Training in progress, step 1
Browse files
adapter_config.json
ADDED
|
@@ -0,0 +1,195 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"alpha_pattern": {},
|
| 3 |
+
"auto_mapping": null,
|
| 4 |
+
"base_model_name_or_path": "kakaobrain/kogpt",
|
| 5 |
+
"bias": "none",
|
| 6 |
+
"fan_in_fan_out": false,
|
| 7 |
+
"inference_mode": true,
|
| 8 |
+
"init_lora_weights": true,
|
| 9 |
+
"layer_replication": null,
|
| 10 |
+
"layers_pattern": null,
|
| 11 |
+
"layers_to_transform": null,
|
| 12 |
+
"loftq_config": {},
|
| 13 |
+
"lora_alpha": 16,
|
| 14 |
+
"lora_dropout": 0.05,
|
| 15 |
+
"megatron_config": null,
|
| 16 |
+
"megatron_core": "megatron.core",
|
| 17 |
+
"modules_to_save": null,
|
| 18 |
+
"peft_type": "LORA",
|
| 19 |
+
"r": 8,
|
| 20 |
+
"rank_pattern": {},
|
| 21 |
+
"revision": null,
|
| 22 |
+
"target_modules": [
|
| 23 |
+
"transformer.h.17.mlp.fc_out",
|
| 24 |
+
"transformer.h.8.mlp.fc_in",
|
| 25 |
+
"transformer.h.11.attn.k_proj",
|
| 26 |
+
"transformer.h.17.attn.q_proj",
|
| 27 |
+
"transformer.h.1.mlp.fc_in",
|
| 28 |
+
"transformer.h.1.mlp.fc_out",
|
| 29 |
+
"transformer.h.16.mlp.fc_out",
|
| 30 |
+
"transformer.h.5.attn.out_proj",
|
| 31 |
+
"transformer.h.15.attn.k_proj",
|
| 32 |
+
"transformer.h.0.attn.q_proj",
|
| 33 |
+
"transformer.h.5.attn.k_proj",
|
| 34 |
+
"transformer.h.14.attn.out_proj",
|
| 35 |
+
"transformer.h.10.attn.q_proj",
|
| 36 |
+
"transformer.h.26.attn.out_proj",
|
| 37 |
+
"transformer.h.6.attn.v_proj",
|
| 38 |
+
"transformer.h.8.attn.v_proj",
|
| 39 |
+
"transformer.h.7.attn.k_proj",
|
| 40 |
+
"transformer.h.22.attn.q_proj",
|
| 41 |
+
"transformer.h.13.attn.out_proj",
|
| 42 |
+
"transformer.h.7.attn.q_proj",
|
| 43 |
+
"transformer.h.26.attn.k_proj",
|
| 44 |
+
"transformer.h.1.attn.out_proj",
|
| 45 |
+
"transformer.h.15.mlp.fc_out",
|
| 46 |
+
"transformer.h.3.attn.out_proj",
|
| 47 |
+
"transformer.h.24.mlp.fc_in",
|
| 48 |
+
"transformer.h.11.attn.q_proj",
|
| 49 |
+
"transformer.h.21.attn.q_proj",
|
| 50 |
+
"transformer.h.6.attn.k_proj",
|
| 51 |
+
"transformer.h.13.attn.v_proj",
|
| 52 |
+
"transformer.h.17.attn.out_proj",
|
| 53 |
+
"transformer.h.24.attn.k_proj",
|
| 54 |
+
"transformer.h.25.attn.q_proj",
|
| 55 |
+
"transformer.h.18.mlp.fc_out",
|
| 56 |
+
"transformer.h.11.attn.v_proj",
|
| 57 |
+
"transformer.h.8.attn.out_proj",
|
| 58 |
+
"transformer.h.2.attn.k_proj",
|
| 59 |
+
"transformer.h.10.attn.out_proj",
|
| 60 |
+
"transformer.h.15.attn.out_proj",
|
| 61 |
+
"transformer.h.27.attn.k_proj",
|
| 62 |
+
"transformer.h.5.mlp.fc_out",
|
| 63 |
+
"transformer.h.0.mlp.fc_out",
|
| 64 |
+
"transformer.h.21.mlp.fc_in",
|
| 65 |
+
"transformer.h.3.mlp.fc_out",
|
| 66 |
+
"transformer.h.18.attn.k_proj",
|
| 67 |
+
"transformer.h.14.mlp.fc_in",
|
| 68 |
+
"transformer.h.4.attn.k_proj",
|
| 69 |
+
"transformer.h.23.attn.v_proj",
|
| 70 |
+
"transformer.h.9.attn.q_proj",
|
| 71 |
+
