evanrichards commited on
Commit
4a3edda
1 Parent(s): e75a856

Training in progress, step 75

Browse files
adapter_config.json CHANGED
@@ -4,7 +4,7 @@
4
  "base_model_class": "LlavaNextForConditionalGeneration",
5
  "parent_library": "transformers.models.llava_next.modeling_llava_next"
6
  },
7
- "base_model_name_or_path": "looppayments/llava-7b",
8
  "bias": "none",
9
  "fan_in_fan_out": false,
10
  "inference_mode": true,
@@ -23,230 +23,230 @@
23
  "rank_pattern": {},
24
  "revision": null,
25
  "target_modules": [
 
 
 
 
 
 
 
 
 
 
 
26
  "language_model.model.layers.6.mlp.down_proj",
27
- "language_model.model.layers.26.mlp.up_proj",
28
- "language_model.model.layers.15.self_attn.o_proj",
29
- "language_model.model.layers.6.mlp.up_proj",
30
- "language_model.model.layers.20.self_attn.v_proj",
31
- "language_model.model.layers.20.self_attn.q_proj",
32
- "language_model.model.layers.20.self_attn.k_proj",
33
- "language_model.model.layers.29.self_attn.k_proj",
34
- "language_model.model.layers.25.self_attn.k_proj",
35
- "language_model.model.layers.12.self_attn.q_proj",
36
- "language_model.model.layers.27.self_attn.v_proj",
37
- "language_model.model.layers.14.mlp.up_proj",
38
  "language_model.model.layers.2.mlp.down_proj",
39
- "language_model.model.layers.2.mlp.gate_proj",
40
- "language_model.model.layers.14.self_attn.q_proj",
41
- "language_model.model.layers.17.self_attn.o_proj",
42
- "language_model.model.layers.23.self_attn.q_proj",
43
- "language_model.model.layers.5.self_attn.k_proj",
44
- "language_model.model.layers.12.mlp.gate_proj",
45
  "language_model.model.layers.4.self_attn.q_proj",
46
- "language_model.model.layers.22.self_attn.v_proj",
47
- "language_model.model.layers.12.self_attn.k_proj",
48
- "language_model.model.layers.24.mlp.gate_proj",
49
- "language_model.model.layers.18.self_attn.k_proj",
50
- "language_model.model.layers.4.mlp.up_proj",
51
- "language_model.model.layers.19.mlp.gate_proj",
52
- "language_model.model.layers.3.mlp.down_proj",
53
- "language_model.model.layers.8.self_attn.v_proj",
54
- "language_model.model.layers.11.self_attn.k_proj",
 
 
 
 
 
