lukealonso commited on
Commit
1be8e9f
·
verified ·
1 Parent(s): 76b1278

Add files using upload-large-folder tool

Browse files
Files changed (1) hide show
  1. config.json +2 -77
config.json CHANGED
@@ -84,233 +84,158 @@
84
  "model.layers.1.mlp*",
85
  "model.layers.1.self_attn*",
86
  "model.layers.1.self_attn.indexer*",
87
- "model.layers.10.mlp.shared_experts*",
88
  "model.layers.10.self_attn*",
89
  "model.layers.10.self_attn.indexer*",
90
- "model.layers.11.mlp.shared_experts*",
91
  "model.layers.11.self_attn*",
92
  "model.layers.11.self_attn.indexer*",
93
- "model.layers.12.mlp.shared_experts*",
94
  "model.layers.12.self_attn*",
95
  "model.layers.12.self_attn.indexer*",
96
- "model.layers.13.mlp.shared_experts*",
97
  "model.layers.13.self_attn*",
98
  "model.layers.13.self_attn.indexer*",
99
- "model.layers.14.mlp.shared_experts*",
100
  "model.layers.14.self_attn*",
101
  "model.layers.14.self_attn.indexer*",
102
- "model.layers.15.mlp.shared_experts*",
103
  "model.layers.15.self_attn*",
104
  "model.layers.15.self_attn.indexer*",
105
- "model.layers.16.mlp.shared_experts*",
106
  "model.layers.16.self_attn*",
107
  "model.layers.16.self_attn.indexer*",
108
- "model.layers.17.mlp.shared_experts*",
109
  "model.layers.17.self_attn*",
110
  "model.layers.17.self_attn.indexer*",
111
- "model.layers.18.mlp.shared_experts*",
112
  "model.layers.18.self_attn*",
113
  "model.layers.18.self_attn.indexer*",
114
- "model.layers.19.mlp.shared_experts*",
115
  "model.layers.19.self_attn*",
116
  "model.layers.19.self_attn.indexer*",
117
  "model.layers.2.mlp*",
118
  "model.layers.2.self_attn*",
119
  "model.layers.2.self_attn.indexer*",
120
- "model.layers.20.mlp.shared_experts*",
121
  "model.layers.20.self_attn*",
122
  "model.layers.20.self_attn.indexer*",
123
- "model.layers.21.mlp.shared_experts*",
124
  "model.layers.21.self_attn*",
125
  "model.layers.21.self_attn.indexer*",
126
- "model.layers.22.mlp.shared_experts*",
127
  "model.layers.22.self_attn*",
128
  "model.layers.22.self_attn.indexer*",
129
- "model.layers.23.mlp.shared_experts*",
130
  "model.layers.23.self_attn*",
131
  "model.layers.23.self_attn.indexer*",
132
- "model.layers.24.mlp.shared_experts*",
133
  "model.layers.24.self_attn*",
134
  "model.layers.24.self_attn.indexer*",
135
- "model.layers.25.mlp.shared_experts*",
136
  "model.layers.25.self_attn*",
137
  "model.layers.25.self_attn.indexer*",
138
- "model.layers.26.mlp.shared_experts*",
139
  "model.layers.26.self_attn*",
140
  "model.layers.26.self_attn.indexer*",
141
- "model.layers.27.mlp.shared_experts*",
142
  "model.layers.27.self_attn*",
143
  "model.layers.27.self_attn.indexer*",
144
- "model.layers.28.mlp.shared_experts*",
145
  "model.layers.28.self_attn*",
146
  "model.layers.28.self_attn.indexer*",
147
- "model.layers.29.mlp.shared_experts*",
148
  "model.layers.29.self_attn*",
149
  "model.layers.29.self_attn.indexer*",
150
- "model.layers.3.mlp.shared_experts*",
151
  "model.layers.3.self_attn*",
152
  "model.layers.3.self_attn.indexer*",
153
- "model.layers.30.mlp.shared_experts*",
154
  "model.layers.30.self_attn*",
155
  "model.layers.30.self_attn.indexer*",
156
- "model.layers.31.mlp.shared_experts*",
157
  "model.layers.31.self_attn*",
158
  "model.layers.31.self_attn.indexer*",
159
