Instructions to use alexantonov/gemma-text-to-sql with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use alexantonov/gemma-text-to-sql with Transformers:
# Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("alexantonov/gemma-text-to-sql", dtype="auto") - Notebooks
- Google Colab
- Kaggle
Training in progress, epoch 1
Browse files- adapter_config.json +190 -190
- adapter_model.safetensors +1 -1
- runs/May09_08-02-47_g21-3e350fb3-8deb-3e4a-ae1a-6dbf2e96917c/events.out.tfevents.1778313767.g21-3e350fb3-8deb-3e4a-ae1a-6dbf2e96917c.17051.0 +2 -2
- runs/May09_08-28-20_g21-3e350fb3-8deb-3e4a-ae1a-6dbf2e96917c/events.out.tfevents.1778315300.g21-3e350fb3-8deb-3e4a-ae1a-6dbf2e96917c.17051.1 +3 -0
- runs/May09_08-34-00_g21-3e350fb3-8deb-3e4a-ae1a-6dbf2e96917c/events.out.tfevents.1778315640.g21-3e350fb3-8deb-3e4a-ae1a-6dbf2e96917c.17051.2 +3 -0
- runs/May09_08-37-46_g21-3e350fb3-8deb-3e4a-ae1a-6dbf2e96917c/events.out.tfevents.1778315866.g21-3e350fb3-8deb-3e4a-ae1a-6dbf2e96917c.17051.3 +3 -0
- runs/May09_08-43-36_g21-3e350fb3-8deb-3e4a-ae1a-6dbf2e96917c/events.out.tfevents.1778316216.g21-3e350fb3-8deb-3e4a-ae1a-6dbf2e96917c.19296.0 +3 -0
adapter_config.json
CHANGED
|
@@ -33,220 +33,220 @@
|
|
| 33 |
"rank_pattern": {},
|
| 34 |
"revision": null,
|
| 35 |
"target_modules": [
|
| 36 |
-
"
|
| 37 |
-
"
|
| 38 |
-
"language_model.layers.
|
| 39 |
-
"
|
| 40 |
-
"language_model.layers.
|
| 41 |
-
"
|
| 42 |
-
"
|
| 43 |
-
"language_model.layers.
|
| 44 |
-
"language_model.layers.
|
| 45 |
-
"
|
| 46 |
-
"per_layer_input_gate",
|
| 47 |
-
"16.self_attn.q_proj",
|
| 48 |
-
"language_model.layers.1.self_attn.o_proj",
|
| 49 |
-
"20.mlp.gate_proj",
|
| 50 |
-
"language_model.layers.8.mlp.up_proj",
|
| 51 |
-
"relative_k_proj",
|
| 52 |
-
"28.mlp.gate_proj",
|
| 53 |
-
"language_model.layers.11.mlp.up_proj",
|
| 54 |
-
"language_model.layers.6.self_attn.q_proj",
|
| 55 |
-
"17.mlp.up_proj",
|
| 56 |
"language_model.layers.9.self_attn.q_proj",
|
| 57 |
-
"
|
| 58 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 59 |
"25.self_attn.q_proj",
|
| 60 |
-
"
|
| 61 |
-
"language_model.layers.
|
| 62 |
-
"language_model.layers.
|
| 63 |
-
"22.mlp.gate_proj",
|
| 64 |
-
"24.mlp.up_proj",
|
| 65 |
-
"language_model.layers.14.self_attn.k_proj",
|
| 66 |
-
"language_model.layers.1.self_attn.q_proj",
|
| 67 |
-
"24.mlp.down_proj",
|
| 68 |
"34.mlp.down_proj",
|
| 69 |
-
"
|
| 70 |
-
"
|
| 71 |
-
"language_model.layers.
|
| 72 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 73 |
"23.mlp.down_proj",
|
| 74 |
-
"language_model.layers.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 75 |
"language_model.layers.1.mlp.up_proj",
|
| 76 |
-
"language_model.layers.8.self_attn.v_proj",
|
| 77 |
-
"33.mlp.gate_proj",
|
| 78 |
-
"language_model.layers.7.self_attn.k_proj",
|
| 79 |
"21.mlp.down_proj",
|
| 80 |
-
"
|
| 81 |
-
"
|
| 82 |
-
"language_model.layers.