55
  "language_model.model.layers.31.self_attn.v_proj",
56
- "language_model.model.layers.3.mlp.up_proj",
57
- "language_model.model.layers.11.mlp.gate_proj",
58
- "language_model.model.layers.14.self_attn.v_proj",
59
- "language_model.model.layers.25.mlp.up_proj",
60
- "language_model.model.layers.18.self_attn.o_proj",
61
- "language_model.model.layers.28.mlp.up_proj",
62
- "language_model.model.layers.5.self_attn.v_proj",
63
- "language_model.model.layers.12.self_attn.v_proj",
64
- "language_model.model.layers.14.self_attn.o_proj",
65
- "language_model.model.layers.0.mlp.gate_proj",
66
- "language_model.model.layers.22.self_attn.o_proj",
67
- "language_model.model.layers.31.self_attn.q_proj",
68
  "language_model.model.layers.19.self_attn.o_proj",
69
- "language_model.model.layers.19.self_attn.q_proj",
70
- "language_model.model.layers.28.mlp.gate_proj",
71
- "language_model.model.layers.0.mlp.up_proj",
72
- "language_model.model.layers.16.self_attn.v_proj",
73
- "language_model.model.layers.15.self_attn.q_proj",
74
- "language_model.model.layers.8.self_attn.o_proj",
75
- "language_model.model.layers.3.self_attn.v_proj",
76
- "language_model.model.layers.10.self_attn.q_proj",
77
- "language_model.model.layers.11.mlp.up_proj",
78
- "language_model.model.layers.19.mlp.down_proj",
79
- "language_model.model.layers.1.mlp.gate_proj",
80
- "language_model.model.layers.18.mlp.gate_proj",
81
- "language_model.model.layers.26.mlp.down_proj",
82
  "language_model.model.layers.7.mlp.down_proj",
83
- "language_model.model.layers.4.mlp.down_proj",
84
- "language_model.model.layers.0.mlp.down_proj",
85
- "language_model.model.layers.3.self_attn.o_proj",
86
- "language_model.model.layers.11.self_attn.o_proj",
87
- "language_model.model.layers.18.self_attn.q_proj",
88
- "language_model.model.layers.6.self_attn.v_proj",
89
- "language_model.model.layers.31.mlp.gate_proj",
90
- "language_model.model.layers.13.mlp.down_proj",
91
- "language_model.model.layers.13.mlp.gate_proj",
92
- "language_model.model.layers.5.mlp.down_proj",
93
  "language_model.model.layers.19.self_attn.v_proj",
94
- "language_model.model.layers.24.mlp.up_proj",
95
- "language_model.model.layers.6.mlp.gate_proj",
96
- "language_model.model.layers.6.self_attn.q_proj",
97
  "language_model.model.layers.27.mlp.down_proj",
98
- "language_model.model.layers.8.mlp.gate_proj",
99
- "language_model.model.layers.25.mlp.down_proj",
100
- "language_model.model.layers.18.self_attn.v_proj",
101
- "language_model.model.layers.31.self_attn.o_proj",
102
- "language_model.model.layers.20.self_attn.o_proj",
103
- "language_model.model.layers.22.self_attn.k_proj",
104
- "language_model.model.layers.16.self_attn.k_proj",
105
- "language_model.model.layers.13.self_attn.o_proj",
106
- "language_model.model.layers.15.mlp.down_proj",
107
- "language_model.model.layers.26.self_attn.k_proj",
108
- "language_model.model.layers.11.self_attn.v_proj",
109
- "language_model.model.layers.3.self_attn.k_proj",
110
- "language_model.model.layers.9.self_attn.o_proj",
111
- "language_model.model.layers.8.mlp.up_proj",
112
  "language_model.model.layers.6.self_attn.k_proj",
113
- "language_model.model.layers.13.self_attn.k_proj",
114
- "language_model.model.layers.13.mlp.up_proj",
115
- "language_model.model.layers.20.mlp.down_proj",
116
- "language_model.model.layers.26.self_attn.v_proj",
117
- "language_model.model.layers.28.mlp.down_proj",
118
- "language_model.model.layers.17.self_attn.v_proj",
119
- "language_model.model.layers.24.self_attn.q_proj",
120
- "language_model.model.layers.26.self_attn.o_proj",
121
- "language_model.model.layers.29.self_attn.o_proj",
122
  "language_model.model.layers.7.self_attn.v_proj",
123
- "language_model.model.layers.21.mlp.gate_proj",
124
- "language_model.model.layers.10.self_attn.k_proj",
125
- "language_model.model.layers.30.mlp.down_proj",
126
- "language_model.model.layers.29.mlp.up_proj",
127
- "language_model.model.layers.23.self_attn.o_proj",
128
- "language_model.model.layers.2.self_attn.o_proj",
 
129
  "language_model.model.layers.28.self_attn.q_proj",
130
- "language_model.model.layers.30.self_attn.v_proj",
131
- "language_model.model.layers.7.self_attn.o_proj",
132
- "language_model.model.layers.21.mlp.up_proj",
133
- "language_model.model.layers.17.self_attn.q_proj",
134
- "language_model.model.layers.0.self_attn.q_proj",
135
- "language_model.model.layers.25.self_attn.o_proj",
136
- "language_model.model.layers.3.self_attn.q_proj",
137
- "language_model.model.layers.0.self_attn.v_proj",
138
- "language_model.model.layers.2.mlp.up_proj",
139
- "language_model.model.layers.21.self_attn.k_proj",
140
- "language_model.model.layers.24.self_attn.o_proj",
141
- "language_model.model.layers.14.self_attn.k_proj",
142
- "language_model.model.layers.15.mlp.up_proj",
143
  "language_model.model.layers.29.mlp.down_proj",
144
- "language_model.model.layers.16.mlp.gate_proj",
145
- "language_model.model.layers.2.self_attn.k_proj",
 
