anonymousatom commited on
Commit
1b75e07
·
verified ·
1 Parent(s): 235b36c

Training in progress, step 1

Browse files
adapter_config.json CHANGED
@@ -23,258 +23,258 @@
23
  "rank_pattern": {},
24
  "revision": null,
25
  "target_modules": [
26
- "model.layers.12.self_attn.q_proj",
27
- "model.layers.9.self_attn.q_proj",
28
- "model.layers.4.self_attn.q_proj",
29
- "model.layers.28.self_attn.q_proj",
30
- "model.layers.26.self_attn.o_proj",
31
  "model.layers.28.mlp.down_proj",
32
- "model.layers.29.self_attn.v_proj",
33
- "model.layers.19.mlp.up_proj",
34
- "model.layers.16.self_attn.q_proj",
35
- "model.layers.30.self_attn.q_proj",
36
- "model.layers.16.mlp.gate_proj",
37
- "model.layers.5.mlp.gate_proj",
38
- "model.layers.32.mlp.up_proj",
39
- "model.layers.32.mlp.down_proj",
40
- "model.layers.20.self_attn.o_proj",
41
- "model.layers.12.self_attn.o_proj",
42
- "model.layers.10.mlp.down_proj",
43
- "model.layers.18.mlp.up_proj",
44
- "model.layers.34.self_attn.k_proj",
45
- "model.layers.3.self_attn.q_proj",
 
 
 
 
 
 
 
 
 
 
 
 
46
  "model.layers.26.mlp.down_proj",
47
- "model.layers.30.mlp.down_proj",
48
- "model.layers.18.self_attn.o_proj",
49
- "model.layers.16.self_attn.k_proj",
50
- "model.layers.13.mlp.up_proj",
51
- "model.layers.35.mlp.gate_proj",
52
  "model.layers.14.mlp.up_proj",
53
- "model.layers.29.mlp.gate_proj",
54
- "model.layers.8.mlp.down_proj",
55
- "model.layers.14.self_attn.v_proj",
56
- "model.layers.24.mlp.up_proj",
57
- "model.layers.28.self_attn.k_proj",
58
- "model.layers.0.mlp.up_proj",
59
- "model.layers.30.mlp.gate_proj",
60
- "model.layers.25.self_attn.q_proj",
61
- "model.layers.22.self_attn.k_proj",
62
- "model.layers.6.self_attn.q_proj",
63
- "model.layers.16.mlp.up_proj",
64
- "model.layers.33.mlp.down_proj",
65
- "model.layers.4.self_attn.o_proj",
66
- "model.layers.11.self_attn.o_proj",
67
- "model.layers.9.self_attn.o_proj",
68
- "model.layers.16.mlp.down_proj",
69
- "model.layers.21.mlp.down_proj",
70
- "model.layers.34.mlp.down_proj",
71
  "model.layers.18.self_attn.q_proj",
72
- "model.layers.27.mlp.up_proj",
73
- "model.layers.34.self_attn.v_proj",
74
- "model.layers.32.self_attn.k_proj",
75
- "model.layers.5.self_attn.o_proj",
76
- "model.layers.21.self_attn.o_proj",
77
- "model.layers.19.self_attn.o_proj",
78
- "model.layers.31.self_attn.q_proj",
79
  "model.layers.23.self_attn.k_proj",
80
- "model.layers.23.self_attn.o_proj",
81
- "model.layers.2.self_attn.q_proj",
82
- "model.layers.7.self_attn.q_proj",
83
- "model.layers.11.self_attn.q_proj",
 
 
 
 
 
 
 
 
 
 
 
 
 
 
84
  "model.layers.28.self_attn.v_proj",
 
 
 
 
 
85
  "model.layers.25.mlp.gate_proj",
86
- "model.layers.9.self_attn.v_proj",
87
- "model.layers.28.mlp.gate_proj",
88
- "model.layers.3.self_attn.v_proj",
 
