decoded-cipher commited on
Commit
b99ab1b
·
verified ·
1 Parent(s): e301149

Upload adapter_config.json with huggingface_hub

Browse files
Files changed (1) hide show
  1. adapter_config.json +15 -411
adapter_config.json CHANGED
@@ -1,416 +1,20 @@
1
  {
2
- "fine_tune_type": "lora",
3
  "peft_type": "LORA",
 
4
  "base_model_name_or_path": "Qwen/Qwen2.5-Coder-1.5B-Instruct",
5
- "learning_rate": 0.0002,
6
- "max_steps": 63,
7
- "max_seq_length": 512,
8
- "use_cce": true,
9
- "num_layers": 28,
10
- "lora_parameters": {
11
- "rank": 16,
12
- "scale": 1.0,
13
- "dropout": 0.0,
14
- "keys": [
15
- "model.layers.27.mlp.up_proj",
16
- "model.layers.27.mlp.down_proj",
17
- "model.layers.27.mlp.gate_proj",
18
- "model.layers.27.self_attn.o_proj",
19
- "model.layers.27.self_attn.v_proj",
20
- "model.layers.27.self_attn.k_proj",
21
- "model.layers.27.self_attn.q_proj",
22
- "model.layers.26.mlp.up_proj",
23
- "model.layers.26.mlp.down_proj",
24
- "model.layers.26.mlp.gate_proj",
25
- "model.layers.26.self_attn.o_proj",
26
- "model.layers.26.self_attn.v_proj",
27
- "model.layers.26.self_attn.k_proj",
28
- "model.layers.26.self_attn.q_proj",
29
- "model.layers.25.mlp.up_proj",
30
- "model.layers.25.mlp.down_proj",
31
- "model.layers.25.mlp.gate_proj",
32
- "model.layers.25.self_attn.o_proj",
33
- "model.layers.25.self_attn.v_proj",
34
- "model.layers.25.self_attn.k_proj",
35
- "model.layers.25.self_attn.q_proj",
36
- "model.layers.24.mlp.up_proj",
37
- "model.layers.24.mlp.down_proj",
38
- "model.layers.24.mlp.gate_proj",
39
- "model.layers.24.self_attn.o_proj",
40
- "model.layers.24.self_attn.v_proj",
41
- "model.layers.24.self_attn.k_proj",
42
- "model.layers.24.self_attn.q_proj",
43
- "model.layers.23.mlp.up_proj",
44
- "model.layers.23.mlp.down_proj",
45
- "model.layers.23.mlp.gate_proj",
46
- "model.layers.23.self_attn.o_proj",
47
- "model.layers.23.self_attn.v_proj",
48
- "model.layers.23.self_attn.k_proj",
49
- "model.layers.23.self_attn.q_proj",
50
- "model.layers.22.mlp.up_proj",
51
- "model.layers.22.mlp.down_proj",
52
- "model.layers.22.mlp.gate_proj",
53
- "model.layers.22.self_attn.o_proj",
54
- "model.layers.22.self_attn.v_proj",
55
- "model.layers.22.self_attn.k_proj",
56
- "model.layers.22.self_attn.q_proj",
57
- "model.layers.21.mlp.up_proj",
58
- "model.layers.21.mlp.down_proj",
59
- "model.layers.21.mlp.gate_proj",
60
- "model.layers.21.self_attn.o_proj",
61
- "model.layers.21.self_attn.v_proj",
62
- "model.layers.21.self_attn.k_proj",
63
- "model.layers.21.self_attn.q_proj",
64
- "model.layers.20.mlp.up_proj",
65
- "model.layers.20.mlp.down_proj",
66
- "model.layers.20.mlp.gate_proj",
67
- "model.layers.20.self_attn.o_proj",
68
- "model.layers.20.self_attn.v_proj",
69
- "model.layers.20.self_attn.k_proj",
70
- "model.layers.20.self_attn.q_proj",
71
- "model.layers.19.mlp.up_proj",