|
| 83 |
-
"
|
| 84 |
-
"language_model.layers.0.self_attn.
|
| 85 |
-
"
|
| 86 |
-
"language_model.layers.
|
| 87 |
-
"
|
| 88 |
-
"language_model.layers.
|
| 89 |
-
"language_model.layers.
|
| 90 |
-
"
|
| 91 |
-
"19.mlp.gate_proj",
|
| 92 |
-
"language_model.layers.13.mlp.gate_proj",
|
| 93 |
-
"28.mlp.up_proj",
|
| 94 |
-
"language_model.layers.13.self_attn.q_proj",
|
| 95 |
-
"language_model.layers.2.mlp.down_proj",
|
| 96 |
-
"29.self_attn.q_proj",
|
| 97 |
-
"language_model.layers.3.self_attn.o_proj",
|
| 98 |
-
"output_proj",
|
| 99 |
-
"34.mlp.gate_proj",
|
| 100 |
-
"language_model.layers.3.mlp.gate_proj",
|
| 101 |
-
"21.self_attn.o_proj",
|
| 102 |
-
"language_model.layers.2.self_attn.o_proj",
|
| 103 |
-
"16.mlp.gate_proj",
|
| 104 |
-
"32.self_attn.o_proj",
|
| 105 |
-
"19.self_attn.q_proj",
|
| 106 |
-
"language_model.layers.7.self_attn.q_proj",
|
| 107 |
-
"per_layer_projection",
|
| 108 |
-
"27.self_attn.o_proj",
|
| 109 |
-
"input_proj_linear",
|
| 110 |
"language_model.layers.12.self_attn.v_proj",
|
| 111 |
-
"
|
| 112 |
-
"language_model.layers.
|
| 113 |
-
"language_model.layers.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 114 |
"language_model.layers.15.mlp.gate_proj",
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 115 |
"29.mlp.up_proj",
|
| 116 |
-
"language_model.layers.
|
| 117 |
-
"
|
| 118 |
-
"
|
| 119 |
-
"language_model.layers.
|
| 120 |
-
"
|
| 121 |
-
"language_model.layers.6.self_attn.v_proj",
|
| 122 |
-
"language_model.layers.15.mlp.up_proj",
|
| 123 |
-
"language_model.layers.11.mlp.gate_proj",
|
| 124 |
-
"24.mlp.gate_proj",
|
| 125 |
-
"language_model.layers.13.mlp.down_proj",
|
| 126 |
-
"language_model.layers.7.self_attn.o_proj",
|
| 127 |
-
"18.mlp.gate_proj",
|
| 128 |
-
"language_model.layers.12.self_attn.o_proj",
|
| 129 |
"23.mlp.up_proj",
|
| 130 |
-
"
|
| 131 |
-
"
|
| 132 |
-
"
|
| 133 |
-
"
|
| 134 |
-
"
|
|
|
|
| 135 |
"language_model.layers.12.mlp.up_proj",
|
| 136 |
-
"
|
| 137 |
"language_model.layers.11.self_attn.k_proj",
|
| 138 |
-
"
|
| 139 |
-
"language_model.layers.
|
| 140 |
-
"
|
| 141 |
-
"
|
| 142 |
-
"
|
| 143 |
-
"
|
| 144 |
-
"language_model.layers.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 145 |
"language_model.layers.14.self_attn.o_proj",
|
| 146 |
-
"
|
| 147 |
-
"
|
| 148 |
-
"language_model.layers.
|
| 149 |
-
"
|
| 150 |
-
"language_model.layers.
|
| 151 |
-
"language_model.layers.5.mlp.down_proj",
|
| 152 |
-
"input_proj",
|
| 153 |
-
"language_model.layers.10.self_attn.q_proj",
|
| 154 |
-
"32.mlp.gate_proj",
|
| 155 |
-
"language_model.layers.0.mlp.gate_proj",
|
| 156 |
-
"21.mlp.up_proj",
|
| 157 |
-
"31.self_attn.o_proj",
|
| 158 |
-
"language_model.layers.1.mlp.down_proj",
|
| 159 |
-
"language_model.layers.4.self_attn.k_proj",
|
| 160 |
-
"language_model.layers.4.self_attn.q_proj",
|
| 161 |
-
"20.mlp.up_proj",
|
| 162 |
"language_model.layers.10.mlp.down_proj",
|
| 163 |
-
"
|
| 164 |
-
"
|
| 165 |
-
"
|
| 166 |
-
"
|
| 167 |
-
"
|
| 168 |
"20.self_attn.q_proj",
|
| 169 |
-
"
|
| 170 |
-
"
|
| 171 |
-
"
|
|
|
|
| 172 |
"language_model.layers.10.mlp.up_proj",
|
| 173 |
-
"language_model.layers.
