sirr
commited on
Commit
·
a5be891
1
Parent(s):
858f6c0
Training in progress, step 200
Browse files- adapter_config.json +211 -0
- adapter_model.bin +3 -0
- training_args.bin +3 -0
adapter_config.json
ADDED
|
@@ -0,0 +1,211 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"auto_mapping": null,
|
| 3 |
+
"base_model_name_or_path": "gpt2-xl",
|
| 4 |
+
"bias": "none",
|
| 5 |
+
"fan_in_fan_out": false,
|
| 6 |
+
"inference_mode": true,
|
| 7 |
+
"init_lora_weights": true,
|
| 8 |
+
"layers_pattern": null,
|
| 9 |
+
"layers_to_transform": null,
|
| 10 |
+
"lora_alpha": 16,
|
| 11 |
+
"lora_dropout": 0.1,
|
| 12 |
+
"modules_to_save": null,
|
| 13 |
+
"peft_type": "LORA",
|
| 14 |
+
"r": 8,
|
| 15 |
+
"revision": null,
|
| 16 |
+
"target_modules": [
|
| 17 |
+
"transformer.h.0.attn.c_attn",
|
| 18 |
+
"transformer.h.1.attn.c_attn",
|
| 19 |
+
"transformer.h.2.attn.c_attn",
|
| 20 |
+
"transformer.h.3.attn.c_attn",
|
| 21 |
+
"transformer.h.4.attn.c_attn",
|
| 22 |
+
"transformer.h.5.attn.c_attn",
|
| 23 |
+
"transformer.h.6.attn.c_attn",
|
| 24 |
+
"transformer.h.7.attn.c_attn",
|
| 25 |
+
"transformer.h.8.attn.c_attn",
|
| 26 |
+
"transformer.h.9.attn.c_attn",
|
| 27 |
+
"transformer.h.10.attn.c_attn",
|
| 28 |
+
"transformer.h.11.attn.c_attn",
|
| 29 |
+
"transformer.h.12.attn.c_attn",
|
| 30 |
+
"transformer.h.13.attn.c_attn",
|
| 31 |
+
"transformer.h.14.attn.c_attn",
|
| 32 |
+
"transformer.h.15.attn.c_attn",
|
| 33 |
+
"transformer.h.16.attn.c_attn",
|
| 34 |
+
"transformer.h.17.attn.c_attn",
|
| 35 |
+
"transformer.h.18.attn.c_attn",
|
| 36 |
+
"transformer.h.19.attn.c_attn",
|
| 37 |
+
"transformer.h.20.attn.c_attn",
|
| 38 |
+
"transformer.h.21.attn.c_attn",
|
| 39 |
+
"transformer.h.22.attn.c_attn",
|
| 40 |
+
"transformer.h.23.attn.c_attn",
|
| 41 |
+
"transformer.h.24.attn.c_attn",
|
| 42 |
+
"transformer.h.25.attn.c_attn",
|
| 43 |
+
"transformer.h.26.attn.c_attn",
|
| 44 |
+
"transformer.h.27.attn.c_attn",
|
| 45 |
+
"transformer.h.28.attn.c_attn",
|
| 46 |
+
"transformer.h.29.attn.c_attn",
|
| 47 |
+
"transformer.h.30.attn.c_attn",
|
| 48 |
+
"transformer.h.31.attn.c_attn",
|
| 49 |
+
"transformer.h.32.attn.c_attn",
|
| 50 |
+
"transformer.h.33.attn.c_attn",
|
| 51 |
+
"transformer.h.34.attn.c_attn",
|
| 52 |
+
"transformer.h.35.attn.c_attn",
|
| 53 |
+
"transformer.h.36.attn.c_attn",
|
| 54 |
+
"transformer.h.37.attn.c_attn",
|
| 55 |
+
"transformer.h.38.attn.c_attn",
|
| 56 |
+
"transformer.h.39.attn.c_attn",
|
| 57 |
+
"transformer.h.40.attn.c_attn",
|
| 58 |
+
"transformer.h.41.attn.c_attn",
|
| 59 |
+
"transformer.h.42.attn.c_attn",
|
| 60 |
+
"transformer.h.43.attn.c_attn",
|
| 61 |
+
"transformer.h.44.attn.c_attn",
|
| 62 |
+
"transformer.h.45.attn.c_attn",
|
| 63 |
+
"transformer.h.46.attn.c_attn",
|
| 64 |
+
"transformer.h.47.attn.c_attn",
|
| 65 |
+
"transformer.h.0.attn.c_proj",
|
| 66 |
+
"transformer.h.1.attn.c_proj",
|
| 67 |
+
"transformer.h.2.attn.c_proj",
|
| 68 |
+
"transformer.h.3.attn.c_proj",
|
| 69 |
+
"transformer.h.4.attn.c_proj",
|
| 70 |
+
"transformer.h.5.attn.c_proj",
|
| 71 |
+
