fkyyy commited on
Commit
5e447ea
·
verified ·
1 Parent(s): 0e2af9f

Add transformer/fp8_quant_config.json

Browse files
Files changed (1) hide show
  1. transformer/fp8_quant_config.json +359 -0
transformer/fp8_quant_config.json ADDED
@@ -0,0 +1,359 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "format": "fp8-e4m3-perrow-v1",
3
+ "scale_suffix": "_scale",
4
+ "source_repo": "fkyyy/MiniMax-H3-slim",
5
+ "component": "transformer",
6
+ "adaln_quantized": true,
7
+ "quantized_keys": [
8
+ "transformer_blocks.0.adaln_proj.linear.weight",
9
+ "transformer_blocks.0.attn.to_k.weight",
10
+ "transformer_blocks.0.attn.to_out.0.weight",
11
+ "transformer_blocks.0.attn.to_q.weight",
12
+ "transformer_blocks.0.attn.to_v.weight",
13
+ "transformer_blocks.0.ff.net.0.proj.weight",
14
+ "transformer_blocks.0.ff.net.2.weight",
15
+ "transformer_blocks.1.adaln_proj.linear.weight",
16
+ "transformer_blocks.1.attn.to_k.weight",
17
+ "transformer_blocks.1.attn.to_out.0.weight",
18
+ "transformer_blocks.1.attn.to_q.weight",
19
+ "transformer_blocks.1.attn.to_v.weight",
20
+ "transformer_blocks.1.ff.net.0.proj.weight",
21
+ "transformer_blocks.1.ff.net.2.weight",
22
+ "transformer_blocks.10.adaln_proj.linear.weight",
23
+ "transformer_blocks.10.attn.to_k.weight",
24
+ "transformer_blocks.10.attn.to_out.0.weight",
25
+ "transformer_blocks.10.attn.to_q.weight",
26
+ "transformer_blocks.10.attn.to_v.weight",
27
+ "transformer_blocks.10.ff.net.0.proj.weight",
28
+ "transformer_blocks.10.ff.net.2.weight",
29
+ "transformer_blocks.11.adaln_proj.linear.weight",
30
+ "transformer_blocks.11.attn.to_k.weight",
31
+ "transformer_blocks.11.attn.to_out.0.weight",
32
+ "transformer_blocks.11.attn.to_q.weight",
33
+ "transformer_blocks.11.attn.to_v.weight",
34
+ "transformer_blocks.11.ff.net.0.proj.weight",
35
+ "transformer_blocks.11.ff.net.2.weight",
36
+ "transformer_blocks.12.adaln_proj.linear.weight",
37
+ "transformer_blocks.12.attn.to_k.weight",
38
+ "transformer_blocks.12.attn.to_out.0.weight",
39
+ "transformer_blocks.12.attn.to_q.weight",
40
+ "transformer_blocks.12.attn.to_v.weight",
41
+ "transformer_blocks.12.ff.net.0.proj.weight",
42
+ "transformer_blocks.12.ff.net.2.weight",
43
+ "transformer_blocks.13.adaln_proj.linear.weight",
44
+ "transformer_blocks.13.attn.to_k.weight",
45
+ "transformer_blocks.13.attn.to_out.0.weight",
46
+ "transformer_blocks.13.attn.to_q.weight",
47
+ "transformer_blocks.13.attn.to_v.weight",
48
+ "transformer_blocks.13.ff.net.0.proj.weight",
49
+ "transformer_blocks.13.ff.net.2.weight",
50
+ "transformer_blocks.14.adaln_proj.linear.weight",
51
+ "transformer_blocks.14.attn.to_k.weight",
52
+ "transformer_blocks.14.attn.to_out.0.weight",
53
+ "transformer_blocks.14.attn.to_q.weight",
54
+ "transformer_blocks.14.attn.to_v.weight",
55
+ "transformer_blocks.14.ff.net.0.proj.weight",
56
+ "transformer_blocks.14.ff.net.2.weight",
57
+ "transformer_blocks.15.adaln_proj.linear.weight",
58
+ "transformer_blocks.15.attn.to_k.weight",