- "model.layers.32.mlp.shared_experts*",
160
  "model.layers.32.self_attn*",
161
  "model.layers.32.self_attn.indexer*",
162
- "model.layers.33.mlp.shared_experts*",
163
  "model.layers.33.self_attn*",
164
  "model.layers.33.self_attn.indexer*",
165
- "model.layers.34.mlp.shared_experts*",
166
  "model.layers.34.self_attn*",
167
  "model.layers.34.self_attn.indexer*",
168
- "model.layers.35.mlp.shared_experts*",
169
  "model.layers.35.self_attn*",
170
  "model.layers.35.self_attn.indexer*",
171
- "model.layers.36.mlp.shared_experts*",
172
  "model.layers.36.self_attn*",
173
  "model.layers.36.self_attn.indexer*",
174
- "model.layers.37.mlp.shared_experts*",
175
  "model.layers.37.self_attn*",
176
  "model.layers.37.self_attn.indexer*",
177
- "model.layers.38.mlp.shared_experts*",
178
  "model.layers.38.self_attn*",
179
  "model.layers.38.self_attn.indexer*",
180
- "model.layers.39.mlp.shared_experts*",
181
  "model.layers.39.self_attn*",
182
  "model.layers.39.self_attn.indexer*",
183
- "model.layers.4.mlp.shared_experts*",
184
  "model.layers.4.self_attn*",
185
  "model.layers.4.self_attn.indexer*",
186
- "model.layers.40.mlp.shared_experts*",
187
  "model.layers.40.self_attn*",
188
  "model.layers.40.self_attn.indexer*",
189
- "model.layers.41.mlp.shared_experts*",
190
  "model.layers.41.self_attn*",
191
  "model.layers.41.self_attn.indexer*",
192
- "model.layers.42.mlp.shared_experts*",
193
  "model.layers.42.self_attn*",
194
  "model.layers.42.self_attn.indexer*",
195
- "model.layers.43.mlp.shared_experts*",
196
  "model.layers.43.self_attn*",
197
  "model.layers.43.self_attn.indexer*",
198
- "model.layers.44.mlp.shared_experts*",
199
  "model.layers.44.self_attn*",
200
  "model.layers.44.self_attn.indexer*",
201
- "model.layers.45.mlp.shared_experts*",
202
  "model.layers.45.self_attn*",
203
  "model.layers.45.self_attn.indexer*",
204
- "model.layers.46.mlp.shared_experts*",
205
  "model.layers.46.self_attn*",
206
  "model.layers.46.self_attn.indexer*",
207
- "model.layers.47.mlp.shared_experts*",
208
  "model.layers.47.self_attn*",
209
  "model.layers.47.self_attn.indexer*",
210
- "model.layers.48.mlp.shared_experts*",
211
  "model.layers.48.self_attn*",
212
  "model.layers.48.self_attn.indexer*",
213
- "model.layers.49.mlp.shared_experts*",
214
  "model.layers.49.self_attn*",
215
  "model.layers.49.self_attn.indexer*",
216
- "model.layers.5.mlp.shared_experts*",
217
  "model.layers.5.self_attn*",
218
  "model.layers.5.self_attn.indexer*",
219
- "model.layers.50.mlp.shared_experts*",
220
  "model.layers.50.self_attn*",
221
  "model.layers.50.self_attn.indexer*",
222
- "model.layers.51.mlp.shared_experts*",
223
  "model.layers.51.self_attn*",
224
  "model.layers.51.self_attn.indexer*",
225
- "model.layers.52.mlp.shared_experts*",
226
  "model.layers.52.self_attn*",
227
  "model.layers.52.self_attn.indexer*",
228
- "model.layers.53.mlp.shared_experts*",
229
  "model.layers.53.self_attn*",
230
  "model.layers.53.self_attn.indexer*",
231
- "model.layers.54.mlp.shared_experts*",
232
  "model.layers.54.self_attn*",
233
  "model.layers.54.self_attn.indexer*",
234
- "model.layers.55.mlp.shared_experts*",
235
  "model.layers.55.self_attn*",
236
  "model.layers.55.self_attn.indexer*",
237
- "model.layers.56.mlp.shared_experts*",
238
  "model.layers.56.self_attn*",
