File size: 100,812 Bytes
68ca2f2 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 |
/opt/conda/envs/py310/bin/python -m mlc_llm gen_config /models/Qwen2-7B-Instruct --quantization q4f16_1 --conv-template chatml --output /models/mlc-delivery/hf/mlc-ai/Qwen2-7B-Instruct-q4f16_1-MLC [2024-06-06 23:42:10] INFO auto_config.py:116: [92mFound[0m model configuration: /models/Qwen2-7B-Instruct/config.json [2024-06-06 23:42:10] INFO auto_config.py:154: [92mFound[0m model type: [1mqwen2[0m. Use `--model-type` to override. [2024-06-06 23:42:10] INFO qwen2_model.py:49: [1mcontext_window_size[0m not found in config.json. Falling back to [1mmax_position_embeddings[0m (32768) [2024-06-06 23:42:10] INFO qwen2_model.py:66: [1mprefill_chunk_size[0m defaults to 2048 [2024-06-06 23:42:10] INFO config.py:107: Overriding [1mmax_batch_size[0m from 1 to 80 [2024-06-06 23:42:10] INFO gen_config.py:143: [generation_config.json] Setting [1mbos_token_id[0m: 151643 [2024-06-06 23:42:10] INFO gen_config.py:143: [generation_config.json] Setting [1mpad_token_id[0m: 151643 [2024-06-06 23:42:10] INFO gen_config.py:143: [generation_config.json] Setting [1meos_token_id[0m: [151645, 151643] [2024-06-06 23:42:10] INFO gen_config.py:143: [generation_config.json] Setting [1mrepetition_penalty[0m: 1.05 [2024-06-06 23:42:10] INFO gen_config.py:143: [generation_config.json] Setting [1mtemperature[0m: 0.7 [2024-06-06 23:42:10] INFO gen_config.py:143: [generation_config.json] Setting [1mtop_p[0m: 0.8 [2024-06-06 23:42:10] INFO gen_config.py:157: [91mNot found[0m tokenizer config: /models/Qwen2-7B-Instruct/tokenizer.model [2024-06-06 23:42:10] INFO gen_config.py:155: [92mFound[0m tokenizer config: /models/Qwen2-7B-Instruct/tokenizer.json. Copying to [1m/models/mlc-delivery/hf/mlc-ai/Qwen2-7B-Instruct-q4f16_1-MLC/tokenizer.json[0m [2024-06-06 23:42:10] INFO gen_config.py:155: [92mFound[0m tokenizer config: /models/Qwen2-7B-Instruct/vocab.json. Copying to [1m/models/mlc-delivery/hf/mlc-ai/Qwen2-7B-Instruct-q4f16_1-MLC/vocab.json[0m [2024-06-06 23:42:10] INFO gen_config.py:155: [92mFound[0m tokenizer config: /models/Qwen2-7B-Instruct/merges.txt. Copying to [1m/models/mlc-delivery/hf/mlc-ai/Qwen2-7B-Instruct-q4f16_1-MLC/merges.txt[0m [2024-06-06 23:42:10] INFO gen_config.py:157: [91mNot found[0m tokenizer config: /models/Qwen2-7B-Instruct/added_tokens.json [2024-06-06 23:42:10] INFO gen_config.py:155: [92mFound[0m tokenizer config: /models/Qwen2-7B-Instruct/tokenizer_config.json. Copying to [1m/models/mlc-delivery/hf/mlc-ai/Qwen2-7B-Instruct-q4f16_1-MLC/tokenizer_config.json[0m [2024-06-06 23:42:10] INFO gen_config.py:216: Detected tokenizer info: {'token_postproc_method': 'byte_level', 'prepend_space_in_encode': False, 'strip_space_in_decode': False} [2024-06-06 23:42:10] INFO gen_config.py:32: [System default] Setting [1mpresence_penalty[0m: 0.0 [2024-06-06 23:42:10] INFO gen_config.py:32: [System default] Setting [1mfrequency_penalty[0m: 0.0 [2024-06-06 23:42:10] INFO gen_config.py:223: Dumping configuration file to: [1m/models/mlc-delivery/hf/mlc-ai/Qwen2-7B-Instruct-q4f16_1-MLC/mlc-chat-config.json[0m /opt/conda/envs/py310/bin/python -m mlc_llm convert_weight /models/Qwen2-7B-Instruct --quantization q4f16_1 --output /models/mlc-delivery/hf/mlc-ai/Qwen2-7B-Instruct-q4f16_1-MLC [2024-06-06 23:42:12] INFO auto_config.py:116: [92mFound[0m model configuration: /models/Qwen2-7B-Instruct/config.json [2024-06-06 23:42:14] INFO auto_device.py:79: [92mFound[0m device: cuda:0 [2024-06-06 23:42:15] INFO auto_device.py:88: [91mNot found[0m device: rocm:0 [2024-06-06 23:42:16] INFO auto_device.py:88: [91mNot found[0m device: metal:0 [2024-06-06 23:42:18] INFO auto_device.py:79: [92mFound[0m device: vulkan:0 [2024-06-06 23:42:18] INFO auto_device.py:79: [92mFound[0m device: vulkan:1 [2024-06-06 23:42:18] INFO auto_device.py:79: [92mFound[0m device: vulkan:2 [2024-06-06 23:42:18] INFO auto_device.py:79: [92mFound[0m device: vulkan:3 [2024-06-06 23:42:20] INFO auto_device.py:88: [91mNot found[0m device: opencl:0 [2024-06-06 23:42:20] INFO auto_device.py:35: Using device: [1mcuda:0[0m [2024-06-06 23:42:20] INFO auto_weight.py:71: Finding weights in: /models/Qwen2-7B-Instruct [2024-06-06 23:42:20] INFO auto_weight.py:137: [91mNot found[0m Huggingface PyTorch [2024-06-06 23:42:20] INFO auto_weight.py:144: [92mFound[0m source weight format: huggingface-safetensor. Source configuration: /models/Qwen2-7B-Instruct/model.safetensors.index.json [2024-06-06 23:42:20] INFO auto_weight.py:107: Using source weight configuration: [1m/models/Qwen2-7B-Instruct/model.safetensors.index.json[0m. Use `--source` to override. [2024-06-06 23:42:20] INFO auto_weight.py:111: Using source weight format: [1mhuggingface-safetensor[0m. Use `--source-format` to override. [2024-06-06 23:42:20] INFO auto_config.py:154: [92mFound[0m model type: [1mqwen2[0m. Use `--model-type` to override. [2024-06-06 23:42:20] INFO qwen2_model.py:49: [1mcontext_window_size[0m not found in config.json. Falling back to [1mmax_position_embeddings[0m (32768) [2024-06-06 23:42:20] INFO qwen2_model.py:66: [1mprefill_chunk_size[0m defaults to 2048 [1mWeight conversion with arguments:[0m [1m--config[0m /models/Qwen2-7B-Instruct/config.json [1m--quantization[0m GroupQuantize(name='q4f16_1', kind='group-quant', group_size=32, quantize_dtype='int4', storage_dtype='uint32', model_dtype='float16', linear_weight_layout='NK', quantize_embedding=True, quantize_final_fc=True, num_elem_per_storage=8, num_storage_per_group=4, max_int_value=7) [1m--model-type[0m qwen2 [1m--device[0m cuda:0 [1m--source[0m /models/Qwen2-7B-Instruct/model.safetensors.index.json [1m--source-format[0m huggingface-safetensor [1m--output[0m /models/mlc-delivery/hf/mlc-ai/Qwen2-7B-Instruct-q4f16_1-MLC Start storing to cache /models/mlc-delivery/hf/mlc-ai/Qwen2-7B-Instruct-q4f16_1-MLC 0%| | 0/199 [00:00<?, ?it/s] [2024-06-06 23:42:22] INFO huggingface_loader.py:185: Loading HF parameters from: /models/Qwen2-7B-Instruct/model-00004-of-00004.safetensors 0%| | 0/199 [00:00<?, ?it/s] [2024-06-06 23:42:30] INFO group_quantization.py:217: Compiling quantize function for key: ((152064, 3584), float16, cuda, axis=1, output_transpose=False) 0%| | 0/199 [00:07<?, ?it/s] [2024-06-06 23:42:31] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mlm_head.q_weight[0m", shape: (152064, 448), dtype: uint32 0%| | 0/199 [00:08<?, ?it/s] [2024-06-06 23:42:31] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mlm_head.q_scale[0m", shape: (152064, 112), dtype: