Hugging Face's logo

chentianqi
/
Qwen2.5-1.5B-Open-R1-Distill-ScaleTuning

Text Generation
Transformers
Safetensors
qwen2
Generated from Trainer
open-r1
trl
sft
conversational
text-generation-inference
4-bit precision
gptq
Model card Files
xet
 Community
chentianqi commited on
Commit
837cf14
·
verified ·
1 Parent(s): 182a9b7

End of training

Browse files
Files changed (6) hide show
  1. README.md +59 -0
  2. adapter_model.safetensors +1 -1
  3. all_results.json +8 -0
  4. config.json +57 -0
  5. train_results.json +8 -0
  6. trainer_state.json +413 -0
README.md ADDED
@@ -0,0 +1,59 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ datasets: open-r1/OpenR1-Math-220k
3
+ library_name: transformers
4
+ model_name: Qwen2.5-1.5B-Open-R1-Distill-ScaleTuning
5
+ tags:
6
+ - generated_from_trainer
7
+ - open-r1
8
+ - trl
9
+ - sft
10
+ licence: license
11
+ ---
12
+
13
+ # Model Card for Qwen2.5-1.5B-Open-R1-Distill-ScaleTuning
14
+
15
+ This model is a fine-tuned version of [None](https://huggingface.co/None) on the [open-r1/OpenR1-Math-220k](https://huggingface.co/datasets/open-r1/OpenR1-Math-220k) dataset.
16
+ It has been trained using [TRL](https://github.com/huggingface/trl).
17
+
18
+ ## Quick start
19
+
20
+ ```python
21
+ from transformers import pipeline
22
+
23
+ question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
24
+ generator = pipeline("text-generation", model="chentianqi/Qwen2.5-1.5B-Open-R1-Distill-ScaleTuning", device="cuda")
25
+ output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
26
+ print(output["generated_text"])
27
+ ```
28
+
29
+ ## Training procedure
30
+
31
+
32
+
33
+
34
+ This model was trained with SFT.
35
+
36
+ ### Framework versions
37
+
38
+ - TRL: 0.16.0.dev0
39
+ - Transformers: 4.49.0
40
+ - Pytorch: 2.5.1
41
+ - Datasets: 3.2.0
42
+ - Tokenizers: 0.21.1
43
+
44
+ ## Citations
45
+
46
+
47
+
48
+ Cite TRL as:
49
+
50
+ ```bibtex
51
+ @misc{vonwerra2022trl,
52
+ title = {{TRL: Transformer Reinforcement Learning}},
53
+ author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallouédec},
54
+ year = 2020,
55
+ journal = {GitHub repository},
56
+ publisher = {GitHub},
57
+ howpublished = {\url{https://github.com/huggingface/trl}}
58
+ }
59
+ ```
adapter_model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:f4a4a0f404c76bd532d05ec1e30900a83d24953ad3dd5069cf54355e57eca9a1
3
  size 20497672
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3d41f45949006dda35784f86e285393efaca36cd602a1f9cf18b39c6fa084c20
3
  size 20497672
all_results.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "total_flos": 4349016447909888.0,
3
+ "train_loss": 0.638175894520176,
4
+ "train_runtime": 10527.0005,
5
+ "train_samples": 93733,
6
+ "train_samples_per_second": 3.258,
7
+ "train_steps_per_second": 0.025
8
+ }
config.json ADDED
@@ -0,0 +1,57 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_attn_implementation_autoset": true,
3
+ "_name_or_path": "/home/chentianqi/model/Qwen/Qwen2.5-1.5B-Instruct-GPTQ-Int4",
4