"transformer.h.6.attn.c_proj",
|
| 72 |
+
"transformer.h.7.attn.c_proj",
|
| 73 |
+
"transformer.h.8.attn.c_proj",
|
| 74 |
+
"transformer.h.9.attn.c_proj",
|
| 75 |
+
"transformer.h.10.attn.c_proj",
|
| 76 |
+
"transformer.h.11.attn.c_proj",
|
| 77 |
+
"transformer.h.12.attn.c_proj",
|
| 78 |
+
"transformer.h.13.attn.c_proj",
|
| 79 |
+
"transformer.h.14.attn.c_proj",
|
| 80 |
+
"transformer.h.15.attn.c_proj",
|
| 81 |
+
"transformer.h.16.attn.c_proj",
|
| 82 |
+
"transformer.h.17.attn.c_proj",
|
| 83 |
+
"transformer.h.18.attn.c_proj",
|
| 84 |
+
"transformer.h.19.attn.c_proj",
|
| 85 |
+
"transformer.h.20.attn.c_proj",
|
| 86 |
+
"transformer.h.21.attn.c_proj",
|
| 87 |
+
"transformer.h.22.attn.c_proj",
|
| 88 |
+
"transformer.h.23.attn.c_proj",
|
| 89 |
+
"transformer.h.24.attn.c_proj",
|
| 90 |
+
"transformer.h.25.attn.c_proj",
|
| 91 |
+
"transformer.h.26.attn.c_proj",
|
| 92 |
+
"transformer.h.27.attn.c_proj",
|
| 93 |
+
"transformer.h.28.attn.c_proj",
|
| 94 |
+
"transformer.h.29.attn.c_proj",
|
| 95 |
+
"transformer.h.30.attn.c_proj",
|
| 96 |
+
"transformer.h.31.attn.c_proj",
|
| 97 |
+
"transformer.h.32.attn.c_proj",
|
| 98 |
+
"transformer.h.33.attn.c_proj",
|
| 99 |
+
"transformer.h.34.attn.c_proj",
|
| 100 |
+
"transformer.h.35.attn.c_proj",
|
| 101 |
+
"transformer.h.36.attn.c_proj",
|
| 102 |
+
"transformer.h.37.attn.c_proj",
|
| 103 |
+
"transformer.h.38.attn.c_proj",
|
| 104 |
+
"transformer.h.39.attn.c_proj",
|
| 105 |
+
"transformer.h.40.attn.c_proj",
|
| 106 |
+
"transformer.h.41.attn.c_proj",
|
| 107 |
+
"transformer.h.42.attn.c_proj",
|
| 108 |
+
"transformer.h.43.attn.c_proj",
|
| 109 |
+
"transformer.h.44.attn.c_proj",
|
| 110 |
+
"transformer.h.45.attn.c_proj",
|
| 111 |
+
"transformer.h.46.attn.c_proj",
|
| 112 |
+
"transformer.h.47.attn.c_proj",
|
| 113 |
+
"transformer.h.0.mlp.c_fc",
|
| 114 |
+
"transformer.h.1.mlp.c_fc",
|
| 115 |
+
"transformer.h.2.mlp.c_fc",
|
| 116 |
+
"transformer.h.3.mlp.c_fc",
|
| 117 |
+
"transformer.h.4.mlp.c_fc",
|
| 118 |
+
"transformer.h.5.mlp.c_fc",
|
| 119 |
+
"transformer.h.6.mlp.c_fc",
|
| 120 |
+
"transformer.h.7.mlp.c_fc",
|
| 121 |
+
"transformer.h.8.mlp.c_fc",
|
| 122 |
+
"transformer.h.9.mlp.c_fc",
|
| 123 |
+
"transformer.h.10.mlp.c_fc",
|
| 124 |
+
"transformer.h.11.mlp.c_fc",
|
| 125 |
+
"transformer.h.12.mlp.c_fc",
|
| 126 |
+
"transformer.h.13.mlp.c_fc",
|
| 127 |
+
"transformer.h.14.mlp.c_fc",
|
| 128 |
+
"transformer.h.15.mlp.c_fc",
|
| 129 |
+
"transformer.h.16.mlp.c_fc",
|
| 130 |
+
"transformer.h.17.mlp.c_fc",
|
| 131 |
+
"transformer.h.18.mlp.c_fc",
|
| 132 |
+
"transformer.h.19.mlp.c_fc",
|
| 133 |
+
"transformer.h.20.mlp.c_fc",
|
| 134 |
+
"transformer.h.21.mlp.c_fc",
|
| 135 |
+
"transformer.h.22.mlp.c_fc",
|
| 136 |
+
"transformer.h.23.mlp.c_fc",
|
| 137 |
+
"transformer.h.24.mlp.c_fc",
|
| 138 |
+
"transformer.h.25.mlp.c_fc",
|
| 139 |
+
"transformer.h.26.mlp.c_fc",
|
| 140 |
+
"transformer.h.27.mlp.c_fc",
|
| 141 |
+
"transformer.h.28.mlp.c_fc",
|
| 142 |
+
"transformer.h.29.mlp.c_fc",
|
| 143 |
+
"transformer.h.30.mlp.c_fc",
|
| 144 |
+