 
 
 
 
 
 
 
 
 
89
  "model.layers.7.self_attn.k_proj",
 
 
 
 
 
 
 
90
  "model.layers.1.self_attn.q_proj",
91
- "model.layers.17.self_attn.q_proj",
92
- "model.layers.31.self_attn.o_proj",
93
- "model.layers.25.self_attn.k_proj",
94
- "model.layers.10.mlp.gate_proj",
95
- "model.layers.20.self_attn.v_proj",
96
- "model.layers.12.self_attn.v_proj",
97
- "model.layers.29.self_attn.q_proj",
98
- "model.layers.27.self_attn.v_proj",
99
  "model.layers.13.mlp.down_proj",
100
- "model.layers.35.self_attn.o_proj",
101
- "model.layers.21.mlp.gate_proj",
102
- "model.layers.8.self_attn.k_proj",
103
- "model.layers.14.mlp.gate_proj",
104
- "model.layers.26.self_attn.v_proj",
105
- "model.layers.15.self_attn.o_proj",
106
- "model.layers.22.mlp.down_proj",
107
- "model.layers.28.mlp.up_proj",
108
- "model.layers.5.mlp.up_proj",
109
- "model.layers.6.self_attn.k_proj",
110
- "model.layers.6.mlp.gate_proj",
111
- "model.layers.22.self_attn.q_proj",
112
- "model.layers.0.self_attn.v_proj",
113
- "model.layers.17.mlp.up_proj",
114
- "model.layers.19.self_attn.q_proj",
115
- "model.layers.2.self_attn.v_proj",
116
- "model.layers.25.self_attn.v_proj",
117
- "model.layers.12.mlp.down_proj",
118
- "model.layers.1.self_attn.k_proj",
119
- "model.layers.4.self_attn.v_proj",
120
- "model.layers.6.mlp.up_proj",
121
  "model.layers.34.self_attn.q_proj",
122
- "model.layers.27.self_attn.o_proj",
123
- "model.layers.33.self_attn.v_proj",
124
- "model.layers.10.self_attn.o_proj",
125
- "model.layers.34.mlp.up_proj",
126
- "model.layers.2.self_attn.o_proj",
127
- "model.layers.19.mlp.gate_proj",
128
- "model.layers.8.self_attn.v_proj",
129
- "model.layers.20.self_attn.q_proj",
130
- "model.layers.3.mlp.down_proj",
131
- "model.layers.8.mlp.gate_proj",
132
- "model.layers.5.mlp.down_proj",
133
- "model.layers.29.self_attn.o_proj",
134
- "model.layers.3.mlp.gate_proj",
135
- "model.layers.25.self_attn.o_proj",
136
- "model.layers.4.mlp.gate_proj",
137
  "model.layers.34.mlp.gate_proj",
138
- "model.layers.26.self_attn.k_proj",
139
- "model.layers.32.self_attn.q_proj",
140
- "model.layers.23.mlp.gate_proj",
141
- "model.layers.11.mlp.down_proj",
142
- "model.layers.1.self_attn.o_proj",
143
- "model.layers.4.mlp.up_proj",
144
- "model.layers.21.self_attn.q_proj",
 
145
  "model.layers.17.self_attn.k_proj",
 
 
 
 
 
 
 
 
 
 
 
 
 
 
146
  "model.layers.19.self_attn.v_proj",
147
- "model.layers.15.mlp.gate_proj",
148
- "model.layers.0.self_attn.o_proj",
149
- "model.layers.3.self_attn.k_proj",
150
- "model.layers.30.mlp.up_proj",
151
- "model.layers.33.self_attn.q_proj",
152
- "model.layers.1.mlp.gate_proj",
153
- "model.layers.12.mlp.gate_proj",
154
- "model.layers.11.self_attn.k_proj",
155
- "model.layers.23.mlp.down_proj",
156
- "model.layers.20.mlp.up_proj",
157
- "model.layers.9.mlp.down_proj",
158
- "model.layers.22.mlp.up_proj",
159
- "model.layers.9.mlp.up_proj",
160
  "model.layers.29.self_attn.k_proj",
161
- "model.layers.7.mlp.up_proj",
 