float16 0%| | 0/199 [00:09<?, ?it/s] 1%| | 1/199 [00:09<32:17, 9.78s/it] [2024-06-06 23:42:31] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.22.input_layernorm.weight[0m", shape: (3584,), dtype: float16 1%| | 1/199 [00:09<32:17, 9.78s/it] [2024-06-06 23:42:32] INFO group_quantization.py:217: Compiling quantize function for key: ((3584, 18944), float16, cuda, axis=1, output_transpose=False) 1%| | 1/199 [00:10<32:17, 9.78s/it] [2024-06-06 23:42:32] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.22.mlp.down_proj.q_weight[0m", shape: (3584, 2368), dtype: uint32 1%| | 1/199 [00:10<32:17, 9.78s/it] [2024-06-06 23:42:32] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.22.mlp.down_proj.q_scale[0m", shape: (3584, 592), dtype: float16 1%| | 1/199 [00:10<32:17, 9.78s/it] 2%|β | 3/199 [00:10<09:21, 2.87s/it] [2024-06-06 23:42:32] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.22.post_attention_layernorm.weight[0m", shape: (3584,), dtype: float16 2%|β | 3/199 [00:10<09:21, 2.87s/it] [2024-06-06 23:42:32] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.23.input_layernorm.weight[0m", shape: (3584,), dtype: float16 2%|β | 3/199 [00:10<09:21, 2.87s/it] [2024-06-06 23:42:33] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.23.mlp.down_proj.q_weight[0m", shape: (3584, 2368), dtype: uint32 2%|β | 3/199 [00:10<09:21, 2.87s/it] [2024-06-06 23:42:33] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.23.mlp.down_proj.q_scale[0m", shape: (3584, 592), dtype: float16 2%|β | 3/199 [00:11<09:21, 2.87s/it] 3%|β | 6/199 [00:11<03:48, 1.19s/it] [2024-06-06 23:42:34] INFO group_quantization.py:217: Compiling quantize function for key: ((37888, 3584), float16, cuda, axis=1, output_transpose=False) 3%|β | 6/199 [00:11<03:48, 1.19s/it] [2024-06-06 23:42:34] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.23.mlp.gate_up_proj.q_weight[0m", shape: (37888, 448), dtype: uint32 3%|β | 6/199 [00:12<03:48, 1.19s/it] [2024-06-06 23:42:34] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.23.mlp.gate_up_proj.q_scale[0m", shape: (37888, 112), dtype: float16 3%|β | 6/199 [00:12<03:48, 1.19s/it] 4%|β | 7/199 [00:12<04:02, 1.26s/it] [2024-06-06 23:42:34] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.23.post_attention_layernorm.weight[0m", shape: (3584,), dtype: float16 4%|β | 7/199 [00:12<04:02, 1.26s/it] [2024-06-06 23:42:34] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.23.self_attn.c_attn.bias[0m", shape: (4608,), dtype: float16 4%|β | 7/199 [00:12<04:02, 1.26s/it] [2024-06-06 23:42:34] INFO group_quantization.py:217: Compiling quantize function for key: ((4608, 3584), float16, cuda, axis=1, output_transpose=False) 4%|β | 7/199 [00:12<04:02, 1.26s/it] [2024-06-06 23:42:35] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.23.self_attn.c_attn.q_weight[0m", shape: (4608, 448), dtype: uint32 4%|β | 7/199 [00:13<04:02, 1.26s/it] [2024-06-06 23:42:35] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.23.self_attn.c_attn.q_scale[0m", shape: (4608, 112), dtype: float16 4%|β | 7/199 [00:13<04:02, 1.26s/it] 5%|β | 10/199 [00:13<02:18, 1.36it/s] [2024-06-06 23:42:35] INFO group_quantization.py:217: Compiling quantize function for key: ((3584, 3584), float16, cuda, axis=1, output_transpose=False) 5%|β | 10/199 [00:13<02:18, 1.36it/s] [2024-06-06 23:42:35] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.23.self_attn.o_proj.q_weight[0m", shape: (3584, 448), dtype: uint32 5%|β | 10/199 [00:13<02:18, 1.36it/s] [2024-06-06 23:42:35] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.23.self_attn.o_proj.q_scale[0m", shape: (3584, 112), dtype: float16 5%|β | 10/199 [00:13<02:18, 1.36it/s] 6%|β | 11/199 [00:13<02:10, 1.44it/s] [2024-06-06 23:42:35] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.24.input_layernorm.weight[0m", shape: (3584,), dtype: float16 6%|β | 11/199 [00:13<02:10, 1.44it/s] [2024-06-06 23:42:36] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.24.mlp.down_proj.q_weight[0m", shape: (3584, 2368), dtype: uint32 6%|β | 11/199 [00:13<02:10, 1.44it/s] [2024-06-06 23:42:36] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.24.mlp.down_proj.q_scale[0m", shape: (3584, 592), dtype: float16 6%|β | 11/199 [00:13<02:10, 1.44it/s] 7%|β | 13/199 [00:14<01:36, 1.93it/s] [2024-06-06 23:42:37] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.24.mlp.gate_up_proj.q_weight[0m", shape: (37888, 448), dtype: uint32 7%|β | 13/199 [00:15<01:36, 1.93it/s] [2024-06-06 23:42:37] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.24.mlp.gate_up_proj.q_scale[0m", shape: (37888, 112), dtype: float16 7%|β | 13/199 [00:15<01:36, 1.93it/s] 7%|β | 14/199 [00:15<02:05, 1.48it/s] [2024-06-06 23:42:37] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.24.post_attention_layernorm.weight[0m", shape: (3584,), dtype: float16 7%|β | 14/199 [00:15<02:05, 1.48it/s] [2024-06-06 23:42:37] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.24.self_attn.c_attn.bias[0m", shape: (4608,), dtype: float16 7%|β | 14/199 [00:15<02:05, 1.48it/s] [2024-06-06 23:42:37] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.24.self_attn.c_attn.q_weight[0m", shape: (4608, 448), dtype: uint32 7%|β | 14/199 [00:15<02:05, 1.48it/s] [2024-06-06 23:42:37] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.24.self_attn.c_attn.q_scale[0m", shape: (4608, 112), dtype: float16 7%|β | 14/199 [00:15<02:05, 1.48it/s] [2024-06-06 23:42:37] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.24.self_attn.o_proj.q_weight[0m", shape: (3584, 448), dtype: uint32 7%|β | 14/199 [00:15<02:05, 1.48it/s] [2024-06-06 23:42:37] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.24.self_attn.o_proj.q_scale[0m", shape: (3584, 112), dtype: float16 7%|β | 14/199 [00:15<02:05, 1.48it/s] 9%|β | 18/199 [00:15<01:00, 3.00it/s] [2024-06-06 23:42:37] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.25.input_layernorm.weight[0m", shape: (3584,), dtype: float16 9%|β | 18/199 [00:15<01:00, 3.00it/s] [2024-06-06 23:42:37] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.25.mlp.down_proj.q_weight[0m", shape: (3584, 2368), dtype: uint32 9%|β | 18/199 [00:15<01:00, 3.00it/s] [2024-06-06 23:42:38] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.25.mlp.down_proj.q_scale[0m", shape: (3584, 592), dtype: float16 9%|β | 18/199 [00:15<01:00, 3.00it/s] 10%|β | 20/199 [00:15<00:52, 3.43it/s] [2024-06-06 23:42:39] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.25.mlp.gate_up_proj.q_weight[0m", shape: (37888, 448), dtype: uint32 10%|β | 