"transformer.h.31.mlp.c_fc",
|
| 145 |
+
"transformer.h.32.mlp.c_fc",
|
| 146 |
+
"transformer.h.33.mlp.c_fc",
|
| 147 |
+
"transformer.h.34.mlp.c_fc",
|
| 148 |
+
"transformer.h.35.mlp.c_fc",
|
| 149 |
+
"transformer.h.36.mlp.c_fc",
|
| 150 |
+
"transformer.h.37.mlp.c_fc",
|
| 151 |
+
"transformer.h.38.mlp.c_fc",
|
| 152 |
+
"transformer.h.39.mlp.c_fc",
|
| 153 |
+
"transformer.h.40.mlp.c_fc",
|
| 154 |
+
"transformer.h.41.mlp.c_fc",
|
| 155 |
+
"transformer.h.42.mlp.c_fc",
|
| 156 |
+
"transformer.h.43.mlp.c_fc",
|
| 157 |
+
"transformer.h.44.mlp.c_fc",
|
| 158 |
+
"transformer.h.45.mlp.c_fc",
|
| 159 |
+
"transformer.h.46.mlp.c_fc",
|
| 160 |
+
"transformer.h.47.mlp.c_fc",
|
| 161 |
+
"transformer.h.0.mlp.c_proj",
|
| 162 |
+
"transformer.h.1.mlp.c_proj",
|
| 163 |
+
"transformer.h.2.mlp.c_proj",
|
| 164 |
+
"transformer.h.3.mlp.c_proj",
|
| 165 |
+
"transformer.h.4.mlp.c_proj",
|
| 166 |
+
"transformer.h.5.mlp.c_proj",
|
| 167 |
+
"transformer.h.6.mlp.c_proj",
|
| 168 |
+
"transformer.h.7.mlp.c_proj",
|
| 169 |
+
"transformer.h.8.mlp.c_proj",
|
| 170 |
+
"transformer.h.9.mlp.c_proj",
|
| 171 |
+
"transformer.h.10.mlp.c_proj",
|
| 172 |
+
"transformer.h.11.mlp.c_proj",
|
| 173 |
+
"transformer.h.12.mlp.c_proj",
|
| 174 |
+
"transformer.h.13.mlp.c_proj",
|
| 175 |
+
"transformer.h.14.mlp.c_proj",
|
| 176 |
+
"transformer.h.15.mlp.c_proj",
|
| 177 |
+
"transformer.h.16.mlp.c_proj",
|
| 178 |
+
"transformer.h.17.mlp.c_proj",
|
| 179 |
+
"transformer.h.18.mlp.c_proj",
|
| 180 |
+
"transformer.h.19.mlp.c_proj",
|
| 181 |
+
"transformer.h.20.mlp.c_proj",
|
| 182 |
+
"transformer.h.21.mlp.c_proj",
|
| 183 |
+
"transformer.h.22.mlp.c_proj",
|
| 184 |
+
"transformer.h.23.mlp.c_proj",
|
| 185 |
+
"transformer.h.24.mlp.c_proj",
|
| 186 |
+
"transformer.h.25.mlp.c_proj",
|
| 187 |
+
"transformer.h.26.mlp.c_proj",
|
| 188 |
+
"transformer.h.27.mlp.c_proj",
|
| 189 |
+
"transformer.h.28.mlp.c_proj",
|
| 190 |
+
"transformer.h.29.mlp.c_proj",
|
| 191 |
+
"transformer.h.30.mlp.c_proj",
|
| 192 |
+
"transformer.h.31.mlp.c_proj",
|
| 193 |
+
"transformer.h.32.mlp.c_proj",
|
| 194 |
+
"transformer.h.33.mlp.c_proj",
|
| 195 |
+
"transformer.h.34.mlp.c_proj",
|
| 196 |
+
"transformer.h.35.mlp.c_proj",
|
| 197 |
+
"transformer.h.36.mlp.c_proj",
|
| 198 |
+
"transformer.h.37.mlp.c_proj",
|
| 199 |
+
"transformer.h.38.mlp.c_proj",
|
| 200 |
+
"transformer.h.39.mlp.c_proj",
|
| 201 |
+
"transformer.h.40.mlp.c_proj",
|
| 202 |
+
"transformer.h.41.mlp.c_proj",
|
| 203 |
+
"transformer.h.42.mlp.c_proj",
|
| 204 |
+
"transformer.h.43.mlp.c_proj",
|
| 205 |
+
"transformer.h.44.mlp.c_proj",
|
| 206 |
+
"transformer.h.45.mlp.c_proj",
|
| 207 |
+
"transformer.h.46.mlp.c_proj",
|
| 208 |
+
"transformer.h.47.mlp.c_proj"
|
| 209 |
+
],
|
| 210 |
+
"task_type": "CAUSAL_LM"
|
| 211 |
+
}
|
adapter_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bce46fcb81bcb2fbf81bd68a05646efaa596565105c53098fd5b96d735804724
|
| 3 |
+
size 39457933
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e4726cfdfe4bae82880d219428dbd216d4ffe556f94a2115f6c4f72ba8eb8170
|
| 3 |
+
size 4027
|