+ "architectures": [
5
+ "Qwen2ForCausalLM"
6
+ ],
7
+ "attention_dropout": 0.0,
8
+ "bos_token_id": 151643,
9
+ "eos_token_id": 151645,
10
+ "hidden_act": "silu",
11
+ "hidden_size": 1536,
12
+ "initializer_range": 0.02,
13
+ "intermediate_size": 8960,
14
+ "max_position_embeddings": 32768,
15
+ "max_window_layers": 21,
16
+ "model_type": "qwen2",
17
+ "num_attention_heads": 12,
18
+ "num_hidden_layers": 28,
19
+ "num_key_value_heads": 2,
20
+ "quantization_config": {
21
+ "backend": "auto",
22
+ "batch_size": 1,
23
+ "bits": 4,
24
+ "block_name_to_quantize": null,
25
+ "cache_block_outputs": true,
26
+ "checkpoint_format": "gptq",
27
+ "damp_percent": 0.01,
28
+ "dataset": null,
29
+ "desc_act": false,
30
+ "exllama_config": {
31
+ "version": 1
32
+ },
33
+ "group_size": 128,
34
+ "max_input_length": null,
35
+ "meta": null,
36
+ "model_seqlen": null,
37
+ "module_name_preceding_first_block": null,
38
+ "modules_in_block_to_quantize": null,
39
+ "pad_token_id": null,
40
+ "quant_method": "gptq",
41
+ "sym": true,
42
+ "tokenizer": null,
43
+ "true_sequential": true,
44
+ "use_cuda_fp16": false,
45
+ "use_exllama": true
46
+ },
47
+ "rms_norm_eps": 1e-06,
48
+ "rope_scaling": null,
49
+ "rope_theta": 1000000.0,
50
+ "sliding_window": 32768,
51
+ "tie_word_embeddings": true,
52
+ "torch_dtype": "bfloat16",
53
+ "transformers_version": "4.49.0",
54
+ "use_cache": true,
55
+ "use_sliding_window": false,
56
+ "vocab_size": 151936
57
+ }
train_results.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "total_flos": 4349016447909888.0,
3
+ "train_loss": 0.638175894520176,
4
+ "train_runtime": 10527.0005,
5
+ "train_samples": 93733,
6
+ "train_samples_per_second": 3.258,
7
+ "train_steps_per_second": 0.025
8
+ }
trainer_state.json ADDED
@@ -0,0 +1,413 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 1.0,
5
+ "eval_steps": 500,
6
+ "global_step": 268,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.018656716417910446,
13
+ "grad_norm": 2.6418865664688487,
14
+ "learning_rate": 1.785714285714286e-05,
15
+ "loss": 0.8869,
16
+ "step": 5
17
+ },
18
+ {
19
+ "epoch": 0.03731343283582089,
20
+ "grad_norm": 2.5407967486574643,
21
+ "learning_rate": 3.571428571428572e-05,
22
+ "loss": 0.8485,
23
+ "step": 10
24
+ },
25
+ {
26
+ "epoch": 0.055970149253731345,
27
+ "grad_norm": 1.206864418084265,
28
+ "learning_rate": 4.999827900623038e-05,
29
+ "loss": 0.7708,
30
+ "step": 15
31
+ },
32
+ {
33
+ "epoch": 0.07462686567164178,
34
+ "grad_norm": 0.7638177576934349,
35
+ "learning_rate": 4.993807186343243e-05,
36
+ "loss": 0.7332,
37
+ "step": 20
38
+ },
39
+ {
40
+ "epoch": 0.09328358208955224,
41
+ "grad_norm": 0.5954972247062816,
42
+ "learning_rate": 4.979207812402531e-05,
43
+ "loss": 0.6995,
44
+ "step": 25
45
+ },
46
+ {
47
+ "epoch": 0.11194029850746269,
48
+ "grad_norm": 0.4813118006100347,
49
+ "learning_rate": 4.956085596012407e-05,
50
+ "loss": 0.6912,
51
+ "step": 30
52
+ },
53
+ {
54
+ "epoch": 0.13059701492537312,
55
+ "grad_norm": 0.445300379853154,
