": 0.2,
143
+ "gradient_accumulation_steps": 1,
144
+ "gradient_checkpointing": false,
145
+ "optim": "adamw_torch",
146
+ "learning_rate": 5e-05,
147
+ "weight_decay": 0.01,
148
+ "warmup_steps": 0,
149
+ "report_to": "tensorboard",
150
+ "save_safetensors": true,
151
+ "packing": false
152
+ },
153
+ "use_peft": true,
154
+ "peft_task_type": "CAUSAL_LM",
155
+ "peft_inference_mode": false,
156
+ "peft_r": 8,
157
+ "peft_lora_alpha": 32,
158
+ "peft_lora_dropout": 0.1,
159
+ "fsdp": "none",
160
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
161
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
162
+ "fsdp_fsdp_min_num_params": 100000000.0,
163
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
164
+ "fsdp_fsdp_forward_prefetch": false,
165
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
166
+ "fsdp_fsdp_offload_params": false,
167
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
168
+ "fsdp_fsdp_use_orig_params": true,
169
+ "fsdp_fsdp_activation_checkpointing": false,
170
+ "fsdp_fsdp_limit_all_gathers": true,
171
+ "fsdp_fsdp_sync_module_states": true,
172
+ "use_git": true,
173
+ "random_seed": 0
174
+ }
175
+ },
176
+ "hellaswag": {
177
+ "acc_norm": 0.6968,
178
+ "acc_norm_stderr": 0.0045966388708370024,
179
+ "args": {
180
+ "exp_name": "results/v1/hellaswag/Qwen/Qwen2.5-1.5B",
181
+ "task_name": "hellaswag",
182
+ "model_name": "Qwen/Qwen2.5-1.5B",
183
+ "dataset_param": {
184
+ "max_num_train": 50000,
185
+ "completion_only": false,
186
+ "apply_chat_template": false
187
+ },
188
+ "resume_from_checkpoint": false,
189
+ "train_param": {
190
+ "max_length": 1024,
191
+ "per_device_train_batch_size": 8,
192
+ "auto_find_batch_size": true,
193
+ "num_train_epochs": 1,
194
+ "max_steps": -1,
195
+ "logging_strategy": "steps",
196
+ "save_strategy": "steps",
197
+ "logging_steps": 0.2,
198
+ "save_steps": 0.2,
199
+ "gradient_accumulation_steps": 1,
200
+ "gradient_checkpointing": false,
201
+ "optim": "adamw_torch",
202
+ "learning_rate": 5e-05,
203
+ "weight_decay": 0.01,
204
+ "warmup_steps": 0,
205
+ "report_to": "tensorboard",
206
+ "save_safetensors": true,
207
+ "packing": false
208
+ },
209
+ "use_peft": true,
210
+ "peft_task_type": "CAUSAL_LM",
211
+ "peft_inference_mode": false,
212
+ "peft_r": 8,
213
+ "peft_lora_alpha": 32,
214
+ "peft_lora_dropout": 0.1,
215
+ "fsdp": "none",
216
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
217
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
218
+ "fsdp_fsdp_min_num_params": 100000000.0,
219
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
220
+ "fsdp_fsdp_forward_prefetch": false,
221
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
222
+ "fsdp_fsdp_offload_params": false,
223
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
224
+ "fsdp_fsdp_use_orig_params": true,
225
+ "fsdp_fsdp_activation_checkpointing": false,
226
+ "fsdp_fsdp_limit_all_gathers": true,
227
+ "fsdp_fsdp_sync_module_states": true,
228
+ "use_git": true,
229
+ "random_seed": 0
230
+ }
231
+ },
232
+ "social_iqa": {
233
+ "acc": 0.5138178096212896,
234
+ "acc_stderr": 0.011309749296679794,
235
+ "args": {
236
+ "exp_name": "results/v1/social_iqa/Qwen/Qwen2.5-1.5B",
237
+ "task_name": "social_iqa",
238
+ "model_name": "Qwen/Qwen2.5-1.5B",
239
+ "dataset_param": {
240
+ "max_num_train": 50000,
241
+ "completion_only": false,
242
+ "apply_chat_template": false
243
+ },
244
+ "resume_from_checkpoint": false,
245
+ "train_param": {
246
+ "max_length": 1024,
247
+ "per_device_train_batch_size": 8,
248
+ "auto_find_batch_size": true,
249
+ "num_train_epochs": 1,
250
+ "max_steps": -1,
251
+ "logging_strategy": "steps",
252
+ "save_strategy": "steps",
253
+ "logging_steps": 0.2,
254
+ "save_steps": 0.2,
255
+ "gradient_accumulation_steps": 1,
256
+ "gradient_checkpointing": false,
257
+ "optim": "adamw_torch",
258
+ "learning_rate": 5e-05,
259
+ "weight_decay": 0.01,
260
+ "warmup_steps": 0,
261
+ "report_to": "tensorboard",
262
+ "save_safetensors": true,
263
+ "packing": false
264
+ },
265
+ "use_peft": true,
266
+ "peft_task_type": "CAUSAL_LM",
267
+ "peft_inference_mode": false,
268
+ "peft_r": 8,
269
+ "peft_lora_alpha": 32,
270
+ "peft_lora_dropout": 0.1,
271
+ "fsdp": "none",
272
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
273
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
274
+ "fsdp_fsdp_min_num_params": 100000000.0,
275
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
276
+ "fsdp_fsdp_forward_prefetch": false,
277
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
278
+ "fsdp_fsdp_offload_params": false,
279
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
280
+ "fsdp_fsdp_use_orig_params": true,
281
+ "fsdp_fsdp_activation_checkpointing": false,
282
+ "fsdp_fsdp_limit_all_gathers": true,
283
+ "fsdp_fsdp_sync_module_states": true,
284
+ "use_git": true,
285
+ "random_seed": 0
286
+ }
287
+ },
288
+ "piqa": {
289
+ "acc_norm": 0.7780195865070729,
290
+ "acc_norm_stderr": 0.009696120744662006,
291
+ "args": {
292
+ "exp_name": "results/v2/piqa/Qwen/Qwen2.5-1.5B",
293
+ "task_name": "piqa",
294
+ "model_name": "Qwen/Qwen2.5-1.5B",
295
+ "dataset_param": {
296
+ "max_num_train": 50000,
297
+ "completion_only": false,
298
+ "apply_chat_template": false
299
+ },
300
+ "resume_from_checkpoint": false,
301
+ "train_param": {
302
+ "max_length": 1024,
303
+ "per_device_train_batch_size": 8,
304
+ "auto_find_batch_size": true,
305
+ "num_train_epochs": 5,
306
+ "max_steps": -1,
307
+ "logging_strategy": "epoch",
308
+ "save_strategy": "epoch",
309
+ "logging_steps": 0.2,
310
+ "save_steps": 0.2,
311
+ "gradient_accumulation_steps": 1,
312
+ "gradient_checkpointing": false,
313
+ "optim": "adamw_torch",
314
+ "learning_rate": 5e-05,
315
+ "weight_decay": 0.01,
316
+ "warmup_steps": 0,
317
+ "report_to": "tensorboard",
318
+ "save_safetensors": true,
319
+ "packing": false
320
+ },
321
+ "use_peft": true,
322
+ "peft_task_type": "CAUSAL_LM",
323
+ "peft_inference_mode": false,
324
+ "peft_r": 8,
325
+ "peft_lora_alpha": 32,
326
+ "peft_lora_dropout": 0.1,
327
+ "fsdp": "none",
328
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
329
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
330
+ "fsdp_fsdp_min_num_params": 100000000.0,
331
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
332
+ "fsdp_fsdp_forward_prefetch": false,
333
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
334
+ "fsdp_fsdp_offload_params": false,
335
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
336
+ "fsdp_fsdp_use_orig_params": true,
337
+ "fsdp_fsdp_activation_checkpointing": false,
338
+ "fsdp_fsdp_limit_all_gathers": true,
339
+ "fsdp_fsdp_sync_module_states": true,
340
+ "use_git": true,
341
+ "random_seed": 0
342
+ }
343
+ },
344
+ "commonsense_qa": {
345
+ "acc": 0.7764127764127764,
346
+ "acc_stderr": 0.011928612008761155,
347
+ "args": {
348
+ "exp_name": "results/v2/commonsense_qa/Qwen/Qwen2.5-1.5B",
349
+ "task_name": "commonsense_qa",
350
+ "model_name": "Qwen/Qwen2.5-1.5B",
351
+ "dataset_param": {
352
+ "max_num_train": 50000,
353
+ "completion_only": false,
354
+ "apply_chat_template": false
355
+ },
356
+ "resume_from_checkpoint": false,
357
+ "train_param": {
358
+ "max_length": 1024,
359
+ "per_device_train_batch_size": 8,
360
+ "auto_find_batch_size": true,
361
+ "num_train_epochs": 5,
362
+ "max_steps": -1,
363
+ "logging_strategy": "epoch",
364
+ "save_strategy": "epoch",
365
+ "logging_steps": 0.2,
366
+ "save_steps": 0.2,
367
+ "gradient_accumulation_steps": 1,
368
+ "gradient_checkpointing": false,
369
+ "optim": "adamw_torch",
370
+ "learning_rate": 5e-05,
371
+ "weight_decay": 0.01,
372
+ "warmup_steps": 0,
373
+ "report_to": "tensorboard",
374
+ "save_safetensors": true,
375
+ "packing": false
376
+ },
377
+ "use_peft": true,
378
+ "peft_task_type": "CAUSAL_LM",
379
+ "peft_inference_mode": false,
380
+ "peft_r": 8,
381
+ "peft_lora_alpha": 32,
382
+ "peft_lora_dropout": 0.1,
383
+ "fsdp": "none",
384
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
385
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
386
+ "fsdp_fsdp_min_num_params": 100000000.0,
387
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
388
+ "fsdp_fsdp_forward_prefetch": false,
389
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
390
+ "fsdp_fsdp_offload_params": false,
391
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
392
+ "fsdp_fsdp_use_orig_params": true,
393
+ "fsdp_fsdp_activation_checkpointing": false,
394
+ "fsdp_fsdp_limit_all_gathers": true,
395
+ "fsdp_fsdp_sync_module_states": true,
396
+ "use_git": true,
397
+ "random_seed": 0
398
+ }
399
+ },
400
+ "gsm8k": {
401
+ "exact_match,flexible-extract": 0.6034874905231236,
402
+ "exact_match,flexible-extract_stderr": 0.013474258584033345,
403
+ "args": {
404
+ "exp_name": "results/v2/gsm8k/Qwen/Qwen2.5-1.5B",
405
+ "task_name": "gsm8k",
406
+ "model_name": "Qwen/Qwen2.5-1.5B",
407
+ "dataset_param": {
408
+ "max_num_train": 50000,
409
+ "completion_only": false,
410
+ "apply_chat_template": false
411
+ },
412
+ "resume_from_checkpoint": false,
413
+ "train_param": {
414
+ "max_length": 1024,
415
+ "per_device_train_batch_size": 8,
416
+ "auto_find_batch_size": true,
417
+ "num_train_epochs": 5,
418
+ "max_steps": -1,
419
+ "logging_strategy": "epoch",
420
+ "save_strategy": "epoch",
421
+ "logging_steps": 0.2,
422
+ "save_steps": 0.2,
423
+ "gradient_accumulation_steps": 1,
424
+ "gradient_checkpointing": false,
425
+ "optim": "adamw_torch",
426
+ "learning_rate": 5e-05,
427
+ "weight_decay": 0.01,
428
+ "warmup_steps": 0,
429
+ "report_to": "tensorboard",
430
+ "save_safetensors": true,
431
+ "packing": false
432
+ },
433
+ "use_peft": true,
434
+ "peft_task_type": "CAUSAL_LM",
435
+ "peft_inference_mode": false,
436
+ "peft_r": 8,
437
+ "peft_lora_alpha": 32,
438
+ "peft_lora_dropout": 0.1,
439
+ "fsdp": "none",
440
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
441
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
442
+ "fsdp_fsdp_min_num_params": 100000000.0,
443
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
444
+ "fsdp_fsdp_forward_prefetch": false,
445
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
446
+ "fsdp_fsdp_offload_params": false,
447
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
448
+ "fsdp_fsdp_use_orig_params": true,
449
+ "fsdp_fsdp_activation_checkpointing": false,
450
+ "fsdp_fsdp_limit_all_gathers": true,
451
+ "fsdp_fsdp_sync_module_states": true,
452
+ "use_git": true,
453
+ "random_seed": 0
454
+ }
455
+ },
456
+ "arc_challenge": {
457
+ "acc_norm": 0.4931740614334471,
458
+ "acc_norm_stderr": 0.014610029151379813,
459
+ "args": {
460
+ "exp_name": "results/v2/arc_challenge/Qwen/Qwen2.5-1.5B",
461
+ "task_name": "arc_challenge",
462
+ "model_name": "Qwen/Qwen2.5-1.5B",
463
+ "dataset_param": {
464
+ "max_num_train": 50000,
465
+ "completion_only": false,
466
+ "apply_chat_template": false
467
+ },
468
+ "resume_from_checkpoint": false,
469
+ "train_param": {
470
+ "max_length": 1024,
471
+ "per_device_train_batch_size": 8,
472
+ "auto_find_batch_size": true,
473
+ "num_train_epochs": 5,
474
+ "max_steps": -1,
475
+ "logging_strategy": "epoch",
476
+ "save_strategy": "epoch",
477
+ "logging_steps": 0.2,
478
+ "save_steps": 0.2,
479
+ "gradient_accumulation_steps": 1,
480
+ "gradient_checkpointing": false,
481
+ "optim": "adamw_torch",
482
+ "learning_rate": 5e-05,
483
+ "weight_decay": 0.01,
484
+ "warmup_steps": 0,
485
+ "report_to": "tensorboard",
486
+ "save_safetensors": true,
487
+ "packing": false
488
+ },
489
+ "use_peft": true,
490
+ "peft_task_type": "CAUSAL_LM",
491
+ "peft_inference_mode": false,
492
+ "peft_r": 8,
493
+ "peft_lora_alpha": 32,
494
+ "peft_lora_dropout": 0.1,
495
+ "fsdp": "none",
496
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
497
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
498
+ "fsdp_fsdp_min_num_params": 100000000.0,
499
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
500
+ "fsdp_fsdp_forward_prefetch": false,
501
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
502
+ "fsdp_fsdp_offload_params": false,
503
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
504
+ "fsdp_fsdp_use_orig_params": true,
505
+ "fsdp_fsdp_activation_checkpointing": false,
506
+ "fsdp_fsdp_limit_all_gathers": true,
507
+ "fsdp_fsdp_sync_module_states": true,
508
+ "use_git": true,
509
+ "random_seed": 0
510
+ }
511
+ }
512
+ }
513
+ }
Qwen/Qwen2.5-14B-Instruct/results_2026-01-05-04-27-51.json ADDED
@@ -0,0 +1,513 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "model_name": "Qwen/Qwen2.5-14B-Instruct",
4
+ "model_type": "instruction-tuned",
5
+ "params": 14.0
6
+ },
7
+ "results": {
8
+ "medmcqa": {
9
+ "acc_norm": 0.6540760219937843,
10
+ "acc_norm_stderr": 0.007355503953762135,
11
+ "args": {
12
+ "exp_name": "results/v1/medmcqa/Qwen/Qwen2.5-14B-Instruct",
13
+ "task_name": "medmcqa",
14
+ "model_name": "Qwen/Qwen2.5-14B-Instruct",
15
+ "dataset_param": {
16
+ "max_num_train": 50000,
17
+ "completion_only": false,
18
+ "apply_chat_template": true
19
+ },
20
+ "resume_from_checkpoint": false,
21
+ "train_param": {
22
+ "max_length": 1024,
23
+ "per_device_train_batch_size": 8,
24
+ "auto_find_batch_size": true,
25
+ "num_train_epochs": 1,
26
+ "max_steps": -1,
27
+ "logging_strategy": "steps",
28
+ "save_strategy": "steps",
29
+ "logging_steps": 0.2,
30
+ "save_steps": 0.2,
31
+ "gradient_accumulation_steps": 1,
32
+ "gradient_checkpointing": false,
33
+ "optim": "adamw_torch",
34
+ "learning_rate": 5e-05,
35
+ "weight_decay": 0.01,
36
+ "warmup_steps": 0,
37
+ "report_to": "tensorboard",
38
+ "save_safetensors": true,
39
+ "packing": false
40
+ },
41
+ "use_peft": true,
42
+ "peft_task_type": "CAUSAL_LM",
43
+ "peft_inference_mode": false,
44
+ "peft_r": 8,
45
+ "peft_lora_alpha": 32,
46
+ "peft_lora_dropout": 0.1,
47
+ "fsdp": "none",
48
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
49
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
50
+ "fsdp_fsdp_min_num_params": 100000000.0,
51
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
52
+ "fsdp_fsdp_forward_prefetch": false,
53
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
54
+ "fsdp_fsdp_offload_params": false,
55
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
56
+ "fsdp_fsdp_use_orig_params": true,
57
+ "fsdp_fsdp_activation_checkpointing": false,
58
+ "fsdp_fsdp_limit_all_gathers": true,
59
+ "fsdp_fsdp_sync_module_states": true,
60
+ "use_git": true,
61
+ "random_seed": 0
62
+ }
63
+ },
64
+ "nq_open": {
65
+ "exact_match,remove_whitespace": 0.32936288088642657,
66
+ "exact_match,remove_whitespace_stderr": 0.007823258471850336,
67
+ "args": {
68
+ "exp_name": "results/v1/nq_open/Qwen/Qwen2.5-14B-Instruct",
69
+ "task_name": "nq_open",
70
+ "model_name": "Qwen/Qwen2.5-14B-Instruct",
71
+ "dataset_param": {
72
+ "max_num_train": 50000,
73
+ "completion_only": false,
74
+ "apply_chat_template": true
75
+ },
76
+ "resume_from_checkpoint": false,
77
+ "train_param": {
78
+ "max_length": 1024,
79
+ "per_device_train_batch_size": 8,
80
+ "auto_find_batch_size": true,
81
+ "num_train_epochs": 1,
82
+ "max_steps": -1,
83
+ "logging_strategy": "steps",
84
+ "save_strategy": "steps",
85
+ "logging_steps": 0.2,
86
+ "save_steps": 0.2,
87
+ "gradient_accumulation_steps": 1,
88
+ "gradient_checkpointing": false,
89
+ "optim": "adamw_torch",
90
+ "learning_rate": 5e-05,
91
+ "weight_decay": 0.01,
92
+ "warmup_steps": 0,
93
+ "report_to": "tensorboard",
94
+ "save_safetensors": true,
95
+ "packing": false
96
+ },
97
+ "use_peft": true,
98
+ "peft_task_type": "CAUSAL_LM",
99
+ "peft_inference_mode": false,
100
+ "peft_r": 8,
101
+ "peft_lora_alpha": 32,
102
+ "peft_lora_dropout": 0.1,
103
+ "fsdp": "none",
104
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
105
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
106
+ "fsdp_fsdp_min_num_params": 100000000.0,
107
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
108
+ "fsdp_fsdp_forward_prefetch": false,
109
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
110
+ "fsdp_fsdp_offload_params": false,
111
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
112
+ "fsdp_fsdp_use_orig_params": true,
113
+ "fsdp_fsdp_activation_checkpointing": false,
114
+ "fsdp_fsdp_limit_all_gathers": true,
115
+ "fsdp_fsdp_sync_module_states": true,
116
+ "use_git": true,
117
+ "random_seed": 0
118
+ }
119
+ },
120
+ "winogrande": {
121
+ "acc": 0.8650355169692187,
122
+ "acc_stderr": 0.009603064913219057,
123
+ "args": {
124
+ "exp_name": "results/v1/winogrande/Qwen/Qwen2.5-14B-Instruct",
125
+ "task_name": "winogrande",
126
+ "model_name": "Qwen/Qwen2.5-14B-Instruct",
127
+ "dataset_param": {
128
+ "max_num_train": 50000,
129
+ "completion_only": false,
130
+ "apply_chat_template": true
131
+ },
132
+ "resume_from_checkpoint": false,
133
+ "train_param": {
134
+ "max_length": 1024,
135
+ "per_device_train_batch_size": 8,
136
+ "auto_find_batch_size": true,
137
+ "num_train_epochs": 1,
138
+ "max_steps": -1,
139
+ "logging_strategy": "steps",
140
+ "save_strategy": "steps",
141
+ "logging_steps": 0.2,
142
+ "save_steps": 0.2,
143
+ "gradient_accumulation_steps": 1,
144
+ "gradient_checkpointing": false,
145
+ "optim": "adamw_torch",
146
+ "learning_rate": 5e-05,
147
+ "weight_decay": 0.01,
148
+ "warmup_steps": 0,
149
+ "report_to": "tensorboard",
150
+ "save_safetensors": true,
151
+ "packing": false
152
+ },
153
+ "use_peft": true,
154
+ "peft_task_type": "CAUSAL_LM",
155
+ "peft_inference_mode": false,
156
+ "peft_r": 8,
157
+ "peft_lora_alpha": 32,
158
+ "peft_lora_dropout": 0.1,
159
+ "fsdp": "none",
160
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
161
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
162
+ "fsdp_fsdp_min_num_params": 100000000.0,
163
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
164
+ "fsdp_fsdp_forward_prefetch": false,
165
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
166
+ "fsdp_fsdp_offload_params": false,
167
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
168
+ "fsdp_fsdp_use_orig_params": true,
169
+ "fsdp_fsdp_activation_checkpointing": false,
170
+ "fsdp_fsdp_limit_all_gathers": true,
171
+ "fsdp_fsdp_sync_module_states": true,
172
+ "use_git": true,
173
+ "random_seed": 0
174
+ }
175
+ },
176
+ "hellaswag": {
177
+ "acc_norm": 0.8534,
178
+ "acc_norm_stderr": 0.003537244013284742,
179
+ "args": {
180
+ "exp_name": "results/v1/hellaswag/Qwen/Qwen2.5-14B-Instruct",
181
+ "task_name": "hellaswag",
182
+ "model_name": "Qwen/Qwen2.5-14B-Instruct",
183
+ "dataset_param": {
184
+ "max_num_train": 50000,
185
+ "completion_only": false,
186
+ "apply_chat_template": true
187
+ },
188
+ "resume_from_checkpoint": false,
189
+ "train_param": {
190
+ "max_length": 1024,
191
+ "per_device_train_batch_size": 8,
192
+ "auto_find_batch_size": true,
193
+ "num_train_epochs": 1,
194
+ "max_steps": -1,
195
+ "logging_strategy": "steps",
196
+ "save_strategy": "steps",
197
+ "logging_steps": 0.2,
198
+ "save_steps": 0.2,
199
+ "gradient_accumulation_steps": 1,
200
+ "gradient_checkpointing": false,
201
+ "optim": "adamw_torch",
202
+ "learning_rate": 5e-05,
203
+ "weight_decay": 0.01,
204
+ "warmup_steps": 0,
205
+ "report_to": "tensorboard",
206
+ "save_safetensors": true,
207
+ "packing": false
208
+ },
209
+ "use_peft": true,
210
+ "peft_task_type": "CAUSAL_LM",
211
+ "peft_inference_mode": false,
212
+ "peft_r": 8,
213
+ "peft_lora_alpha": 32,
214
+ "peft_lora_dropout": 0.1,
215
+ "fsdp": "none",
216
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
217
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
218
+ "fsdp_fsdp_min_num_params": 100000000.0,
219
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
220
+ "fsdp_fsdp_forward_prefetch": false,
221
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
222
+ "fsdp_fsdp_offload_params": false,
223
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
224
+ "fsdp_fsdp_use_orig_params": true,
225
+ "fsdp_fsdp_activation_checkpointing": false,
226
+ "fsdp_fsdp_limit_all_gathers": true,
227
+ "fsdp_fsdp_sync_module_states": true,
228
+ "use_git": true,
229
+ "random_seed": 0
230
+ }
231
+ },
232
+ "social_iqa": {
233
+ "acc": 0.5834186284544524,
234
+ "acc_stderr": 0.011155497599417979,
235
+ "args": {
236
+ "exp_name": "results/v1/social_iqa/Qwen/Qwen2.5-14B-Instruct",
237
+ "task_name": "social_iqa",
238
+ "model_name": "Qwen/Qwen2.5-14B-Instruct",
239
+ "dataset_param": {
240
+ "max_num_train": 50000,
241
+ "completion_only": false,
242
+ "apply_chat_template": true
243
+ },
244
+ "resume_from_checkpoint": false,
245
+ "train_param": {
246
+ "max_length": 1024,
247
+ "per_device_train_batch_size": 8,
248
+ "auto_find_batch_size": true,
249
+ "num_train_epochs": 1,
250
+ "max_steps": -1,
251
+ "logging_strategy": "steps",
252
+ "save_strategy": "steps",
253
+ "logging_steps": 0.2,
254
+ "save_steps": 0.2,
255
+ "gradient_accumulation_steps": 1,
256
+ "gradient_checkpointing": false,
257
+ "optim": "adamw_torch",
258
+ "learning_rate": 5e-05,
259
+ "weight_decay": 0.01,
260
+ "warmup_steps": 0,
261
+ "report_to": "tensorboard",
262
+ "save_safetensors": true,
263
+ "packing": false
264
+ },
265
+ "use_peft": true,
266
+ "peft_task_type": "CAUSAL_LM",
267
+ "peft_inference_mode": false,
268
+ "peft_r": 8,
269
+ "peft_lora_alpha": 32,
270
+ "peft_lora_dropout": 0.1,
271
+ "fsdp": "none",
272
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
273
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
274
+ "fsdp_fsdp_min_num_params": 100000000.0,
275
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
276
+ "fsdp_fsdp_forward_prefetch": false,
277
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
278
+ "fsdp_fsdp_offload_params": false,
279
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
280
+ "fsdp_fsdp_use_orig_params": true,
281
+ "fsdp_fsdp_activation_checkpointing": false,
282
+ "fsdp_fsdp_limit_all_gathers": true,
283
+ "fsdp_fsdp_sync_module_states": true,
284
+ "use_git": true,
285
+ "random_seed": 0
286
+ }
287
+ },
288
+ "piqa": {
289
+ "acc_norm": 0.8427638737758433,
290
+ "acc_norm_stderr": 0.008493262462603447,
291
+ "args": {
292
+ "exp_name": "results/v2/piqa/Qwen/Qwen2.5-14B-Instruct",
293
+ "task_name": "piqa",
294
+ "model_name": "Qwen/Qwen2.5-14B-Instruct",
295
+ "dataset_param": {
296
+ "max_num_train": 50000,
297
+ "completion_only": false,
298
+ "apply_chat_template": true
299
+ },
300
+ "resume_from_checkpoint": false,
301
+ "train_param": {
302
+ "max_length": 1024,
303
+ "per_device_train_batch_size": 8,
304
+ "auto_find_batch_size": true,
305
+ "num_train_epochs": 5,
306
+ "max_steps": -1,
307
+ "logging_strategy": "epoch",
308
+ "save_strategy": "epoch",
309
+ "logging_steps": 0.2,
310
+ "save_steps": 0.2,
311
+ "gradient_accumulation_steps": 1,
312
+ "gradient_checkpointing": false,
313
+ "optim": "adamw_torch",
314
+ "learning_rate": 5e-05,
315