59
+ "transformer_blocks.15.attn.to_out.0.weight",
60
+ "transformer_blocks.15.attn.to_q.weight",
61
+ "transformer_blocks.15.attn.to_v.weight",
62
+ "transformer_blocks.15.ff.net.0.proj.weight",
63
+ "transformer_blocks.15.ff.net.2.weight",
64
+ "transformer_blocks.16.adaln_proj.linear.weight",
65
+ "transformer_blocks.16.attn.to_k.weight",
66
+ "transformer_blocks.16.attn.to_out.0.weight",
67
+ "transformer_blocks.16.attn.to_q.weight",
68
+ "transformer_blocks.16.attn.to_v.weight",
69
+ "transformer_blocks.16.ff.net.0.proj.weight",
70
+ "transformer_blocks.16.ff.net.2.weight",
71
+ "transformer_blocks.17.adaln_proj.linear.weight",
72
+ "transformer_blocks.17.attn.to_k.weight",
73
+ "transformer_blocks.17.attn.to_out.0.weight",
74
+ "transformer_blocks.17.attn.to_q.weight",
75
+ "transformer_blocks.17.attn.to_v.weight",
76
+ "transformer_blocks.17.ff.net.0.proj.weight",
77
+ "transformer_blocks.17.ff.net.2.weight",
78
+ "transformer_blocks.18.adaln_proj.linear.weight",
79
+ "transformer_blocks.18.attn.to_k.weight",
80
+ "transformer_blocks.18.attn.to_out.0.weight",
81
+ "transformer_blocks.18.attn.to_q.weight",
82
+ "transformer_blocks.18.attn.to_v.weight",
83
+ "transformer_blocks.18.ff.net.0.proj.weight",
84
+ "transformer_blocks.18.ff.net.2.weight",
85
+ "transformer_blocks.19.adaln_proj.linear.weight",
86
+ "transformer_blocks.19.attn.to_k.weight",
87
+ "transformer_blocks.19.attn.to_out.0.weight",
88
+ "transformer_blocks.19.attn.to_q.weight",
89
+ "transformer_blocks.19.attn.to_v.weight",
90
+ "transformer_blocks.19.ff.net.0.proj.weight",
91
+ "transformer_blocks.19.ff.net.2.weight",
92
+ "transformer_blocks.2.adaln_proj.linear.weight",
93
+ "transformer_blocks.2.attn.to_k.weight",
94
+ "transformer_blocks.2.attn.to_out.0.weight",
95
+ "transformer_blocks.2.attn.to_q.weight",
96
+ "transformer_blocks.2.attn.to_v.weight",
97
+ "transformer_blocks.2.ff.net.0.proj.weight",
98
+ "transformer_blocks.2.ff.net.2.weight",
99
+ "transformer_blocks.20.adaln_proj.linear.weight",
100
+ "transformer_blocks.20.attn.to_k.weight",
101
+ "transformer_blocks.20.attn.to_out.0.weight",
102
+ "transformer_blocks.20.attn.to_q.weight",
103
+ "transformer_blocks.20.attn.to_v.weight",
104
+ "transformer_blocks.20.ff.net.0.proj.weight",
105
+ "transformer_blocks.20.ff.net.2.weight",
106
+ "transformer_blocks.21.adaln_proj.linear.weight",
107
+ "transformer_blocks.21.attn.to_k.weight",
108
+ "transformer_blocks.21.attn.to_out.0.weight",
109
+ "transformer_blocks.21.attn.to_q.weight",
110
+ "transformer_blocks.21.attn.to_v.weight",
111
+ "transformer_blocks.21.ff.net.0.proj.weight",
112
+ "transformer_blocks.21.ff.net.2.weight",
113
+ "transformer_blocks.22.adaln_proj.linear.weight",
114
+ "transformer_blocks.22.attn.to_k.weight",
115
+ "transformer_blocks.22.attn.to_out.0.weight",
116
+ "transformer_blocks.22.attn.to_q.weight",
117
+ "transformer_blocks.22.attn.to_v.weight",
118
+ "transformer_blocks.22.ff.net.0.proj.weight",
119