|
| 174 |
-
"
|
| 175 |
-
"
|
| 176 |
-
"
|
| 177 |
-
"
|
| 178 |
-
"
|
| 179 |
-
"
|
| 180 |
-
"language_model.layers.
|
| 181 |
-
"
|
| 182 |
-
"
|
| 183 |
-
"
|
| 184 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 185 |
"language_model.layers.2.mlp.up_proj",
|
| 186 |
-
"language_model.layers.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 187 |
"language_model.layers.13.self_attn.v_proj",
|
| 188 |
-
"language_model.layers.3.self_attn.v_proj",
|
| 189 |
-
"29.mlp.down_proj",
|
| 190 |
"28.mlp.down_proj",
|
| 191 |
-
"
|
| 192 |
-
"
|
| 193 |
-
"
|
| 194 |
-
"
|
| 195 |
-
"
|
| 196 |
-
"language_model.layers.
|
| 197 |
-
"
|
| 198 |
-
"
|
| 199 |
-
"language_model.layers.
|
| 200 |
-
"language_model.layers.
|
| 201 |
-
"language_model.layers.0.self_attn.k_proj",
|
| 202 |
-
"22.self_attn.q_proj",
|
| 203 |
-
"26.self_attn.q_proj",
|
| 204 |
-
"19.mlp.up_proj",
|
| 205 |
"25.mlp.up_proj",
|
| 206 |
-
"language_model.layers.
|
| 207 |
-
"
|
| 208 |
-
"
|
| 209 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 210 |
"17.self_attn.q_proj",
|
| 211 |
-
"
|
| 212 |
-
"
|
| 213 |
-
"
|
| 214 |
-
"
|
| 215 |
"language_model.layers.1.self_attn.k_proj",
|
| 216 |
-
"language_model.layers.12.self_attn.q_proj",
|
| 217 |
-
"26.mlp.down_proj",
|
| 218 |
-
"language_model.layers.10.self_attn.v_proj",
|
| 219 |
-
"23.mlp.gate_proj",
|
| 220 |
-
"28.self_attn.o_proj",
|
| 221 |
-
"language_model.layers.7.mlp.up_proj",
|
| 222 |
"language_model.layers.14.self_attn.q_proj",
|
| 223 |
-
"
|
| 224 |
-
"27.mlp.up_proj",
|
| 225 |
"16.mlp.down_proj",
|
| 226 |
-
"
|
| 227 |
-
"
|
| 228 |
-
"
|
| 229 |
-
"
|
| 230 |
-
"
|
| 231 |
-
"language_model.layers.14.mlp.
|
| 232 |
-
"
|
| 233 |
-
"
|
| 234 |
-
"language_model.layers.
|
| 235 |
-
"language_model.layers.