20/199 [00:16<00:52, 3.43it/s] [2024-06-06 23:42:39] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.25.mlp.gate_up_proj.q_scale[0m", shape: (37888, 112), dtype: float16 10%|β | 20/199 [00:17<00:52, 3.43it/s] 11%|β | 21/199 [00:17<01:22, 2.16it/s] [2024-06-06 23:42:39] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.25.post_attention_layernorm.weight[0m", shape: (3584,), dtype: float16 11%|β | 21/199 [00:17<01:22, 2.16it/s] [2024-06-06 23:42:39] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.25.self_attn.c_attn.bias[0m", shape: (4608,), dtype: float16 11%|β | 21/199 [00:17<01:22, 2.16it/s] [2024-06-06 23:42:39] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.25.self_attn.c_attn.q_weight[0m", shape: (4608, 448), dtype: uint32 11%|β | 21/199 [00:17<01:22, 2.16it/s] [2024-06-06 23:42:39] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.25.self_attn.c_attn.q_scale[0m", shape: (4608, 112), dtype: float16 11%|β | 21/199 [00:17<01:22, 2.16it/s] [2024-06-06 23:42:39] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.25.self_attn.o_proj.q_weight[0m", shape: (3584, 448), dtype: uint32 11%|β | 21/199 [00:17<01:22, 2.16it/s] [2024-06-06 23:42:39] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.25.self_attn.o_proj.q_scale[0m", shape: (3584, 112), dtype: float16 11%|β | 21/199 [00:17<01:22, 2.16it/s] 13%|ββ | 25/199 [00:17<00:44, 3.93it/s] [2024-06-06 23:42:39] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.26.input_layernorm.weight[0m", shape: (3584,), dtype: float16 13%|ββ | 25/199 [00:17<00:44, 3.93it/s] [2024-06-06 23:42:39] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.26.mlp.down_proj.q_weight[0m", shape: (3584, 2368), dtype: uint32 13%|ββ | 25/199 [00:17<00:44, 3.93it/s] [2024-06-06 23:42:39] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.26.mlp.down_proj.q_scale[0m", shape: (3584, 592), dtype: float16 13%|ββ | 25/199 [00:17<00:44, 3.93it/s] 14%|ββ | 27/199 [00:17<00:40, 4.27it/s] [2024-06-06 23:42:40] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.26.mlp.gate_up_proj.q_weight[0m", shape: (37888, 448), dtype: uint32 14%|ββ | 27/199 [00:18<00:40, 4.27it/s] [2024-06-06 23:42:41] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.26.mlp.gate_up_proj.q_scale[0m", shape: (37888, 112), dtype: float16 14%|ββ | 27/199 [00:18<00:40, 4.27it/s] [2024-06-06 23:42:41] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.26.post_attention_layernorm.weight[0m", shape: (3584,), dtype: float16 14%|ββ | 27/199 [00:18<00:40, 4.27it/s] 15%|ββ | 29/199 [00:18<00:57, 2.96it/s] [2024-06-06 23:42:41] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.26.self_attn.c_attn.bias[0m", shape: (4608,), dtype: float16 15%|ββ | 29/199 [00:18<00:57, 2.96it/s] [2024-06-06 23:42:41] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.26.self_attn.c_attn.q_weight[0m", shape: (4608, 448), dtype: uint32 15%|ββ | 29/199 [00:18<00:57, 2.96it/s] [2024-06-06 23:42:41] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.26.self_attn.c_attn.q_scale[0m", shape: (4608, 112), dtype: float16 15%|ββ | 29/199 [00:18<00:57, 2.96it/s] [2024-06-06 23:42:41] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.26.self_attn.o_proj.q_weight[0m", shape: (3584, 448), dtype: uint32 15%|ββ | 29/199 [00:19<00:57, 2.96it/s] [2024-06-06 23:42:41] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.26.self_attn.o_proj.q_scale[0m", shape: (3584, 112), dtype: float16 15%|ββ | 29/199 [00:19<00:57, 2.96it/s] 16%|ββ | 32/199 [00:19<00:38, 4.37it/s] [2024-06-06 23:42:41] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.27.input_layernorm.weight[0m", shape: (3584,), dtype: float16 16%|ββ | 32/199 [00:19<00:38, 4.37it/s] [2024-06-06 23:42:41] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.27.mlp.down_proj.q_weight[0m", shape: (3584, 2368), dtype: uint32 16%|ββ | 32/199 [00:19<00:38, 4.37it/s] [2024-06-06 23:42:41] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.27.mlp.down_proj.q_scale[0m", shape: (3584, 592), dtype: float16 16%|ββ | 32/199 [00:19<00:38, 4.37it/s] 17%|ββ | 34/199 [00:19<00:35, 4.67it/s] [2024-06-06 23:42:42] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.27.mlp.gate_up_proj.q_weight[0m", shape: (37888, 448), dtype: uint32 17%|ββ | 34/199 [00:20<00:35, 4.67it/s] [2024-06-06 23:42:42] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.27.mlp.gate_up_proj.q_scale[0m", shape: (37888, 112), dtype: float16 17%|ββ | 34/199 [00:20<00:35, 4.67it/s] [2024-06-06 23:42:42] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.27.post_attention_layernorm.weight[0m", shape: (3584,), dtype: float16 17%|ββ | 34/199 [00:20<00:35, 4.67it/s] 18%|ββ | 36/199 [00:20<00:50, 3.25it/s] [2024-06-06 23:42:42] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.27.self_attn.c_attn.bias[0m", shape: (4608,), dtype: float16 18%|ββ | 36/199 [00:20<00:50, 3.25it/s] [2024-06-06 23:42:42] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.27.self_attn.c_attn.q_weight[0m", shape: (4608, 448), dtype: uint32 18%|ββ | 36/199 [00:20<00:50, 3.25it/s] [2024-06-06 23:42:42] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.27.self_attn.c_attn.q_scale[0m", shape: (4608, 112), dtype: float16 18%|ββ | 36/199 [00:20<00:50, 3.25it/s] [2024-06-06 23:42:42] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.27.self_attn.o_proj.q_weight[0m", shape: (3584, 448), dtype: uint32 18%|ββ | 36/199 [00:20<00:50, 3.25it/s] [2024-06-06 23:42:42] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.27.self_attn.o_proj.q_scale[0m", shape: (3584, 112), dtype: float16 18%|ββ | 36/199 [00:20<00:50, 3.25it/s] 20%|ββ | 39/199 [00:20<00:33, 4.76it/s] [2024-06-06 23:42:42] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.norm.weight[0m", shape: (3584,), dtype: float16 20%|ββ | 39/199 [00:20<00:33, 4.76it/s] [2024-06-06 23:42:42] INFO huggingface_loader.py:197: Unloading HF weight file: /models/Qwen2-7B-Instruct/model-00004-of-00004.safetensors 20%|ββ | 39/199 [00:20<00:33, 4.76it/s] [2024-06-06 23:42:43] INFO huggingface_loader.py:185: Loading HF parameters from: /models/Qwen2-7B-Instruct/model-00001-of-00004.safetensors 20%|ββ | 39/199 [00:21<00:33, 4.76it/s] [2024-06-06 23:42:50] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.embed_tokens.q_weight[0m", shape: (152064, 448), dtype: uint32 20%|ββ | 39/199 [00:28<00:33, 4.76it/s] [2024-06-06 23:42:51] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.embed_tokens.q_scale[0m", shape: (152064, 112), dtype: float16 20%|ββ | 39/199 [00:29<00:33, 4.76it/s] 21%|ββ | 41/199 [00:29<03:25, 1.30s/it] [2024-06-06 23:42:51] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.0.input_layernorm.weight[0m", shape: (3584,), dtype: float16 21%|ββ | 41/199 [00:29<03:25, 1.30s/it] [2024-06-06 23:42:52] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.0.mlp.down_proj.q_weight[0m", shape: (3584, 2368), dtype: uint32 21%|ββ | 41/199 [00:29<03:25, 1.30s/it] [2024-06-06 23:42:52] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.0.mlp.down_proj.q_scale[0m", shape: (3584, 592), dtype: float16 21%|ββ | 41/199 [00:29<03:25, 1.30s/it] 22%|βββ | 43/199 [00:29<02:38, 1.02s/it] [2024-06-06 23:42:53] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.0.mlp.gate_up_proj.q_weight[0m", shape: (37888, 448), dtype: uint32 22%|βββ | 43/199 [00:31<02:38, 1.02s/it] [2024-06-06 23:42:53] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.0.mlp.gate_up_proj.q_scale[0m", shape: (37888, 112), dtype: float16 22%|βββ | 43/199 [00:31<02:38, 1.02s/it] 22%|βββ | 44/199 [00:31<02:55, 1.13s/it] [2024-06-06 23:42:53] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.0.post_attention_layernorm.weight[0m", shape: (3584,), dtype: float16 22%|βββ | 44/199 [00:31<02:55, 1.13s/it] [2024-06-06 23:42:53] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.0.self_attn.c_attn.bias[0m", shape: (4608,), dtype: float16 22%|βββ | 44/199 [00:31<02:55, 1.13s/it] [2024-06-06 23:42:53] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.0.self_attn.c_attn.q_weight[0m", shape: (4608, 448), dtype: uint32 22%|βββ | 44/199 [00:31<02:55, 1.13s/it] [2024-06-06 23:42:53] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.0.self_attn.c_attn.q_scale[0m", shape: (4608, 112), dtype: float16 22%|βββ | 44/199 [00:31<02:55, 1.13s/it] [2024-06-06 23:42:54] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.0.self_attn.o_proj.q_weight[0m", shape: (3584, 448), dtype: uint32 22%|βββ | 44/199 [00:31<02:55, 1.13s/it] [2024-06-06 23:42:54] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.0.self_attn.o_proj.q_scale[0m", shape: (3584, 112), dtype: float16 22%|βββ | 44/199 [00:31<02:55, 1.13s/it] 24%|βββ | 48/199 [00:31<01:30, 1.66it/s] [2024-06-06 23:42:54] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.1.input_layernorm.weight[0m", shape: (3584,), dtype: float16 24%|βββ | 48/199 [00:31<01:30, 1.66it/s] [2024-06-06 23:42:54] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.1.mlp.down_proj.q_weight[0m", shape: (3584, 2368), dtype: uint32 24%|βββ | 48/199 [00:32<01:30, 1.66it/s] [2024-06-06 23:42:54] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.1.mlp.down_proj.q_scale[0m", shape: (3584, 592), dtype: float16 24%|βββ | 48/199 [00:32<01:30, 1.66it/s] 25%|βββ | 50/199 [00:32<01:14, 2.00it/s] [2024-06-06 23:42:56] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.1.mlp.gate_up_proj.q_weight[0m", shape: (37888, 448), dtype: uint32 25%|βββ | 50/199 [00:33<01:14, 2.00it/s] [2024-06-06 23:42:56] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.1.mlp.gate_up_proj.q_scale[0m", shape: (37888, 112), dtype: float16 25%|βββ | 50/199 [00:34<01:14, 2.00it/s] [2024-06-06 23:42:56] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.1.post_attention_layernorm.weight[0m", shape: (3584,), dtype: float16 25%|βββ | 50/199 [00:34<01:14, 2.00it/s] 26%|βββ | 52/199 [00:34<01:30, 1.62it/s] [2024-06-06 23:42:56] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.1.self_attn.c_attn.bias[0m", shape: (4608,), dtype: float16 26%|βββ | 52/199 [00:34<01:30, 1.62it/s] [2024-06-06 23:42:56] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.1.self_attn.c_attn.q_weight[0m", shape: (4608, 448), dtype: uint32 26%|βββ | 52/199 [00:34<01:30, 1.62it/s] [2024-06-06 23:42:56] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.1.self_attn.c_attn.q_scale[0m", shape: (4608, 112), dtype: float16 26%|βββ | 52/199 [00:34<01:30, 1.62it/s] [2024-06-06 23:42:56] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.1.self_attn.o_proj.q_weight[0m", shape: (3584, 448), dtype: uint32 26%|βββ | 52/199 [00:34<01:30, 1.62it/s] [2024-06-06 23:42:56] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.1.self_attn.o_proj.q_scale[0m", shape: (3584, 112), dtype: float16 26%|βββ | 52/199 [00:34<01:30, 1.62it/s] 28%|βββ | 55/199 [00:34<00:58, 2.44it/s] [2024-06-06 23:42:56] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.2.input_layernorm.weight[0m", shape: (3584,), dtype: float16 28%|βββ | 55/199 [00:34<00:58, 2.44it/s] [2024-06-06 23:42:56] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.2.mlp.down_proj.q_weight[0m", shape: (3584, 2368), dtype: uint32 28%|βββ | 55/199 [00:34<00:58, 2.44it/s] [2024-06-06 23:42:56] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.2.mlp.down_proj.q_scale[0m", shape: (3584, 592), dtype: float16 28%|βββ | 55/199 [00:34<00:58, 2.44it/s] 29%|βββ | 57/199 [00:34<00:50, 2.80it/s] [2024-06-06 23:42:58] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.2.mlp.gate_up_proj.q_weight[0m", shape: (37888, 448), dtype: uint32 29%|βββ | 57/199 [00:36<00:50, 2.80it/s] [2024-06-06 23:42:58] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.2.mlp.gate_up_proj.q_scale[0m", shape: (37888, 112), dtype: float16 29%|βββ | 57/199 [00:36<00:50, 2.80it/s] 29%|βββ | 58/199 [00:36<01:22, 1.71it/s] [2024-06-06 23:42:58] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.2.post_attention_layernorm.weight[0m", shape: (3584,), dtype: float16 29%|βββ | 58/199 [00:36<01:22, 1.71it/s] [2024-06-06 23:42:58] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.2.self_attn.c_attn.bias[0m", shape: (4608,), dtype: float16 29%|βββ | 58/199 [00:36<01:22, 1.71it/s] [2024-06-06 23:42:58] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.2.self_attn.c_attn.q_weight[0m", shape: (4608, 448), dtype: uint32 29%|βββ | 58/199 [00:36<01:22, 1.71it/s] [2024-06-06 23:42:58] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.2.self_attn.c_attn.q_scale[0m", shape: (4608, 112), dtype: float16 29%|βββ | 58/199 [00:36<01:22, 1.71it/s] [2024-06-06 23:42:58] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.2.self_attn.o_proj.q_weight[0m", shape: (3584, 448), dtype: uint32 29%|βββ | 58/199 [00:36<01:22, 1.71it/s] [2024-06-06 23:42:58] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.2.self_attn.o_proj.q_scale[0m", shape: (3584, 112), dtype: float16 29%|βββ | 58/199 [00:36<01:22, 1.71it/s] 31%|βββ | 62/199 [00:36<00:44, 3.08it/s] [2024-06-06 23:42:58] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.3.input_layernorm.weight[0m", shape: (3584,), dtype: float16 31%|βββ | 62/199 [00:36<00:44, 3.08it/s] [2024-06-06 23:42:59] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.3.mlp.down_proj.q_weight[0m", shape: (3584, 2368), dtype: uint32 31%|βββ | 62/199 [00:36<00:44, 3.08it/s] [2024-06-06 23:42:59] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.3.mlp.down_proj.q_scale[0m", shape: (3584, 592), dtype: float16 31%|βββ | 62/199 [00:36<00:44, 3.08it/s] 32%|ββββ | 64/199 [00:36<00:39, 3.42it/s] [2024-06-06 23:43:00] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.3.mlp.gate_up_proj.q_weight[0m", shape: (37888, 448), dtype: uint32 32%|ββββ | 64/199 [00:38<00:39, 3.42it/s] [2024-06-06 23:43:00] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.3.mlp.gate_up_proj.q_scale[0m", shape: (37888, 112), dtype: float16 32%|ββββ | 64/199 [00:38<00:39, 3.42it/s] [2024-06-06 23:43:01] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.3.post_attention_layernorm.weight[0m", shape: (3584,), dtype: float16 32%|ββββ | 64/199 [00:38<00:39, 3.42it/s] 33%|ββββ | 66/199 [00:38<01:00, 2.18it/s] [2024-06-06 23:43:01] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.3.self_attn.c_attn.bias[0m", shape: (4608,), dtype: float16 33%|ββββ | 66/199 [00:38<01:00, 2.18it/s] [2024-06-06 23:43:01] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.3.self_attn.c_attn.q_weight[0m", shape: (4608, 448), dtype: uint32 33%|ββββ | 66/199 [00:38<01:00, 2.18it/s] [2024-06-06 23:43:01] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.3.self_attn.c_attn.q_scale[0m", shape: (4608, 112), dtype: float16 33%|ββββ | 66/199 [00:38<01:00, 2.18it/s] [2024-06-06 23:43:01] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.3.self_attn.o_proj.q_weight[0m", shape: (3584, 448), dtype: uint32 33%|ββββ | 66/199 [00:38<01:00, 2.18it/s] [2024-06-06 23:43:01] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.3.self_attn.o_proj.q_scale[0m", shape: (3584, 112), dtype: float16 33%|ββββ | 66/199 [00:38<01:00, 2.18it/s] 35%|ββββ | 69/199 [00:38<00:40, 3.25it/s] [2024-06-06 23:43:01] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.4.input_layernorm.weight[0m", shape: (3584,), dtype: float16 35%|ββββ | 69/199 [00:38<00:40, 3.25it/s] [2024-06-06 23:43:01] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.4.mlp.down_proj.q_weight[0m", shape: (3584, 2368), dtype: uint32 35%|ββββ | 69/199 [00:39<00:40, 3.25it/s] [2024-06-06 23:43:01] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.4.mlp.down_proj.q_scale[0m", shape: (3584, 592), dtype: float16 35%|ββββ | 69/199 [00:39<00:40, 3.25it/s] 36%|ββββ | 71/199 [00:39<00:35, 3.58it/s] [2024-06-06 23:43:02] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.4.mlp.gate_up_proj.q_weight[0m", shape: (37888, 448), dtype: uint32 36%|ββββ | 71/199 [00:40<00:35, 3.58it/s] [2024-06-06 23:43:03] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.4.mlp.gate_up_proj.q_scale[0m", shape: (37888, 112), dtype: float16 36%|ββββ | 71/199 [00:40<00:35, 3.58it/s] [2024-06-06 23:43:03] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.4.post_attention_layernorm.weight[0m", shape: (3584,), dtype: float16 36%|ββββ | 71/199 [00:40<00:35, 3.58it/s] 37%|ββββ | 73/199 [00:40<00:51, 2.43it/s] [2024-06-06 23:43:03] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.4.self_attn.c_attn.bias[0m", shape: (4608,), dtype: float16 37%|ββββ | 73/199 [00:40<00:51, 2.43it/s] [2024-06-06 23:43:03] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.4.self_attn.c_attn.q_weight[0m", shape: (4608, 448), dtype: uint32 37%|ββββ | 73/199 [00:40<00:51, 2.43it/s] [2024-06-06 23:43:03] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.4.self_attn.c_attn.q_scale[0m", shape: (4608, 112), dtype: float16 37%|ββββ | 73/199 [00:40<00:51, 2.43it/s] [2024-06-06 23:43:03] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.4.self_attn.o_proj.q_weight[0m", shape: (3584, 448), dtype: uint32 37%|ββββ | 73/199 [00:41<00:51, 2.43it/s] [2024-06-06 23:43:03] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.4.self_attn.o_proj.q_scale[0m", shape: (3584, 112), dtype: float16 37%|ββββ | 73/199 [00:41<00:51, 2.43it/s] 38%|ββββ | 76/199 [00:41<00:33, 3.62it/s] [2024-06-06 23:43:03] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.5.input_layernorm.weight[0m", shape: (3584,), dtype: float16 38%|ββββ | 76/199 [00:41<00:33, 3.62it/s] [2024-06-06 23:43:03] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.5.mlp.down_proj.q_weight[0m", shape: (3584, 2368), dtype: uint32 38%|ββββ | 76/199 [00:41<00:33, 3.62it/s] [2024-06-06 23:43:03] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.5.mlp.down_proj.q_scale[0m", shape: (3584, 592), dtype: float16 38%|ββββ | 76/199 [00:41<00:33, 3.62it/s] 39%|ββββ | 78/199 [00:41<00:30, 4.00it/s] [2024-06-06 23:43:04] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.5.mlp.gate_up_proj.q_weight[0m", shape: (37888, 448), dtype: uint32 39%|ββββ | 78/199 [00:42<00:30, 4.00it/s] [2024-06-06 23:43:05] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.5.mlp.gate_up_proj.q_scale[0m", shape: (37888, 112), dtype: float16 39%|ββββ | 78/199 [00:42<00:30, 4.00it/s] [2024-06-06 23:43:05] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.5.post_attention_layernorm.weight[0m", shape: (3584,), dtype: float16 39%|ββββ | 78/199 [00:43<00:30, 4.00it/s] 40%|ββββ | 80/199 [00:43<00:48, 2.47it/s] [2024-06-06 23:43:05] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.5.self_attn.c_attn.bias[0m", shape: (4608,), dtype: float16 40%|ββββ | 80/199 [00:43<00:48, 2.47it/s] [2024-06-06 23:43:05] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.5.self_attn.c_attn.q_weight[0m", shape: (4608, 448), dtype: uint32 40%|ββββ | 80/199 [00:43<00:48, 2.47it/s] [2024-06-06 23:43:05] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.5.self_attn.c_attn.q_scale[0m", shape: (4608, 112), dtype: float16 40%|ββββ | 80/199 [00:43<00:48, 2.47it/s] [2024-06-06 23:43:05] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.5.self_attn.o_proj.q_weight[0m", shape: (3584, 448), dtype: uint32 40%|ββββ | 80/199 [00:43<00:48, 2.47it/s] [2024-06-06 23:43:05] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.5.self_attn.o_proj.q_scale[0m", shape: (3584, 112), dtype: float16 40%|ββββ | 80/199 [00:43<00:48, 2.47it/s] 42%|βββββ | 83/199 [00:43<00:31, 3.67it/s] [2024-06-06 23:43:05] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.6.self_attn.c_attn.bias[0m", shape: (4608,), dtype: float16 42%|βββββ | 83/199 [00:43<00:31, 3.67it/s] [2024-06-06 