+ "transformer_blocks.22.ff.net.2.weight",
120
+ "transformer_blocks.23.adaln_proj.linear.weight",
121
+ "transformer_blocks.23.attn.to_k.weight",
122
+ "transformer_blocks.23.attn.to_out.0.weight",
123
+ "transformer_blocks.23.attn.to_q.weight",
124
+ "transformer_blocks.23.attn.to_v.weight",
125
+ "transformer_blocks.23.ff.net.0.proj.weight",
126
+ "transformer_blocks.23.ff.net.2.weight",
127
+ "transformer_blocks.24.adaln_proj.linear.weight",
128
+ "transformer_blocks.24.attn.to_k.weight",
129
+ "transformer_blocks.24.attn.to_out.0.weight",
130
+ "transformer_blocks.24.attn.to_q.weight",
131
+ "transformer_blocks.24.attn.to_v.weight",
132
+ "transformer_blocks.24.ff.net.0.proj.weight",
133
+ "transformer_blocks.24.ff.net.2.weight",
134
+ "transformer_blocks.25.adaln_proj.linear.weight",
135
+ "transformer_blocks.25.attn.to_k.weight",
136
+ "transformer_blocks.25.attn.to_out.0.weight",
137
+ "transformer_blocks.25.attn.to_q.weight",
138
+ "transformer_blocks.25.attn.to_v.weight",
139
+ "transformer_blocks.25.ff.net.0.proj.weight",
140
+ "transformer_blocks.25.ff.net.2.weight",
141
+ "transformer_blocks.26.adaln_proj.linear.weight",
142
+ "transformer_blocks.26.attn.to_k.weight",
143
+ "transformer_blocks.26.attn.to_out.0.weight",
144
+ "transformer_blocks.26.attn.to_q.weight",
145
+ "transformer_blocks.26.attn.to_v.weight",
146
+ "transformer_blocks.26.ff.net.0.proj.weight",
147
+ "transformer_blocks.26.ff.net.2.weight",
148
+ "transformer_blocks.27.adaln_proj.linear.weight",
149
+ "transformer_blocks.27.attn.to_k.weight",
150
+ "transformer_blocks.27.attn.to_out.0.weight",
151
+ "transformer_blocks.27.attn.to_q.weight",
152
+ "transformer_blocks.27.attn.to_v.weight",
153
+ "transformer_blocks.27.ff.net.0.proj.weight",
154
+ "transformer_blocks.27.ff.net.2.weight",
155
+ "transformer_blocks.28.adaln_proj.linear.weight",
156
+ "transformer_blocks.28.attn.to_k.weight",
157
+ "transformer_blocks.28.attn.to_out.0.weight",
158
+ "transformer_blocks.28.attn.to_q.weight",
159
+ "transformer_blocks.28.attn.to_v.weight",
160
+ "transformer_blocks.28.ff.net.0.proj.weight",
161
+ "transformer_blocks.28.ff.net.2.weight",
162
+ "transformer_blocks.29.adaln_proj.linear.weight",
163
+ "transformer_blocks.29.attn.to_k.weight",
164
+ "transformer_blocks.29.attn.to_out.0.weight",
165
+ "transformer_blocks.29.attn.to_q.weight",
166
+ "transformer_blocks.29.attn.to_v.weight",
167
+ "transformer_blocks.29.ff.net.0.proj.weight",
168
+ "transformer_blocks.29.ff.net.2.weight",
169
+ "transformer_blocks.3.adaln_proj.linear.weight",
170
+ "transformer_blocks.3.attn.to_k.weight",
171
+ "transformer_blocks.3.attn.to_out.0.weight",
172
+ "transformer_blocks.3.attn.to_q.weight",
173
+ "transformer_blocks.3.attn.to_v.weight",
174
+ "transformer_blocks.3.ff.net.0.proj.weight",
175
+ "transformer_blocks.3.ff.net.2.weight",
176
+ "transformer_blocks.30.adaln_proj.linear.weight",
177
+ "transformer_blocks.30.attn.to_k.weight",
178
+ "transformer_blocks.30.attn.to_out.0.weight",