+ "weight_decay": 0.01,
316
+ "warmup_steps": 0,
317
+ "report_to": "tensorboard",
318
+ "save_safetensors": true,
319
+ "packing": false
320
+ },
321
+ "use_peft": true,
322
+ "peft_task_type": "CAUSAL_LM",
323
+ "peft_inference_mode": false,
324
+ "peft_r": 8,
325
+ "peft_lora_alpha": 32,
326
+ "peft_lora_dropout": 0.1,
327
+ "fsdp": "none",
328
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
329
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
330
+ "fsdp_fsdp_min_num_params": 100000000.0,
331
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
332
+ "fsdp_fsdp_forward_prefetch": false,
333
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
334
+ "fsdp_fsdp_offload_params": false,
335
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
336
+ "fsdp_fsdp_use_orig_params": true,
337
+ "fsdp_fsdp_activation_checkpointing": false,
338
+ "fsdp_fsdp_limit_all_gathers": true,
339
+ "fsdp_fsdp_sync_module_states": true,
340
+ "use_git": true,
341
+ "random_seed": 0
342
+ }
343
+ },
344
+ "commonsense_qa": {
345
+ "acc": 0.8828828828828829,
346
+ "acc_stderr": 0.009206229982351257,
347
+ "args": {
348
+ "exp_name": "results/v2/commonsense_qa/Qwen/Qwen2.5-14B-Instruct",
349
+ "task_name": "commonsense_qa",
350
+ "model_name": "Qwen/Qwen2.5-14B-Instruct",
351
+ "dataset_param": {
352
+ "max_num_train": 50000,
353
+ "completion_only": false,
354
+ "apply_chat_template": true
355
+ },
356
+ "resume_from_checkpoint": false,
357
+ "train_param": {
358
+ "max_length": 1024,
359
+ "per_device_train_batch_size": 8,
360
+ "auto_find_batch_size": true,
361
+ "num_train_epochs": 5,
362
+ "max_steps": -1,
363
+ "logging_strategy": "epoch",
364
+ "save_strategy": "epoch",
365
+ "logging_steps": 0.2,
366
+ "save_steps": 0.2,
367
+ "gradient_accumulation_steps": 1,
368
+ "gradient_checkpointing": false,
369
+ "optim": "adamw_torch",
370
+ "learning_rate": 5e-05,
371
+ "weight_decay": 0.01,
372
+ "warmup_steps": 0,
373
+ "report_to": "tensorboard",
374
+ "save_safetensors": true,
375
+ "packing": false
376
+ },
377
+ "use_peft": true,
378
+ "peft_task_type": "CAUSAL_LM",
379
+ "peft_inference_mode": false,
380
+ "peft_r": 8,
381
+ "peft_lora_alpha": 32,
382
+ "peft_lora_dropout": 0.1,
383
+ "fsdp": "none",
384
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
385
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
386
+ "fsdp_fsdp_min_num_params": 100000000.0,
387
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
388
+ "fsdp_fsdp_forward_prefetch": false,
389
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
390
+ "fsdp_fsdp_offload_params": false,
391
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
392
+ "fsdp_fsdp_use_orig_params": true,
393
+ "fsdp_fsdp_activation_checkpointing": false,
394
+ "fsdp_fsdp_limit_all_gathers": true,
395
+ "fsdp_fsdp_sync_module_states": true,
396
+ "use_git": true,
397
+ "random_seed": 0
398
+ }
399
+ },
400
+ "gsm8k": {
401
+ "exact_match,flexible-extract": 0.8248673237300985,
402
+ "exact_match,flexible-extract_stderr": 0.010469307043157905,
403
+ "args": {
404
+ "exp_name": "results/v2/gsm8k/Qwen/Qwen2.5-14B-Instruct",
405
+ "task_name": "gsm8k",
406
+ "model_name": "Qwen/Qwen2.5-14B-Instruct",
407
+ "dataset_param": {
408
+ "max_num_train": 50000,
409
+ "completion_only": false,
410
+ "apply_chat_template": true
411
+ },
412
+ "resume_from_checkpoint": false,
413
+ "train_param": {
414
+ "max_length": 1024,
415
+ "per_device_train_batch_size": 8,
416
+ "auto_find_batch_size": true,
417
+ "num_train_epochs": 5,
418
+ "max_steps": -1,
419
+ "logging_strategy": "epoch",
420
+ "save_strategy": "epoch",
421
+ "logging_steps": 0.2,
422
+ "save_steps": 0.2,
423
+ "gradient_accumulation_steps": 1,
424
+ "gradient_checkpointing": false,
425
+ "optim": "adamw_torch",
426
+ "learning_rate": 5e-05,
427
+ "weight_decay": 0.01,
428
+ "warmup_steps": 0,
429
+ "report_to": "tensorboard",
430
+ "save_safetensors": true,
431
+ "packing": false
432
+ },
433
+ "use_peft": true,
434
+ "peft_task_type": "CAUSAL_LM",
435
+ "peft_inference_mode": false,
436
+ "peft_r": 8,
437
+ "peft_lora_alpha": 32,
438
+ "peft_lora_dropout": 0.1,
439
+ "fsdp": "none",
440
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
441
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
442
+ "fsdp_fsdp_min_num_params": 100000000.0,
443
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
444
+ "fsdp_fsdp_forward_prefetch": false,
445
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
446
+ "fsdp_fsdp_offload_params": false,
447
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
448
+ "fsdp_fsdp_use_orig_params": true,
449
+ "fsdp_fsdp_activation_checkpointing": false,
450
+ "fsdp_fsdp_limit_all_gathers": true,
451
+ "fsdp_fsdp_sync_module_states": true,
452
+ "use_git": true,
453
+ "random_seed": 0
454
+ }
455
+ },
456
+ "arc_challenge": {
457
+ "acc_norm": 0.6629692832764505,
458
+ "acc_norm_stderr": 0.013813476652902269,
459
+ "args": {
460
+ "exp_name": "results/v2/arc_challenge/Qwen/Qwen2.5-14B-Instruct",
461
+ "task_name": "arc_challenge",
462
+ "model_name": "Qwen/Qwen2.5-14B-Instruct",
463
+ "dataset_param": {
464
+ "max_num_train": 50000,
465
+ "completion_only": false,
466
+ "apply_chat_template": true
467
+ },
468
+ "resume_from_checkpoint": false,
469
+ "train_param": {
470
+ "max_length": 1024,
471
+ "per_device_train_batch_size": 8,
472
+ "auto_find_batch_size": true,
473
+ "num_train_epochs": 5,
474
+ "max_steps": -1,
475
+ "logging_strategy": "epoch",
476
+ "save_strategy": "epoch",
477
+ "logging_steps": 0.2,
478
+ "save_steps": 0.2,
479
+ "gradient_accumulation_steps": 1,
480
+ "gradient_checkpointing": false,
481
+ "optim": "adamw_torch",
482
+ "learning_rate": 5e-05,
483
+ "weight_decay": 0.01,
484
+ "warmup_steps": 0,
485
+ "report_to": "tensorboard",
486
+ "save_safetensors": true,
487
+ "packing": false
488
+ },
489
+ "use_peft": true,
490
+ "peft_task_type": "CAUSAL_LM",
491
+ "peft_inference_mode": false,
492
+ "peft_r": 8,
493
+ "peft_lora_alpha": 32,
494
+ "peft_lora_dropout": 0.1,
495
+ "fsdp": "none",
496
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
497
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
498
+ "fsdp_fsdp_min_num_params": 100000000.0,
499
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
500
+ "fsdp_fsdp_forward_prefetch": false,
501
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
502
+ "fsdp_fsdp_offload_params": false,
503
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
504
+ "fsdp_fsdp_use_orig_params": true,
505
+ "fsdp_fsdp_activation_checkpointing": false,
506
+ "fsdp_fsdp_limit_all_gathers": true,
507
+ "fsdp_fsdp_sync_module_states": true,
508
+ "use_git": true,
509
+ "random_seed": 0
510
+ }
511
+ }
512
+ }
513
+ }
Qwen/Qwen2.5-14B/results_2026-01-05-04-27-50.json ADDED
@@ -0,0 +1,513 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "model_name": "Qwen/Qwen2.5-14B",
4
+ "model_type": "pretrained",
5
+ "params": 14.0
6
+ },
7
+ "results": {
8
+ "medmcqa": {
9
+ "acc_norm": 0.6564666507291418,
10
+ "acc_norm_stderr": 0.007343426826332978,
11
+ "args": {
12
+ "exp_name": "results/v1/medmcqa/Qwen/Qwen2.5-14B",
13
+ "task_name": "medmcqa",
14
+ "model_name": "Qwen/Qwen2.5-14B",
15
+ "dataset_param": {
16
+ "max_num_train": 50000,
17
+ "completion_only": false,
18
+ "apply_chat_template": false
19
+ },
20
+ "resume_from_checkpoint": false,
21
+ "train_param": {
22
+ "max_length": 1024,
23
+ "per_device_train_batch_size": 8,
24
+ "auto_find_batch_size": true,
25
+ "num_train_epochs": 1,
26
+ "max_steps": -1,
27
+ "logging_strategy": "steps",
28
+ "save_strategy": "steps",
29
+ "logging_steps": 0.2,
30
+ "save_steps": 0.2,
31
+ "gradient_accumulation_steps": 1,
32
+ "gradient_checkpointing": false,
33
+ "optim": "adamw_torch",
34
+ "learning_rate": 5e-05,
35
+ "weight_decay": 0.01,
36
+ "warmup_steps": 0,
37
+ "report_to": "tensorboard",
38
+ "save_safetensors": true,
39
+ "packing": false
40
+ },
41
+ "use_peft": true,
42
+ "peft_task_type": "CAUSAL_LM",
43
+ "peft_inference_mode": false,
44
+ "peft_r": 8,
45
+ "peft_lora_alpha": 32,
46
+ "peft_lora_dropout": 0.1,
47
+ "fsdp": "none",
48
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
49
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
50
+ "fsdp_fsdp_min_num_params": 100000000.0,
51
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
52
+ "fsdp_fsdp_forward_prefetch": false,
53
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
54
+ "fsdp_fsdp_offload_params": false,
55
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
56
+ "fsdp_fsdp_use_orig_params": true,
57
+ "fsdp_fsdp_activation_checkpointing": false,
58
+ "fsdp_fsdp_limit_all_gathers": true,
59
+ "fsdp_fsdp_sync_module_states": true,
60
+ "use_git": true,
61
+ "random_seed": 0
62
+ }
63
+ },
64
+ "nq_open": {
65
+ "exact_match,remove_whitespace": 0.3257617728531856,
66
+ "exact_match,remove_whitespace_stderr": 0.007801234000204628,
67
+ "args": {
68
+ "exp_name": "results/v1/nq_open/Qwen/Qwen2.5-14B",
69
+ "task_name": "nq_open",
70
+ "model_name": "Qwen/Qwen2.5-14B",
71
+ "dataset_param": {
72
+ "max_num_train": 50000,
73
+ "completion_only": false,
74
+ "apply_chat_template": false
75
+ },
76
+ "resume_from_checkpoint": false,
77
+ "train_param": {
78
+ "max_length": 1024,
79
+ "per_device_train_batch_size": 8,
80
+ "auto_find_batch_size": true,
81
+ "num_train_epochs": 1,
82
+ "max_steps": -1,
83
+ "logging_strategy": "steps",
84
+ "save_strategy": "steps",
85
+ "logging_steps": 0.2,
86
+ "save_steps": 0.2,
87
+ "gradient_accumulation_steps": 1,
88
+ "gradient_checkpointing": false,
89
+ "optim": "adamw_torch",
90
+ "learning_rate": 5e-05,
91
+ "weight_decay": 0.01,
92
+ "warmup_steps": 0,
93
+ "report_to": "tensorboard",
94
+ "save_safetensors": true,
95
+ "packing": false
96
+ },
97
+ "use_peft": true,
98
+ "peft_task_type": "CAUSAL_LM",
99
+ "peft_inference_mode": false,
100
+ "peft_r": 8,
101
+ "peft_lora_alpha": 32,
102
+ "peft_lora_dropout": 0.1,
103
+ "fsdp": "none",
104
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
105
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
106
+ "fsdp_fsdp_min_num_params": 100000000.0,
107
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
108
+ "fsdp_fsdp_forward_prefetch": false,
109
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
110
+ "fsdp_fsdp_offload_params": false,
111
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
112
+ "fsdp_fsdp_use_orig_params": true,
113
+ "fsdp_fsdp_activation_checkpointing": false,
114
+ "fsdp_fsdp_limit_all_gathers": true,
115
+ "fsdp_fsdp_sync_module_states": true,
116
+ "use_git": true,
117
+ "random_seed": 0
118
+ }
119
+ },
120
+ "winogrande": {
121
+ "acc": 0.8555643251775849,
122
+ "acc_stderr": 0.009879767358079246,
123
+ "args": {
124
+ "exp_name": "results/v1/winogrande/Qwen/Qwen2.5-14B",
125
+ "task_name": "winogrande",
126
+ "model_name": "Qwen/Qwen2.5-14B",
127
+ "dataset_param": {
128
+ "max_num_train": 50000,
129
+ "completion_only": false,
130
+ "apply_chat_template": false
131
+ },
132
+ "resume_from_checkpoint": false,
133
+ "train_param": {
134
+ "max_length": 1024,
135
+ "per_device_train_batch_size": 8,
136
+ "auto_find_batch_size": true,
137
+ "num_train_epochs": 1,
138
+ "max_steps": -1,
139
+ "logging_strategy": "steps",
140
+ "save_strategy": "steps",
141
+ "logging_steps": 0.2,
142
+ "save_steps": 0.2,
143
+ "gradient_accumulation_steps": 1,
144
+ "gradient_checkpointing": false,
145
+ "optim": "adamw_torch",
146
+ "learning_rate": 5e-05,
147
+ "weight_decay": 0.01,
148
+ "warmup_steps": 0,
149
+ "report_to": "tensorboard",
150
+ "save_safetensors": true,
151
+ "packing": false
152
+ },
153
+ "use_peft": true,
154
+ "peft_task_type": "CAUSAL_LM",
155
+ "peft_inference_mode": false,
156
+ "peft_r": 8,
157
+ "peft_lora_alpha": 32,
158
+ "peft_lora_dropout": 0.1,
159
+ "fsdp": "none",
160
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
161
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
162
+ "fsdp_fsdp_min_num_params": 100000000.0,
163
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
164
+ "fsdp_fsdp_forward_prefetch": false,
165
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
166
+ "fsdp_fsdp_offload_params": false,
167
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
168
+ "fsdp_fsdp_use_orig_params": true,
169
+ "fsdp_fsdp_activation_checkpointing": false,
170
+ "fsdp_fsdp_limit_all_gathers": true,
171
+ "fsdp_fsdp_sync_module_states": true,
172
+ "use_git": true,
173
+ "random_seed": 0
174
+ }
175
+ },
176
+ "hellaswag": {
177
+ "acc_norm": 0.8544,
178
+ "acc_norm_stderr": 0.0035272238560677465,
179
+ "args": {
180
+ "exp_name": "results/v1/hellaswag/Qwen/Qwen2.5-14B",
181
+ "task_name": "hellaswag",
182
+ "model_name": "Qwen/Qwen2.5-14B",
183
+ "dataset_param": {
184
+ "max_num_train": 50000,
185
+ "completion_only": false,
186
+ "apply_chat_template": false
187
+ },
188
+ "resume_from_checkpoint": false,
189
+ "train_param": {
190
+ "max_length": 1024,
191
+ "per_device_train_batch_size": 8,
192
+ "auto_find_batch_size": true,
193
+ "num_train_epochs": 1,
194
+ "max_steps": -1,
195
+ "logging_strategy": "steps",
196
+ "save_strategy": "steps",
197
+ "logging_steps": 0.2,
198
+ "save_steps": 0.2,
199
+ "gradient_accumulation_steps": 1,
200
+ "gradient_checkpointing": false,
201
+ "optim": "adamw_torch",
202
+ "learning_rate": 5e-05,
203
+ "weight_decay": 0.01,
204
+ "warmup_steps": 0,
205
+ "report_to": "tensorboard",
206
+ "save_safetensors": true,
207
+ "packing": false
208
+ },
209
+ "use_peft": true,
210
+ "peft_task_type": "CAUSAL_LM",
211
+ "peft_inference_mode": false,
212
+ "peft_r": 8,
213
+ "peft_lora_alpha": 32,
214
+ "peft_lora_dropout": 0.1,
215
+ "fsdp": "none",
216
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
217
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
218
+ "fsdp_fsdp_min_num_params": 100000000.0,
219
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
220
+ "fsdp_fsdp_forward_prefetch": false,
221
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
222
+ "fsdp_fsdp_offload_params": false,
223
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
224
+ "fsdp_fsdp_use_orig_params": true,
225
+ "fsdp_fsdp_activation_checkpointing": false,
226
+ "fsdp_fsdp_limit_all_gathers": true,
227
+ "fsdp_fsdp_sync_module_states": true,
228
+ "use_git": true,
229
+ "random_seed": 0
230
+ }
231
+ },
232
+ "social_iqa": {
233
+ "acc": 0.5757420675537359,
234
+ "acc_stderr": 0.011183502662341794,
235
+ "args": {
236
+ "exp_name": "results/v1/social_iqa/Qwen/Qwen2.5-14B",
237
+ "task_name": "social_iqa",
238
+ "model_name": "Qwen/Qwen2.5-14B",
239
+ "dataset_param": {
240
+ "max_num_train": 50000,
241
+ "completion_only": false,
242
+ "apply_chat_template": false
243
+ },
244
+ "resume_from_checkpoint": false,
245
+ "train_param": {
246
+ "max_length": 1024,
247
+ "per_device_train_batch_size": 8,
248
+ "auto_find_batch_size": true,
249
+ "num_train_epochs": 1,
250
+ "max_steps": -1,
251
+ "logging_strategy": "steps",
252
+ "save_strategy": "steps",
253
+ "logging_steps": 0.2,
254
+ "save_steps": 0.2,
255
+ "gradient_accumulation_steps": 1,
256
+ "gradient_checkpointing": false,
257
+ "optim": "adamw_torch",
258
+ "learning_rate": 5e-05,
259
+ "weight_decay": 0.01,
260
+ "warmup_steps": 0,
261
+ "report_to": "tensorboard",
262
+ "save_safetensors": true,
263
+ "packing": false
264
+ },
265
+ "use_peft": true,
266
+ "peft_task_type": "CAUSAL_LM",
267
+ "peft_inference_mode": false,
268
+ "peft_r": 8,
269
+ "peft_lora_alpha": 32,
270
+ "peft_lora_dropout": 0.1,
271
+ "fsdp": "none",
272
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
273
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
274
+ "fsdp_fsdp_min_num_params": 100000000.0,
275
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
276
+ "fsdp_fsdp_forward_prefetch": false,
277
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
278
+ "fsdp_fsdp_offload_params": false,
279
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
280
+ "fsdp_fsdp_use_orig_params": true,
281
+ "fsdp_fsdp_activation_checkpointing": false,
282
+ "fsdp_fsdp_limit_all_gathers": true,
283
+ "fsdp_fsdp_sync_module_states": true,
284
+ "use_git": true,
285
+ "random_seed": 0
286
+ }
287
+ },
288
+ "piqa": {
289
+ "acc_norm": 0.8411316648531012,
290
+ "acc_norm_stderr": 0.008528960151153363,
291
+ "args": {
292
+ "exp_name": "results/v2/piqa/Qwen/Qwen2.5-14B",
293
+ "task_name": "piqa",
294
+ "model_name": "Qwen/Qwen2.5-14B",
295
+ "dataset_param": {
296
+ "max_num_train": 50000,
297
+ "completion_only": false,
298
+ "apply_chat_template": false
299
+ },
300
+ "resume_from_checkpoint": false,
301
+ "train_param": {
302
+ "max_length": 1024,
303
+ "per_device_train_batch_size": 8,
304
+ "auto_find_batch_size": true,
305
+ "num_train_epochs": 5,
306
+ "max_steps": -1,
307
+ "logging_strategy": "epoch",
308
+ "save_strategy": "epoch",
309
+ "logging_steps": 0.2,
310
+ "save_steps": 0.2,
311
+ "gradient_accumulation_steps": 1,
312
+ "gradient_checkpointing": false,
313
+ "optim": "adamw_torch",
314
+ "learning_rate": 5e-05,
315
+ "weight_decay": 0.01,
316
+ "warmup_steps": 0,
317
+ "report_to": "tensorboard",
318
+ "save_safetensors": true,
319
+ "packing": false
320
+ },
321
+ "use_peft": true,
322
+ "peft_task_type": "CAUSAL_LM",
323
+ "peft_inference_mode": false,
324
+ "peft_r": 8,
325
+ "peft_lora_alpha": 32,
326
+ "peft_lora_dropout": 0.1,
327
+ "fsdp": "none",
328
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
329
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
330
+ "fsdp_fsdp_min_num_params": 100000000.0,
331
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
332
+ "fsdp_fsdp_forward_prefetch": false,
333
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
334
+ "fsdp_fsdp_offload_params": false,
335
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
336
+ "fsdp_fsdp_use_orig_params": true,
337
+ "fsdp_fsdp_activation_checkpointing": false,
338
+ "fsdp_fsdp_limit_all_gathers": true,
339
+ "fsdp_fsdp_sync_module_states": true,
340
+ "use_git": true,
341
+ "random_seed": 0
342
+ }
343
+ },
344
+ "commonsense_qa": {
345
+ "acc": 0.8755118755118755,
346
+ "acc_stderr": 0.009451811837305528,
347
+ "args": {
348
+ "exp_name": "results/v2/commonsense_qa/Qwen/Qwen2.5-14B",
349
+ "task_name": "commonsense_qa",
350
+ "model_name": "Qwen/Qwen2.5-14B",
351
+ "dataset_param": {
352
+ "max_num_train": 50000,
353
+ "completion_only": false,
354
+ "apply_chat_template": false
355
+ },
356
+ "resume_from_checkpoint": false,
357
+ "train_param": {
358
+ "max_length": 1024,
359
+ "per_device_train_batch_size": 8,
360
+ "auto_find_batch_size": true,
361
+ "num_train_epochs": 5,
362
+ "max_steps": -1,
363
+ "logging_strategy": "epoch",
364
+ "save_strategy": "epoch",
365
+ "logging_steps": 0.2,
366
+ "save_steps": 0.2,
367
+ "gradient_accumulation_steps": 1,
368
+ "gradient_checkpointing": false,
369
+ "optim": "adamw_torch",
370
+ "learning_rate": 5e-05,
371
+ "weight_decay": 0.01,
372
+ "warmup_steps": 0,
373
+ "report_to": "tensorboard",
374
+ "save_safetensors": true,
375
+ "packing": false
376
+ },
377
+ "use_peft": true,
378
+ "peft_task_type": "CAUSAL_LM",
379
+ "peft_inference_mode": false,
380
+ "peft_r": 8,
381
+ "peft_lora_alpha": 32,
382
+ "peft_lora_dropout": 0.1,
383
+ "fsdp": "none",
384
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
385
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
386
+ "fsdp_fsdp_min_num_params": 100000000.0,
387
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
388
+ "fsdp_fsdp_forward_prefetch": false,
389
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
390
+ "fsdp_fsdp_offload_params": false,
391
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
392
+ "fsdp_fsdp_use_orig_params": true,
393
+ "fsdp_fsdp_activation_checkpointing": false,
394
+ "fsdp_fsdp_limit_all_gathers": true,
395
+ "fsdp_fsdp_sync_module_states": true,
396
+ "use_git": true,
397
+ "random_seed": 0
398
+ }
399
+ },
400
+ "gsm8k": {
401
+ "exact_match,flexible-extract": 0.8430629264594389,
402
+ "exact_match,flexible-extract_stderr": 0.010019246595616161,
403
+ "args": {
404
+ "exp_name": "results/v2/gsm8k/Qwen/Qwen2.5-14B",
405
+ "task_name": "gsm8k",
406
+ "model_name": "Qwen/Qwen2.5-14B",
407
+ "dataset_param": {
408
+ "max_num_train": 50000,
409
+ "completion_only": false,
410
+ "apply_chat_template": false
411
+ },
412
+ "resume_from_checkpoint": false,
413
+ "train_param": {
414
+ "max_length": 1024,
415
+ "per_device_train_batch_size": 8,
416
+ "auto_find_batch_size": true,
417
+ "num_train_epochs": 5,
418
+ "max_steps": -1,
419
+ "logging_strategy": "epoch",
420
+ "save_strategy": "epoch",
421
+ "logging_steps": 0.2,
422
+ "save_steps": 0.2,
423
+ "gradient_accumulation_steps": 1,
424
+ "gradient_checkpointing": false,
425
+ "optim": "adamw_torch",
426
+ "learning_rate": 5e-05,
427
+ "weight_decay": 0.01,
428
+ "warmup_steps": 0,
429
+ "report_to": "tensorboard",
430
+ "save_safetensors": true,
431
+ "packing": false
432
+ },
433
+ "use_peft": true,
434
+ "peft_task_type": "CAUSAL_LM",
435
+ "peft_inference_mode": false,
436
+ "peft_r": 8,
437
+ "peft_lora_alpha": 32,
438
+ "peft_lora_dropout": 0.1,
439
+ "fsdp": "none",
440
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
441
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
442
+ "fsdp_fsdp_min_num_params": 100000000.0,
443
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
444
+ "fsdp_fsdp_forward_prefetch": false,
445
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
446
+ "fsdp_fsdp_offload_params": false,
447
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
448
+ "fsdp_fsdp_use_orig_params": true,
449
+ "fsdp_fsdp_activation_checkpointing": false,
450
+ "fsdp_fsdp_limit_all_gathers": true,
451
+ "fsdp_fsdp_sync_module_states": true,
452
+ "use_git": true,
453
+ "random_seed": 0
454
+ }
455
+ },
456
+ "arc_challenge": {
457
+ "acc_norm": 0.6825938566552902,
458
+ "acc_norm_stderr": 0.01360223908803817,
459
+ "args": {
460
+ "exp_name": "results/v2/arc_challenge/Qwen/Qwen2.5-14B",
461
+ "task_name": "arc_challenge",
462
+ "model_name": "Qwen/Qwen2.5-14B",
463
+ "dataset_param": {
464
+ "max_num_train": 50000,
465
+ "completion_only": false,
466
+ "apply_chat_template": false
467
+ },
468
+ "resume_from_checkpoint": false,
469
+ "train_param": {
470
+ "max_length": 1024,
471
+ "per_device_train_batch_size": 8,
472
+ "auto_find_batch_size": true,
473
+ "num_train_epochs": 5,
474
+ "max_steps": -1,
475
+ "logging_strategy": "epoch",
476
+ "save_strategy": "epoch",
477
+ "logging_steps": 0.2,
478
+ "save_steps": 0.2,
479
+ "gradient_accumulation_steps": 1,
480
+ "gradient_checkpointing": false,
481
+ "optim": "adamw_torch",
482
+ "learning_rate": 5e-05,
483
+ "weight_decay": 0.01,
484
+ "warmup_steps": 0,
485
+ "report_to": "tensorboard",
486
+ "save_safetensors": true,
487
+ "packing": false
488
+ },
489
+ "use_peft": true,
490