23:43:05] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.6.self_attn.c_attn.q_weight[0m", shape: (4608, 448), dtype: uint32 42%|βββββ | 83/199 [00:43<00:31, 3.67it/s] [2024-06-06 23:43:05] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.6.self_attn.c_attn.q_scale[0m", shape: (4608, 112), dtype: float16 42%|βββββ | 83/199 [00:43<00:31, 3.67it/s] 43%|βββββ | 85/199 [00:43<00:25, 4.51it/s] [2024-06-06 23:43:05] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.6.self_attn.o_proj.q_weight[0m", shape: (3584, 448), dtype: uint32 43%|βββββ | 85/199 [00:43<00:25, 4.51it/s] [2024-06-06 23:43:05] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.6.self_attn.o_proj.q_scale[0m", shape: (3584, 112), dtype: float16 43%|βββββ | 85/199 [00:43<00:25, 4.51it/s] [2024-06-06 23:43:05] INFO huggingface_loader.py:197: Unloading HF weight file: /models/Qwen2-7B-Instruct/model-00001-of-00004.safetensors 43%|βββββ | 85/199 [00:43<00:25, 4.51it/s] [2024-06-06 23:43:06] INFO huggingface_loader.py:185: Loading HF parameters from: /models/Qwen2-7B-Instruct/model-00002-of-00004.safetensors 43%|βββββ | 85/199 [00:43<00:25, 4.51it/s] [2024-06-06 23:43:10] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.10.input_layernorm.weight[0m", shape: (3584,), dtype: float16 43%|βββββ | 85/199 [00:48<00:25, 4.51it/s] 44%|βββββ | 87/199 [00:48<01:34, 1.19it/s] [2024-06-06 23:43:10] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.10.mlp.down_proj.q_weight[0m", shape: (3584, 2368), dtype: uint32 44%|βββββ | 87/199 [00:48<01:34, 1.19it/s] [2024-06-06 23:43:10] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.10.mlp.down_proj.q_scale[0m", shape: (3584, 592), dtype: float16 44%|βββββ | 87/199 [00:48<01:34, 1.19it/s] 44%|βββββ | 88/199 [00:48<01:24, 1.31it/s] [2024-06-06 23:43:12] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.10.mlp.gate_up_proj.q_weight[0m", shape: (37888, 448), dtype: uint32 44%|βββββ | 88/199 [00:49<01:24, 1.31it/s] [2024-06-06 23:43:12] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.10.mlp.gate_up_proj.q_scale[0m", shape: (37888, 112), dtype: float16 44%|βββββ | 88/199 [00:50<01:24, 1.31it/s] 45%|βββββ | 89/199 [00:50<01:38, 1.11it/s] [2024-06-06 23:43:12] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.10.post_attention_layernorm.weight[0m", shape: (3584,), dtype: float16 45%|βββββ | 89/199 [00:50<01:38, 1.11it/s] [2024-06-06 23:43:12] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.10.self_attn.c_attn.bias[0m", shape: (4608,), dtype: float16 45%|βββββ | 89/199 [00:50<01:38, 1.11it/s] [2024-06-06 23:43:12] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.10.self_attn.c_attn.q_weight[0m", shape: (4608, 448), dtype: uint32 45%|βββββ | 89/199 [00:50<01:38, 1.11it/s] [2024-06-06 23:43:12] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.10.self_attn.c_attn.q_scale[0m", shape: (4608, 112), dtype: float16 45%|βββββ | 89/199 [00:50<01:38, 1.11it/s] [2024-06-06 23:43:12] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.10.self_attn.o_proj.q_weight[0m", shape: (3584, 448), dtype: uint32 45%|βββββ | 89/199 [00:50<01:38, 1.11it/s] [2024-06-06 23:43:12] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.10.self_attn.o_proj.q_scale[0m", shape: (3584, 112), dtype: float16 45%|βββββ | 89/199 [00:50<01:38, 1.11it/s] 47%|βββββ | 93/199 [00:50<00:48, 2.18it/s] [2024-06-06 23:43:12] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.11.input_layernorm.weight[0m", shape: (3584,), dtype: float16 47%|βββββ | 93/199 [00:50<00:48, 2.18it/s] [2024-06-06 23:43:12] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.11.mlp.down_proj.q_weight[0m", shape: (3584, 2368), dtype: uint32 47%|βββββ | 93/199 [00:50<00:48, 2.18it/s] [2024-06-06 23:43:12] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.11.mlp.down_proj.q_scale[0m", shape: (3584, 592), dtype: float16 47%|βββββ | 93/199 [00:50<00:48, 2.18it/s] 48%|βββββ | 95/199 [00:50<00:39, 2.62it/s] [2024-06-06 23:43:13] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.11.mlp.gate_up_proj.q_weight[0m", shape: (37888, 448), dtype: uint32 48%|βββββ | 95/199 [00:51<00:39, 2.62it/s] [2024-06-06 23:43:14] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.11.mlp.gate_up_proj.q_scale[0m", shape: (37888, 112), dtype: float16 48%|βββββ | 95/199 [00:51<00:39, 2.62it/s] 48%|βββββ | 96/199 [00:51<00:51, 1.99it/s] [2024-06-06 23:43:14] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.11.post_attention_layernorm.weight[0m", shape: (3584,), dtype: float16 48%|βββββ | 96/199 [00:51<00:51, 1.99it/s] [2024-06-06 23:43:14] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.11.self_attn.c_attn.bias[0m", shape: (4608,), dtype: float16 48%|βββββ | 96/199 [00:51<00:51, 1.99it/s] [2024-06-06 23:43:14] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.11.self_attn.c_attn.q_weight[0m", shape: (4608, 448), dtype: uint32 48%|βββββ | 96/199 [00:51<00:51, 1.99it/s] [2024-06-06 23:43:14] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.11.self_attn.c_attn.q_scale[0m", shape: (4608, 112), dtype: float16 48%|βββββ | 96/199 [00:51<00:51, 1.99it/s] [2024-06-06 23:43:14] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.11.self_attn.o_proj.q_weight[0m", shape: (3584, 448), dtype: uint32 48%|βββββ | 96/199 [00:51<00:51, 1.99it/s] [2024-06-06 23:43:14] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.11.self_attn.o_proj.q_scale[0m", shape: (3584, 112), dtype: float16 48%|βββββ | 96/199 [00:52<00:51, 1.99it/s] 50%|βββββ | 100/199 [00:52<00:28, 3.50it/s] [2024-06-06 23:43:14] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.12.input_layernorm.weight[0m", shape: (3584,), dtype: float16 50%|βββββ | 100/199 [00:52<00:28, 3.50it/s] [2024-06-06 23:43:14] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.12.mlp.down_proj.q_weight[0m", shape: (3584, 2368), dtype: uint32 50%|βββββ | 100/199 [00:52<00:28, 3.50it/s] [2024-06-06 23:43:14] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.12.mlp.down_proj.q_scale[0m", shape: (3584, 592), dtype: float16 50%|βββββ | 100/199 [00:52<00:28, 3.50it/s] 51%|ββββββ | 102/199 [00:52<00:24, 3.89it/s] [2024-06-06 23:43:15] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.12.mlp.gate_up_proj.q_weight[0m", shape: (37888, 448), dtype: uint32 51%|ββββββ | 102/199 [00:53<00:24, 3.89it/s] [2024-06-06 23:43:15] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.12.mlp.gate_up_proj.q_scale[0m", shape: (37888, 112), dtype: float16 51%|ββββββ | 102/199 [00:53<00:24, 3.89it/s] 52%|ββββββ | 103/199 [00:53<00:34, 2.79it/s] [2024-06-06 