 
 
 
 
 
 
 
 
 
 
146
  "language_model.model.layers.23.self_attn.v_proj",
147
- "language_model.model.layers.30.self_attn.o_proj",
148
- "language_model.model.layers.30.self_attn.k_proj",
149
- "language_model.model.layers.4.self_attn.v_proj",
150
- "language_model.model.layers.6.self_attn.o_proj",
151
- "language_model.model.layers.15.self_attn.k_proj",
152
- "language_model.model.layers.18.mlp.up_proj",
153
- "language_model.model.layers.12.mlp.up_proj",
154
- "language_model.model.layers.30.mlp.up_proj",
155
- "language_model.model.layers.10.self_attn.o_proj",
156
- "language_model.model.layers.24.self_attn.v_proj",
157
  "language_model.model.layers.28.self_attn.o_proj",
158
- "language_model.model.layers.16.self_attn.q_proj",
159
- "language_model.model.layers.23.mlp.down_proj",
160
- "language_model.model.layers.29.self_attn.q_proj",
161
- "language_model.model.layers.29.mlp.gate_proj",
162
  "language_model.model.layers.12.mlp.down_proj",
163
- "language_model.model.layers.15.self_attn.v_proj",
 
 
 
 
 
164
  "language_model.model.layers.25.mlp.gate_proj",
165
- "language_model.model.layers.21.self_attn.q_proj",
166
- "language_model.model.layers.26.self_attn.q_proj",
167
- "language_model.model.layers.22.mlp.down_proj",
168
- "language_model.model.layers.27.self_attn.q_proj",
169
- "language_model.model.layers.4.mlp.gate_proj",
170
- "language_model.model.layers.0.self_attn.k_proj",
171
- "language_model.model.layers.30.self_attn.q_proj",
172
- "language_model.model.layers.7.mlp.gate_proj",
173
- "language_model.model.layers.2.self_attn.q_proj",
174
- "language_model.model.layers.10.mlp.gate_proj",
175
- "language_model.model.layers.0.self_attn.o_proj",
176
- "language_model.model.layers.9.self_attn.q_proj",
177
- "language_model.model.layers.1.mlp.down_proj",
178
- "language_model.model.layers.30.mlp.gate_proj",
179
- "language_model.model.layers.27.mlp.gate_proj",
180
- "language_model.model.layers.10.mlp.up_proj",
181
- "language_model.model.layers.27.mlp.up_proj",
182
- "language_model.model.layers.4.self_attn.k_proj",
183
- "language_model.model.layers.17.mlp.down_proj",
184
- "language_model.model.layers.17.mlp.gate_proj",
185
- "language_model.model.layers.9.self_attn.v_proj",
186
- "language_model.model.layers.20.mlp.up_proj",
187
  "language_model.model.layers.21.self_attn.v_proj",
188
- "language_model.model.layers.7.self_attn.q_proj",
189
- "language_model.model.layers.31.self_attn.k_proj",
190
- "language_model.model.layers.9.mlp.up_proj",
191
- "language_model.model.layers.11.self_attn.q_proj",
192
- "language_model.model.layers.23.self_attn.k_proj",
193
- "language_model.model.layers.25.self_attn.v_proj",
 
 
 
 
 
 
 
 
194
  "language_model.model.layers.26.mlp.gate_proj",
195
- "language_model.model.layers.10.self_attn.v_proj",
196
- "language_model.model.layers.5.mlp.up_proj",
197
- "language_model.model.layers.1.self_attn.v_proj",
198
- "language_model.model.layers.1.self_attn.q_proj",
199
- "language_model.model.layers.27.self_attn.k_proj",
200
  "language_model.model.layers.28.self_attn.v_proj",
201
- "language_model.model.layers.21.self_attn.o_proj",
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
202
  "language_model.model.layers.31.mlp.down_proj",
203
- "language_model.model.layers.8.self_attn.q_proj",
204
- "language_model.model.layers.22.mlp.gate_proj",
 