179
+ "transformer_blocks.30.attn.to_q.weight",
180
+ "transformer_blocks.30.attn.to_v.weight",
181
+ "transformer_blocks.30.ff.net.0.proj.weight",
182
+ "transformer_blocks.30.ff.net.2.weight",
183
+ "transformer_blocks.31.adaln_proj.linear.weight",
184
+ "transformer_blocks.31.attn.to_k.weight",
185
+ "transformer_blocks.31.attn.to_out.0.weight",
186
+ "transformer_blocks.31.attn.to_q.weight",
187
+ "transformer_blocks.31.attn.to_v.weight",
188
+ "transformer_blocks.31.ff.net.0.proj.weight",
189
+ "transformer_blocks.31.ff.net.2.weight",
190
+ "transformer_blocks.32.adaln_proj.linear.weight",
191
+ "transformer_blocks.32.attn.to_k.weight",
192
+ "transformer_blocks.32.attn.to_out.0.weight",
193
+ "transformer_blocks.32.attn.to_q.weight",
194
+ "transformer_blocks.32.attn.to_v.weight",
195
+ "transformer_blocks.32.ff.net.0.proj.weight",
196
+ "transformer_blocks.32.ff.net.2.weight",
197
+ "transformer_blocks.33.adaln_proj.linear.weight",
198
+ "transformer_blocks.33.attn.to_k.weight",
199
+ "transformer_blocks.33.attn.to_out.0.weight",
200
+ "transformer_blocks.33.attn.to_q.weight",
201
+ "transformer_blocks.33.attn.to_v.weight",
202
+ "transformer_blocks.33.ff.net.0.proj.weight",
203
+ "transformer_blocks.33.ff.net.2.weight",
204
+ "transformer_blocks.34.adaln_proj.linear.weight",
205
+ "transformer_blocks.34.attn.to_k.weight",
206
+ "transformer_blocks.34.attn.to_out.0.weight",
207
+ "transformer_blocks.34.attn.to_q.weight",
208
+ "transformer_blocks.34.attn.to_v.weight",
209
+ "transformer_blocks.34.ff.net.0.proj.weight",
210
+ "transformer_blocks.34.ff.net.2.weight",
211
+ "transformer_blocks.35.adaln_proj.linear.weight",
212
+ "transformer_blocks.35.attn.to_k.weight",
213
+ "transformer_blocks.35.attn.to_out.0.weight",
214
+ "transformer_blocks.35.attn.to_q.weight",
215
+ "transformer_blocks.35.attn.to_v.weight",
216
+ "transformer_blocks.35.ff.net.0.proj.weight",
217
+ "transformer_blocks.35.ff.net.2.weight",
218
+ "transformer_blocks.36.adaln_proj.linear.weight",
219
+ "transformer_blocks.36.attn.to_k.weight",
220
+ "transformer_blocks.36.attn.to_out.0.weight",
221
+ "transformer_blocks.36.attn.to_q.weight",
222
+ "transformer_blocks.36.attn.to_v.weight",
223
+ "transformer_blocks.36.ff.net.0.proj.weight",
224
+ "transformer_blocks.36.ff.net.2.weight",
225
+ "transformer_blocks.37.adaln_proj.linear.weight",
226
+ "transformer_blocks.37.attn.to_k.weight",
227
+ "transformer_blocks.37.attn.to_out.0.weight",
228
+ "transformer_blocks.37.attn.to_q.weight",
229
+ "transformer_blocks.37.attn.to_v.weight",
230
+ "transformer_blocks.37.ff.net.0.proj.weight",
231
+ "transformer_blocks.37.ff.net.2.weight",
232
+ "transformer_blocks.38.adaln_proj.linear.weight",
233
+ "transformer_blocks.38.attn.to_k.weight",
234
+ "transformer_blocks.38.attn.to_out.0.weight",
235
+ "transformer_blocks.38.attn.to_q.weight",
236
+ "transformer_blocks.38.attn.to_v.weight",
237
+ "transformer_blocks.38.ff.net.0.proj.weight",
238