 
162
  "model.layers.25.mlp.up_proj",
163
- "model.layers.12.self_attn.k_proj",
164
- "model.layers.26.mlp.gate_proj",
 
 
 
 
 
 
 
 
165
  "model.layers.21.self_attn.v_proj",
166
- "model.layers.1.self_attn.v_proj",
167
- "model.layers.23.mlp.up_proj",
168
- "model.layers.35.mlp.down_proj",
169
- "model.layers.14.self_attn.o_proj",
170
- "model.layers.16.self_attn.o_proj",
171
- "model.layers.7.mlp.down_proj",
172
- "model.layers.30.self_attn.o_proj",
173
- "model.layers.17.mlp.down_proj",
174
- "model.layers.30.self_attn.v_proj",
175
- "model.layers.5.self_attn.k_proj",
176
- "model.layers.15.mlp.up_proj",
177
- "model.layers.3.mlp.up_proj",
178
  "model.layers.29.mlp.down_proj",
179
- "model.layers.24.self_attn.v_proj",
180
- "model.layers.23.self_attn.q_proj",
181
- "model.layers.23.self_attn.v_proj",
182
- "model.layers.31.self_attn.k_proj",
183
- "model.layers.22.self_attn.o_proj",
184
- "model.layers.26.mlp.up_proj",
185
- "model.layers.4.mlp.down_proj",
186
- "model.layers.31.self_attn.v_proj",
187
- "model.layers.13.self_attn.o_proj",
188
- "model.layers.24.mlp.down_proj",
189
- "model.layers.11.mlp.up_proj",
190
- "model.layers.2.self_attn.k_proj",
191
- "model.layers.6.self_attn.v_proj",
192
- "model.layers.24.self_attn.q_proj",
193
- "model.layers.13.self_attn.v_proj",
194
- "model.layers.3.self_attn.o_proj",
195
- "model.layers.13.self_attn.k_proj",
196
- "model.layers.4.self_attn.k_proj",
197
- "model.layers.0.mlp.down_proj",
198
- "model.layers.8.mlp.up_proj",
199
- "model.layers.20.self_attn.k_proj",
200
- "model.layers.15.self_attn.k_proj",
201
- "model.layers.21.mlp.up_proj",
202
- "model.layers.17.self_attn.v_proj",
203
  "model.layers.33.mlp.gate_proj",
204
- "model.layers.35.self_attn.k_proj",
205
  "model.layers.13.mlp.gate_proj",
206
- "model.layers.20.mlp.gate_proj",
207
- "model.layers.7.self_attn.o_proj",
208
- "model.layers.18.mlp.down_proj",
 
209
  "model.layers.29.mlp.up_proj",
210
- "model.layers.17.mlp.gate_proj",
211
- "model.layers.31.mlp.gate_proj",
212
- "model.layers.10.self_attn.v_proj",
 
 
213
  "model.layers.28.self_attn.o_proj",
214
- "model.layers.2.mlp.gate_proj",
215
- "model.layers.25.mlp.down_proj",
216
- "model.layers.6.mlp.down_proj",
217
- "model.layers.32.self_attn.o_proj",
218
- "model.layers.1.mlp.up_proj",
219
- "model.layers.9.self_attn.k_proj",
220
- "model.layers.22.mlp.gate_proj",
221
- "model.layers.8.self_attn.o_proj",
222
- "model.layers.32.self_attn.v_proj",
223
- "model.layers.34.self_attn.o_proj",
224
- "model.layers.0.self_attn.k_proj",
225
- "model.layers.35.self_attn.q_proj",
226
- "model.layers.14.self_attn.q_proj",
227
  "model.layers.8.self_attn.q_proj",
228
- "model.layers.27.mlp.down_proj",
229
- "model.layers.0.self_attn.q_proj",
230
- "model.layers.32.mlp.gate_proj",
231
- "model.layers.10.self_attn.k_proj",
232
- "model.layers.18.self_attn.v_proj",
233
- "model.layers.35.self_attn.v_proj",
234
- "model.layers.22.self_attn.v_proj",
 