 
 
 
 
 
 
 
 
 
 
 
205
  "language_model.model.layers.23.mlp.gate_proj",
206
- "language_model.model.layers.24.mlp.down_proj",
207
- "language_model.model.layers.14.mlp.down_proj",
208
- "language_model.model.layers.3.mlp.gate_proj",
209
- "language_model.model.layers.27.self_attn.o_proj",
210
- "language_model.model.layers.1.self_attn.o_proj",
211
- "language_model.model.layers.13.self_attn.q_proj",
212
- "language_model.model.layers.21.mlp.down_proj",
213
- "language_model.model.layers.16.mlp.up_proj",
214
- "language_model.model.layers.9.mlp.down_proj",
215
- "language_model.model.layers.23.mlp.up_proj",
216
- "language_model.model.layers.31.mlp.up_proj",
217
  "language_model.model.layers.5.self_attn.q_proj",
218
- "language_model.model.layers.7.self_attn.k_proj",
219
- "language_model.model.layers.25.self_attn.q_proj",
220
- "language_model.model.layers.16.self_attn.o_proj",
221
- "language_model.model.layers.22.mlp.up_proj",
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
222
  "language_model.model.layers.7.mlp.up_proj",
223
- "language_model.model.layers.12.self_attn.o_proj",
224
- "language_model.model.layers.20.mlp.gate_proj",
225
- "language_model.model.layers.5.mlp.gate_proj",
 
 
 
 
 
 
 
 
 
 
226
  "language_model.model.layers.1.mlp.up_proj",
227
- "language_model.model.layers.5.self_attn.o_proj",
228
- "language_model.model.layers.17.self_attn.k_proj",
229
- "language_model.model.layers.19.mlp.up_proj",
230
  "language_model.model.layers.8.mlp.down_proj",
231
- "language_model.model.layers.11.mlp.down_proj",
232
- "language_model.model.layers.13.self_attn.v_proj",
233
- "language_model.model.layers.16.mlp.down_proj",
234
- "language_model.model.layers.28.self_attn.k_proj",
235
- "language_model.model.layers.9.self_attn.k_proj",
236
- "language_model.model.layers.1.self_attn.k_proj",
237
- "language_model.model.layers.9.mlp.gate_proj",
 
 
238
  "language_model.model.layers.8.self_attn.k_proj",
239
- "language_model.model.layers.22.self_attn.q_proj",
240
- "language_model.model.layers.4.self_attn.o_proj",
 
 
 
 
 
 
241
  "language_model.model.layers.2.self_attn.v_proj",
242
- "language_model.model.layers.18.mlp.down_proj",
 
 
 
 
 
 
 
 
243
  "language_model.model.layers.29.self_attn.v_proj",
244
- "language_model.model.layers.10.mlp.down_proj",
245
- "language_model.model.layers.24.self_attn.k_proj",
246
- "language_model.model.layers.19.self_attn.k_proj",
247
- "language_model.model.layers.14.mlp.gate_proj",
248
- "language_model.model.layers.17.mlp.up_proj",
249
- "language_model.model.layers.15.mlp.gate_proj"
250
  ],
251
  "task_type": null,
252
  "use_dora": false,
 
4
  "base_model_class": "LlavaNextForConditionalGeneration",
5
  "parent_library": "transformers.models.llava_next.modeling_llava_next"
6
  },
7
+ "base_model_name_or_path": "looppayments/llava-7b-addresses-only",
8
  "bias": "none",
9
  "fan_in_fan_out": false,
10
  "inference_mode": true,
 