23:43:15] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.12.post_attention_layernorm.weight[0m", shape: (3584,), dtype: float16 52%|ββββββ | 103/199 [00:53<00:34, 2.79it/s] [2024-06-06 23:43:15] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.12.self_attn.c_attn.bias[0m", shape: (4608,), dtype: float16 52%|ββββββ | 103/199 [00:53<00:34, 2.79it/s] [2024-06-06 23:43:15] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.12.self_attn.c_attn.q_weight[0m", shape: (4608, 448), dtype: uint32 52%|ββββββ | 103/199 [00:53<00:34, 2.79it/s] [2024-06-06 23:43:15] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.12.self_attn.c_attn.q_scale[0m", shape: (4608, 112), dtype: float16 52%|ββββββ | 103/199 [00:53<00:34, 2.79it/s] [2024-06-06 23:43:15] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.12.self_attn.o_proj.q_weight[0m", shape: (3584, 448), dtype: uint32 52%|ββββββ | 103/199 [00:53<00:34, 2.79it/s] [2024-06-06 23:43:15] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.12.self_attn.o_proj.q_scale[0m", shape: (3584, 112), dtype: float16 52%|ββββββ | 103/199 [00:53<00:34, 2.79it/s] 54%|ββββββ | 107/199 [00:53<00:19, 4.71it/s] [2024-06-06 23:43:15] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.13.input_layernorm.weight[0m", shape: (3584,), dtype: float16 54%|ββββββ | 107/199 [00:53<00:19, 4.71it/s] [2024-06-06 23:43:15] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.13.mlp.down_proj.q_weight[0m", shape: (3584, 2368), dtype: uint32 54%|ββββββ | 107/199 [00:53<00:19, 4.71it/s] [2024-06-06 23:43:16] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.13.mlp.down_proj.q_scale[0m", shape: (3584, 592), dtype: float16 54%|ββββββ | 107/199 [00:53<00:19, 4.71it/s] 55%|ββββββ | 109/199 [00:53<00:18, 4.94it/s] [2024-06-06 23:43:16] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.13.mlp.gate_up_proj.q_weight[0m", shape: (37888, 448), dtype: uint32 55%|ββββββ | 109/199 [00:54<00:18, 4.94it/s] [2024-06-06 23:43:17] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.13.mlp.gate_up_proj.q_scale[0m", shape: (37888, 112), dtype: float16 55%|ββββββ | 109/199 [00:54<00:18, 4.94it/s] 55%|ββββββ | 110/199 [00:54<00:29, 3.00it/s] [2024-06-06 23:43:17] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.13.post_attention_layernorm.weight[0m", shape: (3584,), dtype: float16 55%|ββββββ | 110/199 [00:54<00:29, 3.00it/s] [2024-06-06 23:43:17] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.13.self_attn.c_attn.bias[0m", shape: (4608,), dtype: float16 55%|ββββββ | 110/199 [00:54<00:29, 3.00it/s] [2024-06-06 23:43:17] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.13.self_attn.c_attn.q_weight[0m", shape: (4608, 448), dtype: uint32 55%|ββββββ | 110/199 [00:54<00:29, 3.00it/s] [2024-06-06 23:43:17] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.13.self_attn.c_attn.q_scale[0m", shape: (4608, 112), dtype: float16 55%|ββββββ | 110/199 [00:54<00:29, 3.00it/s] [2024-06-06 23:43:17] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.13.self_attn.o_proj.q_weight[0m", shape: (3584, 448), dtype: uint32 55%|ββββββ | 110/199 [00:55<00:29, 3.00it/s] [2024-06-06 23:43:17] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.13.self_attn.o_proj.q_scale[0m", shape: (3584, 112), dtype: float16 55%|ββββββ | 110/199 [00:55<00:29, 3.00it/s] 57%|ββββββ | 114/199 [00:55<00:17, 4.96it/s] [2024-06-06 23:43:17] INFO huggingface_loader.py:185: Loading HF parameters from: /models/Qwen2-7B-Instruct/model-00003-of-00004.safetensors 57%|ββββββ | 114/199 [00:55<00:17, 4.96it/s] [2024-06-06 23:43:22] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.14.mlp.gate_up_proj.q_weight[0m", shape: (37888, 448), dtype: uint32 57%|ββββββ | 114/199 [01:00<00:17, 4.96it/s] [2024-06-06 23:43:22] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.14.mlp.gate_up_proj.q_scale[0m", shape: (37888, 112), dtype: float16 57%|ββββββ | 114/199 [01:00<00:17, 4.96it/s] [2024-06-06 23:43:22] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.14.self_attn.c_attn.bias[0m", shape: (4608,), dtype: float16 57%|ββββββ | 114/199 [01:00<00:17, 4.96it/s] 58%|ββββββ | 116/199 [01:00<01:09, 1.20it/s] [2024-06-06 23:43:22] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.14.self_attn.c_attn.q_weight[0m", shape: (4608, 448), dtype: uint32 58%|ββββββ | 116/199 [01:00<01:09, 1.20it/s] [2024-06-06 23:43:22] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.14.self_attn.c_attn.q_scale[0m", shape: (4608, 112), dtype: float16 58%|ββββββ | 116/199 [01:00<01:09, 1.20it/s] [2024-06-06 23:43:22] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.14.self_attn.o_proj.q_weight[0m", shape: (3584, 448), dtype: uint32 58%|ββββββ | 116/199 [01:00<01:09, 1.20it/s] [2024-06-06 23:43:22] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.14.self_attn.o_proj.q_scale[0m", shape: (3584, 112), dtype: float16 58%|ββββββ | 116/199 [01:00<01:09, 1.20it/s] 59%|ββββββ | 118/199 [01:00<00:51, 1.59it/s] [2024-06-06 23:43:22] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.6.input_layernorm.weight[0m", shape: (3584,), dtype: float16 59%|ββββββ | 118/199 [01:00<00:51, 1.59it/s] [2024-06-06 23:43:23] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.6.mlp.down_proj.q_weight[0m", shape: (3584, 2368), dtype: uint32 59%|ββββββ | 118/199 [01:00<00:51, 1.59it/s] [2024-06-06 23:43:23] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.6.mlp.down_proj.q_scale[0m", shape: (3584, 592), dtype: float16 59%|ββββββ | 118/199 [01:01<00:51, 1.59it/s] 60%|ββββββ | 120/199 [01:01<00:40, 1.94it/s] [2024-06-06 23:43:24] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.6.mlp.gate_up_proj.q_weight[0m", shape: (37888, 448), dtype: uint32 60%|ββββββ | 120/199 [01:02<00:40, 1.94it/s] [2024-06-06 23:43:24] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.6.mlp.gate_up_proj.q_scale[0m", shape: (37888, 112), dtype: float16 60%|ββββββ | 120/199 [01:02<00:40, 1.94it/s] 61%|ββββββ | 121/199 [01:02<00:53, 1.47it/s] [2024-06-06 23:43:24] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.6.post_attention_layernorm.weight[0m", shape: (3584,), dtype: float16 61%|ββββββ | 121/199 [01:02<00:53, 1.47it/s] [2024-06-06 23:43:24] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.7.input_layernorm.weight[0m", shape: (3584,), dtype: float16 61%|ββββββ | 121/199 [01:02<00:53, 1.47it/s] [2024-06-06 23:43:25] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.7.mlp.down_proj.q_weight[0m", shape: (3584, 2368), dtype: uint32 61%|ββββββ | 121/199 [01:02<00:53, 1.47it/s] [2024-06-06 