|
| 236 |
-
"23.self_attn.q_proj",
|
| 237 |
-
"language_model.layers.2.mlp.gate_proj",
|
| 238 |
-
"language_model.layers.5.self_attn.k_proj",
|
| 239 |
-
"24.self_attn.q_proj",
|
| 240 |
-
"20.mlp.down_proj",
|
| 241 |
-
"language_model.layers.5.self_attn.q_proj",
|
| 242 |
-
"language_model.layers.9.self_attn.o_proj",
|
| 243 |
-
"29.self_attn.o_proj",
|
| 244 |
-
"language_model.layers.3.mlp.down_proj",
|
| 245 |
-
"language_model.layers.7.mlp.down_proj",
|
| 246 |
-
"17.mlp.down_proj",
|
| 247 |
-
"16.mlp.up_proj",
|
| 248 |
-
"20.self_attn.o_proj",
|
| 249 |
-
"19.mlp.down_proj"
|
| 250 |
],
|
| 251 |
"target_parameters": null,
|
| 252 |
"task_type": "CAUSAL_LM",
|
|
|
|
| 33 |
"rank_pattern": {},
|
| 34 |
"revision": null,
|
| 35 |
"target_modules": [
|
| 36 |
+
"21.mlp.gate_proj",
|
| 37 |
+
"language_model.layers.15.mlp.up_proj",
|
| 38 |
+
"language_model.layers.4.self_attn.o_proj",
|
| 39 |
+
"language_model.layers.13.self_attn.k_proj",
|
| 40 |
+
"language_model.layers.9.mlp.gate_proj",
|
| 41 |
+
"24.self_attn.o_proj",
|
| 42 |
+
"30.self_attn.o_proj",
|
| 43 |
+
"language_model.layers.7.self_attn.o_proj",
|
| 44 |
+
"language_model.layers.0.self_attn.q_proj",
|
| 45 |
+
"29.self_attn.q_proj",
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 46 |
"language_model.layers.9.self_attn.q_proj",
|
| 47 |
+
"language_model.layers.2.self_attn.q_proj",
|
| 48 |
+
"input_proj_linear",
|
| 49 |
+
"25.self_attn.o_proj",
|
| 50 |
+
"language_model.layers.3.mlp.down_proj",
|
| 51 |
+
"31.self_attn.q_proj",
|
| 52 |
+
"18.mlp.gate_proj",
|
| 53 |
+
"29.mlp.down_proj",
|
| 54 |
+
"33.mlp.up_proj",
|
| 55 |
+
"language_model.layers.2.mlp.gate_proj",
|
| 56 |
+
"language_model.layers.9.self_attn.o_proj",
|
| 57 |
+
"input_proj",
|
| 58 |
+
"language_model.layers.15.mlp.down_proj",
|
| 59 |
+
"language_model.layers.10.self_attn.k_proj",
|
| 60 |
"25.self_attn.q_proj",
|
| 61 |
+
"language_model.layers.15.self_attn.o_proj",
|
| 62 |
+
"language_model.layers.6.mlp.up_proj",
|
| 63 |
+
"language_model.layers.9.self_attn.k_proj",
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 64 |
"34.mlp.down_proj",
|
| 65 |
+
"18.self_attn.o_proj",
|
| 66 |
+
"27.self_attn.o_proj",
|
| 67 |
+
"language_model.layers.11.self_attn.o_proj",
|
| 68 |
+
"23.self_attn.o_proj",
|
| 69 |
+
"30.mlp.down_proj",
|
| 70 |
+
"language_model.layers.14.mlp.up_proj",
|
| 71 |
+
"linear",
|
| 72 |
+
"language_model.layers.5.self_attn.k_proj",
|
| 73 |
+
"16.mlp.gate_proj",
|
| 74 |
+
"language_model.layers.3.mlp.up_proj",
|
| 75 |
+
"18.mlp.down_proj",
|
| 76 |
+
"language_model.layers.5.mlp.down_proj",
|
| 77 |
+
"32.self_attn.o_proj",
|
| 78 |
+
"26.self_attn.o_proj",
|
| 79 |
+
"16.self_attn.o_proj",
|
| 80 |
+
"25.mlp.gate_proj",
|
| 81 |
"23.mlp.down_proj",
|
| 82 |
+
"language_model.layers.1.mlp.down_proj",
|
| 83 |
+
"language_model.layers.14.self_attn.k_proj",
|
| 84 |
+
"20.mlp.gate_proj",
|
| 85 |
+
"28.self_attn.o_proj",
|
| 86 |
+
"34.mlp.gate_proj",
|
| 87 |
+
"34.self_attn.o_proj",
|
| 88 |
+