23
  "rank_pattern": {},
24
  "revision": null,
25
  "target_modules": [
26
+ "language_model.model.layers.4.mlp.up_proj",
27
+ "language_model.model.layers.15.self_attn.v_proj",
28
+ "language_model.model.layers.17.self_attn.q_proj",
29
+ "language_model.model.layers.0.mlp.up_proj",
30
+ "language_model.model.layers.18.self_attn.v_proj",
31
+ "language_model.model.layers.23.self_attn.k_proj",
32
+ "language_model.model.layers.13.self_attn.q_proj",
33
+ "language_model.model.layers.26.self_attn.k_proj",
34
+ "language_model.model.layers.31.mlp.up_proj",
35
+ "language_model.model.layers.4.self_attn.k_proj",
36
+ "language_model.model.layers.3.mlp.down_proj",
37
  "language_model.model.layers.6.mlp.down_proj",
38
+ "language_model.model.layers.22.mlp.up_proj",
 
 
 
 
 
 
 
 
 
 
39
  "language_model.model.layers.2.mlp.down_proj",
 
 
 
 
 
 
40
  "language_model.model.layers.4.self_attn.q_proj",
41
+ "language_model.model.layers.18.self_attn.q_proj",
42
+ "language_model.model.layers.10.mlp.up_proj",
43
+ "language_model.model.layers.5.self_attn.o_proj",
44
+ "language_model.model.layers.31.self_attn.k_proj",
45
+ "language_model.model.layers.22.self_attn.k_proj",
46
+ "language_model.model.layers.13.mlp.down_proj",
47
+ "language_model.model.layers.11.mlp.down_proj",
48
+ "language_model.model.layers.31.mlp.gate_proj",
49
+ "language_model.model.layers.11.self_attn.q_proj",
50
+ "language_model.model.layers.27.self_attn.o_proj",
51
+ "language_model.model.layers.4.mlp.down_proj",
52
+ "language_model.model.layers.13.self_attn.v_proj",
53
+ "language_model.model.layers.14.mlp.down_proj",
54
+ "language_model.model.layers.29.self_attn.q_proj",
55
  "language_model.model.layers.31.self_attn.v_proj",
56
+ "language_model.model.layers.4.self_attn.o_proj",
57
+ "language_model.model.layers.26.self_attn.q_proj",
58
+ "language_model.model.layers.20.mlp.gate_proj",
59
+ "language_model.model.layers.25.mlp.down_proj",
60
+ "language_model.model.layers.7.self_attn.k_proj",
 
 
 
 
 
 
 
61
  "language_model.model.layers.19.self_attn.o_proj",
 
 
 
 
 
 
 
 
 
 
 
 
 
62
  "language_model.model.layers.7.mlp.down_proj",
63
+ "language_model.model.layers.8.self_attn.v_proj",
 
 
 
 
 
 
 
 
 
64
  "language_model.model.layers.19.self_attn.v_proj",
65
+ "language_model.model.layers.10.self_attn.o_proj",
 
 
66
  "language_model.model.layers.27.mlp.down_proj",
67
+ "language_model.model.layers.5.self_attn.k_proj",
68
+ "language_model.model.layers.10.self_attn.v_proj",
69
+ "language_model.model.layers.6.self_attn.q_proj",
70
+ "language_model.model.layers.30.self_attn.v_proj",
71
+ "language_model.model.layers.3.mlp.gate_proj",
72
+ "language_model.model.layers.14.mlp.up_proj",
73
+ "language_model.model.layers.0.mlp.gate_proj",
 
 
 
 
 
 
 
74
  "language_model.model.layers.6.self_attn.k_proj",
75
+ "language_model.model.layers.2.self_attn.k_proj",
76
+ "language_model.model.layers.19.mlp.gate_proj",
 
 
 
 
 
 
 
77
  "language_model.model.layers.7.self_attn.v_proj",
78
+ "language_model.model.layers.9.mlp.up_proj",
79
+ "language_model.model.layers.15.self_attn.o_proj",
80
+ "language_model.model.layers.29.self_attn.k_proj",
81
+ "language_model.model.layers.18.self_attn.k_proj",
82
+ "language_model.model.layers.18.mlp.up_proj",
83
+ "language_model.model.layers.5.mlp.gate_proj",
84
+ "language_model.model.layers.16.mlp.up_proj",
85
  "language_model.model.layers.28.self_attn.q_proj",
 