239
  "model.layers.56.self_attn.indexer*",
240
- "model.layers.57.mlp.shared_experts*",
241
  "model.layers.57.self_attn*",
242
  "model.layers.57.self_attn.indexer*",
243
- "model.layers.58.mlp.shared_experts*",
244
  "model.layers.58.self_attn*",
245
  "model.layers.58.self_attn.indexer*",
246
- "model.layers.59.mlp.shared_experts*",
247
  "model.layers.59.self_attn*",
248
  "model.layers.59.self_attn.indexer*",
249
- "model.layers.6.mlp.shared_experts*",
250
  "model.layers.6.self_attn*",
251
  "model.layers.6.self_attn.indexer*",
252
- "model.layers.60.mlp.shared_experts*",
253
  "model.layers.60.self_attn*",
254
  "model.layers.60.self_attn.indexer*",
255
- "model.layers.61.mlp.shared_experts*",
256
  "model.layers.61.self_attn*",
257
  "model.layers.61.self_attn.indexer*",
258
- "model.layers.62.mlp.shared_experts*",
259
  "model.layers.62.self_attn*",
260
  "model.layers.62.self_attn.indexer*",
261
- "model.layers.63.mlp.shared_experts*",
262
  "model.layers.63.self_attn*",
263
  "model.layers.63.self_attn.indexer*",
264
- "model.layers.64.mlp.shared_experts*",
265
  "model.layers.64.self_attn*",
266
  "model.layers.64.self_attn.indexer*",
267
- "model.layers.65.mlp.shared_experts*",
268
  "model.layers.65.self_attn*",
269
  "model.layers.65.self_attn.indexer*",
270
- "model.layers.66.mlp.shared_experts*",
271
  "model.layers.66.self_attn*",
272
  "model.layers.66.self_attn.indexer*",
273
- "model.layers.67.mlp.shared_experts*",
274
  "model.layers.67.self_attn*",
275
  "model.layers.67.self_attn.indexer*",
276
- "model.layers.68.mlp.shared_experts*",
277
  "model.layers.68.self_attn*",
278
  "model.layers.68.self_attn.indexer*",
279
- "model.layers.69.mlp.shared_experts*",
280
  "model.layers.69.self_attn*",
281
  "model.layers.69.self_attn.indexer*",
282
- "model.layers.7.mlp.shared_experts*",
283
  "model.layers.7.self_attn*",
284
  "model.layers.7.self_attn.indexer*",
285
- "model.layers.70.mlp.shared_experts*",
286
  "model.layers.70.self_attn*",
287
  "model.layers.70.self_attn.indexer*",
288
- "model.layers.71.mlp.shared_experts*",
289
  "model.layers.71.self_attn*",
290
  "model.layers.71.self_attn.indexer*",
291
- "model.layers.72.mlp.shared_experts*",
292
  "model.layers.72.self_attn*",
293
  "model.layers.72.self_attn.indexer*",
294
- "model.layers.73.mlp.shared_experts*",
295
  "model.layers.73.self_attn*",
296
  "model.layers.73.self_attn.indexer*",
297
- "model.layers.74.mlp.shared_experts*",
298
  "model.layers.74.self_attn*",
299
  "model.layers.74.self_attn.indexer*",
300
- "model.layers.75.mlp.shared_experts*",
301
  "model.layers.75.self_attn*",
302
  "model.layers.75.self_attn.indexer*",
303
- "model.layers.76.mlp.shared_experts*",
304
  "model.layers.76.self_attn*",
305
  "model.layers.76.self_attn.indexer*",
306
- "model.layers.77.mlp.shared_experts*",
307
  "model.layers.77.self_attn*",
308
  "model.layers.77.self_attn.indexer*",
309
- "model.layers.78.mlp.shared_experts*",
310
- "model.layers.8.mlp.shared_experts*",
311
  "model.layers.8.self_attn*",
312
  "model.layers.8.self_attn.indexer*",
313
- "model.layers.9.mlp.shared_experts*",
314
  "model.layers.9.self_attn*",
315
  "model.layers.9.self_attn.indexer*"
316
  ],
@@ -321,4 +246,4 @@
321
  },
322
  "quant_method": "modelopt"
323
  }
324
- }
 