+ "transformer_blocks.38.ff.net.2.weight",
239
+ "transformer_blocks.39.adaln_proj.linear.weight",
240
+ "transformer_blocks.39.attn.to_k.weight",
241
+ "transformer_blocks.39.attn.to_out.0.weight",
242
+ "transformer_blocks.39.attn.to_q.weight",
243
+ "transformer_blocks.39.attn.to_v.weight",
244
+ "transformer_blocks.39.ff.net.0.proj.weight",
245
+ "transformer_blocks.39.ff.net.2.weight",
246
+ "transformer_blocks.4.adaln_proj.linear.weight",
247
+ "transformer_blocks.4.attn.to_k.weight",
248
+ "transformer_blocks.4.attn.to_out.0.weight",
249
+ "transformer_blocks.4.attn.to_q.weight",
250
+ "transformer_blocks.4.attn.to_v.weight",
251
+ "transformer_blocks.4.ff.net.0.proj.weight",
252
+ "transformer_blocks.4.ff.net.2.weight",
253
+ "transformer_blocks.40.adaln_proj.linear.weight",
254
+ "transformer_blocks.40.attn.to_k.weight",
255
+ "transformer_blocks.40.attn.to_out.0.weight",
256
+ "transformer_blocks.40.attn.to_q.weight",
257
+ "transformer_blocks.40.attn.to_v.weight",
258
+ "transformer_blocks.40.ff.net.0.proj.weight",
259
+ "transformer_blocks.40.ff.net.2.weight",
260
+ "transformer_blocks.41.adaln_proj.linear.weight",
261
+ "transformer_blocks.41.attn.to_k.weight",
262
+ "transformer_blocks.41.attn.to_out.0.weight",
263
+ "transformer_blocks.41.attn.to_q.weight",
264
+ "transformer_blocks.41.attn.to_v.weight",
265
+ "transformer_blocks.41.ff.net.0.proj.weight",
266
+ "transformer_blocks.41.ff.net.2.weight",
267
+ "transformer_blocks.42.adaln_proj.linear.weight",
268
+ "transformer_blocks.42.attn.to_k.weight",
269
+ "transformer_blocks.42.attn.to_out.0.weight",
270
+ "transformer_blocks.42.attn.to_q.weight",
271
+ "transformer_blocks.42.attn.to_v.weight",
272
+ "transformer_blocks.42.ff.net.0.proj.weight",
273
+ "transformer_blocks.42.ff.net.2.weight",
274
+ "transformer_blocks.43.adaln_proj.linear.weight",
275
+ "transformer_blocks.43.attn.to_k.weight",
276
+ "transformer_blocks.43.attn.to_out.0.weight",
277
+ "transformer_blocks.43.attn.to_q.weight",
278
+ "transformer_blocks.43.attn.to_v.weight",
279
+ "transformer_blocks.43.ff.net.0.proj.weight",
280
+ "transformer_blocks.43.ff.net.2.weight",
281
+ "transformer_blocks.44.adaln_proj.linear.weight",
282
+ "transformer_blocks.44.attn.to_k.weight",
283
+ "transformer_blocks.44.attn.to_out.0.weight",
284
+ "transformer_blocks.44.attn.to_q.weight",
285
+ "transformer_blocks.44.attn.to_v.weight",
286
+ "transformer_blocks.44.ff.net.0.proj.weight",
287
+ "transformer_blocks.44.ff.net.2.weight",
288
+ "transformer_blocks.45.adaln_proj.linear.weight",
289
+ "transformer_blocks.45.attn.to_k.weight",
290
+ "transformer_blocks.45.attn.to_out.0.weight",
291
+ "transformer_blocks.45.attn.to_q.weight",
292
+ "transformer_blocks.45.attn.to_v.weight",
293
+ "transformer_blocks.45.ff.net.0.proj.weight",
294
+ "transformer_blocks.45.ff.net.2.weight",
295
+ "transformer_blocks.46.adaln_proj.linear.weight",
296
+ "transformer_blocks.46.attn.to_k.weight",
297
+ "transformer_blocks.46.attn.to_out.0.weight",