72
- "model.layers.19.mlp.down_proj",
73
- "model.layers.19.mlp.gate_proj",
74
- "model.layers.19.self_attn.o_proj",
75
- "model.layers.19.self_attn.v_proj",
76
- "model.layers.19.self_attn.k_proj",
77
- "model.layers.19.self_attn.q_proj",
78
- "model.layers.18.mlp.up_proj",
79
- "model.layers.18.mlp.down_proj",
80
- "model.layers.18.mlp.gate_proj",
81
- "model.layers.18.self_attn.o_proj",
82
- "model.layers.18.self_attn.v_proj",
83
- "model.layers.18.self_attn.k_proj",
84
- "model.layers.18.self_attn.q_proj",
85
- "model.layers.17.mlp.up_proj",
86
- "model.layers.17.mlp.down_proj",
87
- "model.layers.17.mlp.gate_proj",
88
- "model.layers.17.self_attn.o_proj",
89
- "model.layers.17.self_attn.v_proj",
90
- "model.layers.17.self_attn.k_proj",
91
- "model.layers.17.self_attn.q_proj",
92
- "model.layers.16.mlp.up_proj",
93
- "model.layers.16.mlp.down_proj",
94
- "model.layers.16.mlp.gate_proj",
95
- "model.layers.16.self_attn.o_proj",
96
- "model.layers.16.self_attn.v_proj",
97
- "model.layers.16.self_attn.k_proj",
98
- "model.layers.16.self_attn.q_proj",
99
- "model.layers.15.mlp.up_proj",
100
- "model.layers.15.mlp.down_proj",
101
- "model.layers.15.mlp.gate_proj",
102
- "model.layers.15.self_attn.o_proj",
103
- "model.layers.15.self_attn.v_proj",
104
- "model.layers.15.self_attn.k_proj",
105
- "model.layers.15.self_attn.q_proj",
106
- "model.layers.14.mlp.up_proj",
107
- "model.layers.14.mlp.down_proj",
108
- "model.layers.14.mlp.gate_proj",
109
- "model.layers.14.self_attn.o_proj",
110
- "model.layers.14.self_attn.v_proj",
111
- "model.layers.14.self_attn.k_proj",
112
- "model.layers.14.self_attn.q_proj",
113
- "model.layers.13.mlp.up_proj",
114
- "model.layers.13.mlp.down_proj",
115
- "model.layers.13.mlp.gate_proj",
116
- "model.layers.13.self_attn.o_proj",
117
- "model.layers.13.self_attn.v_proj",
118
- "model.layers.13.self_attn.k_proj",
119
- "model.layers.13.self_attn.q_proj",
120
- "model.layers.12.mlp.up_proj",
121
- "model.layers.12.mlp.down_proj",
122
- "model.layers.12.mlp.gate_proj",
123
- "model.layers.12.self_attn.o_proj",
124
- "model.layers.12.self_attn.v_proj",
125
- "model.layers.12.self_attn.k_proj",
126
- "model.layers.12.self_attn.q_proj",
127
- "model.layers.11.mlp.up_proj",
128
- "model.layers.11.mlp.down_proj",
129
- "model.layers.11.mlp.gate_proj",
130
- "model.layers.11.self_attn.o_proj",
131
- "model.layers.11.self_attn.v_proj",
132
- "model.layers.11.self_attn.k_proj",
133
- "model.layers.11.self_attn.q_proj",
134
- "model.layers.10.mlp.up_proj",
135
- "model.layers.10.mlp.down_proj",
136
- "model.layers.10.mlp.gate_proj",
137
- "model.layers.10.self_attn.o_proj",
138
- "model.layers.10.self_attn.v_proj",
139
- "model.layers.10.self_attn.k_proj",
140
- "model.layers.10.self_attn.q_proj",
141
- "model.layers.9.mlp.up_proj",
142