+ "peft_task_type": "CAUSAL_LM",
491
+ "peft_inference_mode": false,
492
+ "peft_r": 8,
493
+ "peft_lora_alpha": 32,
494
+ "peft_lora_dropout": 0.1,
495
+ "fsdp": "none",
496
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
497
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
498
+ "fsdp_fsdp_min_num_params": 100000000.0,
499
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
500
+ "fsdp_fsdp_forward_prefetch": false,
501
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
502
+ "fsdp_fsdp_offload_params": false,
503
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
504
+ "fsdp_fsdp_use_orig_params": true,
505
+ "fsdp_fsdp_activation_checkpointing": false,
506
+ "fsdp_fsdp_limit_all_gathers": true,
507
+ "fsdp_fsdp_sync_module_states": true,
508
+ "use_git": true,
509
+ "random_seed": 0
510
+ }
511
+ }
512
+ }
513
+ }
Qwen/Qwen2.5-3B-Instruct/results_2026-01-05-04-27-51.json ADDED
@@ -0,0 +1,513 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "model_name": "Qwen/Qwen2.5-3B-Instruct",
4
+ "model_type": "instruction-tuned",
5
+ "params": 3.0
6
+ },
7
+ "results": {
8
+ "medmcqa": {
9
+ "acc_norm": 0.544107100167344,
10
+ "acc_norm_stderr": 0.007701611309710366,
11
+ "args": {
12
+ "exp_name": "results/v1/medmcqa/Qwen/Qwen2.5-3B-Instruct",
13
+ "task_name": "medmcqa",
14
+ "model_name": "Qwen/Qwen2.5-3B-Instruct",
15
+ "dataset_param": {
16
+ "max_num_train": 50000,
17
+ "completion_only": false,
18
+ "apply_chat_template": true
19
+ },
20
+ "resume_from_checkpoint": false,
21
+ "train_param": {
22
+ "max_length": 1024,
23
+ "per_device_train_batch_size": 8,
24
+ "auto_find_batch_size": true,
25
+ "num_train_epochs": 1,
26
+ "max_steps": -1,
27
+ "logging_strategy": "steps",
28
+ "save_strategy": "steps",
29
+ "logging_steps": 0.2,
30
+ "save_steps": 0.2,
31
+ "gradient_accumulation_steps": 1,
32
+ "gradient_checkpointing": false,
33
+ "optim": "adamw_torch",
34
+ "learning_rate": 5e-05,
35
+ "weight_decay": 0.01,
36
+ "warmup_steps": 0,
37
+ "report_to": "tensorboard",
38
+ "save_safetensors": true,
39
+ "packing": false
40
+ },
41
+ "use_peft": true,
42
+ "peft_task_type": "CAUSAL_LM",
43
+ "peft_inference_mode": false,
44
+ "peft_r": 8,
45
+ "peft_lora_alpha": 32,
46
+ "peft_lora_dropout": 0.1,
47
+ "fsdp": "none",
48
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
49
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
50
+ "fsdp_fsdp_min_num_params": 100000000.0,
51
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
52
+ "fsdp_fsdp_forward_prefetch": false,
53
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
54
+ "fsdp_fsdp_offload_params": false,
55
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
56
+ "fsdp_fsdp_use_orig_params": true,
57
+ "fsdp_fsdp_activation_checkpointing": false,
58
+ "fsdp_fsdp_limit_all_gathers": true,
59
+ "fsdp_fsdp_sync_module_states": true,
60
+ "use_git": true,
61
+ "random_seed": 0
62
+ }
63
+ },
64
+ "nq_open": {
65
+ "exact_match,remove_whitespace": 0.1922437673130194,
66
+ "exact_match,remove_whitespace_stderr": 0.006559531955420279,
67
+ "args": {
68
+ "exp_name": "results/v1/nq_open/Qwen/Qwen2.5-3B-Instruct",
69
+ "task_name": "nq_open",
70
+ "model_name": "Qwen/Qwen2.5-3B-Instruct",
71
+ "dataset_param": {
72
+ "max_num_train": 50000,
73
+ "completion_only": false,
74
+ "apply_chat_template": true
75
+ },
76
+ "resume_from_checkpoint": false,
77
+ "train_param": {
78
+ "max_length": 1024,
79
+ "per_device_train_batch_size": 8,
80
+ "auto_find_batch_size": true,
81
+ "num_train_epochs": 1,
82
+ "max_steps": -1,
83
+ "logging_strategy": "steps",
84
+ "save_strategy": "steps",
85
+ "logging_steps": 0.2,
86
+ "save_steps": 0.2,
87
+ "gradient_accumulation_steps": 1,
88
+ "gradient_checkpointing": false,
89
+ "optim": "adamw_torch",
90
+ "learning_rate": 5e-05,
91
+ "weight_decay": 0.01,
92
+ "warmup_steps": 0,
93
+ "report_to": "tensorboard",
94
+ "save_safetensors": true,
95
+ "packing": false
96
+ },
97
+ "use_peft": true,
98
+ "peft_task_type": "CAUSAL_LM",
99
+ "peft_inference_mode": false,
100
+ "peft_r": 8,
101
+ "peft_lora_alpha": 32,
102
+ "peft_lora_dropout": 0.1,
103
+ "fsdp": "none",
104
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
105
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
106
+ "fsdp_fsdp_min_num_params": 100000000.0,
107
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
108
+ "fsdp_fsdp_forward_prefetch": false,
109
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
110
+ "fsdp_fsdp_offload_params": false,
111
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
112
+ "fsdp_fsdp_use_orig_params": true,
113
+ "fsdp_fsdp_activation_checkpointing": false,
114
+ "fsdp_fsdp_limit_all_gathers": true,
115
+ "fsdp_fsdp_sync_module_states": true,
116
+ "use_git": true,
117
+ "random_seed": 0
118
+ }
119
+ },
120
+ "winogrande": {
121
+ "acc": 0.7466456195737964,
122
+ "acc_stderr": 0.012223754434233626,
123
+ "args": {
124
+ "exp_name": "results/v1/winogrande/Qwen/Qwen2.5-3B-Instruct",
125
+ "task_name": "winogrande",
126
+ "model_name": "Qwen/Qwen2.5-3B-Instruct",
127
+ "dataset_param": {
128
+ "max_num_train": 50000,
129
+ "completion_only": false,
130
+ "apply_chat_template": true
131
+ },
132
+ "resume_from_checkpoint": false,
133
+ "train_param": {
134
+ "max_length": 1024,
135
+ "per_device_train_batch_size": 8,
136
+ "auto_find_batch_size": true,
137
+ "num_train_epochs": 1,
138
+ "max_steps": -1,
139
+ "logging_strategy": "steps",
140
+ "save_strategy": "steps",
141
+ "logging_steps": 0.2,
142
+ "save_steps": 0.2,
143
+ "gradient_accumulation_steps": 1,
144
+ "gradient_checkpointing": false,
145
+ "optim": "adamw_torch",
146
+ "learning_rate": 5e-05,
147
+ "weight_decay": 0.01,
148
+ "warmup_steps": 0,
149
+ "report_to": "tensorboard",
150
+ "save_safetensors": true,
151
+ "packing": false
152
+ },
153
+ "use_peft": true,
154
+ "peft_task_type": "CAUSAL_LM",
155
+ "peft_inference_mode": false,
156
+ "peft_r": 8,
157
+ "peft_lora_alpha": 32,
158
+ "peft_lora_dropout": 0.1,
159
+ "fsdp": "none",
160
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
161
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
162
+ "fsdp_fsdp_min_num_params": 100000000.0,
163
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
164
+ "fsdp_fsdp_forward_prefetch": false,
165
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
166
+ "fsdp_fsdp_offload_params": false,
167
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
168
+ "fsdp_fsdp_use_orig_params": true,
169
+ "fsdp_fsdp_activation_checkpointing": false,
170
+ "fsdp_fsdp_limit_all_gathers": true,
171
+ "fsdp_fsdp_sync_module_states": true,
172
+ "use_git": true,
173
+ "random_seed": 0
174
+ }
175
+ },
176
+ "hellaswag": {
177
+ "acc_norm": 0.7554,
178
+ "acc_norm_stderr": 0.004298712818180255,
179
+ "args": {
180
+ "exp_name": "results/v1/hellaswag/Qwen/Qwen2.5-3B-Instruct",
181
+ "task_name": "hellaswag",
182
+ "model_name": "Qwen/Qwen2.5-3B-Instruct",
183
+ "dataset_param": {
184
+ "max_num_train": 50000,
185
+ "completion_only": false,
186
+ "apply_chat_template": true
187
+ },
188
+ "resume_from_checkpoint": false,
189
+ "train_param": {
190
+ "max_length": 1024,
191
+ "per_device_train_batch_size": 8,
192
+ "auto_find_batch_size": true,
193
+ "num_train_epochs": 1,
194
+ "max_steps": -1,
195
+ "logging_strategy": "steps",
196
+ "save_strategy": "steps",
197
+ "logging_steps": 0.2,
198
+ "save_steps": 0.2,
199
+ "gradient_accumulation_steps": 1,
200
+ "gradient_checkpointing": false,
201
+ "optim": "adamw_torch",
202
+ "learning_rate": 5e-05,
203
+ "weight_decay": 0.01,
204
+ "warmup_steps": 0,
205
+ "report_to": "tensorboard",
206
+ "save_safetensors": true,
207
+ "packing": false
208
+ },
209
+ "use_peft": true,
210
+ "peft_task_type": "CAUSAL_LM",
211
+ "peft_inference_mode": false,
212
+ "peft_r": 8,
213
+ "peft_lora_alpha": 32,
214
+ "peft_lora_dropout": 0.1,
215
+ "fsdp": "none",
216
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
217
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
218
+ "fsdp_fsdp_min_num_params": 100000000.0,
219
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
220
+ "fsdp_fsdp_forward_prefetch": false,
221
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
222
+ "fsdp_fsdp_offload_params": false,
223
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
224
+ "fsdp_fsdp_use_orig_params": true,
225
+ "fsdp_fsdp_activation_checkpointing": false,
226
+ "fsdp_fsdp_limit_all_gathers": true,
227
+ "fsdp_fsdp_sync_module_states": true,
228
+ "use_git": true,
229
+ "random_seed": 0
230
+ }
231
+ },
232
+ "social_iqa": {
233
+ "acc": 0.5358239508700102,
234
+ "acc_stderr": 0.011284993242142344,
235
+ "args": {
236
+ "exp_name": "results/v1/social_iqa/Qwen/Qwen2.5-3B-Instruct",
237
+ "task_name": "social_iqa",
238
+ "model_name": "Qwen/Qwen2.5-3B-Instruct",
239
+ "dataset_param": {
240
+ "max_num_train": 50000,
241
+ "completion_only": false,
242
+ "apply_chat_template": true
243
+ },
244
+ "resume_from_checkpoint": false,
245
+ "train_param": {
246
+ "max_length": 1024,
247
+ "per_device_train_batch_size": 8,
248
+ "auto_find_batch_size": true,
249
+ "num_train_epochs": 1,
250
+ "max_steps": -1,
251
+ "logging_strategy": "steps",
252
+ "save_strategy": "steps",
253
+ "logging_steps": 0.2,
254
+ "save_steps": 0.2,
255
+ "gradient_accumulation_steps": 1,
256
+ "gradient_checkpointing": false,
257
+ "optim": "adamw_torch",
258
+ "learning_rate": 5e-05,
259
+ "weight_decay": 0.01,
260
+ "warmup_steps": 0,
261
+ "report_to": "tensorboard",
262
+ "save_safetensors": true,
263
+ "packing": false
264
+ },
265
+ "use_peft": true,
266
+ "peft_task_type": "CAUSAL_LM",
267
+ "peft_inference_mode": false,
268
+ "peft_r": 8,
269
+ "peft_lora_alpha": 32,
270
+ "peft_lora_dropout": 0.1,
271
+ "fsdp": "none",
272
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
273
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
274
+ "fsdp_fsdp_min_num_params": 100000000.0,
275
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
276
+ "fsdp_fsdp_forward_prefetch": false,
277
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
278
+ "fsdp_fsdp_offload_params": false,
279
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
280
+ "fsdp_fsdp_use_orig_params": true,
281
+ "fsdp_fsdp_activation_checkpointing": false,
282
+ "fsdp_fsdp_limit_all_gathers": true,
283
+ "fsdp_fsdp_sync_module_states": true,
284
+ "use_git": true,
285
+ "random_seed": 0
286
+ }
287
+ },
288
+ "piqa": {
289
+ "acc_norm": 0.7905331882480957,
290
+ "acc_norm_stderr": 0.009494302979819801,
291
+ "args": {
292
+ "exp_name": "results/v2/piqa/Qwen/Qwen2.5-3B-Instruct",
293
+ "task_name": "piqa",
294
+ "model_name": "Qwen/Qwen2.5-3B-Instruct",
295
+ "dataset_param": {
296
+ "max_num_train": 50000,
297
+ "completion_only": false,
298
+ "apply_chat_template": true
299
+ },
300
+ "resume_from_checkpoint": false,
301
+ "train_param": {
302
+ "max_length": 1024,
303
+ "per_device_train_batch_size": 8,
304
+ "auto_find_batch_size": true,
305
+ "num_train_epochs": 5,
306
+ "max_steps": -1,
307
+ "logging_strategy": "epoch",
308
+ "save_strategy": "epoch",
309
+ "logging_steps": 0.2,
310
+ "save_steps": 0.2,
311
+ "gradient_accumulation_steps": 1,
312
+ "gradient_checkpointing": false,
313
+ "optim": "adamw_torch",
314
+ "learning_rate": 5e-05,
315
+ "weight_decay": 0.01,
316
+ "warmup_steps": 0,
317
+ "report_to": "tensorboard",
318
+ "save_safetensors": true,
319
+ "packing": false
320
+ },
321
+ "use_peft": true,
322
+ "peft_task_type": "CAUSAL_LM",
323
+ "peft_inference_mode": false,
324
+ "peft_r": 8,
325
+ "peft_lora_alpha": 32,
326
+ "peft_lora_dropout": 0.1,
327
+ "fsdp": "none",
328
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
329
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
330
+ "fsdp_fsdp_min_num_params": 100000000.0,
331
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
332
+ "fsdp_fsdp_forward_prefetch": false,
333
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
334
+ "fsdp_fsdp_offload_params": false,
335
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
336
+ "fsdp_fsdp_use_orig_params": true,
337
+ "fsdp_fsdp_activation_checkpointing": false,
338
+ "fsdp_fsdp_limit_all_gathers": true,
339
+ "fsdp_fsdp_sync_module_states": true,
340
+ "use_git": true,
341
+ "random_seed": 0
342
+ }
343
+ },
344
+ "commonsense_qa": {
345
+ "acc": 0.814086814086814,
346
+ "acc_stderr": 0.011138085349810693,
347
+ "args": {
348
+ "exp_name": "results/v2/commonsense_qa/Qwen/Qwen2.5-3B-Instruct",
349
+ "task_name": "commonsense_qa",
350
+ "model_name": "Qwen/Qwen2.5-3B-Instruct",
351
+ "dataset_param": {
352
+ "max_num_train": 50000,
353
+ "completion_only": false,
354
+ "apply_chat_template": true
355
+ },
356
+ "resume_from_checkpoint": false,
357
+ "train_param": {
358
+ "max_length": 1024,
359
+ "per_device_train_batch_size": 8,
360
+ "auto_find_batch_size": true,
361
+ "num_train_epochs": 5,
362
+ "max_steps": -1,
363
+ "logging_strategy": "epoch",
364
+ "save_strategy": "epoch",
365
+ "logging_steps": 0.2,
366
+ "save_steps": 0.2,
367
+ "gradient_accumulation_steps": 1,
368
+ "gradient_checkpointing": false,
369
+ "optim": "adamw_torch",
370
+ "learning_rate": 5e-05,
371
+ "weight_decay": 0.01,
372
+ "warmup_steps": 0,
373
+ "report_to": "tensorboard",
374
+ "save_safetensors": true,
375
+ "packing": false
376
+ },
377
+ "use_peft": true,
378
+ "peft_task_type": "CAUSAL_LM",
379
+ "peft_inference_mode": false,
380
+ "peft_r": 8,
381
+ "peft_lora_alpha": 32,
382
+ "peft_lora_dropout": 0.1,
383
+ "fsdp": "none",
384
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
385
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
386
+ "fsdp_fsdp_min_num_params": 100000000.0,
387
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
388
+ "fsdp_fsdp_forward_prefetch": false,
389
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
390
+ "fsdp_fsdp_offload_params": false,
391
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
392
+ "fsdp_fsdp_use_orig_params": true,
393
+ "fsdp_fsdp_activation_checkpointing": false,
394
+ "fsdp_fsdp_limit_all_gathers": true,
395
+ "fsdp_fsdp_sync_module_states": true,
396
+ "use_git": true,
397
+ "random_seed": 0
398
+ }
399
+ },
400
+ "gsm8k": {
401
+ "exact_match,flexible-extract": 0.6588324488248674,
402
+ "exact_match,flexible-extract_stderr": 0.013059111935831494,
403
+ "args": {
404
+ "exp_name": "results/v2/gsm8k/Qwen/Qwen2.5-3B-Instruct",
405
+ "task_name": "gsm8k",
406
+ "model_name": "Qwen/Qwen2.5-3B-Instruct",
407
+ "dataset_param": {
408
+ "max_num_train": 50000,
409
+ "completion_only": false,
410
+ "apply_chat_template": true
411
+ },
412
+ "resume_from_checkpoint": false,
413
+ "train_param": {
414
+ "max_length": 1024,
415
+ "per_device_train_batch_size": 8,
416
+ "auto_find_batch_size": true,
417
+ "num_train_epochs": 5,
418
+ "max_steps": -1,
419
+ "logging_strategy": "epoch",
420
+ "save_strategy": "epoch",
421
+ "logging_steps": 0.2,
422
+ "save_steps": 0.2,
423
+ "gradient_accumulation_steps": 1,
424
+ "gradient_checkpointing": false,
425
+ "optim": "adamw_torch",
426
+ "learning_rate": 5e-05,
427
+ "weight_decay": 0.01,
428
+ "warmup_steps": 0,
429
+ "report_to": "tensorboard",
430
+ "save_safetensors": true,
431
+ "packing": false
432
+ },
433
+ "use_peft": true,
434
+ "peft_task_type": "CAUSAL_LM",
435
+ "peft_inference_mode": false,
436
+ "peft_r": 8,
437
+ "peft_lora_alpha": 32,
438
+ "peft_lora_dropout": 0.1,
439
+ "fsdp": "none",
440
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
441
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
442
+ "fsdp_fsdp_min_num_params": 100000000.0,
443
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
444
+ "fsdp_fsdp_forward_prefetch": false,
445
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
446
+ "fsdp_fsdp_offload_params": false,
447
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
448
+ "fsdp_fsdp_use_orig_params": true,
449
+ "fsdp_fsdp_activation_checkpointing": false,
450
+ "fsdp_fsdp_limit_all_gathers": true,
451
+ "fsdp_fsdp_sync_module_states": true,
452
+ "use_git": true,
453
+ "random_seed": 0
454
+ }
455
+ },
456
+ "arc_challenge": {
457
+ "acc_norm": 0.5093856655290102,
458
+ "acc_norm_stderr": 0.014608816322065003,
459
+ "args": {
460
+ "exp_name": "results/v2/arc_challenge/Qwen/Qwen2.5-3B-Instruct",
461
+ "task_name": "arc_challenge",
462
+ "model_name": "Qwen/Qwen2.5-3B-Instruct",
463
+ "dataset_param": {
464
+ "max_num_train": 50000,
465
+ "completion_only": false,
466
+ "apply_chat_template": true
467
+ },
468
+ "resume_from_checkpoint": false,
469
+ "train_param": {
470
+ "max_length": 1024,
471
+ "per_device_train_batch_size": 8,
472
+ "auto_find_batch_size": true,
473
+ "num_train_epochs": 5,
474
+ "max_steps": -1,
475
+ "logging_strategy": "epoch",
476
+ "save_strategy": "epoch",
477
+ "logging_steps": 0.2,
478
+ "save_steps": 0.2,
479
+ "gradient_accumulation_steps": 1,
480
+ "gradient_checkpointing": false,
481
+ "optim": "adamw_torch",
482
+ "learning_rate": 5e-05,
483
+ "weight_decay": 0.01,
484
+ "warmup_steps": 0,
485
+ "report_to": "tensorboard",
486
+ "save_safetensors": true,
487
+ "packing": false
488
+ },
489
+ "use_peft": true,
490
+ "peft_task_type": "CAUSAL_LM",
491
+ "peft_inference_mode": false,
492
+ "peft_r": 8,
493
+ "peft_lora_alpha": 32,
494
+ "peft_lora_dropout": 0.1,
495
+ "fsdp": "none",
496
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
497
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
498
+ "fsdp_fsdp_min_num_params": 100000000.0,
499
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
500
+ "fsdp_fsdp_forward_prefetch": false,
501
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
502
+ "fsdp_fsdp_offload_params": false,
503
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
504
+ "fsdp_fsdp_use_orig_params": true,
505
+ "fsdp_fsdp_activation_checkpointing": false,
506
+ "fsdp_fsdp_limit_all_gathers": true,
507
+ "fsdp_fsdp_sync_module_states": true,
508
+ "use_git": true,
509
+ "random_seed": 0
510
+ }
511
+ }
512
+ }
513
+ }
Qwen/Qwen2.5-3B/results_2026-01-05-04-27-50.json ADDED
@@ -0,0 +1,513 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "model_name": "Qwen/Qwen2.5-3B",
4
+ "model_type": "pretrained",
5
+ "params": 3.0
6
+ },
7
+ "results": {
8
+ "medmcqa": {
9
+ "acc_norm": 0.5460196031556299,
10
+ "acc_norm_stderr": 0.007698935005654058,
11
+ "args": {
12
+ "exp_name": "results/v1/medmcqa/Qwen/Qwen2.5-3B",
13
+ "task_name": "medmcqa",
14
+ "model_name": "Qwen/Qwen2.5-3B",
15
+ "dataset_param": {
16
+ "max_num_train": 50000,
17
+ "completion_only": false,
18
+ "apply_chat_template": false
19
+ },
20
+ "resume_from_checkpoint": false,
21
+ "train_param": {
22
+ "max_length": 1024,
23
+ "per_device_train_batch_size": 8,
24
+ "auto_find_batch_size": true,
25
+ "num_train_epochs": 1,
26
+ "max_steps": -1,
27
+ "logging_strategy": "steps",
28
+ "save_strategy": "steps",
29
+ "logging_steps": 0.2,
30
+ "save_steps": 0.2,
31
+ "gradient_accumulation_steps": 1,
32
+ "gradient_checkpointing": false,
33
+ "optim": "adamw_torch",
34
+ "learning_rate": 5e-05,
35
+ "weight_decay": 0.01,
36
+ "warmup_steps": 0,
37
+ "report_to": "tensorboard",
38
+ "save_safetensors": true,
39
+ "packing": false
40
+ },
41
+ "use_peft": true,
42
+ "peft_task_type": "CAUSAL_LM",
43
+ "peft_inference_mode": false,
44
+ "peft_r": 8,
45
+ "peft_lora_alpha": 32,
46
+ "peft_lora_dropout": 0.1,
47
+ "fsdp": "none",
48
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
49
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
50
+ "fsdp_fsdp_min_num_params": 100000000.0,
51
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
52
+ "fsdp_fsdp_forward_prefetch": false,
53
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
54
+ "fsdp_fsdp_offload_params": false,
55
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
56
+ "fsdp_fsdp_use_orig_params": true,
57
+ "fsdp_fsdp_activation_checkpointing": false,
58
+ "fsdp_fsdp_limit_all_gathers": true,
59
+ "fsdp_fsdp_sync_module_states": true,
60
+ "use_git": true,
61
+ "random_seed": 0
62
+ }
63
+ },
64
+ "nq_open": {
65
+ "exact_match,remove_whitespace": 0.20221606648199447,
66
+ "exact_match,remove_whitespace_stderr": 0.006685856192186142,
67
+ "args": {
68
+ "exp_name": "results/v1/nq_open/Qwen/Qwen2.5-3B",
69
+ "task_name": "nq_open",
70
+ "model_name": "Qwen/Qwen2.5-3B",
71
+ "dataset_param": {
72
+ "max_num_train": 50000,
73
+ "completion_only": false,
74
+ "apply_chat_template": false
75
+ },
76
+ "resume_from_checkpoint": false,
77
+ "train_param": {
78
+ "max_length": 1024,
79
+ "per_device_train_batch_size": 8,
80
+ "auto_find_batch_size": true,
81
+ "num_train_epochs": 1,
82
+ "max_steps": -1,
83
+ "logging_strategy": "steps",
84
+ "save_strategy": "steps",
85
+ "logging_steps": 0.2,
86
+ "save_steps": 0.2,
87
+ "gradient_accumulation_steps": 1,
88
+ "gradient_checkpointing": false,
89
+ "optim": "adamw_torch",
90
+ "learning_rate": 5e-05,
91
+ "weight_decay": 0.01,
92
+ "warmup_steps": 0,
93
+ "report_to": "tensorboard",
94
+ "save_safetensors": true,
95
+ "packing": false
96
+ },
97
+ "use_peft": true,
98
+ "peft_task_type": "CAUSAL_LM",
99
+ "peft_inference_mode": false,
100
+ "peft_r": 8,
101
+ "peft_lora_alpha": 32,
102
+ "peft_lora_dropout": 0.1,
103
+ "fsdp": "none",
104
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
105
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
106
+ "fsdp_fsdp_min_num_params": 100000000.0,
107
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
108
+ "fsdp_fsdp_forward_prefetch": false,
109
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
110
+ "fsdp_fsdp_offload_params": false,
111
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
112
+ "fsdp_fsdp_use_orig_params": true,
113
+ "fsdp_fsdp_activation_checkpointing": false,
114
+ "fsdp_fsdp_limit_all_gathers": true,
115
+ "fsdp_fsdp_sync_module_states": true,
116
+ "use_git": true,
117
+ "random_seed": 0
118
+ }
119
+ },
120
+ "winogrande": {
121
+ "acc": 0.7458563535911602,
122
+ "acc_stderr": 0.012236307219708264,
123
+ "args": {
124
+ "exp_name": "results/v1/winogrande/Qwen/Qwen2.5-3B",
125
+ "task_name": "winogrande",
126
+ "model_name": "Qwen/Qwen2.5-3B",
127
+ "dataset_param": {
128
+ "max_num_train": 50000,
129
+ "completion_only": false,
130
+ "apply_chat_template": false
131
+ },
132
+ "resume_from_checkpoint": false,
133
+ "train_param": {
134
+ "max_length": 1024,
135
+ "per_device_train_batch_size": 8,
136
+ "auto_find_batch_size": true,
137
+ "num_train_epochs": 1,
138
+ "max_steps": -1,
139
+ "logging_strategy": "steps",
140
+ "save_strategy": "steps",
141
+ "logging_steps": 0.2,
142
+ "save_steps": 0.2,
143
+ "gradient_accumulation_steps": 1,
144
+ "gradient_checkpointing": false,
145
+ "optim": "adamw_torch",
146
+ "learning_rate": 5e-05,
147
+ "weight_decay": 0.01,
148
+ "warmup_steps": 0,
149
+ "report_to": "tensorboard",
150
+ "save_safetensors": true,