 
 
 
235
  "model.layers.9.mlp.gate_proj",
236
- "model.layers.12.mlp.up_proj",
237
- "model.layers.33.self_attn.k_proj",
238
- "model.layers.11.mlp.gate_proj",
239
- "model.layers.15.mlp.down_proj",
240
- "model.layers.24.self_attn.o_proj",
241
- "model.layers.2.mlp.down_proj",
242
- "model.layers.2.mlp.up_proj",
243
- "model.layers.10.mlp.up_proj",
244
- "model.layers.14.mlp.down_proj",
245
- "model.layers.31.mlp.up_proj",
246
- "model.layers.33.self_attn.o_proj",
247
- "model.layers.6.self_attn.o_proj",
248
- "model.layers.5.self_attn.q_proj",
249
- "model.layers.16.self_attn.v_proj",
 
250
  "model.layers.7.mlp.gate_proj",
251
- "model.layers.18.mlp.gate_proj",
252
- "model.layers.20.mlp.down_proj",
253
- "model.layers.13.self_attn.q_proj",
254
  "model.layers.11.self_attn.v_proj",
255
- "model.layers.15.self_attn.v_proj",
256
- "model.layers.26.self_attn.q_proj",
257
- "model.layers.27.self_attn.q_proj",
258
- "model.layers.33.mlp.up_proj",
259
- "model.layers.35.mlp.up_proj",
260
- "model.layers.10.self_attn.q_proj",
261
- "model.layers.0.mlp.gate_proj",
262
- "model.layers.19.self_attn.k_proj",
263
- "model.layers.24.mlp.gate_proj",
264
- "model.layers.27.self_attn.k_proj",
265
- "model.layers.15.self_attn.q_proj",
266
- "model.layers.18.self_attn.k_proj",
267
- "model.layers.14.self_attn.k_proj",
268
- "model.layers.30.self_attn.k_proj",
269
- "model.layers.7.self_attn.v_proj",
270
  "model.layers.17.self_attn.o_proj",
271
- "model.layers.24.self_attn.k_proj",
272
- "model.layers.5.self_attn.v_proj",
273
- "model.layers.31.mlp.down_proj",
274
- "model.layers.27.mlp.gate_proj",
275
- "model.layers.1.mlp.down_proj",
276
- "model.layers.21.self_attn.k_proj",
277
- "model.layers.19.mlp.down_proj"
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
278
  ],
279
  "task_type": null,
280
  "use_dora": false,
 
23
  "rank_pattern": {},
24
  "revision": null,
25
  "target_modules": [
26
+ "model.layers.20.self_attn.q_proj",
27
+ "model.layers.31.self_attn.k_proj",
28
+ "model.layers.16.mlp.down_proj",
29
+ "model.layers.24.mlp.down_proj",
 