56
+ "learning_rate": 4.924528939432311e-05,
57
+ "loss": 0.674,
58
+ "step": 35
59
+ },
60
+ {
61
+ "epoch": 0.14925373134328357,
62
+ "grad_norm": 0.4020802905302009,
63
+ "learning_rate": 4.884658491984735e-05,
64
+ "loss": 0.665,
65
+ "step": 40
66
+ },
67
+ {
68
+ "epoch": 0.16791044776119404,
69
+ "grad_norm": 0.40863213996310194,
70
+ "learning_rate": 4.8366266887814235e-05,
71
+ "loss": 0.6659,
72
+ "step": 45
73
+ },
74
+ {
75
+ "epoch": 0.1865671641791045,
76
+ "grad_norm": 0.3527401341977073,
77
+ "learning_rate": 4.780617167924209e-05,
78
+ "loss": 0.648,
79
+ "step": 50
80
+ },
81
+ {
82
+ "epoch": 0.20522388059701493,
83
+ "grad_norm": 0.3951425883208962,
84
+ "learning_rate": 4.716844068408693e-05,
85
+ "loss": 0.651,
86
+ "step": 55
87
+ },
88
+ {
89
+ "epoch": 0.22388059701492538,
90
+ "grad_norm": 0.38406053219373687,
91
+ "learning_rate": 4.6455512114150546e-05,
92
+ "loss": 0.6465,
93
+ "step": 60
94
+ },
95
+ {
96
+ "epoch": 0.24253731343283583,
97
+ "grad_norm": 0.3801139364450689,
98
+ "learning_rate": 4.5670111681161296e-05,
99
+ "loss": 0.6368,
100
+ "step": 65
101
+ },
102
+ {
103
+ "epoch": 0.26119402985074625,
104
+ "grad_norm": 0.346192663338788,
105
+ "learning_rate": 4.481524217566783e-05,
106
+ "loss": 0.6336,
107
+ "step": 70
108
+ },
109
+ {
110
+ "epoch": 0.2798507462686567,
111
+ "grad_norm": 0.39271666638999647,
112
+ "learning_rate": 4.3894171986588217e-05,
113
+ "loss": 0.6322,
114
+ "step": 75
115
+ },
116
+ {
117
+ "epoch": 0.29850746268656714,
118
+ "grad_norm": 0.3647408038419753,
119
+ "learning_rate": 4.29104226053073e-05,
120
+ "loss": 0.6316,
121
+ "step": 80
122
+ },
123
+ {
124
+ "epoch": 0.31716417910447764,
125
+ "grad_norm": 0.3631004101545322,
126
+ "learning_rate": 4.186775516209732e-05,
127
+ "loss": 0.6276,
128
+ "step": 85
129
+ },
130
+ {
131
+ "epoch": 0.3358208955223881,
132
+ "grad_norm": 0.3910048503901255,
133
+ "learning_rate": 4.077015604633669e-05,
134
+ "loss": 0.631,
135
+ "step": 90
136
+ },
137
+ {
138
+ "epoch": 0.35447761194029853,
139
+ "grad_norm": 0.35963367987288647,
140
+ "learning_rate": 3.962182166550441e-05,
141
+ "loss": 0.6296,
142
+ "step": 95
143
+ },
144
+ {
145
+ "epoch": 0.373134328358209,
146
+ "grad_norm": 0.41095954078127345,
147
+ "learning_rate": 3.8427142401220634e-05,
148
+ "loss": 0.6256,
149
+ "step": 100
150
+ },
151
+ {
152
+ "epoch": 0.3917910447761194,
153
+ "grad_norm": 0.392459358109425,
154
+ "learning_rate": 3.71906858236735e-05,
155
+ "loss": 0.6223,
156
+ "step": 105
157
+ },
158
+ {
159
+ "epoch": 0.41044776119402987,
160
+ "grad_norm": 0.34802029972198373,
161
+ "learning_rate": 3.591717922860785e-05,
162
+ "loss": 0.63,
163
+ "step": 110
164
+ },
165
+ {
166
+ "epoch": 0.4291044776119403,
167
+ "grad_norm": 0.3539677382034743,
168
+ "learning_rate": 3.46114915636416e-05,
169
+ "loss": 0.6206,
170
+ "step": 115
171
+ },
172
+ {
173
+ "epoch": 0.44776119402985076,
174
+ "grad_norm": 0.3634540444833806,
175
+ "learning_rate": 3.3278614813010034e-05,