151
+ "packing": false
152
+ },
153
+ "use_peft": true,
154
+ "peft_task_type": "CAUSAL_LM",
155
+ "peft_inference_mode": false,
156
+ "peft_r": 8,
157
+ "peft_lora_alpha": 32,
158
+ "peft_lora_dropout": 0.1,
159
+ "fsdp": "none",
160
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
161
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
162
+ "fsdp_fsdp_min_num_params": 100000000.0,
163
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
164
+ "fsdp_fsdp_forward_prefetch": false,
165
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
166
+ "fsdp_fsdp_offload_params": false,
167
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
168
+ "fsdp_fsdp_use_orig_params": true,
169
+ "fsdp_fsdp_activation_checkpointing": false,
170
+ "fsdp_fsdp_limit_all_gathers": true,
171
+ "fsdp_fsdp_sync_module_states": true,
172
+ "use_git": true,
173
+ "random_seed": 0
174
+ }
175
+ },
176
+ "hellaswag": {
177
+ "acc_norm": 0.7593,
178
+ "acc_norm_stderr": 0.004275298681715799,
179
+ "args": {
180
+ "exp_name": "results/v1/hellaswag/Qwen/Qwen2.5-3B",
181
+ "task_name": "hellaswag",
182
+ "model_name": "Qwen/Qwen2.5-3B",
183
+ "dataset_param": {
184
+ "max_num_train": 50000,
185
+ "completion_only": false,
186
+ "apply_chat_template": false
187
+ },
188
+ "resume_from_checkpoint": false,
189
+ "train_param": {
190
+ "max_length": 1024,
191
+ "per_device_train_batch_size": 8,
192
+ "auto_find_batch_size": true,
193
+ "num_train_epochs": 1,
194
+ "max_steps": -1,
195
+ "logging_strategy": "steps",
196
+ "save_strategy": "steps",
197
+ "logging_steps": 0.2,
198
+ "save_steps": 0.2,
199
+ "gradient_accumulation_steps": 1,
200
+ "gradient_checkpointing": false,
201
+ "optim": "adamw_torch",
202
+ "learning_rate": 5e-05,
203
+ "weight_decay": 0.01,
204
+ "warmup_steps": 0,
205
+ "report_to": "tensorboard",
206
+ "save_safetensors": true,
207
+ "packing": false
208
+ },
209
+ "use_peft": true,
210
+ "peft_task_type": "CAUSAL_LM",
211
+ "peft_inference_mode": false,
212
+ "peft_r": 8,
213
+ "peft_lora_alpha": 32,
214
+ "peft_lora_dropout": 0.1,
215
+ "fsdp": "none",
216
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
217
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
218
+ "fsdp_fsdp_min_num_params": 100000000.0,
219
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
220
+ "fsdp_fsdp_forward_prefetch": false,
221
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
222
+ "fsdp_fsdp_offload_params": false,
223
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
224
+ "fsdp_fsdp_use_orig_params": true,
225
+ "fsdp_fsdp_activation_checkpointing": false,
226
+ "fsdp_fsdp_limit_all_gathers": true,
227
+ "fsdp_fsdp_sync_module_states": true,
228
+ "use_git": true,
229
+ "random_seed": 0
230
+ }
231
+ },
232
+ "social_iqa": {
233
+ "acc": 0.5373592630501536,
234
+ "acc_stderr": 0.011282443926045702,
235
+ "args": {
236
+ "exp_name": "results/v1/social_iqa/Qwen/Qwen2.5-3B",
237
+ "task_name": "social_iqa",
238
+ "model_name": "Qwen/Qwen2.5-3B",
239
+ "dataset_param": {
240
+ "max_num_train": 50000,
241
+ "completion_only": false,
242
+ "apply_chat_template": false
243
+ },
244
+ "resume_from_checkpoint": false,
245
+ "train_param": {
246
+ "max_length": 1024,
247
+ "per_device_train_batch_size": 8,
248
+ "auto_find_batch_size": true,
249
+ "num_train_epochs": 1,
250
+ "max_steps": -1,
251
+ "logging_strategy": "steps",
252
+ "save_strategy": "steps",
253
+ "logging_steps": 0.2,
254
+ "save_steps": 0.2,
255
+ "gradient_accumulation_steps": 1,
256
+ "gradient_checkpointing": false,
257
+ "optim": "adamw_torch",
258
+ "learning_rate": 5e-05,
259
+ "weight_decay": 0.01,
260
+ "warmup_steps": 0,
261
+ "report_to": "tensorboard",
262
+ "save_safetensors": true,
263
+ "packing": false
264
+ },
265
+ "use_peft": true,
266
+ "peft_task_type": "CAUSAL_LM",
267
+ "peft_inference_mode": false,
268
+ "peft_r": 8,
269
+ "peft_lora_alpha": 32,
270
+ "peft_lora_dropout": 0.1,
271
+ "fsdp": "none",
272
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
273
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
274
+ "fsdp_fsdp_min_num_params": 100000000.0,
275
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
276
+ "fsdp_fsdp_forward_prefetch": false,
277
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
278
+ "fsdp_fsdp_offload_params": false,
279
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
280
+ "fsdp_fsdp_use_orig_params": true,
281
+ "fsdp_fsdp_activation_checkpointing": false,
282
+ "fsdp_fsdp_limit_all_gathers": true,
283
+ "fsdp_fsdp_sync_module_states": true,
284
+ "use_git": true,
285
+ "random_seed": 0
286
+ }
287
+ },
288
+ "piqa": {
289
+ "acc_norm": 0.7965179542981502,
290
+ "acc_norm_stderr": 0.009393041784049923,
291
+ "args": {
292
+ "exp_name": "results/v2/piqa/Qwen/Qwen2.5-3B",
293
+ "task_name": "piqa",
294
+ "model_name": "Qwen/Qwen2.5-3B",
295
+ "dataset_param": {
296
+ "max_num_train": 50000,
297
+ "completion_only": false,
298
+ "apply_chat_template": false
299
+ },
300
+ "resume_from_checkpoint": false,
301
+ "train_param": {
302
+ "max_length": 1024,
303
+ "per_device_train_batch_size": 8,
304
+ "auto_find_batch_size": true,
305
+ "num_train_epochs": 5,
306
+ "max_steps": -1,
307
+ "logging_strategy": "epoch",
308
+ "save_strategy": "epoch",
309
+ "logging_steps": 0.2,
310
+ "save_steps": 0.2,
311
+ "gradient_accumulation_steps": 1,
312
+ "gradient_checkpointing": false,
313
+ "optim": "adamw_torch",
314
+ "learning_rate": 5e-05,
315
+ "weight_decay": 0.01,
316
+ "warmup_steps": 0,
317
+ "report_to": "tensorboard",
318
+ "save_safetensors": true,
319
+ "packing": false
320
+ },
321
+ "use_peft": true,
322
+ "peft_task_type": "CAUSAL_LM",
323
+ "peft_inference_mode": false,
324
+ "peft_r": 8,
325
+ "peft_lora_alpha": 32,
326
+ "peft_lora_dropout": 0.1,
327
+ "fsdp": "none",
328
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
329
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
330
+ "fsdp_fsdp_min_num_params": 100000000.0,
331
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
332
+ "fsdp_fsdp_forward_prefetch": false,
333
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
334
+ "fsdp_fsdp_offload_params": false,
335
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
336
+ "fsdp_fsdp_use_orig_params": true,
337
+ "fsdp_fsdp_activation_checkpointing": false,
338
+ "fsdp_fsdp_limit_all_gathers": true,
339
+ "fsdp_fsdp_sync_module_states": true,
340
+ "use_git": true,
341
+ "random_seed": 0
342
+ }
343
+ },
344
+ "commonsense_qa": {
345
+ "acc": 0.8157248157248157,
346
+ "acc_stderr": 0.011100060579347633,
347
+ "args": {
348
+ "exp_name": "results/v2/commonsense_qa/Qwen/Qwen2.5-3B",
349
+ "task_name": "commonsense_qa",
350
+ "model_name": "Qwen/Qwen2.5-3B",
351
+ "dataset_param": {
352
+ "max_num_train": 50000,
353
+ "completion_only": false,
354
+ "apply_chat_template": false
355
+ },
356
+ "resume_from_checkpoint": false,
357
+ "train_param": {
358
+ "max_length": 1024,
359
+ "per_device_train_batch_size": 8,
360
+ "auto_find_batch_size": true,
361
+ "num_train_epochs": 5,
362
+ "max_steps": -1,
363
+ "logging_strategy": "epoch",
364
+ "save_strategy": "epoch",
365
+ "logging_steps": 0.2,
366
+ "save_steps": 0.2,
367
+ "gradient_accumulation_steps": 1,
368
+ "gradient_checkpointing": false,
369
+ "optim": "adamw_torch",
370
+ "learning_rate": 5e-05,
371
+ "weight_decay": 0.01,
372
+ "warmup_steps": 0,
373
+ "report_to": "tensorboard",
374
+ "save_safetensors": true,
375
+ "packing": false
376
+ },
377
+ "use_peft": true,
378
+ "peft_task_type": "CAUSAL_LM",
379
+ "peft_inference_mode": false,
380
+ "peft_r": 8,
381
+ "peft_lora_alpha": 32,
382
+ "peft_lora_dropout": 0.1,
383
+ "fsdp": "none",
384
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
385
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
386
+ "fsdp_fsdp_min_num_params": 100000000.0,
387
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
388
+ "fsdp_fsdp_forward_prefetch": false,
389
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
390
+ "fsdp_fsdp_offload_params": false,
391
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
392
+ "fsdp_fsdp_use_orig_params": true,
393
+ "fsdp_fsdp_activation_checkpointing": false,
394
+ "fsdp_fsdp_limit_all_gathers": true,
395
+ "fsdp_fsdp_sync_module_states": true,
396
+ "use_git": true,
397
+ "random_seed": 0
398
+ }
399
+ },
400
+ "gsm8k": {
401
+ "exact_match,flexible-extract": 0.690674753601213,
402
+ "exact_match,flexible-extract_stderr": 0.012731710925078138,
403
+ "args": {
404
+ "exp_name": "results/v2/gsm8k/Qwen/Qwen2.5-3B",
405
+ "task_name": "gsm8k",
406
+ "model_name": "Qwen/Qwen2.5-3B",
407
+ "dataset_param": {
408
+ "max_num_train": 50000,
409
+ "completion_only": false,
410
+ "apply_chat_template": false
411
+ },
412
+ "resume_from_checkpoint": false,
413
+ "train_param": {
414
+ "max_length": 1024,
415
+ "per_device_train_batch_size": 8,
416
+ "auto_find_batch_size": true,
417
+ "num_train_epochs": 5,
418
+ "max_steps": -1,
419
+ "logging_strategy": "epoch",
420
+ "save_strategy": "epoch",
421
+ "logging_steps": 0.2,
422
+ "save_steps": 0.2,
423
+ "gradient_accumulation_steps": 1,
424
+ "gradient_checkpointing": false,
425
+ "optim": "adamw_torch",
426
+ "learning_rate": 5e-05,
427
+ "weight_decay": 0.01,
428
+ "warmup_steps": 0,
429
+ "report_to": "tensorboard",
430
+ "save_safetensors": true,
431
+ "packing": false
432
+ },
433
+ "use_peft": true,
434
+ "peft_task_type": "CAUSAL_LM",
435
+ "peft_inference_mode": false,
436
+ "peft_r": 8,
437
+ "peft_lora_alpha": 32,
438
+ "peft_lora_dropout": 0.1,
439
+ "fsdp": "none",
440
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
441
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
442
+ "fsdp_fsdp_min_num_params": 100000000.0,
443
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
444
+ "fsdp_fsdp_forward_prefetch": false,
445
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
446
+ "fsdp_fsdp_offload_params": false,
447
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
448
+ "fsdp_fsdp_use_orig_params": true,
449
+ "fsdp_fsdp_activation_checkpointing": false,
450
+ "fsdp_fsdp_limit_all_gathers": true,
451
+ "fsdp_fsdp_sync_module_states": true,
452
+ "use_git": true,
453
+ "random_seed": 0
454
+ }
455
+ },
456
+ "arc_challenge": {
457
+ "acc_norm": 0.5537542662116041,
458
+ "acc_norm_stderr": 0.014526705548539982,
459
+ "args": {
460
+ "exp_name": "results/v2/arc_challenge/Qwen/Qwen2.5-3B",
461
+ "task_name": "arc_challenge",
462
+ "model_name": "Qwen/Qwen2.5-3B",
463
+ "dataset_param": {
464
+ "max_num_train": 50000,
465
+ "completion_only": false,
466
+ "apply_chat_template": false
467
+ },
468
+ "resume_from_checkpoint": false,
469
+ "train_param": {
470
+ "max_length": 1024,
471
+ "per_device_train_batch_size": 8,
472
+ "auto_find_batch_size": true,
473
+ "num_train_epochs": 5,
474
+ "max_steps": -1,
475
+ "logging_strategy": "epoch",
476
+ "save_strategy": "epoch",
477
+ "logging_steps": 0.2,
478
+ "save_steps": 0.2,
479
+ "gradient_accumulation_steps": 1,
480
+ "gradient_checkpointing": false,
481
+ "optim": "adamw_torch",
482
+ "learning_rate": 5e-05,
483
+ "weight_decay": 0.01,
484
+ "warmup_steps": 0,
485
+ "report_to": "tensorboard",
486
+ "save_safetensors": true,
487
+ "packing": false
488
+ },
489
+ "use_peft": true,
490
+ "peft_task_type": "CAUSAL_LM",
491
+ "peft_inference_mode": false,
492
+ "peft_r": 8,
493
+ "peft_lora_alpha": 32,
494
+ "peft_lora_dropout": 0.1,
495
+ "fsdp": "none",
496
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
497
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
498
+ "fsdp_fsdp_min_num_params": 100000000.0,
499
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
500
+ "fsdp_fsdp_forward_prefetch": false,
501
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
502
+ "fsdp_fsdp_offload_params": false,
503
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
504
+ "fsdp_fsdp_use_orig_params": true,
505
+ "fsdp_fsdp_activation_checkpointing": false,
506
+ "fsdp_fsdp_limit_all_gathers": true,
507
+ "fsdp_fsdp_sync_module_states": true,
508
+ "use_git": true,
509
+ "random_seed": 0
510
+ }
511
+ }
512
+ }
513
+ }
Qwen/Qwen2.5-7B-Instruct/results_2026-01-05-04-27-51.json ADDED
@@ -0,0 +1,513 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "model_name": "Qwen/Qwen2.5-7B-Instruct",
4
+ "model_type": "instruction-tuned",
5
+ "params": 7.0
6
+ },
7
+ "results": {
8
+ "medmcqa": {
9
+ "acc_norm": 0.6117618933779584,
10
+ "acc_norm_stderr": 0.007536128178745705,
11
+ "args": {
12
+ "exp_name": "results/v1/medmcqa/Qwen/Qwen2.5-7B-Instruct",
13
+ "task_name": "medmcqa",
14
+ "model_name": "Qwen/Qwen2.5-7B-Instruct",
15
+ "dataset_param": {
16
+ "max_num_train": 50000,
17
+ "completion_only": false,
18
+ "apply_chat_template": true
19
+ },
20
+ "resume_from_checkpoint": false,
21
+ "train_param": {
22
+ "max_length": 1024,
23
+ "per_device_train_batch_size": 8,
24
+ "auto_find_batch_size": true,
25
+ "num_train_epochs": 1,
26
+ "max_steps": -1,
27
+ "logging_strategy": "steps",
28
+ "save_strategy": "steps",
29
+ "logging_steps": 0.2,
30
+ "save_steps": 0.2,
31
+ "gradient_accumulation_steps": 1,
32
+ "gradient_checkpointing": false,
33
+ "optim": "adamw_torch",
34
+ "learning_rate": 5e-05,
35
+ "weight_decay": 0.01,
36
+ "warmup_steps": 0,
37
+ "report_to": "tensorboard",
38
+ "save_safetensors": true,
39
+ "packing": false
40
+ },
41
+ "use_peft": true,
42
+ "peft_task_type": "CAUSAL_LM",
43
+ "peft_inference_mode": false,
44
+ "peft_r": 8,
45
+ "peft_lora_alpha": 32,
46
+ "peft_lora_dropout": 0.1,
47
+ "fsdp": "none",
48
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
49
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
50
+ "fsdp_fsdp_min_num_params": 100000000.0,
51
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
52
+ "fsdp_fsdp_forward_prefetch": false,
53
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
54
+ "fsdp_fsdp_offload_params": false,
55
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
56
+ "fsdp_fsdp_use_orig_params": true,
57
+ "fsdp_fsdp_activation_checkpointing": false,
58
+ "fsdp_fsdp_limit_all_gathers": true,
59
+ "fsdp_fsdp_sync_module_states": true,
60
+ "use_git": true,
61
+ "random_seed": 0
62
+ }
63
+ },
64
+ "nq_open": {
65
+ "exact_match,remove_whitespace": 0.2515235457063712,
66
+ "exact_match,remove_whitespace_stderr": 0.007222456813452498,
67
+ "args": {
68
+ "exp_name": "results/v1/nq_open/Qwen/Qwen2.5-7B-Instruct",
69
+ "task_name": "nq_open",
70
+ "model_name": "Qwen/Qwen2.5-7B-Instruct",
71
+ "dataset_param": {
72
+ "max_num_train": 50000,
73
+ "completion_only": false,
74
+ "apply_chat_template": true
75
+ },
76
+ "resume_from_checkpoint": false,
77
+ "train_param": {
78
+ "max_length": 1024,
79
+ "per_device_train_batch_size": 8,
80
+ "auto_find_batch_size": true,
81
+ "num_train_epochs": 1,
82
+ "max_steps": -1,
83
+ "logging_strategy": "steps",
84
+ "save_strategy": "steps",
85
+ "logging_steps": 0.2,
86
+ "save_steps": 0.2,
87
+ "gradient_accumulation_steps": 1,
88
+ "gradient_checkpointing": false,
89
+ "optim": "adamw_torch",
90
+ "learning_rate": 5e-05,
91
+ "weight_decay": 0.01,
92
+ "warmup_steps": 0,
93
+ "report_to": "tensorboard",
94
+ "save_safetensors": true,
95
+ "packing": false
96
+ },
97
+ "use_peft": true,
98
+ "peft_task_type": "CAUSAL_LM",
99
+ "peft_inference_mode": false,
100
+ "peft_r": 8,
101
+ "peft_lora_alpha": 32,
102
+ "peft_lora_dropout": 0.1,
103
+ "fsdp": "none",
104
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
105
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
106
+ "fsdp_fsdp_min_num_params": 100000000.0,
107
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
108
+ "fsdp_fsdp_forward_prefetch": false,
109
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
110
+ "fsdp_fsdp_offload_params": false,
111
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
112
+ "fsdp_fsdp_use_orig_params": true,
113
+ "fsdp_fsdp_activation_checkpointing": false,
114
+ "fsdp_fsdp_limit_all_gathers": true,
115
+ "fsdp_fsdp_sync_module_states": true,
116
+ "use_git": true,
117
+ "random_seed": 0
118
+ }
119
+ },
120
+ "winogrande": {
121
+ "acc": 0.8113654301499605,
122
+ "acc_stderr": 0.010995172318019792,
123
+ "args": {
124
+ "exp_name": "results/v1/winogrande/Qwen/Qwen2.5-7B-Instruct",
125
+ "task_name": "winogrande",
126
+ "model_name": "Qwen/Qwen2.5-7B-Instruct",
127
+ "dataset_param": {
128
+ "max_num_train": 50000,
129
+ "completion_only": false,
130
+ "apply_chat_template": true
131
+ },
132
+ "resume_from_checkpoint": false,
133
+ "train_param": {
134
+ "max_length": 1024,
135
+ "per_device_train_batch_size": 8,
136
+ "auto_find_batch_size": true,
137
+ "num_train_epochs": 1,
138
+ "max_steps": -1,
139
+ "logging_strategy": "steps",
140
+ "save_strategy": "steps",
141
+ "logging_steps": 0.2,
142
+ "save_steps": 0.2,
143
+ "gradient_accumulation_steps": 1,
144
+ "gradient_checkpointing": false,
145
+ "optim": "adamw_torch",
146
+ "learning_rate": 5e-05,
147
+ "weight_decay": 0.01,
148
+ "warmup_steps": 0,
149
+ "report_to": "tensorboard",
150
+ "save_safetensors": true,
151
+ "packing": false
152
+ },
153
+ "use_peft": true,
154
+ "peft_task_type": "CAUSAL_LM",
155
+ "peft_inference_mode": false,
156
+ "peft_r": 8,
157
+ "peft_lora_alpha": 32,
158
+ "peft_lora_dropout": 0.1,
159
+ "fsdp": "none",
160
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
161
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
162
+ "fsdp_fsdp_min_num_params": 100000000.0,
163
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
164
+ "fsdp_fsdp_forward_prefetch": false,
165
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
166
+ "fsdp_fsdp_offload_params": false,
167
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
168
+ "fsdp_fsdp_use_orig_params": true,
169
+ "fsdp_fsdp_activation_checkpointing": false,
170
+ "fsdp_fsdp_limit_all_gathers": true,
171
+ "fsdp_fsdp_sync_module_states": true,
172
+ "use_git": true,
173
+ "random_seed": 0
174
+ }
175
+ },
176
+ "hellaswag": {
177
+ "acc_norm": 0.8099,
178
+ "acc_norm_stderr": 0.003923995256607442,
179
+ "args": {
180
+ "exp_name": "results/v1/hellaswag/Qwen/Qwen2.5-7B-Instruct",
181
+ "task_name": "hellaswag",
182
+ "model_name": "Qwen/Qwen2.5-7B-Instruct",
183
+ "dataset_param": {
184
+ "max_num_train": 50000,
185
+ "completion_only": false,
186
+ "apply_chat_template": true
187
+ },
188
+ "resume_from_checkpoint": false,
189
+ "train_param": {
190
+ "max_length": 1024,
191
+ "per_device_train_batch_size": 8,
192
+ "auto_find_batch_size": true,
193
+ "num_train_epochs": 1,
194
+ "max_steps": -1,
195
+ "logging_strategy": "steps",
196
+ "save_strategy": "steps",
197
+ "logging_steps": 0.2,
198
+ "save_steps": 0.2,
199
+ "gradient_accumulation_steps": 1,
200
+ "gradient_checkpointing": false,
201
+ "optim": "adamw_torch",
202
+ "learning_rate": 5e-05,
203
+ "weight_decay": 0.01,
204
+ "warmup_steps": 0,
205
+ "report_to": "tensorboard",
206
+ "save_safetensors": true,
207
+ "packing": false
208
+ },
209
+ "use_peft": true,
210
+ "peft_task_type": "CAUSAL_LM",
211
+ "peft_inference_mode": false,
212
+ "peft_r": 8,
213
+ "peft_lora_alpha": 32,
214
+ "peft_lora_dropout": 0.1,
215
+ "fsdp": "none",
216
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
217
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
218
+ "fsdp_fsdp_min_num_params": 100000000.0,
219
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
220
+ "fsdp_fsdp_forward_prefetch": false,
221
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
222
+ "fsdp_fsdp_offload_params": false,
223
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
224
+ "fsdp_fsdp_use_orig_params": true,
225
+ "fsdp_fsdp_activation_checkpointing": false,
226
+ "fsdp_fsdp_limit_all_gathers": true,
227
+ "fsdp_fsdp_sync_module_states": true,
228
+ "use_git": true,
229
+ "random_seed": 0
230
+ }
231
+ },
232
+ "social_iqa": {
233
+ "acc": 0.5609007164790174,
234
+ "acc_stderr": 0.011229831502847811,
235
+ "args": {
236
+ "exp_name": "results/v1/social_iqa/Qwen/Qwen2.5-7B-Instruct",
237
+ "task_name": "social_iqa",
238
+ "model_name": "Qwen/Qwen2.5-7B-Instruct",
239
+ "dataset_param": {
240
+ "max_num_train": 50000,
241
+ "completion_only": false,
242
+ "apply_chat_template": true
243
+ },
244
+ "resume_from_checkpoint": false,
245
+ "train_param": {
246
+ "max_length": 1024,
247
+ "per_device_train_batch_size": 8,
248
+ "auto_find_batch_size": true,
249
+ "num_train_epochs": 1,
250
+ "max_steps": -1,
251
+ "logging_strategy": "steps",
252
+ "save_strategy": "steps",
253
+ "logging_steps": 0.2,
254
+ "save_steps": 0.2,
255
+ "gradient_accumulation_steps": 1,
256
+ "gradient_checkpointing": false,
257
+ "optim": "adamw_torch",
258
+ "learning_rate": 5e-05,
259
+ "weight_decay": 0.01,
260
+ "warmup_steps": 0,
261
+ "report_to": "tensorboard",
262
+ "save_safetensors": true,
263
+ "packing": false
264
+ },
265
+ "use_peft": true,
266
+ "peft_task_type": "CAUSAL_LM",
267
+ "peft_inference_mode": false,
268
+ "peft_r": 8,
269
+ "peft_lora_alpha": 32,
270
+ "peft_lora_dropout": 0.1,
271
+ "fsdp": "none",
272
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
273
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
274
+ "fsdp_fsdp_min_num_params": 100000000.0,
275
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
276
+ "fsdp_fsdp_forward_prefetch": false,
277
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
278
+ "fsdp_fsdp_offload_params": false,
279
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
280
+ "fsdp_fsdp_use_orig_params": true,
281
+ "fsdp_fsdp_activation_checkpointing": false,
282
+ "fsdp_fsdp_limit_all_gathers": true,
283
+ "fsdp_fsdp_sync_module_states": true,
284
+ "use_git": true,
285
+ "random_seed": 0
286
+ }
287
+ },
288
+ "piqa": {
289
+ "acc_norm": 0.8166485310119695,
290
+ "acc_norm_stderr": 0.00902828398468943,
291
+ "args": {
292
+ "exp_name": "results/v2/piqa/Qwen/Qwen2.5-7B-Instruct",
293
+ "task_name": "piqa",
294
+ "model_name": "Qwen/Qwen2.5-7B-Instruct",
295
+ "dataset_param": {
296
+ "max_num_train": 50000,
297
+ "completion_only": false,
298
+ "apply_chat_template": true
299
+ },
300
+ "resume_from_checkpoint": false,
301
+ "train_param": {
302
+ "max_length": 1024,
303
+ "per_device_train_batch_size": 8,
304
+ "auto_find_batch_size": true,
305
+ "num_train_epochs": 5,
306
+ "max_steps": -1,
307
+ "logging_strategy": "epoch",
308
+ "save_strategy": "epoch",
309
+ "logging_steps": 0.2,
310
+ "save_steps": 0.2,
311
+ "gradient_accumulation_steps": 1,
312
+ "gradient_checkpointing": false,
313
+ "optim": "adamw_torch",
314
+ "learning_rate": 5e-05,
315
+ "weight_decay": 0.01,
316
+ "warmup_steps": 0,
317
+ "report_to": "tensorboard",
318
+ "save_safetensors": true,
319
+ "packing": false
320
+ },
321
+ "use_peft": true,
322
+ "peft_task_type": "CAUSAL_LM",
323
+ "peft_inference_mode": false,
324
+ "peft_r": 8,
325
+ "peft_lora_alpha": 32,
326
+ "peft_lora_dropout": 0.1,