30
  "model.layers.28.mlp.down_proj",
31
+ "model.layers.22.self_attn.v_proj",
32
+ "model.layers.7.mlp.down_proj",
33
+ "model.layers.24.self_attn.v_proj",
34
+ "model.layers.26.self_attn.q_proj",
35
+ "model.layers.10.mlp.gate_proj",
36
+ "model.layers.28.mlp.gate_proj",
37
+ "model.layers.34.self_attn.o_proj",
38
+ "model.layers.31.self_attn.q_proj",
39
+ "model.layers.28.self_attn.q_proj",
40
+ "model.layers.12.self_attn.q_proj",
41
+ "model.layers.11.self_attn.k_proj",
42
+ "model.layers.13.self_attn.v_proj",
43
+ "model.layers.2.self_attn.q_proj",
44
+ "model.layers.21.self_attn.q_proj",
45
+ "model.layers.6.self_attn.o_proj",
46
+ "model.layers.12.self_attn.k_proj",
47
+ "model.layers.9.self_attn.o_proj",
48
+ "model.layers.5.self_attn.k_proj",
49
+ "model.layers.22.mlp.down_proj",
50
+ "model.layers.1.mlp.up_proj",
51
+ "model.layers.23.self_attn.o_proj",
52
+ "model.layers.13.self_attn.o_proj",
53
+ "model.layers.0.self_attn.q_proj",
54
+ "model.layers.5.mlp.up_proj",
55
+ "model.layers.3.mlp.gate_proj",
56
+ "model.layers.16.self_attn.v_proj",
57
  "model.layers.26.mlp.down_proj",
58
+ "model.layers.32.self_attn.v_proj",
 
 
 
 
59
  "model.layers.14.mlp.up_proj",
60
+ "model.layers.13.mlp.up_proj",
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
61
  "model.layers.18.self_attn.q_proj",
62
+ "model.layers.31.mlp.up_proj",
63
+ "model.layers.10.self_attn.o_proj",
64
+ "model.layers.9.self_attn.v_proj",
 
 
 
 
65
  "model.layers.23.self_attn.k_proj",
66
+ "model.layers.0.mlp.gate_proj",
67
+ "model.layers.25.self_attn.q_proj",
68
+ "model.layers.33.self_attn.q_proj",
69
+ "model.layers.18.self_attn.o_proj",
70
+ "model.layers.8.mlp.down_proj",
71
+ "model.layers.31.self_attn.o_proj",
72
+ "model.layers.15.self_attn.o_proj",
73
+ "model.layers.23.self_attn.q_proj",
74
+ "model.layers.8.self_attn.o_proj",
75
+ "model.layers.28.self_attn.k_proj",
76
+ "model.layers.12.mlp.gate_proj",
77
+ "model.layers.33.self_attn.k_proj",
78
+ "model.layers.34.self_attn.k_proj",
79
+ "model.layers.17.self_attn.v_proj",
80
+ "model.layers.33.self_attn.v_proj",
81
+ "model.layers.35.self_attn.k_proj",
82
+ "model.layers.32.mlp.up_proj",
83
+ "model.layers.10.self_attn.k_proj",
84
  "model.layers.28.self_attn.v_proj",
85
+ "model.layers.33.self_attn.o_proj",
86
+ "model.layers.21.self_attn.o_proj",
87
+ "model.layers.9.self_attn.q_proj",
88
+ "model.layers.19.mlp.gate_proj",
89
+ "model.layers.12.self_attn.o_proj",
90
  "model.layers.25.mlp.gate_proj",
91
+ "model.layers.9.self_attn.k_proj",
92
+ "model.layers.6.self_attn.q_proj",
93
+ "model.layers.16.self_attn.q_proj",
94
+ "model.layers.23.mlp.gate_proj",
95
+ "model.layers.35.self_attn.o_proj",
96
+ "model.layers.23.self_attn.v_proj",
97
+ "model.layers.33.mlp.down_proj",
98
+ "model.layers.2.mlp.gate_proj",
99
+ "model.layers.20.self_attn.o_proj",
100
+ "model.layers.33.mlp.up_proj",
101
+ "model.layers.30.self_attn.v_proj",
102
+ "model.layers.20.self_attn.v_proj",
103
+ "model.layers.27.mlp.gate_proj",
104
  "model.layers.7.self_attn.k_proj",
105
+ "model.layers.4.self_attn.o_proj",
106
+ "model.layers.14.mlp.gate_proj",
107
+ "model.layers.17.mlp.gate_proj",
108
+ "model.layers.5.mlp.gate_proj",
109
+ "model.layers.10.mlp.up_proj",
110
+ "model.layers.18.self_attn.k_proj",
111
+ "model.layers.4.self_attn.k_proj",
112
  "model.layers.1.self_attn.q_proj",
113
+ "model.layers.35.mlp.down_proj",
 