176
+ "loss": 0.6225,
177
+ "step": 120
178
+ },
179
+ {
180
+ "epoch": 0.4664179104477612,
181
+ "grad_norm": 0.3438985797382505,
182
+ "learning_rate": 3.1923644911909e-05,
183
+ "loss": 0.6197,
184
+ "step": 125
185
+ },
186
+ {
187
+ "epoch": 0.48507462686567165,
188
+ "grad_norm": 0.34921179180879464,
189
+ "learning_rate": 3.0551762263406576e-05,
190
+ "loss": 0.6189,
191
+ "step": 130
192
+ },
193
+ {
194
+ "epoch": 0.503731343283582,
195
+ "grad_norm": 0.356834046659713,
196
+ "learning_rate": 2.9168211932412042e-05,
197
+ "loss": 0.616,
198
+ "step": 135
199
+ },
200
+ {
201
+ "epoch": 0.5223880597014925,
202
+ "grad_norm": 0.3892649353775712,
203
+ "learning_rate": 2.777828359242567e-05,
204
+ "loss": 0.6222,
205
+ "step": 140
206
+ },
207
+ {
208
+ "epoch": 0.5410447761194029,
209
+ "grad_norm": 0.32280726565127904,
210
+ "learning_rate": 2.6387291301738377e-05,
211
+ "loss": 0.6144,
212
+ "step": 145
213
+ },
214
+ {
215
+ "epoch": 0.5597014925373134,
216
+ "grad_norm": 0.3699986253190016,
217
+ "learning_rate": 2.50005531864019e-05,
218
+ "loss": 0.6135,
219
+ "step": 150
220
+ },
221
+ {
222
+ "epoch": 0.5783582089552238,
223
+ "grad_norm": 0.3290301183716731,
224
+ "learning_rate": 2.362337110764688e-05,
225
+ "loss": 0.6136,
226
+ "step": 155
227
+ },
228
+ {
229
+ "epoch": 0.5970149253731343,
230
+ "grad_norm": 0.35057798873508234,
231
+ "learning_rate": 2.226101039148557e-05,
232
+ "loss": 0.6118,
233
+ "step": 160
234
+ },
235
+ {
236
+ "epoch": 0.6156716417910447,
237
+ "grad_norm": 0.3200003908203416,
238
+ "learning_rate": 2.0918679697998252e-05,
239
+ "loss": 0.6112,
240
+ "step": 165
241
+ },
242
+ {
243
+ "epoch": 0.6343283582089553,
244
+ "grad_norm": 0.33116371844391157,
245
+ "learning_rate": 1.9601511107268255e-05,
246
+ "loss": 0.6117,
247
+ "step": 170
248
+ },
249
+ {
250
+ "epoch": 0.6529850746268657,
251
+ "grad_norm": 0.3188985080282637,
252
+ "learning_rate": 1.8314540498102216e-05,
253
+ "loss": 0.6122,
254
+ "step": 175
255
+ },
256
+ {
257
+ "epoch": 0.6716417910447762,
258
+ "grad_norm": 0.28580845205711797,
259
+ "learning_rate": 1.7062688294552992e-05,
260
+ "loss": 0.6037,
261
+ "step": 180
262
+ },
263
+ {
264
+ "epoch": 0.6902985074626866,
265
+ "grad_norm": 0.3314261158913308,
266
+ "learning_rate": 1.5850740653856096e-05,
267
+ "loss": 0.6076,
268
+ "step": 185
269
+ },
270
+ {
271
+ "epoch": 0.7089552238805971,
272
+ "grad_norm": 0.3095107101560917,
273
+ "learning_rate": 1.4683331167703218e-05,
274
+ "loss": 0.6118,
275
+ "step": 190
276
+ },
277
+ {
278
+ "epoch": 0.7276119402985075,
279
+ "grad_norm": 0.29080935963111504,
280
+ "learning_rate": 1.356492314681356e-05,
281
+ "loss": 0.615,
282
+ "step": 195
283
+ },
284
+ {
285
+ "epoch": 0.746268656716418,
286
+ "grad_norm": 0.29758716955662884,
287
+ "learning_rate": 1.2499792556533716e-05,
288
+ "loss": 0.6095,
289
+ "step": 200
290
+ },
291
+ {
292
+ "epoch": 0.7649253731343284,
293
+ "grad_norm": 0.30963667322516397,
294