298
+ "transformer_blocks.46.attn.to_q.weight",
299
+ "transformer_blocks.46.attn.to_v.weight",
300
+ "transformer_blocks.46.ff.net.0.proj.weight",
301
+ "transformer_blocks.46.ff.net.2.weight",
302
+ "transformer_blocks.47.adaln_proj.linear.weight",
303
+ "transformer_blocks.47.attn.to_k.weight",
304
+ "transformer_blocks.47.attn.to_out.0.weight",
305
+ "transformer_blocks.47.attn.to_q.weight",
306
+ "transformer_blocks.47.attn.to_v.weight",
307
+ "transformer_blocks.47.ff.net.0.proj.weight",
308
+ "transformer_blocks.47.ff.net.2.weight",
309
+ "transformer_blocks.48.adaln_proj.linear.weight",
310
+ "transformer_blocks.48.attn.to_k.weight",
311
+ "transformer_blocks.48.attn.to_out.0.weight",
312
+ "transformer_blocks.48.attn.to_q.weight",
313
+ "transformer_blocks.48.attn.to_v.weight",
314
+ "transformer_blocks.48.ff.net.0.proj.weight",
315
+ "transformer_blocks.48.ff.net.2.weight",
316
+ "transformer_blocks.49.adaln_proj.linear.weight",
317
+ "transformer_blocks.49.attn.to_k.weight",
318
+ "transformer_blocks.49.attn.to_out.0.weight",
319
+ "transformer_blocks.49.attn.to_q.weight",
320
+ "transformer_blocks.49.attn.to_v.weight",
321
+ "transformer_blocks.49.ff.net.0.proj.weight",
322
+ "transformer_blocks.49.ff.net.2.weight",
323
+ "transformer_blocks.5.adaln_proj.linear.weight",
324
+ "transformer_blocks.5.attn.to_k.weight",
325
+ "transformer_blocks.5.attn.to_out.0.weight",
326
+ "transformer_blocks.5.attn.to_q.weight",
327
+ "transformer_blocks.5.attn.to_v.weight",
328
+ "transformer_blocks.5.ff.net.0.proj.weight",
329
+ "transformer_blocks.5.ff.net.2.weight",
330
+ "transformer_blocks.6.adaln_proj.linear.weight",
331
+ "transformer_blocks.6.attn.to_k.weight",
332
+ "transformer_blocks.6.attn.to_out.0.weight",
333
+ "transformer_blocks.6.attn.to_q.weight",
334
+ "transformer_blocks.6.attn.to_v.weight",
335
+ "transformer_blocks.6.ff.net.0.proj.weight",
336
+ "transformer_blocks.6.ff.net.2.weight",
337
+ "transformer_blocks.7.adaln_proj.linear.weight",
338
+ "transformer_blocks.7.attn.to_k.weight",
339
+ "transformer_blocks.7.attn.to_out.0.weight",
340
+ "transformer_blocks.7.attn.to_q.weight",
341
+ "transformer_blocks.7.attn.to_v.weight",
342
+ "transformer_blocks.7.ff.net.0.proj.weight",
343
+ "transformer_blocks.7.ff.net.2.weight",
344
+ "transformer_blocks.8.adaln_proj.linear.weight",
345
+ "transformer_blocks.8.attn.to_k.weight",
346
+ "transformer_blocks.8.attn.to_out.0.weight",
347
+ "transformer_blocks.8.attn.to_q.weight",
348
+ "transformer_blocks.8.attn.to_v.weight",
349
+ "transformer_blocks.8.ff.net.0.proj.weight",
350
+ "transformer_blocks.8.ff.net.2.weight",
351
+ "transformer_blocks.9.adaln_proj.linear.weight",
352
+ "transformer_blocks.9.attn.to_k.weight",
353
+ "transformer_blocks.9.attn.to_out.0.weight",
354
+ "transformer_blocks.9.attn.to_q.weight",
355
+ "transformer_blocks.9.attn.to_v.weight",
356
+ "transformer_blocks.9.ff.net.0.proj.weight",
357
+ "transformer_blocks.9.ff.net.2.weight"
358
+ ]
359
+ }