"language_model.layers.5.mlp.up_proj",
|
| 89 |
"language_model.layers.1.mlp.up_proj",
|
|
|
|
|
|
|
|
|
|
| 90 |
"21.mlp.down_proj",
|
| 91 |
+
"32.self_attn.q_proj",
|
| 92 |
+
"29.self_attn.o_proj",
|
| 93 |
+
"language_model.layers.9.mlp.up_proj",
|
| 94 |
+
"16.mlp.up_proj",
|
| 95 |
+
"language_model.layers.0.self_attn.k_proj",
|
| 96 |
+
"25.mlp.down_proj",
|
| 97 |
+
"language_model.layers.14.self_attn.v_proj",
|
| 98 |
+
"language_model.layers.6.self_attn.k_proj",
|
| 99 |
+
"language_model.layers.7.self_attn.v_proj",
|
| 100 |
+
"language_model.layers.1.mlp.gate_proj",
|
| 101 |
+
"language_model.layers.7.mlp.down_proj",
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 102 |
"language_model.layers.12.self_attn.v_proj",
|
| 103 |
+
"language_model.layers.13.self_attn.o_proj",
|
| 104 |
+
"language_model.layers.6.self_attn.v_proj",
|
| 105 |
+
"language_model.layers.4.self_attn.v_proj",
|
| 106 |
+
"19.self_attn.o_proj",
|
| 107 |
+
"language_model.layers.1.self_attn.o_proj",
|
| 108 |
+
"language_model.layers.10.self_attn.o_proj",
|
| 109 |
+
"language_model.layers.2.self_attn.k_proj",
|
| 110 |
+
"language_model.layers.10.self_attn.v_proj",
|
| 111 |
+
"language_model.layers.11.self_attn.v_proj",
|
| 112 |
+
"language_model.layers.7.self_attn.k_proj",
|
| 113 |
"language_model.layers.15.mlp.gate_proj",
|
| 114 |
+
"30.mlp.gate_proj",
|
| 115 |
+
"31.mlp.up_proj",
|
| 116 |
+
"23.self_attn.q_proj",
|
| 117 |
+
"language_model.layers.10.self_attn.q_proj",
|
| 118 |
+
"language_model.layers.8.self_attn.v_proj",
|
| 119 |
+
"17.mlp.up_proj",
|
| 120 |
"29.mlp.up_proj",
|
| 121 |
+
"language_model.layers.10.mlp.gate_proj",
|
| 122 |
+
"21.self_attn.o_proj",
|
| 123 |
+
"26.mlp.gate_proj",
|
| 124 |
+
"language_model.layers.4.self_attn.q_proj",
|
| 125 |
+
"relative_k_proj",
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 126 |
"23.mlp.up_proj",
|
| 127 |
+
"23.mlp.gate_proj",
|
| 128 |
+
"per_layer_model_projection",
|
| 129 |
+
"language_model.layers.4.mlp.gate_proj",
|
| 130 |
+
"language_model.layers.0.self_attn.o_proj",
|
| 131 |
+
"22.self_attn.q_proj",
|
| 132 |
+
"language_model.layers.8.self_attn.o_proj",
|
| 133 |
"language_model.layers.12.mlp.up_proj",
|
| 134 |
+
"language_model.layers.5.self_attn.o_proj",
|
| 135 |
"language_model.layers.11.self_attn.k_proj",
|
| 136 |
+
"28.mlp.up_proj",
|
| 137 |
+
"language_model.layers.8.mlp.up_proj",
|
| 138 |
+
"22.mlp.up_proj",
|
| 139 |
+
"27.self_attn.q_proj",
|
| 140 |
+
"output_proj",
|
| 141 |
+
"26.mlp.down_proj",
|
| 142 |
+
"language_model.layers.3.self_attn.q_proj",
|
| 143 |
+
"22.self_attn.o_proj",
|
| 144 |
+
"language_model.layers.11.mlp.up_proj",
|
| 145 |
+
"language_model.layers.0.self_attn.v_proj",
|
| 146 |
+
"language_model.layers.5.self_attn.q_proj",
|
| 147 |
+
"language_model.layers.9.self_attn.v_proj",
|
| 148 |
+
"31.mlp.down_proj",
|
| 149 |
+
"language_model.layers.5.mlp.gate_proj",
|
| 150 |
+
"per_layer_input_gate",
|
| 151 |
"language_model.layers.14.self_attn.o_proj",
|
| 152 |
+
"24.self_attn.q_proj",
|
| 153 |
+