 
 
 
 
 
 
 
 
 
 
 
 
86
  "language_model.model.layers.29.mlp.down_proj",
87
+ "language_model.model.layers.10.mlp.down_proj",
88
+ "language_model.model.layers.1.mlp.down_proj",
89
+ "language_model.model.layers.28.self_attn.k_proj",
90
+ "language_model.model.layers.20.self_attn.q_proj",
91
+ "language_model.model.layers.9.self_attn.q_proj",
92
+ "language_model.model.layers.30.mlp.down_proj",
93
+ "language_model.model.layers.4.mlp.gate_proj",
94
+ "language_model.model.layers.19.mlp.up_proj",
95
+ "language_model.model.layers.25.self_attn.v_proj",
96
+ "language_model.model.layers.17.self_attn.o_proj",
97
+ "language_model.model.layers.21.self_attn.q_proj",
98
+ "language_model.model.layers.23.mlp.down_proj",
99
+ "language_model.model.layers.6.self_attn.v_proj",
100
  "language_model.model.layers.23.self_attn.v_proj",
 
 
 
 
 
 
 
 
 
 
101
  "language_model.model.layers.28.self_attn.o_proj",
 
 
 
 
102
  "language_model.model.layers.12.mlp.down_proj",
103
+ "language_model.model.layers.24.mlp.gate_proj",
104
+ "language_model.model.layers.2.mlp.gate_proj",
105
+ "language_model.model.layers.3.mlp.up_proj",
106
+ "language_model.model.layers.25.self_attn.k_proj",
107
+ "language_model.model.layers.10.self_attn.k_proj",
108
+ "language_model.model.layers.14.self_attn.o_proj",
109
  "language_model.model.layers.25.mlp.gate_proj",
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
110
  "language_model.model.layers.21.self_attn.v_proj",
111
+ "language_model.model.layers.26.mlp.down_proj",
112
+ "language_model.model.layers.3.self_attn.k_proj",
113
+ "language_model.model.layers.23.self_attn.o_proj",
114
+ "language_model.model.layers.17.self_attn.k_proj",
115
+ "language_model.model.layers.22.self_attn.o_proj",
116
+ "language_model.model.layers.26.self_attn.o_proj",
117
+ "language_model.model.layers.16.self_attn.k_proj",
118
+ "language_model.model.layers.14.self_attn.q_proj",
119
+ "language_model.model.layers.31.self_attn.q_proj",
120
+ "language_model.model.layers.12.self_attn.k_proj",
121
+ "language_model.model.layers.29.mlp.up_proj",
122
+ "language_model.model.layers.20.mlp.up_proj",
123
+ "language_model.model.layers.11.mlp.gate_proj",
124
+ "language_model.model.layers.15.mlp.gate_proj",
125
  "language_model.model.layers.26.mlp.gate_proj",
126
+ "language_model.model.layers.23.self_attn.q_proj",
127
+ "language_model.model.layers.0.self_attn.v_proj",
128
+ "language_model.model.layers.27.mlp.gate_proj",
 