- "model.layers.9.mlp.down_proj",
143
- "model.layers.9.mlp.gate_proj",
144
- "model.layers.9.self_attn.o_proj",
145
- "model.layers.9.self_attn.v_proj",
146
- "model.layers.9.self_attn.k_proj",
147
- "model.layers.9.self_attn.q_proj",
148
- "model.layers.8.mlp.up_proj",
149
- "model.layers.8.mlp.down_proj",
150
- "model.layers.8.mlp.gate_proj",
151
- "model.layers.8.self_attn.o_proj",
152
- "model.layers.8.self_attn.v_proj",
153
- "model.layers.8.self_attn.k_proj",
154
- "model.layers.8.self_attn.q_proj",
155
- "model.layers.7.mlp.up_proj",
156
- "model.layers.7.mlp.down_proj",
157
- "model.layers.7.mlp.gate_proj",
158
- "model.layers.7.self_attn.o_proj",
159
- "model.layers.7.self_attn.v_proj",
160
- "model.layers.7.self_attn.k_proj",
161
- "model.layers.7.self_attn.q_proj",
162
- "model.layers.6.mlp.up_proj",
163
- "model.layers.6.mlp.down_proj",
164
- "model.layers.6.mlp.gate_proj",
165
- "model.layers.6.self_attn.o_proj",
166
- "model.layers.6.self_attn.v_proj",
167
- "model.layers.6.self_attn.k_proj",
168
- "model.layers.6.self_attn.q_proj",
169
- "model.layers.5.mlp.up_proj",
170
- "model.layers.5.mlp.down_proj",
171
- "model.layers.5.mlp.gate_proj",
172
- "model.layers.5.self_attn.o_proj",
173
- "model.layers.5.self_attn.v_proj",
174
- "model.layers.5.self_attn.k_proj",
175
- "model.layers.5.self_attn.q_proj",
176
- "model.layers.4.mlp.up_proj",
177
- "model.layers.4.mlp.down_proj",
178
- "model.layers.4.mlp.gate_proj",
179
- "model.layers.4.self_attn.o_proj",
180
- "model.layers.4.self_attn.v_proj",
181
- "model.layers.4.self_attn.k_proj",
182
- "model.layers.4.self_attn.q_proj",
183
- "model.layers.3.mlp.up_proj",
184
- "model.layers.3.mlp.down_proj",
185
- "model.layers.3.mlp.gate_proj",
186
- "model.layers.3.self_attn.o_proj",
187
- "model.layers.3.self_attn.v_proj",
188
- "model.layers.3.self_attn.k_proj",
189
- "model.layers.3.self_attn.q_proj",
190
- "model.layers.2.mlp.up_proj",
191
- "model.layers.2.mlp.down_proj",
192
- "model.layers.2.mlp.gate_proj",
193
- "model.layers.2.self_attn.o_proj",
194
- "model.layers.2.self_attn.v_proj",
195
- "model.layers.2.self_attn.k_proj",
196
- "model.layers.2.self_attn.q_proj",
197
- "model.layers.1.mlp.up_proj",
198
- "model.layers.1.mlp.down_proj",
199
- "model.layers.1.mlp.gate_proj",
200
- "model.layers.1.self_attn.o_proj",
201
- "model.layers.1.self_attn.v_proj",
202
- "model.layers.1.self_attn.k_proj",
203
- "model.layers.1.self_attn.q_proj",
204
- "model.layers.0.mlp.up_proj",
205
- "model.layers.0.mlp.down_proj",
206
- "model.layers.0.mlp.gate_proj",
207
- "model.layers.0.self_attn.o_proj",
208
- "model.layers.0.self_attn.v_proj",
209
- "model.layers.0.self_attn.k_proj",
210
- "model.layers.0.self_attn.q_proj"
211
- ]
212
- },
213
- "rank": 16,
214
- "scale": 1.0,
215
- "dropout": 0.0,
216