"language_model.layers.5.self_attn.v_proj",
|
| 154 |
+
"language_model.layers.3.self_attn.v_proj",
|
| 155 |
+
"30.self_attn.q_proj",
|
| 156 |
+
"language_model.layers.7.self_attn.q_proj",
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 157 |
"language_model.layers.10.mlp.down_proj",
|
| 158 |
+
"27.mlp.down_proj",
|
| 159 |
+
"22.mlp.gate_proj",
|
| 160 |
+
"24.mlp.down_proj",
|
| 161 |
+
"language_model.layers.8.self_attn.q_proj",
|
| 162 |
+
"31.self_attn.o_proj",
|
| 163 |
"20.self_attn.q_proj",
|
| 164 |
+
"19.mlp.up_proj",
|
| 165 |
+
"19.mlp.gate_proj",
|
| 166 |
+
"16.self_attn.q_proj",
|
| 167 |
+
"21.mlp.up_proj",
|
| 168 |
"language_model.layers.10.mlp.up_proj",
|
| 169 |
+
"language_model.layers.6.self_attn.o_proj",
|
| 170 |
+
"19.self_attn.q_proj",
|
| 171 |
+
"24.mlp.up_proj",
|
| 172 |
+
"22.mlp.down_proj",
|
| 173 |
+
"28.mlp.gate_proj",
|
| 174 |
+
"per_layer_projection",
|
| 175 |
+
"31.mlp.gate_proj",
|
| 176 |
+
"language_model.layers.11.self_attn.q_proj",
|
| 177 |
+
"30.mlp.up_proj",
|
| 178 |
+
"19.mlp.down_proj",
|
| 179 |
+
"language_model.layers.3.self_attn.o_proj",
|
| 180 |
+
"language_model.layers.4.mlp.up_proj",
|
| 181 |
+
"language_model.layers.8.self_attn.k_proj",
|
| 182 |
+
"34.self_attn.q_proj",
|
| 183 |
+
"language_model.layers.9.mlp.down_proj",
|
| 184 |
+
"language_model.layers.1.self_attn.v_proj",
|
| 185 |
+
"20.mlp.up_proj",
|
| 186 |
+
"language_model.layers.7.mlp.up_proj",
|
| 187 |
+
"language_model.layers.2.self_attn.v_proj",
|
| 188 |
+
"language_model.layers.12.self_attn.o_proj",
|
| 189 |
+
"33.self_attn.q_proj",
|
| 190 |
+
"21.self_attn.q_proj",
|
| 191 |
+
"language_model.layers.15.self_attn.q_proj",
|
| 192 |
+
"26.mlp.up_proj",
|
| 193 |
+
"language_model.layers.6.mlp.gate_proj",
|
| 194 |
+
"language_model.layers.3.self_attn.k_proj",
|
| 195 |
+
"language_model.layers.4.self_attn.k_proj",
|
| 196 |
+
"28.self_attn.q_proj",
|
| 197 |
"language_model.layers.2.mlp.up_proj",
|
| 198 |
+
"language_model.layers.0.mlp.gate_proj",
|
| 199 |
+
"language_model.layers.0.mlp.down_proj",
|
| 200 |
+
"language_model.layers.2.mlp.down_proj",
|
| 201 |
+
"17.mlp.gate_proj",
|
| 202 |
+
"language_model.layers.0.mlp.up_proj",
|
| 203 |
+
"language_model.layers.3.mlp.gate_proj",
|
| 204 |
+
"27.mlp.up_proj",
|
| 205 |
"language_model.layers.13.self_attn.v_proj",
|
|
|
|
|
|
|
| 206 |
"28.mlp.down_proj",
|
| 207 |
+
"33.self_attn.o_proj",
|
| 208 |
+
"language_model.layers.11.mlp.gate_proj",
|
| 209 |
+
"24.mlp.gate_proj",
|
| 210 |
+
"language_model.layers.7.mlp.gate_proj",
|
| 211 |
+
"32.mlp.gate_proj",
|
| 212 |
+
"language_model.layers.12.self_attn.k_proj",
|
| 213 |
+
"33.mlp.gate_proj",
|
| 214 |
+
"language_model.layers.12.mlp.down_proj",
|
| 215 |
+
"language_model.layers.8.mlp.down_proj",
|
| 216 |
+
"language_model.layers.6.self_attn.q_proj",
|
|
|
|
|
|
|
|
|
|
|
|
|
| 217 |
"25.mlp.up_proj",
|
| 218 |
+
"language_model.layers.13.mlp.down_proj",
|
| 219 |
+
"32.mlp.down_proj",
|
| 220 |
+
"17.mlp.down_proj",
|
| 221 |
+
"33.mlp.down_proj",
|