 
 
 
 
 
 
114
  "model.layers.13.mlp.down_proj",
115
+ "model.layers.31.mlp.gate_proj",
116
+ "model.layers.10.self_attn.v_proj",
117
+ "model.layers.35.mlp.gate_proj",
118
+ "model.layers.35.self_attn.q_proj",
119
+ "model.layers.5.self_attn.v_proj",
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
120
  "model.layers.34.self_attn.q_proj",
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
121
  "model.layers.34.mlp.gate_proj",
122
+ "model.layers.1.mlp.gate_proj",
123
+ "model.layers.6.mlp.up_proj",
124
+ "model.layers.15.self_attn.k_proj",
125
+ "model.layers.0.mlp.up_proj",
126
+ "model.layers.18.self_attn.v_proj",
127
+ "model.layers.20.self_attn.k_proj",
128
+ "model.layers.27.self_attn.k_proj",
129
+ "model.layers.14.self_attn.v_proj",
130
  "model.layers.17.self_attn.k_proj",
131
+ "model.layers.26.self_attn.o_proj",
132
+ "model.layers.32.self_attn.o_proj",
133
+ "model.layers.1.self_attn.k_proj",
134
+ "model.layers.13.self_attn.q_proj",
135
+ "model.layers.14.self_attn.q_proj",
136
+ "model.layers.11.self_attn.o_proj",
137
+ "model.layers.16.self_attn.k_proj",
138
+ "model.layers.5.self_attn.q_proj",
139
+ "model.layers.30.self_attn.k_proj",
140
+ "model.layers.25.mlp.down_proj",
141
+ "model.layers.4.mlp.gate_proj",
142
+ "model.layers.0.self_attn.v_proj",
143
+ "model.layers.4.self_attn.q_proj",
144
+ "model.layers.7.self_attn.o_proj",
145
  "model.layers.19.self_attn.v_proj",
146
+ "model.layers.22.mlp.gate_proj",
147
+ "model.layers.2.self_attn.o_proj",
148
+ "model.layers.27.self_attn.v_proj",
149
+ "model.layers.15.mlp.up_proj",
150
+ "model.layers.25.self_attn.k_proj",
151
+ "model.layers.1.self_attn.o_proj",
152
+ "model.layers.28.mlp.up_proj",
153
+ "model.layers.30.mlp.down_proj",
154
+ "model.layers.19.mlp.down_proj",
155
+ "model.layers.24.mlp.gate_proj",
 
 
 
156
  "model.layers.29.self_attn.k_proj",
157
+ "model.layers.34.mlp.down_proj",
158
+ "model.layers.29.self_attn.q_proj",
159
+ "model.layers.4.mlp.down_proj",
160
  "model.layers.25.mlp.up_proj",
161
+ "model.layers.20.mlp.gate_proj",
162
+ "model.layers.24.mlp.up_proj",
163
+ "model.layers.3.self_attn.q_proj",
164
+ "model.layers.27.mlp.down_proj",
165
+ "model.layers.0.mlp.down_proj",
166
+ "model.layers.12.mlp.down_proj",
167
+ "model.layers.35.mlp.up_proj",
168
+ "model.layers.8.self_attn.v_proj",
169
+ "model.layers.6.mlp.down_proj",
170
+ "model.layers.3.self_attn.v_proj",
171
  "model.layers.21.self_attn.v_proj",
172
+ "model.layers.22.self_attn.q_proj",
173
+ "model.layers.25.self_attn.v_proj",
174
+ "model.layers.27.self_attn.q_proj",
 
 
 