- "base_model_commit_hash": "2e1fd397ee46e1388853d2af2c993145b0f1098a",
217
- "requires_unsloth_mlx_runtime_quantization": false,
218
- "unsloth_mlx_lora_module_paths": [
219
- "model.layers.27.mlp.up_proj",
220
- "model.layers.27.mlp.down_proj",
221
- "model.layers.27.mlp.gate_proj",
222
- "model.layers.27.self_attn.o_proj",
223
- "model.layers.27.self_attn.v_proj",
224
- "model.layers.27.self_attn.k_proj",
225
- "model.layers.27.self_attn.q_proj",
226
- "model.layers.26.mlp.up_proj",
227
- "model.layers.26.mlp.down_proj",
228
- "model.layers.26.mlp.gate_proj",
229
- "model.layers.26.self_attn.o_proj",
230
- "model.layers.26.self_attn.v_proj",
231
- "model.layers.26.self_attn.k_proj",
232
- "model.layers.26.self_attn.q_proj",
233
- "model.layers.25.mlp.up_proj",
234
- "model.layers.25.mlp.down_proj",
235
- "model.layers.25.mlp.gate_proj",
236
- "model.layers.25.self_attn.o_proj",
237
- "model.layers.25.self_attn.v_proj",
238
- "model.layers.25.self_attn.k_proj",
239
- "model.layers.25.self_attn.q_proj",
240
- "model.layers.24.mlp.up_proj",
241
- "model.layers.24.mlp.down_proj",
242
- "model.layers.24.mlp.gate_proj",
243
- "model.layers.24.self_attn.o_proj",
244
- "model.layers.24.self_attn.v_proj",
245
- "model.layers.24.self_attn.k_proj",
246
- "model.layers.24.self_attn.q_proj",
247
- "model.layers.23.mlp.up_proj",
248
- "model.layers.23.mlp.down_proj",
249
- "model.layers.23.mlp.gate_proj",
250
- "model.layers.23.self_attn.o_proj",
251
- "model.layers.23.self_attn.v_proj",
252
- "model.layers.23.self_attn.k_proj",
253
- "model.layers.23.self_attn.q_proj",
254
- "model.layers.22.mlp.up_proj",
255
- "model.layers.22.mlp.down_proj",
256
- "model.layers.22.mlp.gate_proj",
257
- "model.layers.22.self_attn.o_proj",
258
- "model.layers.22.self_attn.v_proj",
259
- "model.layers.22.self_attn.k_proj",
260
- "model.layers.22.self_attn.q_proj",
261
- "model.layers.21.mlp.up_proj",
262
- "model.layers.21.mlp.down_proj",
263
- "model.layers.21.mlp.gate_proj",
264
- "model.layers.21.self_attn.o_proj",
265
- "model.layers.21.self_attn.v_proj",
266
- "model.layers.21.self_attn.k_proj",
267
- "model.layers.21.self_attn.q_proj",
268
- "model.layers.20.mlp.up_proj",
269
- "model.layers.20.mlp.down_proj",
270
- "model.layers.20.mlp.gate_proj",
271
- "model.layers.20.self_attn.o_proj",
272
- "model.layers.20.self_attn.v_proj",
273
- "model.layers.20.self_attn.k_proj",
274
- "model.layers.20.self_attn.q_proj",
275
- "model.layers.19.mlp.up_proj",
276
- "model.layers.19.mlp.down_proj",
277
- "model.layers.19.mlp.gate_proj",
278
- "model.layers.19.self_attn.o_proj",
279
- "model.layers.19.self_attn.v_proj",
280
- "model.layers.19.self_attn.k_proj",
281
- "model.layers.19.self_attn.q_proj",
282
- "model.layers.18.mlp.up_proj",
283
- "model.layers.18.mlp.down_proj",
284
- "model.layers.18.mlp.gate_proj",
285
- "model.layers.18.self_attn.o_proj",