84
  "model.layers.1.mlp*",
85
  "model.layers.1.self_attn*",
86
  "model.layers.1.self_attn.indexer*",
87
+ "*shared_experts*",
88
  "model.layers.10.self_attn*",
89
  "model.layers.10.self_attn.indexer*",
 
90
  "model.layers.11.self_attn*",
91
  "model.layers.11.self_attn.indexer*",
 
92
  "model.layers.12.self_attn*",
93
  "model.layers.12.self_attn.indexer*",
 
94
  "model.layers.13.self_attn*",
95
  "model.layers.13.self_attn.indexer*",
 
96
  "model.layers.14.self_attn*",
97
  "model.layers.14.self_attn.indexer*",
 
98
  "model.layers.15.self_attn*",
99
  "model.layers.15.self_attn.indexer*",
 
100
  "model.layers.16.self_attn*",
101
  "model.layers.16.self_attn.indexer*",
 
102
  "model.layers.17.self_attn*",
103
  "model.layers.17.self_attn.indexer*",
 
104
  "model.layers.18.self_attn*",
105
  "model.layers.18.self_attn.indexer*",
 
106
  "model.layers.19.self_attn*",
107
  "model.layers.19.self_attn.indexer*",
108
  "model.layers.2.mlp*",
109
  "model.layers.2.self_attn*",
110
  "model.layers.2.self_attn.indexer*",
 