 
129
  "language_model.model.layers.28.self_attn.v_proj",
130
+ "language_model.model.layers.11.self_attn.o_proj",
131
+ "language_model.model.layers.24.mlp.up_proj",
132
+ "language_model.model.layers.8.mlp.gate_proj",
133
+ "language_model.model.layers.11.self_attn.k_proj",
134
+ "language_model.model.layers.14.self_attn.k_proj",
135
+ "language_model.model.layers.21.self_attn.k_proj",
136
+ "language_model.model.layers.1.self_attn.v_proj",
137
+ "language_model.model.layers.15.mlp.up_proj",
138
+ "language_model.model.layers.17.mlp.gate_proj",
139
+ "language_model.model.layers.18.mlp.gate_proj",
140
+ "language_model.model.layers.19.mlp.down_proj",
141
+ "language_model.model.layers.30.self_attn.q_proj",
142
+ "language_model.model.layers.6.mlp.gate_proj",
143
+ "language_model.model.layers.21.mlp.down_proj",
144
+ "language_model.model.layers.5.mlp.up_proj",
145
+ "language_model.model.layers.14.mlp.gate_proj",
146
+ "language_model.model.layers.22.mlp.down_proj",
147
+ "language_model.model.layers.25.mlp.up_proj",
148
+ "language_model.model.layers.7.self_attn.o_proj",
149
+ "language_model.model.layers.17.mlp.down_proj",
150
+ "language_model.model.layers.7.mlp.gate_proj",
151
+ "language_model.model.layers.22.self_attn.q_proj",
152
+ "language_model.model.layers.1.self_attn.k_proj",
153
+ "language_model.model.layers.10.self_attn.q_proj",
154
+ "language_model.model.layers.20.self_attn.v_proj",
155
+ "language_model.model.layers.15.mlp.down_proj",
156
  "language_model.model.layers.31.mlp.down_proj",
157
+ "language_model.model.layers.26.self_attn.v_proj",
158
+ "language_model.model.layers.0.self_attn.q_proj",
159
+ "language_model.model.layers.30.mlp.up_proj",
160
+ "language_model.model.layers.12.self_attn.v_proj",
161
+ "language_model.model.layers.0.self_attn.k_proj",
162
+ "language_model.model.layers.16.self_attn.o_proj",
163
+ "language_model.model.layers.3.self_attn.q_proj",
164
+ "language_model.model.layers.12.self_attn.o_proj",
165
+ "language_model.model.layers.20.self_attn.o_proj",
166
+ "language_model.model.layers.15.self_attn.k_proj",
167
+ "language_model.model.layers.29.mlp.gate_proj",
168
+ "language_model.model.layers.8.mlp.up_proj",
169
+ "language_model.model.layers.28.mlp.up_proj",
170
+ "language_model.model.layers.22.self_attn.v_proj",
171
  "language_model.model.layers.23.mlp.gate_proj",
172
+ "language_model.model.layers.27.mlp.up_proj",
 
 
 
 
 
 
 
 
 