286
- "model.layers.18.self_attn.v_proj",
287
- "model.layers.18.self_attn.k_proj",
288
- "model.layers.18.self_attn.q_proj",
289
- "model.layers.17.mlp.up_proj",
290
- "model.layers.17.mlp.down_proj",
291
- "model.layers.17.mlp.gate_proj",
292
- "model.layers.17.self_attn.o_proj",
293
- "model.layers.17.self_attn.v_proj",
294
- "model.layers.17.self_attn.k_proj",
295
- "model.layers.17.self_attn.q_proj",
296
- "model.layers.16.mlp.up_proj",
297
- "model.layers.16.mlp.down_proj",
298
- "model.layers.16.mlp.gate_proj",
299
- "model.layers.16.self_attn.o_proj",
300
- "model.layers.16.self_attn.v_proj",
301
- "model.layers.16.self_attn.k_proj",
302
- "model.layers.16.self_attn.q_proj",
303
- "model.layers.15.mlp.up_proj",
304
- "model.layers.15.mlp.down_proj",
305
- "model.layers.15.mlp.gate_proj",
306
- "model.layers.15.self_attn.o_proj",
307
- "model.layers.15.self_attn.v_proj",
308
- "model.layers.15.self_attn.k_proj",
309
- "model.layers.15.self_attn.q_proj",
310
- "model.layers.14.mlp.up_proj",
311
- "model.layers.14.mlp.down_proj",
312
- "model.layers.14.mlp.gate_proj",
313
- "model.layers.14.self_attn.o_proj",
314
- "model.layers.14.self_attn.v_proj",
315
- "model.layers.14.self_attn.k_proj",
316
- "model.layers.14.self_attn.q_proj",
317
- "model.layers.13.mlp.up_proj",
318
- "model.layers.13.mlp.down_proj",
319
- "model.layers.13.mlp.gate_proj",
320
- "model.layers.13.self_attn.o_proj",
321
- "model.layers.13.self_attn.v_proj",
322
- "model.layers.13.self_attn.k_proj",
323
- "model.layers.13.self_attn.q_proj",
324
- "model.layers.12.mlp.up_proj",
325
- "model.layers.12.mlp.down_proj",
326
- "model.layers.12.mlp.gate_proj",
327
- "model.layers.12.self_attn.o_proj",
328
- "model.layers.12.self_attn.v_proj",
329
- "model.layers.12.self_attn.k_proj",
330
- "model.layers.12.self_attn.q_proj",
331
- "model.layers.11.mlp.up_proj",
332
- "model.layers.11.mlp.down_proj",
333
- "model.layers.11.mlp.gate_proj",
334
- "model.layers.11.self_attn.o_proj",
335
- "model.layers.11.self_attn.v_proj",
336
- "model.layers.11.self_attn.k_proj",
337
- "model.layers.11.self_attn.q_proj",
338
- "model.layers.10.mlp.up_proj",
339
- "model.layers.10.mlp.down_proj",
340
- "model.layers.10.mlp.gate_proj",
341
- "model.layers.10.self_attn.o_proj",
342
- "model.layers.10.self_attn.v_proj",
343
- "model.layers.10.self_attn.k_proj",
344
- "model.layers.10.self_attn.q_proj",
345
- "model.layers.9.mlp.up_proj",
346
- "model.layers.9.mlp.down_proj",
347
- "model.layers.9.mlp.gate_proj",
348
- "model.layers.9.self_attn.o_proj",
349
- "model.layers.9.self_attn.v_proj",
350
- "model.layers.9.self_attn.k_proj",
351
- "model.layers.9.self_attn.q_proj",
352
- "model.layers.8.mlp.up_proj",
353
- "model.layers.8.mlp.down_proj",
354
- "model.layers.8.mlp.gate_proj",
355
- "model.layers.8.self_attn.o_proj",