 
 
 
 
 
 
175
  "model.layers.29.mlp.down_proj",
176
+ "model.layers.7.self_attn.v_proj",
177
+ "model.layers.30.mlp.gate_proj",
178
+ "model.layers.32.mlp.down_proj",
179
+ "model.layers.0.self_attn.k_proj",
180
+ "model.layers.2.mlp.down_proj",
181
+ "model.layers.18.mlp.gate_proj",
182
+ "model.layers.6.mlp.gate_proj",
183
+ "model.layers.19.self_attn.k_proj",
184
+ "model.layers.22.mlp.up_proj",
185
+ "model.layers.29.self_attn.o_proj",
186
+ "model.layers.8.mlp.gate_proj",
187
+ "model.layers.9.mlp.up_proj",
188
+ "model.layers.32.mlp.gate_proj",
189
+ "model.layers.11.mlp.gate_proj",
 
 
 
 
 
 
 
 
 
 
190
  "model.layers.33.mlp.gate_proj",
191
+ "model.layers.20.mlp.down_proj",
192
  "model.layers.13.mlp.gate_proj",
193
+ "model.layers.24.self_attn.k_proj",
194
+ "model.layers.14.self_attn.o_proj",
195
+ "model.layers.11.mlp.up_proj",
196
+ "model.layers.21.mlp.down_proj",
197
  "model.layers.29.mlp.up_proj",
198
+ "model.layers.31.mlp.down_proj",
199
+ "model.layers.3.self_attn.k_proj",
200
+ "model.layers.19.self_attn.q_proj",
201
+ "model.layers.32.self_attn.q_proj",
202
+ "model.layers.26.mlp.up_proj",
203
  "model.layers.28.self_attn.o_proj",
204
+ "model.layers.30.self_attn.o_proj",
205
+ "model.layers.10.self_attn.q_proj",
206
+ "model.layers.21.self_attn.k_proj",
207
+ "model.layers.16.mlp.gate_proj",
208
+ "model.layers.17.self_attn.q_proj",
209
+ "model.layers.12.self_attn.v_proj",
 
 
 
 
 
 
 
210
  "model.layers.8.self_attn.q_proj",
211
+ "model.layers.20.mlp.up_proj",
212
+ "model.layers.15.mlp.gate_proj",
213
+ "model.layers.7.self_attn.q_proj",
214
+ "model.layers.18.mlp.down_proj",
215
+ "model.layers.27.self_attn.o_proj",
216
+ "model.layers.26.self_attn.v_proj",
217
+ "model.layers.21.mlp.gate_proj",
218
+ "model.layers.25.self_attn.o_proj",
219
+ "model.layers.23.mlp.down_proj",
220
+ "model.layers.30.self_attn.q_proj",
221
+ "model.layers.5.self_attn.o_proj",
222
  "model.layers.9.mlp.gate_proj",
223
+ "model.layers.14.self_attn.k_proj",
224
+ "model.layers.5.mlp.down_proj",
225
+ "model.layers.22.self_attn.k_proj",
226
+ "model.layers.4.mlp.up_proj",
227
+ "model.layers.0.self_attn.o_proj",
228
+ "model.layers.31.self_attn.v_proj",
229
+ "model.layers.23.mlp.up_proj",
230
+ "model.layers.1.mlp.down_proj",
231
+ "model.layers.19.mlp.up_proj",
232
+ "model.layers.26.mlp.gate_proj",
233
+ "model.layers.18.mlp.up_proj",
234
+ "model.layers.6.self_attn.v_proj",
235
+ "model.layers.15.self_attn.q_proj",
236
+ "model.layers.19.self_attn.o_proj",
237
+ "model.layers.10.mlp.down_proj",
238
  "model.layers.7.mlp.gate_proj",
239
+ "model.layers.6.self_attn.k_proj",
 