"transformer.h.19.attn.out_proj",
|
| 72 |
+
"transformer.h.5.attn.v_proj",
|
| 73 |
+
"transformer.h.25.attn.v_proj",
|
| 74 |
+
"transformer.h.15.attn.q_proj",
|
| 75 |
+
"transformer.h.2.attn.out_proj",
|
| 76 |
+
"transformer.h.23.mlp.fc_in",
|
| 77 |
+
"transformer.h.3.mlp.fc_in",
|
| 78 |
+
"transformer.h.16.attn.v_proj",
|
| 79 |
+
"transformer.h.3.attn.v_proj",
|
| 80 |
+
"transformer.h.27.mlp.fc_out",
|
| 81 |
+
"transformer.h.14.attn.v_proj",
|
| 82 |
+
"transformer.h.8.attn.k_proj",
|
| 83 |
+
"transformer.h.26.attn.v_proj",
|
| 84 |
+
"transformer.h.25.mlp.fc_out",
|
| 85 |
+
"transformer.h.5.attn.q_proj",
|
| 86 |
+
"transformer.h.16.mlp.fc_in",
|
| 87 |
+
"transformer.h.18.mlp.fc_in",
|
| 88 |
+
"transformer.h.14.attn.q_proj",
|
| 89 |
+
"transformer.h.5.mlp.fc_in",
|
| 90 |
+
"transformer.h.8.attn.q_proj",
|
| 91 |
+
"transformer.h.9.attn.k_proj",
|
| 92 |
+
"transformer.h.27.attn.v_proj",
|
| 93 |
+
"transformer.h.18.attn.v_proj",
|
| 94 |
+
"transformer.h.24.attn.v_proj",
|
| 95 |
+
"transformer.h.0.mlp.fc_in",
|
| 96 |
+
"transformer.h.22.attn.v_proj",
|
| 97 |
+
"transformer.h.22.mlp.fc_out",
|
| 98 |
+
"transformer.h.20.attn.out_proj",
|
| 99 |
+
"transformer.h.7.attn.v_proj",
|
| 100 |
+
"transformer.h.16.attn.q_proj",
|
| 101 |
+
"transformer.h.3.attn.q_proj",
|
| 102 |
+
"transformer.h.3.attn.k_proj",
|
| 103 |
+
"transformer.h.25.mlp.fc_in",
|
| 104 |
+
"transformer.h.21.attn.k_proj",
|
| 105 |
+
"transformer.h.9.mlp.fc_out",
|
| 106 |
+
"transformer.h.22.attn.out_proj",
|
| 107 |
+
"transformer.h.1.attn.q_proj",
|
| 108 |
+
"transformer.h.2.mlp.fc_in",
|
| 109 |
+
"transformer.h.20.attn.q_proj",
|
| 110 |
+
"transformer.h.21.attn.v_proj",
|
| 111 |
+
"transformer.h.6.mlp.fc_in",
|
| 112 |
+
"transformer.h.8.mlp.fc_out",
|
| 113 |
+
"transformer.h.26.attn.q_proj",
|
| 114 |
+
"transformer.h.17.mlp.fc_in",
|
| 115 |
+
"transformer.h.0.attn.out_proj",
|
| 116 |
+
"transformer.h.23.attn.out_proj",
|
| 117 |
+
"transformer.h.10.attn.v_proj",
|
| 118 |
+
"transformer.h.13.mlp.fc_out",
|
| 119 |
+
"transformer.h.25.attn.out_proj",
|
| 120 |
+
"transformer.h.7.mlp.fc_in",
|
| 121 |
+
"transformer.h.10.mlp.fc_in",
|
| 122 |
+
"transformer.h.26.mlp.fc_in",
|
| 123 |
+
"transformer.h.21.attn.out_proj",
|
| 124 |
+
"transformer.h.13.attn.k_proj",
|
| 125 |
+
"transformer.h.14.attn.k_proj",
|
| 126 |
+
"transformer.h.11.mlp.fc_out",
|
| 127 |
+
"transformer.h.16.attn.out_proj",
|
| 128 |
+
"transformer.h.6.attn.q_proj",
|
| 129 |
+
"transformer.h.14.mlp.fc_out",
|
| 130 |
+
"transformer.h.20.mlp.fc_out",
|
| 131 |
+
"transformer.h.22.attn.k_proj",
|
| 132 |
+
"transformer.h.20.attn.v_proj",
|
| 133 |
+
"transformer.h.19.mlp.fc_out",
|
| 134 |
+
"transformer.h.22.mlp.fc_in",
|
| 135 |
+
"transformer.h.4.attn.out_proj",
|
| 136 |
+
"transformer.h.12.mlp.fc_in",
|
| 137 |
+
"transformer.h.19.attn.v_proj",
|
| 138 |
+
"transformer.h.15.attn.v_proj",
|
| 139 |
+