 
173
  "language_model.model.layers.5.self_attn.q_proj",
174
+ "language_model.model.layers.7.self_attn.q_proj",
175
+ "language_model.model.layers.22.mlp.gate_proj",
176
+ "language_model.model.layers.18.self_attn.o_proj",
177
+ "language_model.model.layers.30.mlp.gate_proj",
178
+ "language_model.model.layers.19.self_attn.q_proj",
179
+ "language_model.model.layers.21.self_attn.o_proj",
180
+ "language_model.model.layers.9.mlp.gate_proj",
181
+ "language_model.model.layers.24.self_attn.o_proj",
182
+ "language_model.model.layers.10.mlp.gate_proj",
183
+ "language_model.model.layers.20.mlp.down_proj",
184
+ "language_model.model.layers.2.self_attn.q_proj",
185
+ "language_model.model.layers.24.self_attn.k_proj",
186
+ "language_model.model.layers.28.mlp.down_proj",
187
+ "language_model.model.layers.9.self_attn.k_proj",
188
+ "language_model.model.layers.9.self_attn.o_proj",
189
+ "language_model.model.layers.24.self_attn.q_proj",
190
+ "language_model.model.layers.0.mlp.down_proj",
191
+ "language_model.model.layers.21.mlp.up_proj",
192
+ "language_model.model.layers.9.self_attn.v_proj",
193
+ "language_model.model.layers.19.self_attn.k_proj",
194
+ "language_model.model.layers.8.self_attn.q_proj",
195
+ "language_model.model.layers.11.mlp.up_proj",
196
  "language_model.model.layers.7.mlp.up_proj",
197
+ "language_model.model.layers.2.self_attn.o_proj",
198
+ "language_model.model.layers.17.mlp.up_proj",
199
+ "language_model.model.layers.21.mlp.gate_proj",
200
+ "language_model.model.layers.6.mlp.up_proj",
201
+ "language_model.model.layers.23.mlp.up_proj",
202
+ "language_model.model.layers.16.mlp.gate_proj",
203
+ "language_model.model.layers.16.mlp.down_proj",
204
+ "language_model.model.layers.17.self_attn.v_proj",
205
+ "language_model.model.layers.14.self_attn.v_proj",
206
+ "language_model.model.layers.3.self_attn.v_proj",
207
+ "language_model.model.layers.8.self_attn.o_proj",
208
+ "language_model.model.layers.12.mlp.up_proj",
209
+ "language_model.model.layers.13.mlp.up_proj",
210
  "language_model.model.layers.1.mlp.up_proj",
211
+ "language_model.model.layers.2.mlp.up_proj",
212
+ "language_model.model.layers.27.self_attn.q_proj",
213
+ "language_model.model.layers.29.self_attn.o_proj",
214
  "language_model.model.layers.8.mlp.down_proj",
215
+ "language_model.model.layers.25.self_attn.q_proj",
216
+ "language_model.model.layers.26.mlp.up_proj",
217
+ "language_model.model.layers.11.self_attn.v_proj",
218
+ "language_model.model.layers.3.self_attn.o_proj",
219
+ "language_model.model.layers.25.self_attn.o_proj",
220
+ "language_model.model.layers.13.mlp.gate_proj",
221
+ "language_model.model.layers.5.self_attn.v_proj",
222
+ "language_model.model.layers.5.mlp.down_proj",
223
+ "language_model.model.layers.1.mlp.gate_proj",
224
  "language_model.model.layers.8.self_attn.k_proj",
225
+ "language_model.model.layers.13.self_attn.k_proj",
226
+ "language_model.model.layers.16.self_attn.q_proj",
227
+ "language_model.model.layers.27.self_attn.v_proj",
228
+ "language_model.model.layers.1.self_attn.q_proj",
229
+ "language_model.model.layers.30.self_attn.o_proj",
230
+ "language_model.model.layers.31.self_attn.o_proj",
231
+ "language_model.model.layers.12.mlp.gate_proj",
232
+ "language_model.model.layers.27.self_attn.k_proj",
233
  "language_model.model.layers.2.self_attn.v_proj",
234
+ "language_model.model.layers.28.mlp.gate_proj",
235
+ "language_model.model.layers.15.self_attn.q_proj",
236
+ "language_model.model.layers.12.self_attn.q_proj",
237
+ "language_model.model.layers.16.self_attn.v_proj",
238
+ "language_model.model.layers.20.self_attn.k_proj",
239
+ "language_model.model.layers.24.self_attn.v_proj",
240
+ "language_model.model.layers.6.self_attn.o_proj",
241
+ "language_model.model.layers.9.mlp.down_proj",
242
+ "language_model.model.layers.0.self_attn.o_proj",
243
  "language_model.model.layers.29.self_attn.v_proj",
244
+ "language_model.model.layers.18.mlp.down_proj",
245
+ "language_model.model.layers.13.self_attn.o_proj",
246
+ "language_model.model.layers.1.self_attn.o_proj",
247
+ "language_model.model.layers.30.self_attn.k_proj",
248
+ "language_model.model.layers.24.mlp.down_proj",
249
+ "language_model.model.layers.4.self_attn.v_proj"
250
  ],
251
  "task_type": null,
252
  "use_dora": false,
adapter_model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:30c8c75369b1fbe98f74ade191ffccbba5ac37e90016e0b3687bd0a52ffd87a1
3
  size 671155888
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9af98778be767d5899cef90dc27e12abd0250a82cea2b7010b033610f090d006
3
  size 671155888
runs/Aug20_07-21-13_0e8ddffc1b5c/events.out.tfevents.1724138481.0e8ddffc1b5c.3026.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:78778827bf4e3c0160ae8ab5abe12b60e5fa483d08ca74ab6414eb56d90750a0
3
+ size 6614
training_args.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:f2156d4e72def29decfe6c554fd404ac954538d65b87e9616e2edae1cab2477f
3
  size 5176
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ceffd3c110a76951234ad707083db79e2acbfd97ed4c9587bed5b5e69c3dd654
3
  size 5176