327
+ "fsdp": "none",
328
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
329
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
330
+ "fsdp_fsdp_min_num_params": 100000000.0,
331
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
332
+ "fsdp_fsdp_forward_prefetch": false,
333
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
334
+ "fsdp_fsdp_offload_params": false,
335
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
336
+ "fsdp_fsdp_use_orig_params": true,
337
+ "fsdp_fsdp_activation_checkpointing": false,
338
+ "fsdp_fsdp_limit_all_gathers": true,
339
+ "fsdp_fsdp_sync_module_states": true,
340
+ "use_git": true,
341
+ "random_seed": 0
342
+ }
343
+ },
344
+ "commonsense_qa": {
345
+ "acc": 0.8624078624078624,
346
+ "acc_stderr": 0.009862187320355239,
347
+ "args": {
348
+ "exp_name": "results/v2/commonsense_qa/Qwen/Qwen2.5-7B-Instruct",
349
+ "task_name": "commonsense_qa",
350
+ "model_name": "Qwen/Qwen2.5-7B-Instruct",
351
+ "dataset_param": {
352
+ "max_num_train": 50000,
353
+ "completion_only": false,
354
+ "apply_chat_template": true
355
+ },
356
+ "resume_from_checkpoint": false,
357
+ "train_param": {
358
+ "max_length": 1024,
359
+ "per_device_train_batch_size": 8,
360
+ "auto_find_batch_size": true,
361
+ "num_train_epochs": 5,
362
+ "max_steps": -1,
363
+ "logging_strategy": "epoch",
364
+ "save_strategy": "epoch",
365
+ "logging_steps": 0.2,
366
+ "save_steps": 0.2,
367
+ "gradient_accumulation_steps": 1,
368
+ "gradient_checkpointing": false,
369
+ "optim": "adamw_torch",
370
+ "learning_rate": 5e-05,
371
+ "weight_decay": 0.01,
372
+ "warmup_steps": 0,
373
+ "report_to": "tensorboard",
374
+ "save_safetensors": true,
375
+ "packing": false
376
+ },
377
+ "use_peft": true,
378
+ "peft_task_type": "CAUSAL_LM",
379
+ "peft_inference_mode": false,
380
+ "peft_r": 8,
381
+ "peft_lora_alpha": 32,
382
+ "peft_lora_dropout": 0.1,
383
+ "fsdp": "none",
384
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
385
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
386
+ "fsdp_fsdp_min_num_params": 100000000.0,
387
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
388
+ "fsdp_fsdp_forward_prefetch": false,
389
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
390
+ "fsdp_fsdp_offload_params": false,
391
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
392
+ "fsdp_fsdp_use_orig_params": true,
393
+ "fsdp_fsdp_activation_checkpointing": false,
394
+ "fsdp_fsdp_limit_all_gathers": true,
395
+ "fsdp_fsdp_sync_module_states": true,
396
+ "use_git": true,
397
+ "random_seed": 0
398
+ }
399
+ },
400
+ "gsm8k": {
401
+ "exact_match,flexible-extract": 0.7460197119029568,
402
+ "exact_match,flexible-extract_stderr": 0.011989952209548075,
403
+ "args": {
404
+ "exp_name": "results/v2/gsm8k/Qwen/Qwen2.5-7B-Instruct",
405
+ "task_name": "gsm8k",
406
+ "model_name": "Qwen/Qwen2.5-7B-Instruct",
407
+ "dataset_param": {
408
+ "max_num_train": 50000,
409
+ "completion_only": false,
410
+ "apply_chat_template": true
411
+ },
412
+ "resume_from_checkpoint": false,
413
+ "train_param": {
414
+ "max_length": 1024,
415
+ "per_device_train_batch_size": 8,
416
+ "auto_find_batch_size": true,
417
+ "num_train_epochs": 5,
418
+ "max_steps": -1,
419
+ "logging_strategy": "epoch",
420
+ "save_strategy": "epoch",
421
+ "logging_steps": 0.2,
422
+ "save_steps": 0.2,
423
+ "gradient_accumulation_steps": 1,
424
+ "gradient_checkpointing": false,
425
+ "optim": "adamw_torch",
426
+ "learning_rate": 5e-05,
427
+ "weight_decay": 0.01,
428
+ "warmup_steps": 0,
429
+ "report_to": "tensorboard",
430
+ "save_safetensors": true,
431
+ "packing": false
432
+ },
433
+ "use_peft": true,
434
+ "peft_task_type": "CAUSAL_LM",
435
+ "peft_inference_mode": false,
436
+ "peft_r": 8,
437
+ "peft_lora_alpha": 32,
438
+ "peft_lora_dropout": 0.1,
439
+ "fsdp": "none",
440
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
441
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
442
+ "fsdp_fsdp_min_num_params": 100000000.0,
443
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
444
+ "fsdp_fsdp_forward_prefetch": false,
445
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
446
+ "fsdp_fsdp_offload_params": false,
447
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
448
+ "fsdp_fsdp_use_orig_params": true,
449
+ "fsdp_fsdp_activation_checkpointing": false,
450
+ "fsdp_fsdp_limit_all_gathers": true,
451
+ "fsdp_fsdp_sync_module_states": true,
452
+ "use_git": true,
453
+ "random_seed": 0
454
+ }
455
+ },
456
+ "arc_challenge": {
457
+ "acc_norm": 0.613481228668942,
458
+ "acc_norm_stderr": 0.014230084761910481,
459
+ "args": {
460
+ "exp_name": "results/v2/arc_challenge/Qwen/Qwen2.5-7B-Instruct",
461
+ "task_name": "arc_challenge",
462
+ "model_name": "Qwen/Qwen2.5-7B-Instruct",
463
+ "dataset_param": {
464
+ "max_num_train": 50000,
465
+ "completion_only": false,
466
+ "apply_chat_template": true
467
+ },
468
+ "resume_from_checkpoint": false,
469
+ "train_param": {
470
+ "max_length": 1024,
471
+ "per_device_train_batch_size": 8,
472
+ "auto_find_batch_size": true,
473
+ "num_train_epochs": 5,
474
+ "max_steps": -1,
475
+ "logging_strategy": "epoch",
476
+ "save_strategy": "epoch",
477
+ "logging_steps": 0.2,
478
+ "save_steps": 0.2,
479
+ "gradient_accumulation_steps": 1,
480
+ "gradient_checkpointing": false,
481
+ "optim": "adamw_torch",
482
+ "learning_rate": 5e-05,
483
+ "weight_decay": 0.01,
484
+ "warmup_steps": 0,
485
+ "report_to": "tensorboard",
486
+ "save_safetensors": true,
487
+ "packing": false
488
+ },
489
+ "use_peft": true,
490
+ "peft_task_type": "CAUSAL_LM",
491
+ "peft_inference_mode": false,
492
+ "peft_r": 8,
493
+ "peft_lora_alpha": 32,
494
+ "peft_lora_dropout": 0.1,
495
+ "fsdp": "none",
496
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
497
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
498
+ "fsdp_fsdp_min_num_params": 100000000.0,
499
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
500
+ "fsdp_fsdp_forward_prefetch": false,
501
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
502
+ "fsdp_fsdp_offload_params": false,
503
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
504
+ "fsdp_fsdp_use_orig_params": true,
505
+ "fsdp_fsdp_activation_checkpointing": false,
506
+ "fsdp_fsdp_limit_all_gathers": true,
507
+ "fsdp_fsdp_sync_module_states": true,
508
+ "use_git": true,
509
+ "random_seed": 0
510
+ }
511
+ }
512
+ }
513
+ }
Qwen/Qwen2.5-7B/results_2026-01-05-04-27-50.json ADDED
@@ -0,0 +1,513 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "model_name": "Qwen/Qwen2.5-7B",
4
+ "model_type": "pretrained",
5
+ "params": 7.0
6
+ },
7
+ "results": {
8
+ "medmcqa": {
9
+ "acc_norm": 0.609371264642601,
10
+ "acc_norm_stderr": 0.007544510460245957,
11
+ "args": {
12
+ "exp_name": "results/v1/medmcqa/Qwen/Qwen2.5-7B",
13
+ "task_name": "medmcqa",
14
+ "model_name": "Qwen/Qwen2.5-7B",
15
+ "dataset_param": {
16
+ "max_num_train": 50000,
17
+ "completion_only": false,
18
+ "apply_chat_template": false
19
+ },
20
+ "resume_from_checkpoint": false,
21
+ "train_param": {
22
+ "max_length": 1024,
23
+ "per_device_train_batch_size": 8,
24
+ "auto_find_batch_size": true,
25
+ "num_train_epochs": 1,
26
+ "max_steps": -1,
27
+ "logging_strategy": "steps",
28
+ "save_strategy": "steps",
29
+ "logging_steps": 0.2,
30
+ "save_steps": 0.2,
31
+ "gradient_accumulation_steps": 1,
32
+ "gradient_checkpointing": false,
33
+ "optim": "adamw_torch",
34
+ "learning_rate": 5e-05,
35
+ "weight_decay": 0.01,
36
+ "warmup_steps": 0,
37
+ "report_to": "tensorboard",
38
+ "save_safetensors": true,
39
+ "packing": false
40
+ },
41
+ "use_peft": true,
42
+ "peft_task_type": "CAUSAL_LM",
43
+ "peft_inference_mode": false,
44
+ "peft_r": 8,
45
+ "peft_lora_alpha": 32,
46
+ "peft_lora_dropout": 0.1,
47
+ "fsdp": "none",
48
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
49
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
50
+ "fsdp_fsdp_min_num_params": 100000000.0,
51
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
52
+ "fsdp_fsdp_forward_prefetch": false,
53
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
54
+ "fsdp_fsdp_offload_params": false,
55
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
56
+ "fsdp_fsdp_use_orig_params": true,
57
+ "fsdp_fsdp_activation_checkpointing": false,
58
+ "fsdp_fsdp_limit_all_gathers": true,
59
+ "fsdp_fsdp_sync_module_states": true,
60
+ "use_git": true,
61
+ "random_seed": 0
62
+ }
63
+ },
64
+ "nq_open": {
65
+ "exact_match,remove_whitespace": 0.2581717451523546,
66
+ "exact_match,remove_whitespace_stderr": 0.0072847155711420015,
67
+ "args": {
68
+ "exp_name": "results/v1/nq_open/Qwen/Qwen2.5-7B",
69
+ "task_name": "nq_open",
70
+ "model_name": "Qwen/Qwen2.5-7B",
71
+ "dataset_param": {
72
+ "max_num_train": 50000,
73
+ "completion_only": false,
74
+ "apply_chat_template": false
75
+ },
76
+ "resume_from_checkpoint": false,
77
+ "train_param": {
78
+ "max_length": 1024,
79
+ "per_device_train_batch_size": 8,
80
+ "auto_find_batch_size": true,
81
+ "num_train_epochs": 1,
82
+ "max_steps": -1,
83
+ "logging_strategy": "steps",
84
+ "save_strategy": "steps",
85
+ "logging_steps": 0.2,
86
+ "save_steps": 0.2,
87
+ "gradient_accumulation_steps": 1,
88
+ "gradient_checkpointing": false,
89
+ "optim": "adamw_torch",
90
+ "learning_rate": 5e-05,
91
+ "weight_decay": 0.01,
92
+ "warmup_steps": 0,
93
+ "report_to": "tensorboard",
94
+ "save_safetensors": true,
95
+ "packing": false
96
+ },
97
+ "use_peft": true,
98
+ "peft_task_type": "CAUSAL_LM",
99
+ "peft_inference_mode": false,
100
+ "peft_r": 8,
101
+ "peft_lora_alpha": 32,
102
+ "peft_lora_dropout": 0.1,
103
+ "fsdp": "none",
104
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
105
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
106
+ "fsdp_fsdp_min_num_params": 100000000.0,
107
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
108
+ "fsdp_fsdp_forward_prefetch": false,
109
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
110
+ "fsdp_fsdp_offload_params": false,
111
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
112
+ "fsdp_fsdp_use_orig_params": true,
113
+ "fsdp_fsdp_activation_checkpointing": false,
114
+ "fsdp_fsdp_limit_all_gathers": true,
115
+ "fsdp_fsdp_sync_module_states": true,
116
+ "use_git": true,
117
+ "random_seed": 0
118
+ }
119
+ },
120
+ "winogrande": {
121
+ "acc": 0.8113654301499605,
122
+ "acc_stderr": 0.010995172318019785,
123
+ "args": {
124
+ "exp_name": "results/v1/winogrande/Qwen/Qwen2.5-7B",
125
+ "task_name": "winogrande",
126
+ "model_name": "Qwen/Qwen2.5-7B",
127
+ "dataset_param": {
128
+ "max_num_train": 50000,
129
+ "completion_only": false,
130
+ "apply_chat_template": false
131
+ },
132
+ "resume_from_checkpoint": false,
133
+ "train_param": {
134
+ "max_length": 1024,
135
+ "per_device_train_batch_size": 8,
136
+ "auto_find_batch_size": true,
137
+ "num_train_epochs": 1,
138
+ "max_steps": -1,
139
+ "logging_strategy": "steps",
140
+ "save_strategy": "steps",
141
+ "logging_steps": 0.2,
142
+ "save_steps": 0.2,
143
+ "gradient_accumulation_steps": 1,
144
+ "gradient_checkpointing": false,
145
+ "optim": "adamw_torch",
146
+ "learning_rate": 5e-05,
147
+ "weight_decay": 0.01,
148
+ "warmup_steps": 0,
149
+ "report_to": "tensorboard",
150
+ "save_safetensors": true,
151
+ "packing": false
152
+ },
153
+ "use_peft": true,
154
+ "peft_task_type": "CAUSAL_LM",
155
+ "peft_inference_mode": false,
156
+ "peft_r": 8,
157
+ "peft_lora_alpha": 32,
158
+ "peft_lora_dropout": 0.1,
159
+ "fsdp": "none",
160
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
161
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
162
+ "fsdp_fsdp_min_num_params": 100000000.0,
163
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
164
+ "fsdp_fsdp_forward_prefetch": false,
165
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
166
+ "fsdp_fsdp_offload_params": false,
167
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
168
+ "fsdp_fsdp_use_orig_params": true,
169
+ "fsdp_fsdp_activation_checkpointing": false,
170
+ "fsdp_fsdp_limit_all_gathers": true,
171
+ "fsdp_fsdp_sync_module_states": true,
172
+ "use_git": true,
173
+ "random_seed": 0
174
+ }
175
+ },
176
+ "hellaswag": {
177
+ "acc_norm": 0.8116,
178
+ "acc_norm_stderr": 0.0039105080497703025,
179
+ "args": {
180
+ "exp_name": "results/v1/hellaswag/Qwen/Qwen2.5-7B",
181
+ "task_name": "hellaswag",
182
+ "model_name": "Qwen/Qwen2.5-7B",
183
+ "dataset_param": {
184
+ "max_num_train": 50000,
185
+ "completion_only": false,
186
+ "apply_chat_template": false
187
+ },
188
+ "resume_from_checkpoint": false,
189
+ "train_param": {
190
+ "max_length": 1024,
191
+ "per_device_train_batch_size": 8,
192
+ "auto_find_batch_size": true,
193
+ "num_train_epochs": 1,
194
+ "max_steps": -1,
195
+ "logging_strategy": "steps",
196
+ "save_strategy": "steps",
197
+ "logging_steps": 0.2,
198
+ "save_steps": 0.2,
199
+ "gradient_accumulation_steps": 1,
200
+ "gradient_checkpointing": false,
201
+ "optim": "adamw_torch",
202
+ "learning_rate": 5e-05,
203
+ "weight_decay": 0.01,
204
+ "warmup_steps": 0,
205
+ "report_to": "tensorboard",
206
+ "save_safetensors": true,
207
+ "packing": false
208
+ },
209
+ "use_peft": true,
210
+ "peft_task_type": "CAUSAL_LM",
211
+ "peft_inference_mode": false,
212
+ "peft_r": 8,
213
+ "peft_lora_alpha": 32,
214
+ "peft_lora_dropout": 0.1,
215
+ "fsdp": "none",
216
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
217
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
218
+ "fsdp_fsdp_min_num_params": 100000000.0,
219
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
220
+ "fsdp_fsdp_forward_prefetch": false,
221
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
222
+ "fsdp_fsdp_offload_params": false,
223
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
224
+ "fsdp_fsdp_use_orig_params": true,
225
+ "fsdp_fsdp_activation_checkpointing": false,
226
+ "fsdp_fsdp_limit_all_gathers": true,
227
+ "fsdp_fsdp_sync_module_states": true,
228
+ "use_git": true,
229
+ "random_seed": 0
230
+ }
231
+ },
232
+ "social_iqa": {
233
+ "acc": 0.5696008188331627,
234
+ "acc_stderr": 0.011203917417496398,
235
+ "args": {
236
+ "exp_name": "results/v1/social_iqa/Qwen/Qwen2.5-7B",
237
+ "task_name": "social_iqa",
238
+ "model_name": "Qwen/Qwen2.5-7B",
239
+ "dataset_param": {
240
+ "max_num_train": 50000,
241
+ "completion_only": false,
242
+ "apply_chat_template": false
243
+ },
244
+ "resume_from_checkpoint": false,
245
+ "train_param": {
246
+ "max_length": 1024,
247
+ "per_device_train_batch_size": 8,
248
+ "auto_find_batch_size": true,
249
+ "num_train_epochs": 1,
250
+ "max_steps": -1,
251
+ "logging_strategy": "steps",
252
+ "save_strategy": "steps",
253
+ "logging_steps": 0.2,
254
+ "save_steps": 0.2,
255
+ "gradient_accumulation_steps": 1,
256
+ "gradient_checkpointing": false,
257
+ "optim": "adamw_torch",
258
+ "learning_rate": 5e-05,
259
+ "weight_decay": 0.01,
260
+ "warmup_steps": 0,
261
+ "report_to": "tensorboard",
262
+ "save_safetensors": true,
263
+ "packing": false
264
+ },
265
+ "use_peft": true,
266
+ "peft_task_type": "CAUSAL_LM",
267
+ "peft_inference_mode": false,
268
+ "peft_r": 8,
269
+ "peft_lora_alpha": 32,
270
+ "peft_lora_dropout": 0.1,
271
+ "fsdp": "none",
272
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
273
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
274
+ "fsdp_fsdp_min_num_params": 100000000.0,
275
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
276
+ "fsdp_fsdp_forward_prefetch": false,
277
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
278
+ "fsdp_fsdp_offload_params": false,
279
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
280
+ "fsdp_fsdp_use_orig_params": true,
281
+ "fsdp_fsdp_activation_checkpointing": false,
282
+ "fsdp_fsdp_limit_all_gathers": true,
283
+ "fsdp_fsdp_sync_module_states": true,
284
+ "use_git": true,
285
+ "random_seed": 0
286
+ }
287
+ },
288
+ "piqa": {
289
+ "acc_norm": 0.8199129488574538,
290
+ "acc_norm_stderr": 0.0089654178989032,
291
+ "args": {
292
+ "exp_name": "results/v2/piqa/Qwen/Qwen2.5-7B",
293
+ "task_name": "piqa",
294
+ "model_name": "Qwen/Qwen2.5-7B",
295
+ "dataset_param": {
296
+ "max_num_train": 50000,
297
+ "completion_only": false,
298
+ "apply_chat_template": false
299
+ },
300
+ "resume_from_checkpoint": false,
301
+ "train_param": {
302
+ "max_length": 1024,
303
+ "per_device_train_batch_size": 8,
304
+ "auto_find_batch_size": true,
305
+ "num_train_epochs": 5,
306
+ "max_steps": -1,
307
+ "logging_strategy": "epoch",
308
+ "save_strategy": "epoch",
309
+ "logging_steps": 0.2,
310
+ "save_steps": 0.2,
311
+ "gradient_accumulation_steps": 1,
312
+ "gradient_checkpointing": false,
313
+ "optim": "adamw_torch",
314
+ "learning_rate": 5e-05,
315
+ "weight_decay": 0.01,
316
+ "warmup_steps": 0,
317
+ "report_to": "tensorboard",
318
+ "save_safetensors": true,
319
+ "packing": false
320
+ },
321
+ "use_peft": true,
322
+ "peft_task_type": "CAUSAL_LM",
323
+ "peft_inference_mode": false,
324
+ "peft_r": 8,
325
+ "peft_lora_alpha": 32,
326
+ "peft_lora_dropout": 0.1,
327
+ "fsdp": "none",
328
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
329
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
330
+ "fsdp_fsdp_min_num_params": 100000000.0,
331
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
332
+ "fsdp_fsdp_forward_prefetch": false,
333
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
334
+ "fsdp_fsdp_offload_params": false,
335
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
336
+ "fsdp_fsdp_use_orig_params": true,
337
+ "fsdp_fsdp_activation_checkpointing": false,
338
+ "fsdp_fsdp_limit_all_gathers": true,
339
+ "fsdp_fsdp_sync_module_states": true,
340
+ "use_git": true,
341
+ "random_seed": 0
342
+ }
343
+ },
344
+ "commonsense_qa": {
345
+ "acc": 0.8665028665028665,
346
+ "acc_stderr": 0.009737356192047398,
347
+ "args": {
348
+ "exp_name": "results/v2/commonsense_qa/Qwen/Qwen2.5-7B",
349
+ "task_name": "commonsense_qa",
350
+ "model_name": "Qwen/Qwen2.5-7B",
351
+ "dataset_param": {
352
+ "max_num_train": 50000,
353
+ "completion_only": false,
354
+ "apply_chat_template": false
355
+ },
356
+ "resume_from_checkpoint": false,
357
+ "train_param": {
358
+ "max_length": 1024,
359
+ "per_device_train_batch_size": 8,
360
+ "auto_find_batch_size": true,
361
+ "num_train_epochs": 5,
362
+ "max_steps": -1,
363
+ "logging_strategy": "epoch",
364
+ "save_strategy": "epoch",
365
+ "logging_steps": 0.2,
366
+ "save_steps": 0.2,
367
+ "gradient_accumulation_steps": 1,
368
+ "gradient_checkpointing": false,
369
+ "optim": "adamw_torch",
370
+ "learning_rate": 5e-05,
371
+ "weight_decay": 0.01,
372
+ "warmup_steps": 0,
373
+ "report_to": "tensorboard",
374
+ "save_safetensors": true,
375
+ "packing": false
376
+ },
377
+ "use_peft": true,
378
+ "peft_task_type": "CAUSAL_LM",
379
+ "peft_inference_mode": false,
380
+ "peft_r": 8,
381
+ "peft_lora_alpha": 32,
382
+ "peft_lora_dropout": 0.1,
383
+ "fsdp": "none",
384
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
385
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
386
+ "fsdp_fsdp_min_num_params": 100000000.0,
387
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
388
+ "fsdp_fsdp_forward_prefetch": false,
389
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
390
+ "fsdp_fsdp_offload_params": false,
391
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
392
+ "fsdp_fsdp_use_orig_params": true,
393
+ "fsdp_fsdp_activation_checkpointing": false,
394
+ "fsdp_fsdp_limit_all_gathers": true,
395
+ "fsdp_fsdp_sync_module_states": true,
396
+ "use_git": true,
397
+ "random_seed": 0
398
+ }
399
+ },
400
+ "gsm8k": {
401
+ "exact_match,flexible-extract": 0.7945413191811979,
402
+ "exact_match,flexible-extract_stderr": 0.011129170248544778,
403
+ "args": {
404
+ "exp_name": "results/v2/gsm8k/Qwen/Qwen2.5-7B",
405
+ "task_name": "gsm8k",
406
+ "model_name": "Qwen/Qwen2.5-7B",
407
+ "dataset_param": {
408
+ "max_num_train": 50000,
409
+ "completion_only": false,
410
+ "apply_chat_template": false
411
+ },
412
+ "resume_from_checkpoint": false,
413
+ "train_param": {
414
+ "max_length": 1024,
415
+ "per_device_train_batch_size": 8,
416
+ "auto_find_batch_size": true,
417
+ "num_train_epochs": 5,
418
+ "max_steps": -1,
419
+ "logging_strategy": "epoch",
420
+ "save_strategy": "epoch",
421
+ "logging_steps": 0.2,
422
+ "save_steps": 0.2,
423
+ "gradient_accumulation_steps": 1,
424
+ "gradient_checkpointing": false,
425
+ "optim": "adamw_torch",
426
+ "learning_rate": 5e-05,
427
+ "weight_decay": 0.01,
428
+ "warmup_steps": 0,
429
+ "report_to": "tensorboard",
430
+ "save_safetensors": true,
431
+ "packing": false
432
+ },
433
+ "use_peft": true,
434
+ "peft_task_type": "CAUSAL_LM",
435
+ "peft_inference_mode": false,
436
+ "peft_r": 8,
437
+ "peft_lora_alpha": 32,
438
+ "peft_lora_dropout": 0.1,
439
+ "fsdp": "none",
440
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
441
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
442
+ "fsdp_fsdp_min_num_params": 100000000.0,
443
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
444
+ "fsdp_fsdp_forward_prefetch": false,
445
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
446
+ "fsdp_fsdp_offload_params": false,
447
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
448
+ "fsdp_fsdp_use_orig_params": true,
449
+ "fsdp_fsdp_activation_checkpointing": false,
450
+ "fsdp_fsdp_limit_all_gathers": true,
451
+ "fsdp_fsdp_sync_module_states": true,
452
+ "use_git": true,
453
+ "random_seed": 0
454
+ }
455
+ },
456
+ "arc_challenge": {
457
+ "acc_norm": 0.6348122866894198,
458
+ "acc_norm_stderr": 0.0140702655192688,
459
+ "args": {
460
+ "exp_name": "results/v2/arc_challenge/Qwen/Qwen2.5-7B",
461
+ "task_name": "arc_challenge",
462
+ "model_name": "Qwen/Qwen2.5-7B",
463
+ "dataset_param": {
464
+ "max_num_train": 50000,
465
+ "completion_only": false,
466
+ "apply_chat_template": false
467
+ },
468
+ "resume_from_checkpoint": false,
469
+ "train_param": {
470
+ "max_length": 1024,
471
+ "per_device_train_batch_size": 8,
472
+ "auto_find_batch_size": true,
473
+ "num_train_epochs": 5,
474
+ "max_steps": -1,
475
+ "logging_strategy": "epoch",
476
+ "save_strategy": "epoch",
477
+ "logging_steps": 0.2,
478
+ "save_steps": 0.2,
479
+ "gradient_accumulation_steps": 1,
480
+ "gradient_checkpointing": false,
481
+ "optim": "adamw_torch",
482
+ "learning_rate": 5e-05,
483
+ "weight_decay": 0.01,
484
+ "warmup_steps": 0,
485
+ "report_to": "tensorboard",
486
+ "save_safetensors": true,
487
+ "packing": false
488
+ },
489
+ "use_peft": true,
490
+ "peft_task_type": "CAUSAL_LM",
491
+ "peft_inference_mode": false,
492
+ "peft_r": 8,
493
+ "peft_lora_alpha": 32,
494
+ "peft_lora_dropout": 0.1,
495
+ "fsdp": "none",
496
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
497
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
498
+ "fsdp_fsdp_min_num_params": 100000000.0,
499