111
  "model.layers.20.self_attn*",
112
  "model.layers.20.self_attn.indexer*",
 
113
  "model.layers.21.self_attn*",
114
  "model.layers.21.self_attn.indexer*",
 
115
  "model.layers.22.self_attn*",
116
  "model.layers.22.self_attn.indexer*",
 
117
  "model.layers.23.self_attn*",
118
  "model.layers.23.self_attn.indexer*",
 
119
  "model.layers.24.self_attn*",
120
  "model.layers.24.self_attn.indexer*",
 
121
  "model.layers.25.self_attn*",
122
  "model.layers.25.self_attn.indexer*",
 
123
  "model.layers.26.self_attn*",
124
  "model.layers.26.self_attn.indexer*",
 
125
  "model.layers.27.self_attn*",
126
  "model.layers.27.self_attn.indexer*",
 
127
  "model.layers.28.self_attn*",
128
  "model.layers.28.self_attn.indexer*",
 
129
  "model.layers.29.self_attn*",
130
  "model.layers.29.self_attn.indexer*",
 
131
  "model.layers.3.self_attn*",
132
  "model.layers.3.self_attn.indexer*",
 
133
  "model.layers.30.self_attn*",
134
  "model.layers.30.self_attn.indexer*",
 
135
  "model.layers.31.self_attn*",
136
  "model.layers.31.self_attn.indexer*",
 
137
  "model.layers.32.self_attn*",
138
  "model.layers.32.self_attn.indexer*",
 
139
  "model.layers.33.self_attn*",
140
  "model.layers.33.self_attn.indexer*",
 
141
  "model.layers.34.self_attn*",
142
  "model.layers.34.self_attn.indexer*",
 
143
  "model.layers.35.self_attn*",
144
  "model.layers.35.self_attn.indexer*",
 
145
  "model.layers.36.self_attn*",
146
  "model.layers.36.self_attn.indexer*",
 
147
  "model.layers.37.self_attn*",
148
  "model.layers.37.self_attn.indexer*",
 
149
  "model.layers.38.self_attn*",
150
  "model.layers.38.self_attn.indexer*",
 
151
  "model.layers.39.self_attn*",
152
  "model.layers.39.self_attn.indexer*",
 
153
  "model.layers.4.self_attn*",
154
  "model.layers.4.self_attn.indexer*",
 
155
  "model.layers.40.self_attn*",
156
  "model.layers.40.self_attn.indexer*",
 
157
  "model.layers.41.self_attn*",
158
  "model.layers.41.self_attn.indexer*",
 
159
  "model.layers.42.self_attn*",
160
  "model.layers.42.self_attn.indexer*",
 
161
  "model.layers.43.self_attn*",
162
  "model.layers.43.self_attn.indexer*",
 
163
  "model.layers.44.self_attn*",
164
  "model.layers.44.self_attn.indexer*",
 
165
  "model.layers.45.self_attn*",
166
  "model.layers.45.self_attn.indexer*",
 
167
  "model.layers.46.self_attn*",
168
  "model.layers.46.self_attn.indexer*",
 
169
  "model.layers.47.self_attn*",
170
  "model.layers.47.self_attn.indexer*",
 
171
  "model.layers.48.self_attn*",
172
  "model.layers.48.self_attn.indexer*",
 
173
  "model.layers.49.self_attn*",
174
  "model.layers.49.self_attn.indexer*",
 
175
  "model.layers.5.self_attn*",
176
  "model.layers.5.self_attn.indexer*",
 
177
  "model.layers.50.self_attn*",
178
  "model.layers.50.self_attn.indexer*",
 
179
  "model.layers.51.self_attn*",
180
  "model.layers.51.self_attn.indexer*",
 
181
  "model.layers.52.self_attn*",
182
  "model.layers.52.self_attn.indexer*",
 
183
  "model.layers.53.self_attn*",
184
  "model.layers.53.self_attn.indexer*",
 
185
  "model.layers.54.self_attn*",
186
  "model.layers.54.self_attn.indexer*",
 
187
  "model.layers.55.self_attn*",
188
  "model.layers.55.self_attn.indexer*",
 
189
  "model.layers.56.self_attn*",
190
  "model.layers.56.self_attn.indexer*",
 
191
  "model.layers.57.self_attn*",
192
  "model.layers.57.self_attn.indexer*",
 
193
  "model.layers.58.self_attn*",
194
  "model.layers.58.self_attn.indexer*",
 
195
  "model.layers.59.self_attn*",
196
  "model.layers.59.self_attn.indexer*",
 
197
  "model.layers.6.self_attn*",
198
  "model.layers.6.self_attn.indexer*",
 
199
  "model.layers.60.self_attn*",
200
  "model.layers.60.self_attn.indexer*",
 
201
  "model.layers.61.self_attn*",
202
  "model.layers.61.self_attn.indexer*",
 
203
  "model.layers.62.self_attn*",
204
  "model.layers.62.self_attn.indexer*",
 
205
  "model.layers.63.self_attn*",
206
  "model.layers.63.self_attn.indexer*",
 
207
  "model.layers.64.self_attn*",
208
  "model.layers.64.self_attn.indexer*",
 
209
  "model.layers.65.self_attn*",
210
  "model.layers.65.self_attn.indexer*",
 
211
  "model.layers.66.self_attn*",
212
  "model.layers.66.self_attn.indexer*",
 
213
  "model.layers.67.self_attn*",
214
  "model.layers.67.self_attn.indexer*",
 
215
  "model.layers.68.self_attn*",
216
  "model.layers.68.self_attn.indexer*",
 
217
  "model.layers.69.self_attn*",
218
  "model.layers.69.self_attn.indexer*",
 
219
  "model.layers.7.self_attn*",
220
  "model.layers.7.self_attn.indexer*",
 
221
  "model.layers.70.self_attn*",
222
  "model.layers.70.self_attn.indexer*",
 
223
  "model.layers.71.self_attn*",
224
  "model.layers.71.self_attn.indexer*",
 
225
  "model.layers.72.self_attn*",
226
  "model.layers.72.self_attn.indexer*",
 
227
  "model.layers.73.self_attn*",
228
  "model.layers.73.self_attn.indexer*",
 
229
  "model.layers.74.self_attn*",
230
  "model.layers.74.self_attn.indexer*",
 
231
  "model.layers.75.self_attn*",
232
  "model.layers.75.self_attn.indexer*",
 
233
  "model.layers.76.self_attn*",
234
  "model.layers.76.self_attn.indexer*",
 
235
  "model.layers.77.self_attn*",
236
  "model.layers.77.self_attn.indexer*",
 
 
237
  "model.layers.8.self_attn*",
238
  "model.layers.8.self_attn.indexer*",
 
239
  "model.layers.9.self_attn*",
240
  "model.layers.9.self_attn.indexer*"
241
  ],
 
246
  },
247
  "quant_method": "modelopt"
248
  }
249
+ }