| 222 |
+
"language_model.layers.11.mlp.down_proj",
|
| 223 |
+
"27.mlp.gate_proj",
|
| 224 |
+
"language_model.layers.6.mlp.down_proj",
|
| 225 |
+
"language_model.layers.2.self_attn.o_proj",
|
| 226 |
+
"20.self_attn.o_proj",
|
| 227 |
+
"20.mlp.down_proj",
|
| 228 |
+
"language_model.layers.13.self_attn.q_proj",
|
| 229 |
+
"32.mlp.up_proj",
|
| 230 |
+
"language_model.layers.12.self_attn.q_proj",
|
| 231 |
"17.self_attn.q_proj",
|
| 232 |
+
"language_model.layers.14.mlp.gate_proj",
|
| 233 |
+
"34.mlp.up_proj",
|
| 234 |
+
"26.self_attn.q_proj",
|
| 235 |
+
"embedding_projection",
|
| 236 |
"language_model.layers.1.self_attn.k_proj",
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 237 |
"language_model.layers.14.self_attn.q_proj",
|
| 238 |
+
"language_model.layers.12.mlp.gate_proj",
|
|
|
|
| 239 |
"16.mlp.down_proj",
|
| 240 |
+
"language_model.layers.13.mlp.up_proj",
|
| 241 |
+
"17.self_attn.o_proj",
|
| 242 |
+
"18.self_attn.q_proj",
|
| 243 |
+
"29.mlp.gate_proj",
|
| 244 |
+
"18.mlp.up_proj",
|
| 245 |
+
"language_model.layers.14.mlp.down_proj",
|
| 246 |
+
"language_model.layers.8.mlp.gate_proj",
|
| 247 |
+
"language_model.layers.1.self_attn.q_proj",
|
| 248 |
+
"language_model.layers.4.mlp.down_proj",
|
| 249 |
+
"language_model.layers.13.mlp.gate_proj"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 250 |
],
|
| 251 |
"target_parameters": null,
|
| 252 |
"task_type": "CAUSAL_LM",
|
adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1686620368
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ad4b6d50dd494308021afecf5d4603b315744b9342ea2d86e5d2c3d27d02710e
|
| 3 |
size 1686620368
|
runs/May09_08-02-47_g21-3e350fb3-8deb-3e4a-ae1a-6dbf2e96917c/events.out.tfevents.1778313767.g21-3e350fb3-8deb-3e4a-ae1a-6dbf2e96917c.17051.0
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:96e33b4e3c35639ce74803bd62343389bd6c67abcb705d15df95ca0f95e9596a
|
| 3 |
+
size 15248
|
runs/May09_08-28-20_g21-3e350fb3-8deb-3e4a-ae1a-6dbf2e96917c/events.out.tfevents.1778315300.g21-3e350fb3-8deb-3e4a-ae1a-6dbf2e96917c.17051.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1f24874a4c404f2c0fbd3ffeabcdf4b33741c9e56d5442b22c964ee4df9c338f
|
| 3 |
+
size 10989
|
runs/May09_08-34-00_g21-3e350fb3-8deb-3e4a-ae1a-6dbf2e96917c/events.out.tfevents.1778315640.g21-3e350fb3-8deb-3e4a-ae1a-6dbf2e96917c.17051.2
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:137441f68e5f342054317213ecb6f67e9f11e43c7a9ebfa63f7f36455ab2bf35
|
| 3 |
+
size 10225
|
runs/May09_08-37-46_g21-3e350fb3-8deb-3e4a-ae1a-6dbf2e96917c/events.out.tfevents.1778315866.g21-3e350fb3-8deb-3e4a-ae1a-6dbf2e96917c.17051.3
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d1492affa50289bb570e5552698c40a3120e2acf4e56ab02ad0c0e7ba369574e
|
| 3 |
+
size 8280
|
runs/May09_08-43-36_g21-3e350fb3-8deb-3e4a-ae1a-6dbf2e96917c/events.out.tfevents.1778316216.g21-3e350fb3-8deb-3e4a-ae1a-6dbf2e96917c.19296.0
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:142fb96edde91718f996b4368723b1929daafae494e417d699df607c2645567d
|
| 3 |
+
size 19832
|