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
500
+ "fsdp_fsdp_forward_prefetch": false,
501
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
502
+ "fsdp_fsdp_offload_params": false,
503
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
504
+ "fsdp_fsdp_use_orig_params": true,
505
+ "fsdp_fsdp_activation_checkpointing": false,
506
+ "fsdp_fsdp_limit_all_gathers": true,
507
+ "fsdp_fsdp_sync_module_states": true,
508
+ "use_git": true,
509
+ "random_seed": 0
510
+ }
511
+ }
512
+ }
513
+ }
Qwen/Qwen3-0.6B-Base/results_2026-01-05-04-27-50.json ADDED
@@ -0,0 +1,513 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "model_name": "Qwen/Qwen3-0.6B-Base",
4
+ "model_type": "pretrained",
5
+ "params": 0.6
6
+ },
7
+ "results": {
8
+ "medmcqa": {
9
+ "acc_norm": 0.43987568730576143,
10
+ "acc_norm_stderr": 0.007675650243011505,
11
+ "args": {
12
+ "exp_name": "results/v1/medmcqa/Qwen/Qwen3-0.6B-Base",
13
+ "task_name": "medmcqa",
14
+ "model_name": "Qwen/Qwen3-0.6B-Base",
15
+ "dataset_param": {
16
+ "max_num_train": 50000,
17
+ "completion_only": false,
18
+ "apply_chat_template": false
19
+ },
20
+ "resume_from_checkpoint": false,
21
+ "train_param": {
22
+ "max_length": 1024,
23
+ "per_device_train_batch_size": 8,
24
+ "auto_find_batch_size": true,
25
+ "num_train_epochs": 1,
26
+ "max_steps": -1,
27
+ "logging_strategy": "steps",
28
+ "save_strategy": "steps",
29
+ "logging_steps": 0.2,
30
+ "save_steps": 0.2,
31
+ "gradient_accumulation_steps": 1,
32
+ "gradient_checkpointing": false,
33
+ "optim": "adamw_torch",
34
+ "learning_rate": 5e-05,
35
+ "weight_decay": 0.01,
36
+ "warmup_steps": 0,
37
+ "report_to": "tensorboard",
38
+ "save_safetensors": true,
39
+ "packing": false
40
+ },
41
+ "use_peft": true,
42
+ "peft_task_type": "CAUSAL_LM",
43
+ "peft_inference_mode": false,
44
+ "peft_r": 8,
45
+ "peft_lora_alpha": 32,
46
+ "peft_lora_dropout": 0.1,
47
+ "fsdp": "none",
48
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
49
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
50
+ "fsdp_fsdp_min_num_params": 100000000.0,
51
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
52
+ "fsdp_fsdp_forward_prefetch": false,
53
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
54
+ "fsdp_fsdp_offload_params": false,
55
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
56
+ "fsdp_fsdp_use_orig_params": true,
57
+ "fsdp_fsdp_activation_checkpointing": false,
58
+ "fsdp_fsdp_limit_all_gathers": true,
59
+ "fsdp_fsdp_sync_module_states": true,
60
+ "use_git": true,
61
+ "random_seed": 0
62
+ }
63
+ },
64
+ "nq_open": {
65
+ "exact_match,remove_whitespace": 0.10083102493074793,
66
+ "exact_match,remove_whitespace_stderr": 0.0050121528490627955,
67
+ "args": {
68
+ "exp_name": "results/v1/nq_open/Qwen/Qwen3-0.6B-Base",
69
+ "task_name": "nq_open",
70
+ "model_name": "Qwen/Qwen3-0.6B-Base",
71
+ "dataset_param": {
72
+ "max_num_train": 50000,
73
+ "completion_only": false,
74
+ "apply_chat_template": false
75
+ },
76
+ "resume_from_checkpoint": false,
77
+ "train_param": {
78
+ "max_length": 1024,
79
+ "per_device_train_batch_size": 8,
80
+ "auto_find_batch_size": true,
81
+ "num_train_epochs": 1,
82
+ "max_steps": -1,
83
+ "logging_strategy": "steps",
84
+ "save_strategy": "steps",
85
+ "logging_steps": 0.2,
86
+ "save_steps": 0.2,
87
+ "gradient_accumulation_steps": 1,
88
+ "gradient_checkpointing": false,
89
+ "optim": "adamw_torch",
90
+ "learning_rate": 5e-05,
91
+ "weight_decay": 0.01,
92
+ "warmup_steps": 0,
93
+ "report_to": "tensorboard",
94
+ "save_safetensors": true,
95
+ "packing": false
96
+ },
97
+ "use_peft": true,
98
+ "peft_task_type": "CAUSAL_LM",
99
+ "peft_inference_mode": false,
100
+ "peft_r": 8,
101
+ "peft_lora_alpha": 32,
102
+ "peft_lora_dropout": 0.1,
103
+ "fsdp": "none",
104
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
105
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
106
+ "fsdp_fsdp_min_num_params": 100000000.0,
107
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
108
+ "fsdp_fsdp_forward_prefetch": false,
109
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
110
+ "fsdp_fsdp_offload_params": false,
111
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
112
+ "fsdp_fsdp_use_orig_params": true,
113
+ "fsdp_fsdp_activation_checkpointing": false,
114
+ "fsdp_fsdp_limit_all_gathers": true,
115
+ "fsdp_fsdp_sync_module_states": true,
116
+ "use_git": true,
117
+ "random_seed": 0
118
+ }
119
+ },
120
+ "winogrande": {
121
+ "acc": 0.6306235201262825,
122
+ "acc_stderr": 0.01356447059605353,
123
+ "args": {
124
+ "exp_name": "results/v1/winogrande/Qwen/Qwen3-0.6B-Base",
125
+ "task_name": "winogrande",
126
+ "model_name": "Qwen/Qwen3-0.6B-Base",
127
+ "dataset_param": {
128
+ "max_num_train": 50000,
129
+ "completion_only": false,
130
+ "apply_chat_template": false
131
+ },
132
+ "resume_from_checkpoint": false,
133
+ "train_param": {
134
+ "max_length": 1024,
135
+ "per_device_train_batch_size": 8,
136
+ "auto_find_batch_size": true,
137
+ "num_train_epochs": 1,
138
+ "max_steps": -1,
139
+ "logging_strategy": "steps",
140
+ "save_strategy": "steps",
141
+ "logging_steps": 0.2,
142
+ "save_steps": 0.2,
143
+ "gradient_accumulation_steps": 1,
144
+ "gradient_checkpointing": false,
145
+ "optim": "adamw_torch",
146
+ "learning_rate": 5e-05,
147
+ "weight_decay": 0.01,
148
+ "warmup_steps": 0,
149
+ "report_to": "tensorboard",
150
+ "save_safetensors": true,
151
+ "packing": false
152
+ },
153
+ "use_peft": true,
154
+ "peft_task_type": "CAUSAL_LM",
155
+ "peft_inference_mode": false,
156
+ "peft_r": 8,
157
+ "peft_lora_alpha": 32,
158
+ "peft_lora_dropout": 0.1,
159
+ "fsdp": "none",
160
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
161
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
162
+ "fsdp_fsdp_min_num_params": 100000000.0,
163
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
164
+ "fsdp_fsdp_forward_prefetch": false,
165
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
166
+ "fsdp_fsdp_offload_params": false,
167
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
168
+ "fsdp_fsdp_use_orig_params": true,
169
+ "fsdp_fsdp_activation_checkpointing": false,
170
+ "fsdp_fsdp_limit_all_gathers": true,
171
+ "fsdp_fsdp_sync_module_states": true,
172
+ "use_git": true,
173
+ "random_seed": 0
174
+ }
175
+ },
176
+ "hellaswag": {
177
+ "acc_norm": 0.5574,
178
+ "acc_norm_stderr": 0.004967191490080854,
179
+ "args": {
180
+ "exp_name": "results/v1/hellaswag/Qwen/Qwen3-0.6B-Base",
181
+ "task_name": "hellaswag",
182
+ "model_name": "Qwen/Qwen3-0.6B-Base",
183
+ "dataset_param": {
184
+ "max_num_train": 50000,
185
+ "completion_only": false,
186
+ "apply_chat_template": false
187
+ },
188
+ "resume_from_checkpoint": false,
189
+ "train_param": {
190
+ "max_length": 1024,
191
+ "per_device_train_batch_size": 8,
192
+ "auto_find_batch_size": true,
193
+ "num_train_epochs": 1,
194
+ "max_steps": -1,
195
+ "logging_strategy": "steps",
196
+ "save_strategy": "steps",
197
+ "logging_steps": 0.2,
198
+ "save_steps": 0.2,
199
+ "gradient_accumulation_steps": 1,
200
+ "gradient_checkpointing": false,
201
+ "optim": "adamw_torch",
202
+ "learning_rate": 5e-05,
203
+ "weight_decay": 0.01,
204
+ "warmup_steps": 0,
205
+ "report_to": "tensorboard",
206
+ "save_safetensors": true,
207
+ "packing": false
208
+ },
209
+ "use_peft": true,
210
+ "peft_task_type": "CAUSAL_LM",
211
+ "peft_inference_mode": false,
212
+ "peft_r": 8,
213
+ "peft_lora_alpha": 32,
214
+ "peft_lora_dropout": 0.1,
215
+ "fsdp": "none",
216
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
217
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
218
+ "fsdp_fsdp_min_num_params": 100000000.0,
219
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
220
+ "fsdp_fsdp_forward_prefetch": false,
221
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
222
+ "fsdp_fsdp_offload_params": false,
223
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
224
+ "fsdp_fsdp_use_orig_params": true,
225
+ "fsdp_fsdp_activation_checkpointing": false,
226
+ "fsdp_fsdp_limit_all_gathers": true,
227
+ "fsdp_fsdp_sync_module_states": true,
228
+ "use_git": true,
229
+ "random_seed": 0
230
+ }
231
+ },
232
+ "social_iqa": {
233
+ "acc": 0.4887410440122825,
234
+ "acc_stderr": 0.011311201755614007,
235
+ "args": {
236
+ "exp_name": "results/v1/social_iqa/Qwen/Qwen3-0.6B-Base",
237
+ "task_name": "social_iqa",
238
+ "model_name": "Qwen/Qwen3-0.6B-Base",
239
+ "dataset_param": {
240
+ "max_num_train": 50000,
241
+ "completion_only": false,
242
+ "apply_chat_template": false
243
+ },
244
+ "resume_from_checkpoint": false,
245
+ "train_param": {
246
+ "max_length": 1024,
247
+ "per_device_train_batch_size": 8,
248
+ "auto_find_batch_size": true,
249
+ "num_train_epochs": 1,
250
+ "max_steps": -1,
251
+ "logging_strategy": "steps",
252
+ "save_strategy": "steps",
253
+ "logging_steps": 0.2,
254
+ "save_steps": 0.2,
255
+ "gradient_accumulation_steps": 1,
256
+ "gradient_checkpointing": false,
257
+ "optim": "adamw_torch",
258
+ "learning_rate": 5e-05,
259
+ "weight_decay": 0.01,
260
+ "warmup_steps": 0,
261
+ "report_to": "tensorboard",
262
+ "save_safetensors": true,
263
+ "packing": false
264
+ },
265
+ "use_peft": true,
266
+ "peft_task_type": "CAUSAL_LM",
267
+ "peft_inference_mode": false,
268
+ "peft_r": 8,
269
+ "peft_lora_alpha": 32,
270
+ "peft_lora_dropout": 0.1,
271
+ "fsdp": "none",
272
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
273
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
274
+ "fsdp_fsdp_min_num_params": 100000000.0,
275
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
276
+ "fsdp_fsdp_forward_prefetch": false,
277
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
278
+ "fsdp_fsdp_offload_params": false,
279
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
280
+ "fsdp_fsdp_use_orig_params": true,
281
+ "fsdp_fsdp_activation_checkpointing": false,
282
+ "fsdp_fsdp_limit_all_gathers": true,
283
+ "fsdp_fsdp_sync_module_states": true,
284
+ "use_git": true,
285
+ "random_seed": 0
286
+ }
287
+ },
288
+ "piqa": {
289
+ "acc_norm": 0.7312295973884657,
290
+ "acc_norm_stderr": 0.010343392940090013,
291
+ "args": {
292
+ "exp_name": "results/v2/piqa/Qwen/Qwen3-0.6B-Base",
293
+ "task_name": "piqa",
294
+ "model_name": "Qwen/Qwen3-0.6B-Base",
295
+ "dataset_param": {
296
+ "max_num_train": 50000,
297
+ "completion_only": false,
298
+ "apply_chat_template": false
299
+ },
300
+ "resume_from_checkpoint": false,
301
+ "train_param": {
302
+ "max_length": 1024,
303
+ "per_device_train_batch_size": 8,
304
+ "auto_find_batch_size": true,
305
+ "num_train_epochs": 5,
306
+ "max_steps": -1,
307
+ "logging_strategy": "epoch",
308
+ "save_strategy": "epoch",
309
+ "logging_steps": 0.2,
310
+ "save_steps": 0.2,
311
+ "gradient_accumulation_steps": 1,
312
+ "gradient_checkpointing": false,
313
+ "optim": "adamw_torch",
314
+ "learning_rate": 5e-05,
315
+ "weight_decay": 0.01,
316
+ "warmup_steps": 0,
317
+ "report_to": "tensorboard",
318
+ "save_safetensors": true,
319
+ "packing": false
320
+ },
321
+ "use_peft": true,
322
+ "peft_task_type": "CAUSAL_LM",
323
+ "peft_inference_mode": false,
324
+ "peft_r": 8,
325
+ "peft_lora_alpha": 32,
326
+ "peft_lora_dropout": 0.1,
327
+ "fsdp": "none",
328
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
329
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
330
+ "fsdp_fsdp_min_num_params": 100000000.0,
331
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
332
+ "fsdp_fsdp_forward_prefetch": false,
333
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
334
+ "fsdp_fsdp_offload_params": false,
335
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
336
+ "fsdp_fsdp_use_orig_params": true,
337
+ "fsdp_fsdp_activation_checkpointing": false,
338
+ "fsdp_fsdp_limit_all_gathers": true,
339
+ "fsdp_fsdp_sync_module_states": true,
340
+ "use_git": true,
341
+ "random_seed": 0
342
+ }
343
+ },
344
+ "commonsense_qa": {
345
+ "acc": 0.7018837018837019,
346
+ "acc_stderr": 0.013096206775112642,
347
+ "args": {
348
+ "exp_name": "results/v2/commonsense_qa/Qwen/Qwen3-0.6B-Base",
349
+ "task_name": "commonsense_qa",
350
+ "model_name": "Qwen/Qwen3-0.6B-Base",
351
+ "dataset_param": {
352
+ "max_num_train": 50000,
353
+ "completion_only": false,
354
+ "apply_chat_template": false
355
+ },
356
+ "resume_from_checkpoint": false,
357
+ "train_param": {
358
+ "max_length": 1024,
359
+ "per_device_train_batch_size": 8,
360
+ "auto_find_batch_size": true,
361
+ "num_train_epochs": 5,
362
+ "max_steps": -1,
363
+ "logging_strategy": "epoch",
364
+ "save_strategy": "epoch",
365
+ "logging_steps": 0.2,
366
+ "save_steps": 0.2,
367
+ "gradient_accumulation_steps": 1,
368
+ "gradient_checkpointing": false,
369
+ "optim": "adamw_torch",
370
+ "learning_rate": 5e-05,
371
+ "weight_decay": 0.01,
372
+ "warmup_steps": 0,
373
+ "report_to": "tensorboard",
374
+ "save_safetensors": true,
375
+ "packing": false
376
+ },
377
+ "use_peft": true,
378
+ "peft_task_type": "CAUSAL_LM",
379
+ "peft_inference_mode": false,
380
+ "peft_r": 8,
381
+ "peft_lora_alpha": 32,
382
+ "peft_lora_dropout": 0.1,
383
+ "fsdp": "none",
384
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
385
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
386
+ "fsdp_fsdp_min_num_params": 100000000.0,
387
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
388
+ "fsdp_fsdp_forward_prefetch": false,
389
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
390
+ "fsdp_fsdp_offload_params": false,
391
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
392
+ "fsdp_fsdp_use_orig_params": true,
393
+ "fsdp_fsdp_activation_checkpointing": false,
394
+ "fsdp_fsdp_limit_all_gathers": true,
395
+ "fsdp_fsdp_sync_module_states": true,
396
+ "use_git": true,
397
+ "random_seed": 0
398
+ }
399
+ },
400
+ "gsm8k": {
401
+ "exact_match,flexible-extract": 0.5072024260803639,
402
+ "exact_match,flexible-extract_stderr": 0.013771055751972872,
403
+ "args": {
404
+ "exp_name": "results/v2/gsm8k/Qwen/Qwen3-0.6B-Base",
405
+ "task_name": "gsm8k",
406
+ "model_name": "Qwen/Qwen3-0.6B-Base",
407
+ "dataset_param": {
408
+ "max_num_train": 50000,
409
+ "completion_only": false,
410
+ "apply_chat_template": false
411
+ },
412
+ "resume_from_checkpoint": false,
413
+ "train_param": {
414
+ "max_length": 1024,
415
+ "per_device_train_batch_size": 8,
416
+ "auto_find_batch_size": true,
417
+ "num_train_epochs": 5,
418
+ "max_steps": -1,
419
+ "logging_strategy": "epoch",
420
+ "save_strategy": "epoch",
421
+ "logging_steps": 0.2,
422
+ "save_steps": 0.2,
423
+ "gradient_accumulation_steps": 1,
424
+ "gradient_checkpointing": false,
425
+ "optim": "adamw_torch",
426
+ "learning_rate": 5e-05,
427
+ "weight_decay": 0.01,
428
+ "warmup_steps": 0,
429
+ "report_to": "tensorboard",
430
+ "save_safetensors": true,
431
+ "packing": false
432
+ },
433
+ "use_peft": true,
434
+ "peft_task_type": "CAUSAL_LM",
435
+ "peft_inference_mode": false,
436
+ "peft_r": 8,
437
+ "peft_lora_alpha": 32,
438
+ "peft_lora_dropout": 0.1,
439
+ "fsdp": "none",
440
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
441
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
442
+ "fsdp_fsdp_min_num_params": 100000000.0,
443
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
444
+ "fsdp_fsdp_forward_prefetch": false,
445
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
446
+ "fsdp_fsdp_offload_params": false,
447
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
448
+ "fsdp_fsdp_use_orig_params": true,
449
+ "fsdp_fsdp_activation_checkpointing": false,
450
+ "fsdp_fsdp_limit_all_gathers": true,
451
+ "fsdp_fsdp_sync_module_states": true,
452
+ "use_git": true,
453
+ "random_seed": 0
454
+ }
455
+ },
456
+ "arc_challenge": {
457
+ "acc_norm": 0.447098976109215,
458
+ "acc_norm_stderr": 0.014529380160526848,
459
+ "args": {
460
+ "exp_name": "results/v2/arc_challenge/Qwen/Qwen3-0.6B-Base",
461
+ "task_name": "arc_challenge",
462
+ "model_name": "Qwen/Qwen3-0.6B-Base",
463
+ "dataset_param": {
464
+ "max_num_train": 50000,
465
+ "completion_only": false,
466
+ "apply_chat_template": false
467
+ },
468
+ "resume_from_checkpoint": false,
469
+ "train_param": {
470
+ "max_length": 1024,
471
+ "per_device_train_batch_size": 8,
472
+ "auto_find_batch_size": true,
473
+ "num_train_epochs": 5,
474
+ "max_steps": -1,
475
+ "logging_strategy": "epoch",
476
+ "save_strategy": "epoch",
477
+ "logging_steps": 0.2,
478
+ "save_steps": 0.2,
479
+ "gradient_accumulation_steps": 1,
480
+ "gradient_checkpointing": false,
481
+ "optim": "adamw_torch",
482
+ "learning_rate": 5e-05,
483
+ "weight_decay": 0.01,
484
+ "warmup_steps": 0,
485
+ "report_to": "tensorboard",
486
+ "save_safetensors": true,
487
+ "packing": false
488
+ },
489
+ "use_peft": true,
490
+ "peft_task_type": "CAUSAL_LM",
491
+ "peft_inference_mode": false,
492
+ "peft_r": 8,
493
+ "peft_lora_alpha": 32,
494
+ "peft_lora_dropout": 0.1,
495
+ "fsdp": "none",
496
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
497
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
498
+ "fsdp_fsdp_min_num_params": 100000000.0,
499
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
500
+ "fsdp_fsdp_forward_prefetch": false,
501
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
502
+ "fsdp_fsdp_offload_params": false,
503
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
504
+ "fsdp_fsdp_use_orig_params": true,
505
+ "fsdp_fsdp_activation_checkpointing": false,
506
+ "fsdp_fsdp_limit_all_gathers": true,
507
+ "fsdp_fsdp_sync_module_states": true,
508
+ "use_git": true,
509
+ "random_seed": 0
510
+ }
511
+ }
512
+ }
513
+ }
Qwen/Qwen3-0.6B/results_2026-01-05-04-27-51.json ADDED
@@ -0,0 +1,513 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "model_name": "Qwen/Qwen3-0.6B",
4
+ "model_type": "instruction-tuned",
5
+ "params": 0.6
6
+ },
7
+ "results": {
8
+ "medmcqa": {
9
+ "acc_norm": 0.3349270858235716,
10
+ "acc_norm_stderr": 0.007298233459029951,
11
+ "args": {
12
+ "exp_name": "results/v1/medmcqa/Qwen/Qwen3-0.6B",
13
+ "task_name": "medmcqa",
14
+ "model_name": "Qwen/Qwen3-0.6B",
15
+ "dataset_param": {
16
+ "max_num_train": 50000,
17
+ "completion_only": false,
18
+ "apply_chat_template": true
19
+ },
20
+ "resume_from_checkpoint": false,
21
+ "train_param": {
22
+ "max_length": 1024,
23
+ "per_device_train_batch_size": 8,
24
+ "auto_find_batch_size": true,
25
+ "num_train_epochs": 1,
26
+ "max_steps": -1,
27
+ "logging_strategy": "steps",
28
+ "save_strategy": "steps",
29
+ "logging_steps": 0.2,
30
+ "save_steps": 0.2,
31
+ "gradient_accumulation_steps": 1,
32
+ "gradient_checkpointing": false,
33
+ "optim": "adamw_torch",
34
+ "learning_rate": 5e-05,
35
+ "weight_decay": 0.01,
36
+ "warmup_steps": 0,
37
+ "report_to": "tensorboard",
38
+ "save_safetensors": true,
39
+ "packing": false
40
+ },
41
+ "use_peft": true,
42
+ "peft_task_type": "CAUSAL_LM",
43
+ "peft_inference_mode": false,
44
+ "peft_r": 8,
45
+ "peft_lora_alpha": 32,
46
+ "peft_lora_dropout": 0.1,
47
+ "fsdp": "none",
48
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
49
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
50
+ "fsdp_fsdp_min_num_params": 100000000.0,
51
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
52
+ "fsdp_fsdp_forward_prefetch": false,
53
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
54
+ "fsdp_fsdp_offload_params": false,
55
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
56
+ "fsdp_fsdp_use_orig_params": true,
57
+ "fsdp_fsdp_activation_checkpointing": false,
58
+ "fsdp_fsdp_limit_all_gathers": true,
59
+ "fsdp_fsdp_sync_module_states": true,
60
+ "use_git": true,
61
+ "random_seed": 0
62
+ }
63
+ },
64
+ "nq_open": {
65
+ "exact_match,remove_whitespace": 0.07285318559556787,
66
+ "exact_match,remove_whitespace_stderr": 0.004326185003171796,
67
+ "args": {
68
+ "exp_name": "results/v1/nq_open/Qwen/Qwen3-0.6B",
69
+ "task_name": "nq_open",
70
+ "model_name": "Qwen/Qwen3-0.6B",
71
+ "dataset_param": {
72
+ "max_num_train": 50000,
73
+ "completion_only": false,
74
+ "apply_chat_template": true
75
+ },
76
+ "resume_from_checkpoint": false,
77
+ "train_param": {
78
+ "max_length": 1024,
79
+ "per_device_train_batch_size": 8,
80
+ "auto_find_batch_size": true,
81
+ "num_train_epochs": 1,
82
+ "max_steps": -1,
83
+ "logging_strategy": "steps",
84
+ "save_strategy": "steps",
85
+ "logging_steps": 0.2,
86
+ "save_steps": 0.2,
87
+ "gradient_accumulation_steps": 1,
88
+ "gradient_checkpointing": false,
89
+ "optim": "adamw_torch",
90
+ "learning_rate": 5e-05,
91
+ "weight_decay": 0.01,
92
+ "warmup_steps": 0,
93
+ "report_to": "tensorboard",
94
+ "save_safetensors": true,
95
+ "packing": false
96
+ },
97
+ "use_peft": true,
98
+ "peft_task_type": "CAUSAL_LM",
99
+ "peft_inference_mode": false,
100
+ "peft_r": 8,
101
+ "peft_lora_alpha": 32,
102
+ "peft_lora_dropout": 0.1,
103
+ "fsdp": "none",
104
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
105
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
106
+ "fsdp_fsdp_min_num_params": 100000000.0,
107
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
108
+ "fsdp_fsdp_forward_prefetch": false,
109
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
110
+ "fsdp_fsdp_offload_params": false,
111
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
112
+ "fsdp_fsdp_use_orig_params": true,
113
+ "fsdp_fsdp_activation_checkpointing": false,
114
+ "fsdp_fsdp_limit_all_gathers": true,
115
+ "fsdp_fsdp_sync_module_states": true,
116
+ "use_git": true,
117
+ "random_seed": 0
118
+ }
119
+ },
120
+ "winogrande": {
121
+ "acc": 0.6274664561957379,
122
+ "acc_stderr": 0.013588173888522457,
123
+ "args": {
124
+ "exp_name": "results/v1/winogrande/Qwen/Qwen3-0.6B",
125
+ "task_name": "winogrande",
126
+ "model_name": "Qwen/Qwen3-0.6B",
127
+ "dataset_param": {
128
+ "max_num_train": 50000,
129
+ "completion_only": false,
130
+ "apply_chat_template": true
131
+ },
132
+ "resume_from_checkpoint": false,
133
+ "train_param": {
134
+ "max_length": 1024,
135
+ "per_device_train_batch_size": 8,
136
+ "auto_find_batch_size": true,
137
+ "num_train_epochs": 1,
138
+ "max_steps": -1,
139
+ "logging_strategy": "steps",
140
+ "save_strategy": "steps",
141
+ "logging_steps": 0.2,
142
+ "save_steps": 0.2,
143
+ "gradient_accumulation_steps": 1,
144
+ "gradient_checkpointing": false,
145
+ "optim": "adamw_torch",
146
+ "learning_rate": 5e-05,
147
+ "weight_decay": 0.01,
148
+ "warmup_steps": 0,
149
+ "report_to": "tensorboard",
150
+ "save_safetensors": true,
151
+ "packing": false
152
+ },
153
+ "use_peft": true,
154
+ "peft_task_type": "CAUSAL_LM",
155
+ "peft_inference_mode": false,
156
+ "peft_r": 8,
157
+ "peft_lora_alpha": 32,
158
+ "peft_lora_dropout": 0.1,
159
+ "fsdp": "none",
160
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
161
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
162
+ "fsdp_fsdp_min_num_params": 100000000.0,
163