+ "learning_rate": 1.1492011668707753e-05,
295
+ "loss": 0.6066,
296
+ "step": 205
297
+ },
298
+ {
299
+ "epoch": 0.7835820895522388,
300
+ "grad_norm": 0.30658536724452773,
301
+ "learning_rate": 1.0545433492320603e-05,
302
+ "loss": 0.6124,
303
+ "step": 210
304
+ },
305
+ {
306
+ "epoch": 0.8022388059701493,
307
+ "grad_norm": 0.29392954849180936,
308
+ "learning_rate": 9.663677042440537e-06,
309
+ "loss": 0.6063,
310
+ "step": 215
311
+ },
312
+ {
313
+ "epoch": 0.8208955223880597,
314
+ "grad_norm": 0.28571113686888133,
315
+ "learning_rate": 8.850113503781367e-06,
316
+ "loss": 0.6068,
317
+ "step": 220
318
+ },
319
+ {
320
+ "epoch": 0.8395522388059702,
321
+ "grad_norm": 0.29795461215348235,
322
+ "learning_rate": 8.107853341784671e-06,
323
+ "loss": 0.6138,
324
+ "step": 225
325
+ },
326
+ {
327
+ "epoch": 0.8582089552238806,
328
+ "grad_norm": 0.2981168221266009,
329
+ "learning_rate": 7.439734410499752e-06,
330
+ "loss": 0.6099,
331
+ "step": 230
332
+ },
333
+ {
334
+ "epoch": 0.8768656716417911,
335
+ "grad_norm": 0.28675297615307055,
336
+ "learning_rate": 6.848311102728011e-06,
337
+ "loss": 0.6104,
338
+ "step": 235
339
+ },
340
+ {
341
+ "epoch": 0.8955223880597015,
342
+ "grad_norm": 0.30159099563177727,
343
+ "learning_rate": 6.335844583913515e-06,
344
+ "loss": 0.6062,
345
+ "step": 240
346
+ },
347
+ {
348
+ "epoch": 0.914179104477612,
349
+ "grad_norm": 0.28607248798503265,
350
+ "learning_rate": 5.904294147118193e-06,
351
+ "loss": 0.6104,
352
+ "step": 245
353
+ },
354
+ {
355
+ "epoch": 0.9328358208955224,
356
+ "grad_norm": 0.28514568319404526,
357
+ "learning_rate": 5.555309722133842e-06,
358
+ "loss": 0.6073,
359
+ "step": 250
360
+ },
361
+ {
362
+ "epoch": 0.9514925373134329,
363
+ "grad_norm": 0.3069942451365108,
364
+ "learning_rate": 5.290225567370509e-06,
365
+ "loss": 0.6029,
366
+ "step": 255
367
+ },
368
+ {
369
+ "epoch": 0.9701492537313433,
370
+ "grad_norm": 0.2831830125198675,
371
+ "learning_rate": 5.110055168638854e-06,
372
+ "loss": 0.607,
373
+ "step": 260
374
+ },
375
+ {
376
+ "epoch": 0.9888059701492538,
377
+ "grad_norm": 0.28919434131937044,
378
+ "learning_rate": 5.0154873643297575e-06,
379
+ "loss": 0.6109,
380
+ "step": 265
381
+ },
382
+ {
383
+ "epoch": 1.0,
384
+ "step": 268,
385
+ "total_flos": 4349016447909888.0,
386
+ "train_loss": 0.638175894520176,
387
+ "train_runtime": 10527.0005,
388
+ "train_samples_per_second": 3.258,
389
+ "train_steps_per_second": 0.025
390
+ }
391
+ ],
392
+ "logging_steps": 5,
393
+ "max_steps": 268,
394
+ "num_input_tokens_seen": 0,
395
+ "num_train_epochs": 1,
396
+ "save_steps": 100,
397
+ "stateful_callbacks": {
398
+ "TrainerControl": {
399
+ "args": {
400
+ "should_epoch_stop": false,
401
+ "should_evaluate": false,
402
+ "should_log": false,
403
+ "should_save": true,
404
+ "should_training_stop": true
405
+ },
406
+ "attributes": {}
407
+ }
408
+ },
409
+ "total_flos": 4349016447909888.0,
410
+ "train_batch_size": 1,
411
+ "trial_name": null,
412
+ "trial_params": null
413
+ }