 
240
  "model.layers.11.self_attn.v_proj",
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
241
  "model.layers.17.self_attn.o_proj",
242
+ "model.layers.9.mlp.down_proj",
243
+ "model.layers.3.mlp.up_proj",
244
+ "model.layers.15.mlp.down_proj",
245
+ "model.layers.24.self_attn.q_proj",
246
+ "model.layers.26.self_attn.k_proj",
247
+ "model.layers.27.mlp.up_proj",
248
+ "model.layers.29.mlp.gate_proj",
249
+ "model.layers.30.mlp.up_proj",
250
+ "model.layers.34.self_attn.v_proj",
251
+ "model.layers.8.mlp.up_proj",
252
+ "model.layers.32.self_attn.k_proj",
253
+ "model.layers.17.mlp.up_proj",
254
+ "model.layers.16.self_attn.o_proj",
255
+ "model.layers.11.mlp.down_proj",
256
+ "model.layers.2.self_attn.v_proj",
257
+ "model.layers.8.self_attn.k_proj",
258
+ "model.layers.3.mlp.down_proj",
259
+ "model.layers.3.self_attn.o_proj",
260
+ "model.layers.29.self_attn.v_proj",
261
+ "model.layers.11.self_attn.q_proj",
262
+ "model.layers.22.self_attn.o_proj",
263
+ "model.layers.24.self_attn.o_proj",
264
+ "model.layers.15.self_attn.v_proj",
265
+ "model.layers.12.mlp.up_proj",
266
+ "model.layers.2.self_attn.k_proj",
267
+ "model.layers.14.mlp.down_proj",
268
+ "model.layers.34.mlp.up_proj",
269
+ "model.layers.21.mlp.up_proj",
270
+ "model.layers.35.self_attn.v_proj",
271
+ "model.layers.1.self_attn.v_proj",
272
+ "model.layers.2.mlp.up_proj",
273
+ "model.layers.4.self_attn.v_proj",
274
+ "model.layers.7.mlp.up_proj",
275
+ "model.layers.16.mlp.up_proj",
276
+ "model.layers.17.mlp.down_proj",
277
+ "model.layers.13.self_attn.k_proj"
278
  ],
279
  "task_type": null,
280
  "use_dora": false,
adapter_model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:acbf0ec9048f27d845a4a59c5423cce0c906f22e1a4116a3fc63522d54ee97b6
3
  size 239536776
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3e4905216bab983746b4f74366f1ea8f8e1f07b23946874fa70d8bfed5f69a7d
3
  size 239536776
runs/Apr26_20-32-11_b23dc37221c4/events.out.tfevents.1745699583.b23dc37221c4 ADDED
File without changes
runs/Apr26_20-38-06_b23dc37221c4/events.out.tfevents.1745699940.b23dc37221c4 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6fef4d3c073e45b22fb1e9fff6d116aa2f240ac8ccadf1df65f6f338e8d777c2
3
+ size 6475
runs/Apr26_20-40-21_b23dc37221c4/events.out.tfevents.1745700074.b23dc37221c4 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:06528326dcd68e8c255f1916c64e85707b401fe987956bd68808ef6917817b9d
3
+ size 6475
runs/Apr26_20-42-32_b23dc37221c4/events.out.tfevents.1745700209.b23dc37221c4 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:80948ac85de2d33e0cdb056feb7cd170cd6b1a7088326c4d751b0060160f31ab
3
+ size 6682
runs/Apr26_20-45-00_b23dc37221c4/events.out.tfevents.1745700352.b23dc37221c4 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:14d75a27c08dcc978072cf5048dae1c55630b96652eaf47e856fe5224a5ac425
3
+ size 6477
training_args.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:bb93fe04a9e4f0de0ab71d46a1e3bb0616cef8feea8cda2e28790b7b139a6b73
3
  size 7736
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:866637a67e4d10cc9a986a56d97dbda1d72a81463e48f94e48feadc855c6facf
3
  size 7736