fix(config): use explicit KDA beta-gate exclusion paths
#8
by iCavan - opened
- config.json +63 -1
config.json
CHANGED
|
@@ -130,7 +130,69 @@
|
|
| 130 |
"modules_to_not_convert": [
|
| 131 |
"lm_head",
|
| 132 |
"model.word_embeddings",
|
| 133 |
-
"attention.b_proj",
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 134 |
"model.layers.7.attention.g_proj",
|
| 135 |
"model.layers.15.attention.g_proj",
|
| 136 |
"model.layers.23.attention.g_proj",
|
|
|
|
| 130 |
"modules_to_not_convert": [
|
| 131 |
"lm_head",
|
| 132 |
"model.word_embeddings",
|
| 133 |
+
"model.layers.0.attention.b_proj",
|
| 134 |
+
"model.layers.1.attention.b_proj",
|
| 135 |
+
"model.layers.2.attention.b_proj",
|
| 136 |
+
"model.layers.3.attention.b_proj",
|
| 137 |
+
"model.layers.4.attention.b_proj",
|
| 138 |
+
"model.layers.5.attention.b_proj",
|
| 139 |
+
"model.layers.6.attention.b_proj",
|
| 140 |
+
"model.layers.8.attention.b_proj",
|
| 141 |
+
"model.layers.9.attention.b_proj",
|
| 142 |
+
"model.layers.10.attention.b_proj",
|
| 143 |
+
"model.layers.11.attention.b_proj",
|
| 144 |
+
"model.layers.12.attention.b_proj",
|
| 145 |
+
"model.layers.13.attention.b_proj",
|
| 146 |
+
"model.layers.14.attention.b_proj",
|
| 147 |
+
"model.layers.16.attention.b_proj",
|
| 148 |
+
"model.layers.17.attention.b_proj",
|
| 149 |
+
"model.layers.18.attention.b_proj",
|
| 150 |
+
"model.layers.19.attention.b_proj",
|
| 151 |
+
"model.layers.20.attention.b_proj",
|
| 152 |
+
"model.layers.21.attention.b_proj",
|
| 153 |
+
"model.layers.22.attention.b_proj",
|
| 154 |
+
"model.layers.24.attention.b_proj",
|
| 155 |
+
"model.layers.25.attention.b_proj",
|
| 156 |
+
"model.layers.26.attention.b_proj",
|
| 157 |
+
"model.layers.27.attention.b_proj",
|
| 158 |
+
"model.layers.28.attention.b_proj",
|
| 159 |
+
"model.layers.29.attention.b_proj",
|
| 160 |
+
"model.layers.30.attention.b_proj",
|
| 161 |
+
"model.layers.32.attention.b_proj",
|
| 162 |
+
"model.layers.33.attention.b_proj",
|
| 163 |
+
"model.layers.34.attention.b_proj",
|
| 164 |
+
"model.layers.35.attention.b_proj",
|
| 165 |
+
"model.layers.36.attention.b_proj",
|
| 166 |
+
"model.layers.37.attention.b_proj",
|
| 167 |
+
"model.layers.38.attention.b_proj",
|
| 168 |
+
"model.layers.40.attention.b_proj",
|
| 169 |
+
"model.layers.41.attention.b_proj",
|
| 170 |
+
"model.layers.42.attention.b_proj",
|
| 171 |
+
"model.layers.43.attention.b_proj",
|
| 172 |
+
"model.layers.44.attention.b_proj",
|
| 173 |
+
"model.layers.45.attention.b_proj",
|
| 174 |
+
"model.layers.46.attention.b_proj",
|
| 175 |
+
"model.layers.48.attention.b_proj",
|
| 176 |
+
"model.layers.49.attention.b_proj",
|
| 177 |
+
"model.layers.50.attention.b_proj",
|
| 178 |
+
"model.layers.51.attention.b_proj",
|
| 179 |
+
"model.layers.52.attention.b_proj",
|
| 180 |
+
"model.layers.53.attention.b_proj",
|
| 181 |
+
"model.layers.54.attention.b_proj",
|
| 182 |
+
"model.layers.56.attention.b_proj",
|
| 183 |
+
"model.layers.57.attention.b_proj",
|
| 184 |
+
"model.layers.58.attention.b_proj",
|
| 185 |
+
"model.layers.59.attention.b_proj",
|
| 186 |
+
"model.layers.60.attention.b_proj",
|
| 187 |
+
"model.layers.61.attention.b_proj",
|
| 188 |
+
"model.layers.62.attention.b_proj",
|
| 189 |
+
"model.layers.64.attention.b_proj",
|
| 190 |
+
"model.layers.65.attention.b_proj",
|
| 191 |
+
"model.layers.66.attention.b_proj",
|
| 192 |
+
"model.layers.67.attention.b_proj",
|
| 193 |
+
"model.layers.68.attention.b_proj",
|
| 194 |
+
"model.layers.69.attention.b_proj",
|
| 195 |
+
"model.layers.70.attention.b_proj",
|
| 196 |
"model.layers.7.attention.g_proj",
|
| 197 |
"model.layers.15.attention.g_proj",
|
| 198 |
"model.layers.23.attention.g_proj",
|