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
164
+ "fsdp_fsdp_forward_prefetch": false,
165
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
166
+ "fsdp_fsdp_offload_params": false,
167
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
168
+ "fsdp_fsdp_use_orig_params": true,
169
+ "fsdp_fsdp_activation_checkpointing": false,
170
+ "fsdp_fsdp_limit_all_gathers": true,
171
+ "fsdp_fsdp_sync_module_states": true,
172
+ "use_git": true,
173
+ "random_seed": 0
174
+ }
175
+ },
176
+ "hellaswag": {
177
+ "acc_norm": 0.5096,
178
+ "acc_norm_stderr": 0.004999328287713789,
179
+ "args": {
180
+ "exp_name": "results/v1/hellaswag/Qwen/Qwen3-0.6B",
181
+ "task_name": "hellaswag",
182
+ "model_name": "Qwen/Qwen3-0.6B",
183
+ "dataset_param": {
184
+ "max_num_train": 50000,
185
+ "completion_only": false,
186
+ "apply_chat_template": true
187
+ },
188
+ "resume_from_checkpoint": false,
189
+ "train_param": {
190
+ "max_length": 1024,
191
+ "per_device_train_batch_size": 8,
192
+ "auto_find_batch_size": true,
193
+ "num_train_epochs": 1,
194
+ "max_steps": -1,
195
+ "logging_strategy": "steps",
196
+ "save_strategy": "steps",
197
+ "logging_steps": 0.2,
198
+ "save_steps": 0.2,
199
+ "gradient_accumulation_steps": 1,
200
+ "gradient_checkpointing": false,
201
+ "optim": "adamw_torch",
202
+ "learning_rate": 5e-05,
203
+ "weight_decay": 0.01,
204
+ "warmup_steps": 0,
205
+ "report_to": "tensorboard",
206
+ "save_safetensors": true,
207
+ "packing": false
208
+ },
209
+ "use_peft": true,
210
+ "peft_task_type": "CAUSAL_LM",
211
+ "peft_inference_mode": false,
212
+ "peft_r": 8,
213
+ "peft_lora_alpha": 32,
214
+ "peft_lora_dropout": 0.1,
215
+ "fsdp": "none",
216
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
217
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
218
+ "fsdp_fsdp_min_num_params": 100000000.0,
219
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
220
+ "fsdp_fsdp_forward_prefetch": false,
221
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
222
+ "fsdp_fsdp_offload_params": false,
223
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
224
+ "fsdp_fsdp_use_orig_params": true,
225
+ "fsdp_fsdp_activation_checkpointing": false,
226
+ "fsdp_fsdp_limit_all_gathers": true,
227
+ "fsdp_fsdp_sync_module_states": true,
228
+ "use_git": true,
229
+ "random_seed": 0
230
+ }
231
+ },
232
+ "social_iqa": {
233
+ "acc": 0.4657113613101331,
234
+ "acc_stderr": 0.011287435053883957,
235
+ "args": {
236
+ "exp_name": "results/v1/social_iqa/Qwen/Qwen3-0.6B",
237
+ "task_name": "social_iqa",
238
+ "model_name": "Qwen/Qwen3-0.6B",
239
+ "dataset_param": {
240
+ "max_num_train": 50000,
241
+ "completion_only": false,
242
+ "apply_chat_template": true
243
+ },
244
+ "resume_from_checkpoint": false,
245
+ "train_param": {
246
+ "max_length": 1024,
247
+ "per_device_train_batch_size": 8,
248
+ "auto_find_batch_size": true,
249
+ "num_train_epochs": 1,
250
+ "max_steps": -1,
251
+ "logging_strategy": "steps",
252
+ "save_strategy": "steps",
253
+ "logging_steps": 0.2,
254
+ "save_steps": 0.2,
255
+ "gradient_accumulation_steps": 1,
256
+ "gradient_checkpointing": false,
257
+ "optim": "adamw_torch",
258
+ "learning_rate": 5e-05,
259
+ "weight_decay": 0.01,
260
+ "warmup_steps": 0,
261
+ "report_to": "tensorboard",
262
+ "save_safetensors": true,
263
+ "packing": false
264
+ },
265
+ "use_peft": true,
266
+ "peft_task_type": "CAUSAL_LM",
267
+ "peft_inference_mode": false,
268
+ "peft_r": 8,
269
+ "peft_lora_alpha": 32,
270
+ "peft_lora_dropout": 0.1,
271
+ "fsdp": "none",
272
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
273
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
274
+ "fsdp_fsdp_min_num_params": 100000000.0,
275
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
276
+ "fsdp_fsdp_forward_prefetch": false,
277
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
278
+ "fsdp_fsdp_offload_params": false,
279
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
280
+ "fsdp_fsdp_use_orig_params": true,
281
+ "fsdp_fsdp_activation_checkpointing": false,
282
+ "fsdp_fsdp_limit_all_gathers": true,
283
+ "fsdp_fsdp_sync_module_states": true,
284
+ "use_git": true,
285
+ "random_seed": 0
286
+ }
287
+ },
288
+ "piqa": {
289
+ "acc_norm": 0.70620239390642,
290
+ "acc_norm_stderr": 0.010627574080514809,
291
+ "args": {
292
+ "exp_name": "results/v2/piqa/Qwen/Qwen3-0.6B",
293
+ "task_name": "piqa",
294
+ "model_name": "Qwen/Qwen3-0.6B",
295
+ "dataset_param": {
296
+ "max_num_train": 50000,
297
+ "completion_only": false,
298
+ "apply_chat_template": true
299
+ },
300
+ "resume_from_checkpoint": false,
301
+ "train_param": {
302
+ "max_length": 1024,
303
+ "per_device_train_batch_size": 8,
304
+ "auto_find_batch_size": true,
305
+ "num_train_epochs": 5,
306
+ "max_steps": -1,
307
+ "logging_strategy": "epoch",
308
+ "save_strategy": "epoch",
309
+ "logging_steps": 0.2,
310
+ "save_steps": 0.2,
311
+ "gradient_accumulation_steps": 1,
312
+ "gradient_checkpointing": false,
313
+ "optim": "adamw_torch",
314
+ "learning_rate": 5e-05,
315
+ "weight_decay": 0.01,
316
+ "warmup_steps": 0,
317
+ "report_to": "tensorboard",
318
+ "save_safetensors": true,
319
+ "packing": false
320
+ },
321
+ "use_peft": true,
322
+ "peft_task_type": "CAUSAL_LM",
323
+ "peft_inference_mode": false,
324
+ "peft_r": 8,
325
+ "peft_lora_alpha": 32,
326
+ "peft_lora_dropout": 0.1,
327
+ "fsdp": "none",
328
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
329
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
330
+ "fsdp_fsdp_min_num_params": 100000000.0,
331
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
332
+ "fsdp_fsdp_forward_prefetch": false,
333
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
334
+ "fsdp_fsdp_offload_params": false,
335
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
336
+ "fsdp_fsdp_use_orig_params": true,
337
+ "fsdp_fsdp_activation_checkpointing": false,
338
+ "fsdp_fsdp_limit_all_gathers": true,
339
+ "fsdp_fsdp_sync_module_states": true,
340
+ "use_git": true,
341
+ "random_seed": 0
342
+ }
343
+ },
344
+ "commonsense_qa": {
345
+ "acc": 0.6535626535626535,
346
+ "acc_stderr": 0.013623103310029453,
347
+ "args": {
348
+ "exp_name": "results/v2/commonsense_qa/Qwen/Qwen3-0.6B",
349
+ "task_name": "commonsense_qa",
350
+ "model_name": "Qwen/Qwen3-0.6B",
351
+ "dataset_param": {
352
+ "max_num_train": 50000,
353
+ "completion_only": false,
354
+ "apply_chat_template": true
355
+ },
356
+ "resume_from_checkpoint": false,
357
+ "train_param": {
358
+ "max_length": 1024,
359
+ "per_device_train_batch_size": 8,
360
+ "auto_find_batch_size": true,
361
+ "num_train_epochs": 5,
362
+ "max_steps": -1,
363
+ "logging_strategy": "epoch",
364
+ "save_strategy": "epoch",
365
+ "logging_steps": 0.2,
366
+ "save_steps": 0.2,
367
+ "gradient_accumulation_steps": 1,
368
+ "gradient_checkpointing": false,
369
+ "optim": "adamw_torch",
370
+ "learning_rate": 5e-05,
371
+ "weight_decay": 0.01,
372
+ "warmup_steps": 0,
373
+ "report_to": "tensorboard",
374
+ "save_safetensors": true,
375
+ "packing": false
376
+ },
377
+ "use_peft": true,
378
+ "peft_task_type": "CAUSAL_LM",
379
+ "peft_inference_mode": false,
380
+ "peft_r": 8,
381
+ "peft_lora_alpha": 32,
382
+ "peft_lora_dropout": 0.1,
383
+ "fsdp": "none",
384
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
385
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
386
+ "fsdp_fsdp_min_num_params": 100000000.0,
387
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
388
+ "fsdp_fsdp_forward_prefetch": false,
389
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
390
+ "fsdp_fsdp_offload_params": false,
391
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
392
+ "fsdp_fsdp_use_orig_params": true,
393
+ "fsdp_fsdp_activation_checkpointing": false,
394
+ "fsdp_fsdp_limit_all_gathers": true,
395
+ "fsdp_fsdp_sync_module_states": true,
396
+ "use_git": true,
397
+ "random_seed": 0
398
+ }
399
+ },
400
+ "gsm8k": {
401
+ "exact_match,flexible-extract": 0.444275966641395,
402
+ "exact_match,flexible-extract_stderr": 0.013686685712261667,
403
+ "args": {
404
+ "exp_name": "results/v2/gsm8k/Qwen/Qwen3-0.6B",
405
+ "task_name": "gsm8k",
406
+ "model_name": "Qwen/Qwen3-0.6B",
407
+ "dataset_param": {
408
+ "max_num_train": 50000,
409
+ "completion_only": false,
410
+ "apply_chat_template": true
411
+ },
412
+ "resume_from_checkpoint": false,
413
+ "train_param": {
414
+ "max_length": 1024,
415
+ "per_device_train_batch_size": 8,
416
+ "auto_find_batch_size": true,
417
+ "num_train_epochs": 5,
418
+ "max_steps": -1,
419
+ "logging_strategy": "epoch",
420
+ "save_strategy": "epoch",
421
+ "logging_steps": 0.2,
422
+ "save_steps": 0.2,
423
+ "gradient_accumulation_steps": 1,
424
+ "gradient_checkpointing": false,
425
+ "optim": "adamw_torch",
426
+ "learning_rate": 5e-05,
427
+ "weight_decay": 0.01,
428
+ "warmup_steps": 0,
429
+ "report_to": "tensorboard",
430
+ "save_safetensors": true,
431
+ "packing": false
432
+ },
433
+ "use_peft": true,
434
+ "peft_task_type": "CAUSAL_LM",
435
+ "peft_inference_mode": false,
436
+ "peft_r": 8,
437
+ "peft_lora_alpha": 32,
438
+ "peft_lora_dropout": 0.1,
439
+ "fsdp": "none",
440
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
441
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
442
+ "fsdp_fsdp_min_num_params": 100000000.0,
443
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
444
+ "fsdp_fsdp_forward_prefetch": false,
445
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
446
+ "fsdp_fsdp_offload_params": false,
447
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
448
+ "fsdp_fsdp_use_orig_params": true,
449
+ "fsdp_fsdp_activation_checkpointing": false,
450
+ "fsdp_fsdp_limit_all_gathers": true,
451
+ "fsdp_fsdp_sync_module_states": true,
452
+ "use_git": true,
453
+ "random_seed": 0
454
+ }
455
+ },
456
+ "arc_challenge": {
457
+ "acc_norm": 0.3430034129692833,
458
+ "acc_norm_stderr": 0.013872423223718167,
459
+ "args": {
460
+ "exp_name": "results/v2/arc_challenge/Qwen/Qwen3-0.6B",
461
+ "task_name": "arc_challenge",
462
+ "model_name": "Qwen/Qwen3-0.6B",
463
+ "dataset_param": {
464
+ "max_num_train": 50000,
465
+ "completion_only": false,
466
+ "apply_chat_template": true
467
+ },
468
+ "resume_from_checkpoint": false,
469
+ "train_param": {
470
+ "max_length": 1024,
471
+ "per_device_train_batch_size": 8,
472
+ "auto_find_batch_size": true,
473
+ "num_train_epochs": 5,
474
+ "max_steps": -1,
475
+ "logging_strategy": "epoch",
476
+ "save_strategy": "epoch",
477
+ "logging_steps": 0.2,
478
+ "save_steps": 0.2,
479
+ "gradient_accumulation_steps": 1,
480
+ "gradient_checkpointing": false,
481
+ "optim": "adamw_torch",
482
+ "learning_rate": 5e-05,
483
+ "weight_decay": 0.01,
484
+ "warmup_steps": 0,
485
+ "report_to": "tensorboard",
486
+ "save_safetensors": true,
487
+ "packing": false
488
+ },
489
+ "use_peft": true,
490
+ "peft_task_type": "CAUSAL_LM",
491
+ "peft_inference_mode": false,
492
+ "peft_r": 8,
493
+ "peft_lora_alpha": 32,
494
+ "peft_lora_dropout": 0.1,
495
+ "fsdp": "none",
496
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
497
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
498
+ "fsdp_fsdp_min_num_params": 100000000.0,
499
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
500
+ "fsdp_fsdp_forward_prefetch": false,
501
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
502
+ "fsdp_fsdp_offload_params": false,
503
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
504
+ "fsdp_fsdp_use_orig_params": true,
505
+ "fsdp_fsdp_activation_checkpointing": false,
506
+ "fsdp_fsdp_limit_all_gathers": true,
507
+ "fsdp_fsdp_sync_module_states": true,
508
+ "use_git": true,
509
+ "random_seed": 0
510
+ }
511
+ }
512
+ }
513
+ }
Qwen/Qwen3-1.7B-Base/results_2026-01-05-04-27-50.json ADDED
@@ -0,0 +1,513 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "model_name": "Qwen/Qwen3-1.7B-Base",
4
+ "model_type": "pretrained",
5
+ "params": 1.7
6
+ },
7
+ "results": {
8
+ "medmcqa": {
9
+ "acc_norm": 0.5096820463781975,
10
+ "acc_norm_stderr": 0.007730303614276418,
11
+ "args": {
12
+ "exp_name": "results/v1/medmcqa/Qwen/Qwen3-1.7B-Base",
13
+ "task_name": "medmcqa",
14
+ "model_name": "Qwen/Qwen3-1.7B-Base",
15
+ "dataset_param": {
16
+ "max_num_train": 50000,
17
+ "completion_only": false,
18
+ "apply_chat_template": false
19
+ },
20
+ "resume_from_checkpoint": false,
21
+ "train_param": {
22
+ "max_length": 1024,
23
+ "per_device_train_batch_size": 8,
24
+ "auto_find_batch_size": true,
25
+ "num_train_epochs": 1,
26
+ "max_steps": -1,
27
+ "logging_strategy": "steps",
28
+ "save_strategy": "steps",
29
+ "logging_steps": 0.2,
30
+ "save_steps": 0.2,
31
+ "gradient_accumulation_steps": 1,
32
+ "gradient_checkpointing": false,
33
+ "optim": "adamw_torch",
34
+ "learning_rate": 5e-05,
35
+ "weight_decay": 0.01,
36
+ "warmup_steps": 0,
37
+ "report_to": "tensorboard",
38
+ "save_safetensors": true,
39
+ "packing": false
40
+ },
41
+ "use_peft": true,
42
+ "peft_task_type": "CAUSAL_LM",
43
+ "peft_inference_mode": false,
44
+ "peft_r": 8,
45
+ "peft_lora_alpha": 32,
46
+ "peft_lora_dropout": 0.1,
47
+ "fsdp": "none",
48
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
49
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
50
+ "fsdp_fsdp_min_num_params": 100000000.0,
51
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
52
+ "fsdp_fsdp_forward_prefetch": false,
53
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
54
+ "fsdp_fsdp_offload_params": false,
55
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
56
+ "fsdp_fsdp_use_orig_params": true,
57
+ "fsdp_fsdp_activation_checkpointing": false,
58
+ "fsdp_fsdp_limit_all_gathers": true,
59
+ "fsdp_fsdp_sync_module_states": true,
60
+ "use_git": true,
61
+ "random_seed": 0
62
+ }
63
+ },
64
+ "nq_open": {
65
+ "exact_match,remove_whitespace": 0.15290858725761772,
66
+ "exact_match,remove_whitespace_stderr": 0.005990838835970625,
67
+ "args": {
68
+ "exp_name": "results/v1/nq_open/Qwen/Qwen3-1.7B-Base",
69
+ "task_name": "nq_open",
70
+ "model_name": "Qwen/Qwen3-1.7B-Base",
71
+ "dataset_param": {
72
+ "max_num_train": 50000,
73
+ "completion_only": false,
74
+ "apply_chat_template": false
75
+ },
76
+ "resume_from_checkpoint": false,
77
+ "train_param": {
78
+ "max_length": 1024,
79
+ "per_device_train_batch_size": 8,
80
+ "auto_find_batch_size": true,
81
+ "num_train_epochs": 1,
82
+ "max_steps": -1,
83
+ "logging_strategy": "steps",
84
+ "save_strategy": "steps",
85
+ "logging_steps": 0.2,
86
+ "save_steps": 0.2,
87
+ "gradient_accumulation_steps": 1,
88
+ "gradient_checkpointing": false,
89
+ "optim": "adamw_torch",
90
+ "learning_rate": 5e-05,
91
+ "weight_decay": 0.01,
92
+ "warmup_steps": 0,
93
+ "report_to": "tensorboard",
94
+ "save_safetensors": true,
95
+ "packing": false
96
+ },
97
+ "use_peft": true,
98
+ "peft_task_type": "CAUSAL_LM",
99
+ "peft_inference_mode": false,
100
+ "peft_r": 8,
101
+ "peft_lora_alpha": 32,
102
+ "peft_lora_dropout": 0.1,
103
+ "fsdp": "none",
104
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
105
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
106
+ "fsdp_fsdp_min_num_params": 100000000.0,
107
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
108
+ "fsdp_fsdp_forward_prefetch": false,
109
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
110
+ "fsdp_fsdp_offload_params": false,
111
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
112
+ "fsdp_fsdp_use_orig_params": true,
113
+ "fsdp_fsdp_activation_checkpointing": false,
114
+ "fsdp_fsdp_limit_all_gathers": true,
115
+ "fsdp_fsdp_sync_module_states": true,
116
+ "use_git": true,
117
+ "random_seed": 0
118
+ }
119
+ },
120
+ "winogrande": {
121
+ "acc": 0.6977111286503551,
122
+ "acc_stderr": 0.012907200361627538,
123
+ "args": {
124
+ "exp_name": "results/v1/winogrande/Qwen/Qwen3-1.7B-Base",
125
+ "task_name": "winogrande",
126
+ "model_name": "Qwen/Qwen3-1.7B-Base",
127
+ "dataset_param": {
128
+ "max_num_train": 50000,
129
+ "completion_only": false,
130
+ "apply_chat_template": false
131
+ },
132
+ "resume_from_checkpoint": false,
133
+ "train_param": {
134
+ "max_length": 1024,
135
+ "per_device_train_batch_size": 8,
136
+ "auto_find_batch_size": true,
137
+ "num_train_epochs": 1,
138
+ "max_steps": -1,
139
+ "logging_strategy": "steps",
140
+ "save_strategy": "steps",
141
+ "logging_steps": 0.2,
142
+ "save_steps": 0.2,
143
+ "gradient_accumulation_steps": 1,
144
+ "gradient_checkpointing": false,
145
+ "optim": "adamw_torch",
146
+ "learning_rate": 5e-05,
147
+ "weight_decay": 0.01,
148
+ "warmup_steps": 0,
149
+ "report_to": "tensorboard",
150
+ "save_safetensors": true,
151
+ "packing": false
152
+ },
153
+ "use_peft": true,
154
+ "peft_task_type": "CAUSAL_LM",
155
+ "peft_inference_mode": false,
156
+ "peft_r": 8,
157
+ "peft_lora_alpha": 32,
158
+ "peft_lora_dropout": 0.1,
159
+ "fsdp": "none",
160
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
161
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
162
+ "fsdp_fsdp_min_num_params": 100000000.0,
163
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
164
+ "fsdp_fsdp_forward_prefetch": false,
165
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
166
+ "fsdp_fsdp_offload_params": false,
167
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
168
+ "fsdp_fsdp_use_orig_params": true,
169
+ "fsdp_fsdp_activation_checkpointing": false,
170
+ "fsdp_fsdp_limit_all_gathers": true,
171
+ "fsdp_fsdp_sync_module_states": true,
172
+ "use_git": true,
173
+ "random_seed": 0
174
+ }
175
+ },
176
+ "hellaswag": {
177
+ "acc_norm": 0.6919,
178
+ "acc_norm_stderr": 0.004617312092234736,
179
+ "args": {
180
+ "exp_name": "results/v1/hellaswag/Qwen/Qwen3-1.7B-Base",
181
+ "task_name": "hellaswag",
182
+ "model_name": "Qwen/Qwen3-1.7B-Base",
183
+ "dataset_param": {
184
+ "max_num_train": 50000,
185
+ "completion_only": false,
186
+ "apply_chat_template": false
187
+ },
188
+ "resume_from_checkpoint": false,
189
+ "train_param": {
190
+ "max_length": 1024,
191
+ "per_device_train_batch_size": 8,
192
+ "auto_find_batch_size": true,
193
+ "num_train_epochs": 1,
194
+ "max_steps": -1,
195
+ "logging_strategy": "steps",
196
+ "save_strategy": "steps",
197
+ "logging_steps": 0.2,
198
+ "save_steps": 0.2,
199
+ "gradient_accumulation_steps": 1,
200
+ "gradient_checkpointing": false,
201
+ "optim": "adamw_torch",
202
+ "learning_rate": 5e-05,
203
+ "weight_decay": 0.01,
204
+ "warmup_steps": 0,
205
+ "report_to": "tensorboard",
206
+ "save_safetensors": true,
207
+ "packing": false
208
+ },
209
+ "use_peft": true,
210
+ "peft_task_type": "CAUSAL_LM",
211
+ "peft_inference_mode": false,
212
+ "peft_r": 8,
213
+ "peft_lora_alpha": 32,
214
+ "peft_lora_dropout": 0.1,
215
+ "fsdp": "none",
216
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
217
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
218
+ "fsdp_fsdp_min_num_params": 100000000.0,
219
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
220
+ "fsdp_fsdp_forward_prefetch": false,
221
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
222
+ "fsdp_fsdp_offload_params": false,
223
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
224
+ "fsdp_fsdp_use_orig_params": true,
225
+ "fsdp_fsdp_activation_checkpointing": false,
226
+ "fsdp_fsdp_limit_all_gathers": true,
227
+ "fsdp_fsdp_sync_module_states": true,
228
+ "use_git": true,
229
+ "random_seed": 0
230
+ }
231
+ },
232
+ "social_iqa": {
233
+ "acc": 0.5317297850562948,
234
+ "acc_stderr": 0.011291266029595206,
235
+ "args": {
236
+ "exp_name": "results/v1/social_iqa/Qwen/Qwen3-1.7B-Base",
237
+ "task_name": "social_iqa",
238
+ "model_name": "Qwen/Qwen3-1.7B-Base",
239
+ "dataset_param": {
240
+ "max_num_train": 50000,
241
+ "completion_only": false,
242
+ "apply_chat_template": false
243
+ },
244
+ "resume_from_checkpoint": false,
245
+ "train_param": {
246
+ "max_length": 1024,
247
+ "per_device_train_batch_size": 8,
248
+ "auto_find_batch_size": true,
249
+ "num_train_epochs": 1,
250
+ "max_steps": -1,
251
+ "logging_strategy": "steps",
252
+ "save_strategy": "steps",
253
+ "logging_steps": 0.2,
254
+ "save_steps": 0.2,
255
+ "gradient_accumulation_steps": 1,
256
+ "gradient_checkpointing": false,
257
+ "optim": "adamw_torch",
258
+ "learning_rate": 5e-05,
259
+ "weight_decay": 0.01,
260
+ "warmup_steps": 0,
261
+ "report_to": "tensorboard",
262
+ "save_safetensors": true,
263
+ "packing": false
264
+ },
265
+ "use_peft": true,
266
+ "peft_task_type": "CAUSAL_LM",
267
+ "peft_inference_mode": false,
268
+ "peft_r": 8,
269
+ "peft_lora_alpha": 32,
270
+ "peft_lora_dropout": 0.1,
271
+ "fsdp": "none",
272
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
273
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
274
+ "fsdp_fsdp_min_num_params": 100000000.0,
275
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
276
+ "fsdp_fsdp_forward_prefetch": false,
277
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
278
+ "fsdp_fsdp_offload_params": false,
279
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
280
+ "fsdp_fsdp_use_orig_params": true,
281
+ "fsdp_fsdp_activation_checkpointing": false,
282
+ "fsdp_fsdp_limit_all_gathers": true,
283
+ "fsdp_fsdp_sync_module_states": true,
284
+ "use_git": true,
285
+ "random_seed": 0
286
+ }
287
+ },
288
+ "piqa": {
289
+ "acc_norm": 0.7714907508161044,
290
+ "acc_norm_stderr": 0.009796313511829524,
291
+ "args": {
292
+ "exp_name": "results/v2/piqa/Qwen/Qwen3-1.7B-Base",
293
+ "task_name": "piqa",
294
+ "model_name": "Qwen/Qwen3-1.7B-Base",
295
+ "dataset_param": {
296
+ "max_num_train": 50000,
297
+ "completion_only": false,
298
+ "apply_chat_template": false
299
+ },
300
+ "resume_from_checkpoint": false,
301
+ "train_param": {
302
+ "max_length": 1024,
303
+ "per_device_train_batch_size": 8,
304
+ "auto_find_batch_size": true,
305
+ "num_train_epochs": 5,
306
+ "max_steps": -1,
307
+ "logging_strategy": "epoch",
308
+ "save_strategy": "epoch",
309
+ "logging_steps": 0.2,
310
+ "save_steps": 0.2,
311
+ "gradient_accumulation_steps": 1,
312
+ "gradient_checkpointing": false,
313
+ "optim": "adamw_torch",
314
+ "learning_rate": 5e-05,
315
+ "weight_decay": 0.01,
316
+ "warmup_steps": 0,
317
+ "report_to": "tensorboard",
318
+ "save_safetensors": true,
319
+ "packing": false
320
+ },
321
+ "use_peft": true,
322
+ "peft_task_type": "CAUSAL_LM",
323
+ "peft_inference_mode": false,
324
+ "peft_r": 8,
325
+ "peft_lora_alpha": 32,
326
+ "peft_lora_dropout": 0.1,
327
+ "fsdp": "none",
328
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
329
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
330
+ "fsdp_fsdp_min_num_params": 100000000.0,
331
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
332
+ "fsdp_fsdp_forward_prefetch": false,
333
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
334
+ "fsdp_fsdp_offload_params": false,
335
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
336
+ "fsdp_fsdp_use_orig_params": true,