356
- "model.layers.8.self_attn.v_proj",
357
- "model.layers.8.self_attn.k_proj",
358
- "model.layers.8.self_attn.q_proj",
359
- "model.layers.7.mlp.up_proj",
360
- "model.layers.7.mlp.down_proj",
361
- "model.layers.7.mlp.gate_proj",
362
- "model.layers.7.self_attn.o_proj",
363
- "model.layers.7.self_attn.v_proj",
364
- "model.layers.7.self_attn.k_proj",
365
- "model.layers.7.self_attn.q_proj",
366
- "model.layers.6.mlp.up_proj",
367
- "model.layers.6.mlp.down_proj",
368
- "model.layers.6.mlp.gate_proj",
369
- "model.layers.6.self_attn.o_proj",
370
- "model.layers.6.self_attn.v_proj",
371
- "model.layers.6.self_attn.k_proj",
372
- "model.layers.6.self_attn.q_proj",
373
- "model.layers.5.mlp.up_proj",
374
- "model.layers.5.mlp.down_proj",
375
- "model.layers.5.mlp.gate_proj",
376
- "model.layers.5.self_attn.o_proj",
377
- "model.layers.5.self_attn.v_proj",
378
- "model.layers.5.self_attn.k_proj",
379
- "model.layers.5.self_attn.q_proj",
380
- "model.layers.4.mlp.up_proj",
381
- "model.layers.4.mlp.down_proj",
382
- "model.layers.4.mlp.gate_proj",
383
- "model.layers.4.self_attn.o_proj",
384
- "model.layers.4.self_attn.v_proj",
385
- "model.layers.4.self_attn.k_proj",
386
- "model.layers.4.self_attn.q_proj",
387
- "model.layers.3.mlp.up_proj",
388
- "model.layers.3.mlp.down_proj",
389
- "model.layers.3.mlp.gate_proj",
390
- "model.layers.3.self_attn.o_proj",
391
- "model.layers.3.self_attn.v_proj",
392
- "model.layers.3.self_attn.k_proj",
393
- "model.layers.3.self_attn.q_proj",
394
- "model.layers.2.mlp.up_proj",
395
- "model.layers.2.mlp.down_proj",
396
- "model.layers.2.mlp.gate_proj",
397
- "model.layers.2.self_attn.o_proj",
398
- "model.layers.2.self_attn.v_proj",
399
- "model.layers.2.self_attn.k_proj",
400
- "model.layers.2.self_attn.q_proj",
401
- "model.layers.1.mlp.up_proj",
402
- "model.layers.1.mlp.down_proj",
403
- "model.layers.1.mlp.gate_proj",
404
- "model.layers.1.self_attn.o_proj",
405
- "model.layers.1.self_attn.v_proj",
406
- "model.layers.1.self_attn.k_proj",
407
- "model.layers.1.self_attn.q_proj",
408
- "model.layers.0.mlp.up_proj",
409
- "model.layers.0.mlp.down_proj",
410
- "model.layers.0.mlp.gate_proj",
411
- "model.layers.0.self_attn.o_proj",
412
- "model.layers.0.self_attn.v_proj",
413
- "model.layers.0.self_attn.k_proj",
414
- "model.layers.0.self_attn.q_proj"
415
  ]
416
  }
 
1
  {
 
2
  "peft_type": "LORA",
3
+ "task_type": "CAUSAL_LM",
4
  "base_model_name_or_path": "Qwen/Qwen2.5-Coder-1.5B-Instruct",
5
+ "r": 16,
6
+ "lora_alpha": 16,
7
+ "lora_dropout": 0.0,
8
+ "bias": "none",
9
+ "inference_mode": true,
10
+ "fan_in_fan_out": false,
11
+ "target_modules": [
12
+ "q_proj",
13
+ "k_proj",
14
+ "v_proj",
15
+ "o_proj",
16
+ "gate_proj",
17
+ "up_proj",
18
+ "down_proj"
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
19
  ]
20
  }