"transformer.h.7.attn.out_proj",
|
| 140 |
+
"transformer.h.4.attn.q_proj",
|
| 141 |
+
"transformer.h.12.attn.k_proj",
|
| 142 |
+
"transformer.h.24.attn.q_proj",
|
| 143 |
+
"transformer.h.2.attn.v_proj",
|
| 144 |
+
"transformer.h.11.mlp.fc_in",
|
| 145 |
+
"transformer.h.10.mlp.fc_out",
|
| 146 |
+
"transformer.h.18.attn.out_proj",
|
| 147 |
+
"transformer.h.2.mlp.fc_out",
|
| 148 |
+
"transformer.h.20.attn.k_proj",
|
| 149 |
+
"transformer.h.26.mlp.fc_out",
|
| 150 |
+
"transformer.h.0.attn.k_proj",
|
| 151 |
+
"transformer.h.12.attn.v_proj",
|
| 152 |
+
"transformer.h.4.attn.v_proj",
|
| 153 |
+
"transformer.h.12.mlp.fc_out",
|
| 154 |
+
"transformer.h.17.attn.k_proj",
|
| 155 |
+
"transformer.h.18.attn.q_proj",
|
| 156 |
+
"transformer.h.27.attn.out_proj",
|
| 157 |
+
"transformer.h.0.attn.v_proj",
|
| 158 |
+
"transformer.h.4.mlp.fc_out",
|
| 159 |
+
"transformer.h.2.attn.q_proj",
|
| 160 |
+
"transformer.h.23.attn.k_proj",
|
| 161 |
+
"transformer.h.25.attn.k_proj",
|
| 162 |
+
"transformer.h.1.attn.k_proj",
|
| 163 |
+
"transformer.h.13.mlp.fc_in",
|
| 164 |
+
"transformer.h.27.attn.q_proj",
|
| 165 |
+
"transformer.h.6.mlp.fc_out",
|
| 166 |
+
"transformer.h.20.mlp.fc_in",
|
| 167 |
+
"transformer.h.9.mlp.fc_in",
|
| 168 |
+
"transformer.h.19.mlp.fc_in",
|
| 169 |
+
"transformer.h.24.attn.out_proj",
|
| 170 |
+
"transformer.h.12.attn.q_proj",
|
| 171 |
+
"transformer.h.23.mlp.fc_out",
|
| 172 |
+
"transformer.h.6.attn.out_proj",
|
| 173 |
+
"transformer.h.13.attn.q_proj",
|
| 174 |
+
"transformer.h.12.attn.out_proj",
|
| 175 |
+
"transformer.h.19.attn.q_proj",
|
| 176 |
+
"transformer.h.15.mlp.fc_in",
|
| 177 |
+
"transformer.h.23.attn.q_proj",
|
| 178 |
+
"transformer.h.9.attn.out_proj",
|
| 179 |
+
"transformer.h.19.attn.k_proj",
|
| 180 |
+
"transformer.h.17.attn.v_proj",
|
| 181 |
+
"transformer.h.21.mlp.fc_out",
|
| 182 |
+
"transformer.h.7.mlp.fc_out",
|
| 183 |
+
"transformer.h.11.attn.out_proj",
|
| 184 |
+
"transformer.h.10.attn.k_proj",
|
| 185 |
+
"transformer.h.1.attn.v_proj",
|
| 186 |
+
"transformer.h.4.mlp.fc_in",
|
| 187 |
+
"transformer.h.9.attn.v_proj",
|
| 188 |
+
"transformer.h.24.mlp.fc_out",
|
| 189 |
+
"transformer.h.27.mlp.fc_in",
|
| 190 |
+
"transformer.h.16.attn.k_proj"
|
| 191 |
+
],
|
| 192 |
+
"task_type": "CAUSAL_LM",
|
| 193 |
+
"use_dora": false,
|
| 194 |
+
"use_rslora": false
|
| 195 |
+
}
|
adapter_model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1497df2f601ce388150a1ff919fec32e69dd92742e093e45f5caee13df5b3e95
|
| 3 |
+
size 66103768
|
runs/Aug18_13-44-44_oem-WS-C621E-SAGE-Series/events.out.tfevents.1723956285.oem-WS-C621E-SAGE-Series.774859.0
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:122f70e4a4fdee24e9cee0f638b87221159ab3a3a1a0363a8e304bcf65a1fd68
|
| 3 |
+
size 6273
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:dbebcea5791d49803334977bda6bea23488a1464dcef5fbf6ac69cbafd7020c7
|
| 3 |
+
size 5240
|