337
+ "fsdp_fsdp_activation_checkpointing": false,
338
+ "fsdp_fsdp_limit_all_gathers": true,
339
+ "fsdp_fsdp_sync_module_states": true,
340
+ "use_git": true,
341
+ "random_seed": 0
342
+ }
343
+ },
344
+ "commonsense_qa": {
345
+ "acc": 0.7772317772317773,
346
+ "acc_stderr": 0.011913022964039576,
347
+ "args": {
348
+ "exp_name": "results/v2/commonsense_qa/Qwen/Qwen3-1.7B-Base",
349
+ "task_name": "commonsense_qa",
350
+ "model_name": "Qwen/Qwen3-1.7B-Base",
351
+ "dataset_param": {
352
+ "max_num_train": 50000,
353
+ "completion_only": false,
354
+ "apply_chat_template": false
355
+ },
356
+ "resume_from_checkpoint": false,
357
+ "train_param": {
358
+ "max_length": 1024,
359
+ "per_device_train_batch_size": 8,
360
+ "auto_find_batch_size": true,
361
+ "num_train_epochs": 5,
362
+ "max_steps": -1,
363
+ "logging_strategy": "epoch",
364
+ "save_strategy": "epoch",
365
+ "logging_steps": 0.2,
366
+ "save_steps": 0.2,
367
+ "gradient_accumulation_steps": 1,
368
+ "gradient_checkpointing": false,
369
+ "optim": "adamw_torch",
370
+ "learning_rate": 5e-05,
371
+ "weight_decay": 0.01,
372
+ "warmup_steps": 0,
373
+ "report_to": "tensorboard",
374
+ "save_safetensors": true,
375
+ "packing": false
376
+ },
377
+ "use_peft": true,
378
+ "peft_task_type": "CAUSAL_LM",
379
+ "peft_inference_mode": false,
380
+ "peft_r": 8,
381
+ "peft_lora_alpha": 32,
382
+ "peft_lora_dropout": 0.1,
383
+ "fsdp": "none",
384
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
385
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
386
+ "fsdp_fsdp_min_num_params": 100000000.0,
387
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
388
+ "fsdp_fsdp_forward_prefetch": false,
389
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
390
+ "fsdp_fsdp_offload_params": false,
391
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
392
+ "fsdp_fsdp_use_orig_params": true,
393
+ "fsdp_fsdp_activation_checkpointing": false,
394
+ "fsdp_fsdp_limit_all_gathers": true,
395
+ "fsdp_fsdp_sync_module_states": true,
396
+ "use_git": true,
397
+ "random_seed": 0
398
+ }
399
+ },
400
+ "gsm8k": {
401
+ "exact_match,flexible-extract": 0.6694465504169825,
402
+ "exact_match,flexible-extract_stderr": 0.012957496367085026,
403
+ "args": {
404
+ "exp_name": "results/v2/gsm8k/Qwen/Qwen3-1.7B-Base",
405
+ "task_name": "gsm8k",
406
+ "model_name": "Qwen/Qwen3-1.7B-Base",
407
+ "dataset_param": {
408
+ "max_num_train": 50000,
409
+ "completion_only": false,
410
+ "apply_chat_template": false
411
+ },
412
+ "resume_from_checkpoint": false,
413
+ "train_param": {
414
+ "max_length": 1024,
415
+ "per_device_train_batch_size": 8,
416
+ "auto_find_batch_size": true,
417
+ "num_train_epochs": 5,
418
+ "max_steps": -1,
419
+ "logging_strategy": "epoch",
420
+ "save_strategy": "epoch",
421
+ "logging_steps": 0.2,
422
+ "save_steps": 0.2,
423
+ "gradient_accumulation_steps": 1,
424
+ "gradient_checkpointing": false,
425
+ "optim": "adamw_torch",
426
+ "learning_rate": 5e-05,
427
+ "weight_decay": 0.01,
428
+ "warmup_steps": 0,
429
+ "report_to": "tensorboard",
430
+ "save_safetensors": true,
431
+ "packing": false
432
+ },
433
+ "use_peft": true,
434
+ "peft_task_type": "CAUSAL_LM",
435
+ "peft_inference_mode": false,
436
+ "peft_r": 8,
437
+ "peft_lora_alpha": 32,
438
+ "peft_lora_dropout": 0.1,
439
+ "fsdp": "none",
440
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
441
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
442
+ "fsdp_fsdp_min_num_params": 100000000.0,
443
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
444
+ "fsdp_fsdp_forward_prefetch": false,
445
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
446
+ "fsdp_fsdp_offload_params": false,
447
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
448
+ "fsdp_fsdp_use_orig_params": true,
449
+ "fsdp_fsdp_activation_checkpointing": false,
450
+ "fsdp_fsdp_limit_all_gathers": true,
451
+ "fsdp_fsdp_sync_module_states": true,
452
+ "use_git": true,
453
+ "random_seed": 0
454
+ }
455
+ },
456
+ "arc_challenge": {
457
+ "acc_norm": 0.5725255972696246,
458
+ "acc_norm_stderr": 0.014456862944650652,
459
+ "args": {
460
+ "exp_name": "results/v2/arc_challenge/Qwen/Qwen3-1.7B-Base",
461
+ "task_name": "arc_challenge",
462
+ "model_name": "Qwen/Qwen3-1.7B-Base",
463
+ "dataset_param": {
464
+ "max_num_train": 50000,
465
+ "completion_only": false,
466
+ "apply_chat_template": false
467
+ },
468
+ "resume_from_checkpoint": false,
469
+ "train_param": {
470
+ "max_length": 1024,
471
+ "per_device_train_batch_size": 8,
472
+ "auto_find_batch_size": true,
473
+ "num_train_epochs": 5,
474
+ "max_steps": -1,
475
+ "logging_strategy": "epoch",
476
+ "save_strategy": "epoch",
477
+ "logging_steps": 0.2,
478
+ "save_steps": 0.2,
479
+ "gradient_accumulation_steps": 1,
480
+ "gradient_checkpointing": false,
481
+ "optim": "adamw_torch",
482
+ "learning_rate": 5e-05,
483
+ "weight_decay": 0.01,
484
+ "warmup_steps": 0,
485
+ "report_to": "tensorboard",
486
+ "save_safetensors": true,
487
+ "packing": false
488
+ },
489
+ "use_peft": true,
490
+ "peft_task_type": "CAUSAL_LM",
491
+ "peft_inference_mode": false,
492
+ "peft_r": 8,
493
+ "peft_lora_alpha": 32,
494
+ "peft_lora_dropout": 0.1,
495
+ "fsdp": "none",
496
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
497
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
498
+ "fsdp_fsdp_min_num_params": 100000000.0,
499
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
500
+ "fsdp_fsdp_forward_prefetch": false,
501
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
502
+ "fsdp_fsdp_offload_params": false,
503
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
504
+ "fsdp_fsdp_use_orig_params": true,
505
+ "fsdp_fsdp_activation_checkpointing": false,
506
+ "fsdp_fsdp_limit_all_gathers": true,
507
+ "fsdp_fsdp_sync_module_states": true,
508
+ "use_git": true,
509
+ "random_seed": 0
510
+ }
511
+ }
512
+ }
513
+ }
Qwen/Qwen3-1.7B/results_2026-01-05-04-27-51.json ADDED
@@ -0,0 +1,513 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "model_name": "Qwen/Qwen3-1.7B",
4
+ "model_type": "instruction-tuned",
5
+ "params": 1.7
6
+ },
7
+ "results": {
8
+ "medmcqa": {
9
+ "acc_norm": 0.4750179297155152,
10
+ "acc_norm_stderr": 0.0077220964660986565,
11
+ "args": {
12
+ "exp_name": "results/v1/medmcqa/Qwen/Qwen3-1.7B",
13
+ "task_name": "medmcqa",
14
+ "model_name": "Qwen/Qwen3-1.7B",
15
+ "dataset_param": {
16
+ "max_num_train": 50000,
17
+ "completion_only": false,
18
+ "apply_chat_template": true
19
+ },
20
+ "resume_from_checkpoint": false,
21
+ "train_param": {
22
+ "max_length": 1024,
23
+ "per_device_train_batch_size": 8,
24
+ "auto_find_batch_size": true,
25
+ "num_train_epochs": 1,
26
+ "max_steps": -1,
27
+ "logging_strategy": "steps",
28
+ "save_strategy": "steps",
29
+ "logging_steps": 0.2,
30
+ "save_steps": 0.2,
31
+ "gradient_accumulation_steps": 1,
32
+ "gradient_checkpointing": false,
33
+ "optim": "adamw_torch",
34
+ "learning_rate": 5e-05,
35
+ "weight_decay": 0.01,
36
+ "warmup_steps": 0,
37
+ "report_to": "tensorboard",
38
+ "save_safetensors": true,
39
+ "packing": false
40
+ },
41
+ "use_peft": true,
42
+ "peft_task_type": "CAUSAL_LM",
43
+ "peft_inference_mode": false,
44
+ "peft_r": 8,
45
+ "peft_lora_alpha": 32,
46
+ "peft_lora_dropout": 0.1,
47
+ "fsdp": "none",
48
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
49
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
50
+ "fsdp_fsdp_min_num_params": 100000000.0,
51
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
52
+ "fsdp_fsdp_forward_prefetch": false,
53
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
54
+ "fsdp_fsdp_offload_params": false,
55
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
56
+ "fsdp_fsdp_use_orig_params": true,
57
+ "fsdp_fsdp_activation_checkpointing": false,
58
+ "fsdp_fsdp_limit_all_gathers": true,
59
+ "fsdp_fsdp_sync_module_states": true,
60
+ "use_git": true,
61
+ "random_seed": 0
62
+ }
63
+ },
64
+ "nq_open": {
65
+ "exact_match,remove_whitespace": 0.1265927977839335,
66
+ "exact_match,remove_whitespace_stderr": 0.00553502316066295,
67
+ "args": {
68
+ "exp_name": "results/v1/nq_open/Qwen/Qwen3-1.7B",
69
+ "task_name": "nq_open",
70
+ "model_name": "Qwen/Qwen3-1.7B",
71
+ "dataset_param": {
72
+ "max_num_train": 50000,
73
+ "completion_only": false,
74
+ "apply_chat_template": true
75
+ },
76
+ "resume_from_checkpoint": false,
77
+ "train_param": {
78
+ "max_length": 1024,
79
+ "per_device_train_batch_size": 8,
80
+ "auto_find_batch_size": true,
81
+ "num_train_epochs": 1,
82
+ "max_steps": -1,
83
+ "logging_strategy": "steps",
84
+ "save_strategy": "steps",
85
+ "logging_steps": 0.2,
86
+ "save_steps": 0.2,
87
+ "gradient_accumulation_steps": 1,
88
+ "gradient_checkpointing": false,
89
+ "optim": "adamw_torch",
90
+ "learning_rate": 5e-05,
91
+ "weight_decay": 0.01,
92
+ "warmup_steps": 0,
93
+ "report_to": "tensorboard",
94
+ "save_safetensors": true,
95
+ "packing": false
96
+ },
97
+ "use_peft": true,
98
+ "peft_task_type": "CAUSAL_LM",
99
+ "peft_inference_mode": false,
100
+ "peft_r": 8,
101
+ "peft_lora_alpha": 32,
102
+ "peft_lora_dropout": 0.1,
103
+ "fsdp": "none",
104
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
105
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
106
+ "fsdp_fsdp_min_num_params": 100000000.0,
107
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
108
+ "fsdp_fsdp_forward_prefetch": false,
109
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
110
+ "fsdp_fsdp_offload_params": false,
111
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
112
+ "fsdp_fsdp_use_orig_params": true,
113
+ "fsdp_fsdp_activation_checkpointing": false,
114
+ "fsdp_fsdp_limit_all_gathers": true,
115
+ "fsdp_fsdp_sync_module_states": true,
116
+ "use_git": true,
117
+ "random_seed": 0
118
+ }
119
+ },
120
+ "winogrande": {
121
+ "acc": 0.6842936069455406,
122
+ "acc_stderr": 0.01306309474300081,
123
+ "args": {
124
+ "exp_name": "results/v1/winogrande/Qwen/Qwen3-1.7B",
125
+ "task_name": "winogrande",
126
+ "model_name": "Qwen/Qwen3-1.7B",
127
+ "dataset_param": {
128
+ "max_num_train": 50000,
129
+ "completion_only": false,
130
+ "apply_chat_template": true
131
+ },
132
+ "resume_from_checkpoint": false,
133
+ "train_param": {
134
+ "max_length": 1024,
135
+ "per_device_train_batch_size": 8,
136
+ "auto_find_batch_size": true,
137
+ "num_train_epochs": 1,
138
+ "max_steps": -1,
139
+ "logging_strategy": "steps",
140
+ "save_strategy": "steps",
141
+ "logging_steps": 0.2,
142
+ "save_steps": 0.2,
143
+ "gradient_accumulation_steps": 1,
144
+ "gradient_checkpointing": false,
145
+ "optim": "adamw_torch",
146
+ "learning_rate": 5e-05,
147
+ "weight_decay": 0.01,
148
+ "warmup_steps": 0,
149
+ "report_to": "tensorboard",
150
+ "save_safetensors": true,
151
+ "packing": false
152
+ },
153
+ "use_peft": true,
154
+ "peft_task_type": "CAUSAL_LM",
155
+ "peft_inference_mode": false,
156
+ "peft_r": 8,
157
+ "peft_lora_alpha": 32,
158
+ "peft_lora_dropout": 0.1,
159
+ "fsdp": "none",
160
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
161
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
162
+ "fsdp_fsdp_min_num_params": 100000000.0,
163
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
164
+ "fsdp_fsdp_forward_prefetch": false,
165
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
166
+ "fsdp_fsdp_offload_params": false,
167
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
168
+ "fsdp_fsdp_use_orig_params": true,
169
+ "fsdp_fsdp_activation_checkpointing": false,
170
+ "fsdp_fsdp_limit_all_gathers": true,
171
+ "fsdp_fsdp_sync_module_states": true,
172
+ "use_git": true,
173
+ "random_seed": 0
174
+ }
175
+ },
176
+ "hellaswag": {
177
+ "acc_norm": 0.6551,
178
+ "acc_norm_stderr": 0.004753594289152844,
179
+ "args": {
180
+ "exp_name": "results/v1/hellaswag/Qwen/Qwen3-1.7B",
181
+ "task_name": "hellaswag",
182
+ "model_name": "Qwen/Qwen3-1.7B",
183
+ "dataset_param": {
184
+ "max_num_train": 50000,
185
+ "completion_only": false,
186
+ "apply_chat_template": true
187
+ },
188
+ "resume_from_checkpoint": false,
189
+ "train_param": {
190
+ "max_length": 1024,
191
+ "per_device_train_batch_size": 8,
192
+ "auto_find_batch_size": true,
193
+ "num_train_epochs": 1,
194
+ "max_steps": -1,
195
+ "logging_strategy": "steps",
196
+ "save_strategy": "steps",
197
+ "logging_steps": 0.2,
198
+ "save_steps": 0.2,
199
+ "gradient_accumulation_steps": 1,
200
+ "gradient_checkpointing": false,
201
+ "optim": "adamw_torch",
202
+ "learning_rate": 5e-05,
203
+ "weight_decay": 0.01,
204
+ "warmup_steps": 0,
205
+ "report_to": "tensorboard",
206
+ "save_safetensors": true,
207
+ "packing": false
208
+ },
209
+ "use_peft": true,
210
+ "peft_task_type": "CAUSAL_LM",
211
+ "peft_inference_mode": false,
212
+ "peft_r": 8,
213
+ "peft_lora_alpha": 32,
214
+ "peft_lora_dropout": 0.1,
215
+ "fsdp": "none",
216
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
217
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
218
+ "fsdp_fsdp_min_num_params": 100000000.0,
219
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
220
+ "fsdp_fsdp_forward_prefetch": false,
221
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
222
+ "fsdp_fsdp_offload_params": false,
223
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
224
+ "fsdp_fsdp_use_orig_params": true,
225
+ "fsdp_fsdp_activation_checkpointing": false,
226
+ "fsdp_fsdp_limit_all_gathers": true,
227
+ "fsdp_fsdp_sync_module_states": true,
228
+ "use_git": true,
229
+ "random_seed": 0
230
+ }
231
+ },
232
+ "social_iqa": {
233
+ "acc": 0.49744114636642783,
234
+ "acc_stderr": 0.011313922391063573,
235
+ "args": {
236
+ "exp_name": "results/v1/social_iqa/Qwen/Qwen3-1.7B",
237
+ "task_name": "social_iqa",
238
+ "model_name": "Qwen/Qwen3-1.7B",
239
+ "dataset_param": {
240
+ "max_num_train": 50000,
241
+ "completion_only": false,
242
+ "apply_chat_template": true
243
+ },
244
+ "resume_from_checkpoint": false,
245
+ "train_param": {
246
+ "max_length": 1024,
247
+ "per_device_train_batch_size": 8,
248
+ "auto_find_batch_size": true,
249
+ "num_train_epochs": 1,
250
+ "max_steps": -1,
251
+ "logging_strategy": "steps",
252
+ "save_strategy": "steps",
253
+ "logging_steps": 0.2,
254
+ "save_steps": 0.2,
255
+ "gradient_accumulation_steps": 1,
256
+ "gradient_checkpointing": false,
257
+ "optim": "adamw_torch",
258
+ "learning_rate": 5e-05,
259
+ "weight_decay": 0.01,
260
+ "warmup_steps": 0,
261
+ "report_to": "tensorboard",
262
+ "save_safetensors": true,
263
+ "packing": false
264
+ },
265
+ "use_peft": true,
266
+ "peft_task_type": "CAUSAL_LM",
267
+ "peft_inference_mode": false,
268
+ "peft_r": 8,
269
+ "peft_lora_alpha": 32,
270
+ "peft_lora_dropout": 0.1,
271
+ "fsdp": "none",
272
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
273
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
274
+ "fsdp_fsdp_min_num_params": 100000000.0,
275
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
276
+ "fsdp_fsdp_forward_prefetch": false,
277
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
278
+ "fsdp_fsdp_offload_params": false,
279
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
280
+ "fsdp_fsdp_use_orig_params": true,
281
+ "fsdp_fsdp_activation_checkpointing": false,
282
+ "fsdp_fsdp_limit_all_gathers": true,
283
+ "fsdp_fsdp_sync_module_states": true,
284
+ "use_git": true,
285
+ "random_seed": 0
286
+ }
287
+ },
288
+ "piqa": {
289
+ "acc_norm": 0.7633297062023939,
290
+ "acc_norm_stderr": 0.009916841655042807,
291
+ "args": {
292
+ "exp_name": "results/v2/piqa/Qwen/Qwen3-1.7B",
293
+ "task_name": "piqa",
294
+ "model_name": "Qwen/Qwen3-1.7B",
295
+ "dataset_param": {
296
+ "max_num_train": 50000,
297
+ "completion_only": false,
298
+ "apply_chat_template": true
299
+ },
300
+ "resume_from_checkpoint": false,
301
+ "train_param": {
302
+ "max_length": 1024,
303
+ "per_device_train_batch_size": 8,
304
+ "auto_find_batch_size": true,
305
+ "num_train_epochs": 5,
306
+ "max_steps": -1,
307
+ "logging_strategy": "epoch",
308
+ "save_strategy": "epoch",
309
+ "logging_steps": 0.2,
310
+ "save_steps": 0.2,
311
+ "gradient_accumulation_steps": 1,
312
+ "gradient_checkpointing": false,
313
+ "optim": "adamw_torch",
314
+ "learning_rate": 5e-05,
315
+ "weight_decay": 0.01,
316
+ "warmup_steps": 0,
317
+ "report_to": "tensorboard",
318
+ "save_safetensors": true,
319
+ "packing": false
320
+ },
321
+ "use_peft": true,
322
+ "peft_task_type": "CAUSAL_LM",
323
+ "peft_inference_mode": false,
324
+ "peft_r": 8,
325
+ "peft_lora_alpha": 32,
326
+ "peft_lora_dropout": 0.1,
327
+ "fsdp": "none",
328
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
329
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
330
+ "fsdp_fsdp_min_num_params": 100000000.0,
331
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
332
+ "fsdp_fsdp_forward_prefetch": false,
333
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
334
+ "fsdp_fsdp_offload_params": false,
335
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
336
+ "fsdp_fsdp_use_orig_params": true,
337
+ "fsdp_fsdp_activation_checkpointing": false,
338
+ "fsdp_fsdp_limit_all_gathers": true,
339
+ "fsdp_fsdp_sync_module_states": true,
340
+ "use_git": true,
341
+ "random_seed": 0
342
+ }
343
+ },
344
+ "commonsense_qa": {
345
+ "acc": 0.7665847665847666,
346
+ "acc_stderr": 0.012110575321206414,
347
+ "args": {
348
+ "exp_name": "results/v2/commonsense_qa/Qwen/Qwen3-1.7B",
349
+ "task_name": "commonsense_qa",
350
+ "model_name": "Qwen/Qwen3-1.7B",
351
+ "dataset_param": {
352
+ "max_num_train": 50000,
353
+ "completion_only": false,
354
+ "apply_chat_template": true
355
+ },
356
+ "resume_from_checkpoint": false,
357
+ "train_param": {
358
+ "max_length": 1024,
359
+ "per_device_train_batch_size": 8,
360
+ "auto_find_batch_size": true,
361
+ "num_train_epochs": 5,
362
+ "max_steps": -1,
363
+ "logging_strategy": "epoch",
364
+ "save_strategy": "epoch",
365
+ "logging_steps": 0.2,
366
+ "save_steps": 0.2,
367
+ "gradient_accumulation_steps": 1,
368
+ "gradient_checkpointing": false,
369
+ "optim": "adamw_torch",
370
+ "learning_rate": 5e-05,
371
+ "weight_decay": 0.01,
372
+ "warmup_steps": 0,
373
+ "report_to": "tensorboard",
374
+ "save_safetensors": true,
375
+ "packing": false
376
+ },
377
+ "use_peft": true,
378
+ "peft_task_type": "CAUSAL_LM",
379
+ "peft_inference_mode": false,
380
+ "peft_r": 8,
381
+ "peft_lora_alpha": 32,
382
+ "peft_lora_dropout": 0.1,
383
+ "fsdp": "none",
384
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
385
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
386
+ "fsdp_fsdp_min_num_params": 100000000.0,
387
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
388
+ "fsdp_fsdp_forward_prefetch": false,
389
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
390
+ "fsdp_fsdp_offload_params": false,
391
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
392
+ "fsdp_fsdp_use_orig_params": true,
393
+ "fsdp_fsdp_activation_checkpointing": false,
394
+ "fsdp_fsdp_limit_all_gathers": true,
395
+ "fsdp_fsdp_sync_module_states": true,
396
+ "use_git": true,
397
+ "random_seed": 0
398
+ }
399
+ },
400
+ "gsm8k": {
401
+ "exact_match,flexible-extract": 0.5989385898407885,
402
+ "exact_match,flexible-extract_stderr": 0.013500158922245539,
403
+ "args": {
404
+ "exp_name": "results/v2/gsm8k/Qwen/Qwen3-1.7B",
405
+ "task_name": "gsm8k",
406
+ "model_name": "Qwen/Qwen3-1.7B",
407
+ "dataset_param": {
408
+ "max_num_train": 50000,
409
+ "completion_only": false,
410
+ "apply_chat_template": true
411
+ },
412
+ "resume_from_checkpoint": false,
413
+ "train_param": {
414
+ "max_length": 1024,
415
+ "per_device_train_batch_size": 8,
416
+ "auto_find_batch_size": true,
417
+ "num_train_epochs": 5,
418
+ "max_steps": -1,
419
+ "logging_strategy": "epoch",
420
+ "save_strategy": "epoch",
421
+ "logging_steps": 0.2,
422
+ "save_steps": 0.2,
423
+ "gradient_accumulation_steps": 1,
424
+ "gradient_checkpointing": false,
425
+ "optim": "adamw_torch",
426
+ "learning_rate": 5e-05,
427
+ "weight_decay": 0.01,
428
+ "warmup_steps": 0,
429
+ "report_to": "tensorboard",
430
+ "save_safetensors": true,
431
+ "packing": false
432
+ },
433
+ "use_peft": true,
434
+ "peft_task_type": "CAUSAL_LM",
435
+ "peft_inference_mode": false,
436
+ "peft_r": 8,
437
+ "peft_lora_alpha": 32,
438
+ "peft_lora_dropout": 0.1,
439
+ "fsdp": "none",
440
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
441
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
442
+ "fsdp_fsdp_min_num_params": 100000000.0,
443
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
444
+ "fsdp_fsdp_forward_prefetch": false,
445
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
446
+ "fsdp_fsdp_offload_params": false,
447
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
448
+ "fsdp_fsdp_use_orig_params": true,
449
+ "fsdp_fsdp_activation_checkpointing": false,
450
+ "fsdp_fsdp_limit_all_gathers": true,
451
+ "fsdp_fsdp_sync_module_states": true,
452
+ "use_git": true,
453
+ "random_seed": 0
454
+ }
455
+ },
456
+ "arc_challenge": {
457
+ "acc_norm": 0.4880546075085324,
458
+ "acc_norm_stderr": 0.014607220340597171,
459
+ "args": {
460
+ "exp_name": "results/v2/arc_challenge/Qwen/Qwen3-1.7B",
461
+ "task_name": "arc_challenge",
462
+ "model_name": "Qwen/Qwen3-1.7B",
463
+ "dataset_param": {
464
+ "max_num_train": 50000,
465
+ "completion_only": false,
466
+ "apply_chat_template": true
467
+ },
468
+ "resume_from_checkpoint": false,
469
+ "train_param": {
470
+ "max_length": 1024,
471
+ "per_device_train_batch_size": 8,
472
+ "auto_find_batch_size": true,
473
+ "num_train_epochs": 5,
474
+ "max_steps": -1,
475
+ "logging_strategy": "epoch",
476
+ "save_strategy": "epoch",
477
+ "logging_steps": 0.2,
478
+ "save_steps": 0.2,
479
+ "gradient_accumulation_steps": 1,
480
+ "gradient_checkpointing": false,
481
+ "optim": "adamw_torch",
482
+ "learning_rate": 5e-05,
483
+ "weight_decay": 0.01,
484
+ "warmup_steps": 0,
485
+ "report_to": "tensorboard",
486
+ "save_safetensors": true,
487
+ "packing": false
488
+ },
489
+ "use_peft": true,
490
+ "peft_task_type": "CAUSAL_LM",
491
+ "peft_inference_mode": false,
492
+ "peft_r": 8,
493
+ "peft_lora_alpha": 32,
494
+ "peft_lora_dropout": 0.1,
495
+ "fsdp": "none",
496
+ "fsdp_fsdp_auto_wrap_policy": "TRANSFORMER_BASED_WRAP",
497
+ "fsdp_fsdp_transformer_layer_cls_to_wrap": null,
498
+ "fsdp_fsdp_min_num_params": 100000000.0,
499
+ "fsdp_fsdp_backward_prefetch": "BACKWARD_PRE",
500
+ "fsdp_fsdp_forward_prefetch": false,
501
+ "fsdp_fsdp_state_dict_type": "FULL_STATE_DICT",
502
+ "fsdp_fsdp_offload_params": false,
503
+ "fsdp_fsdp_cpu_ram_efficient_loading": false,
504
+ "fsdp_fsdp_use_orig_params": true,
505
+ "fsdp_fsdp_activation_checkpointing": false,
506
+ "fsdp_fsdp_limit_all_gathers": true,
507
+ "fsdp_fsdp_sync_module_states": true,
508
+ "use_git": true,
509
+ "random_seed": 0
510
+ }
511
+ }
512
+ }
513
+ }