23:43:25] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.7.mlp.down_proj.q_scale[0m", shape: (3584, 592), dtype: float16 61%|ββββββ | 121/199 [01:02<00:53, 1.47it/s] 62%|βββββββ | 124/199 [01:02<00:33, 2.23it/s] [2024-06-06 23:43:26] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.7.mlp.gate_up_proj.q_weight[0m", shape: (37888, 448), dtype: uint32 62%|βββββββ | 124/199 [01:04<00:33, 2.23it/s] [2024-06-06 23:43:26] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.7.mlp.gate_up_proj.q_scale[0m", shape: (37888, 112), dtype: float16 62%|βββββββ | 124/199 [01:04<00:33, 2.23it/s] 63%|βββββββ | 125/199 [01:04<00:43, 1.68it/s] [2024-06-06 23:43:26] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.7.post_attention_layernorm.weight[0m", shape: (3584,), dtype: float16 63%|βββββββ | 125/199 [01:04<00:43, 1.68it/s] [2024-06-06 23:43:26] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.7.self_attn.c_attn.bias[0m", shape: (4608,), dtype: float16 63%|βββββββ | 125/199 [01:04<00:43, 1.68it/s] [2024-06-06 23:43:26] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.7.self_attn.c_attn.q_weight[0m", shape: (4608, 448), dtype: uint32 63%|βββββββ | 125/199 [01:04<00:43, 1.68it/s] [2024-06-06 23:43:26] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.7.self_attn.c_attn.q_scale[0m", shape: (4608, 112), dtype: float16 63%|βββββββ | 125/199 [01:04<00:43, 1.68it/s] 64%|βββββββ | 128/199 [01:04<00:26, 2.73it/s] [2024-06-06 23:43:26] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.7.self_attn.o_proj.q_weight[0m", shape: (3584, 448), dtype: uint32 64%|βββββββ | 128/199 [01:04<00:26, 2.73it/s] [2024-06-06 23:43:26] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.7.self_attn.o_proj.q_scale[0m", shape: (3584, 112), dtype: float16 64%|βββββββ | 128/199 [01:04<00:26, 2.73it/s] [2024-06-06 23:43:26] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.8.input_layernorm.weight[0m", shape: (3584,), dtype: float16 64%|βββββββ | 128/199 [01:04<00:26, 2.73it/s] [2024-06-06 23:43:26] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.8.mlp.down_proj.q_weight[0m", shape: (3584, 2368), dtype: uint32 64%|βββββββ | 128/199 [01:04<00:26, 2.73it/s] [2024-06-06 23:43:27] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.8.mlp.down_proj.q_scale[0m", shape: (3584, 592), dtype: float16 64%|βββββββ | 128/199 [01:04<00:26, 2.73it/s] 66%|βββββββ | 131/199 [01:04<00:19, 3.56it/s] [2024-06-06 23:43:28] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.8.mlp.gate_up_proj.q_weight[0m", shape: (37888, 448), dtype: uint32 66%|βββββββ | 131/199 [01:06<00:19, 3.56it/s] [2024-06-06 23:43:28] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.8.mlp.gate_up_proj.q_scale[0m", shape: (37888, 112), dtype: float16 66%|βββββββ | 131/199 [01:06<00:19, 3.56it/s] 66%|βββββββ | 132/199 [01:06<00:30, 2.21it/s] [2024-06-06 23:43:28] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.8.post_attention_layernorm.weight[0m", shape: (3584,), dtype: float16 66%|βββββββ | 132/199 [01:06<00:30, 2.21it/s] [2024-06-06 23:43:28] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.8.self_attn.c_attn.bias[0m", shape: (4608,), dtype: float16 66%|βββββββ | 132/199 [01:06<00:30, 2.21it/s] [2024-06-06 23:43:28] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.8.self_attn.c_attn.q_weight[0m", shape: (4608, 448), dtype: uint32 66%|βββββββ | 132/199 [01:06<00:30, 2.21it/s] [2024-06-06 23:43:28] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.8.self_attn.c_attn.q_scale[0m", shape: (4608, 112), dtype: float16 66%|βββββββ | 132/199 [01:06<00:30, 2.21it/s] 68%|βββββββ | 135/199 [01:06<00:18, 3.41it/s] [2024-06-06 23:43:28] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.8.self_attn.o_proj.q_weight[0m", shape: (3584, 448), dtype: uint32 68%|βββββββ | 135/199 [01:06<00:18, 3.41it/s] [2024-06-06 23:43:28] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.8.self_attn.o_proj.q_scale[0m", shape: (3584, 112), dtype: float16 68%|βββββββ | 135/199 [01:06<00:18, 3.41it/s] [2024-06-06 23:43:28] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.9.input_layernorm.weight[0m", shape: (3584,), dtype: float16 68%|βββββββ | 135/199 [01:06<00:18, 3.41it/s] [2024-06-06 23:43:28] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.9.mlp.down_proj.q_weight[0m", shape: (3584, 2368), dtype: uint32 68%|βββββββ | 135/199 [01:06<00:18, 3.41it/s] [2024-06-06 23:43:29] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.9.mlp.down_proj.q_scale[0m", shape: (3584, 592), dtype: float16 68%|βββββββ | 135/199 [01:06<00:18, 3.41it/s] 69%|βββββββ | 138/199 [01:06<00:14, 4.25it/s] [2024-06-06 23:43:29] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.9.mlp.gate_up_proj.q_weight[0m", shape: (37888, 448), dtype: uint32 69%|βββββββ | 138/199 [01:07<00:14, 4.25it/s] [2024-06-06 23:43:30] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.9.mlp.gate_up_proj.q_scale[0m", shape: (37888, 112), dtype: float16 69%|βββββββ | 138/199 [01:07<00:14, 4.25it/s] [2024-06-06 23:43:30] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.9.post_attention_layernorm.weight[0m", shape: (3584,), dtype: float16 69%|βββββββ | 138/199 [01:07<00:14, 4.25it/s] 70%|βββββββ | 140/199 [01:07<00:19, 3.08it/s] [2024-06-06 23:43:30] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.9.self_attn.c_attn.bias[0m", shape: (4608,), dtype: float16 70%|βββββββ | 140/199 [01:07<00:19, 3.08it/s] [2024-06-06 23:43:30] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.9.self_attn.c_attn.q_weight[0m", shape: (4608, 448), dtype: uint32 70%|βββββββ | 140/199 [01:08<00:19, 3.08it/s] [2024-06-06 23:43:30] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.9.self_attn.c_attn.q_scale[0m", shape: (4608, 112), dtype: float16 70%|βββββββ | 140/199 [01:08<00:19, 3.08it/s] 71%|ββββββββ | 142/199 [01:08<00:14, 3.92it/s] [2024-06-06 23:43:30] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.9.self_attn.o_proj.q_weight[0m", shape: (3584, 448), dtype: uint32 71%|ββββββββ | 142/199 [01:08<00:14, 3.92it/s] [2024-06-06 23:43:30] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.9.self_attn.o_proj.q_scale[0m", shape: (3584, 112), dtype: float16 71%|ββββββββ | 142/199 [01:08<00:14, 3.92it/s] [2024-06-06 23:43:30] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.14.input_layernorm.weight[0m", shape: (3584,), dtype: float16 71%|ββββββββ | 142/199 [01:08<00:14, 3.92it/s] [2024-06-06 23:43:30] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.14.mlp.down_proj.q_weight[0m", shape: (3584, 2368), dtype: uint32 71%|ββββββββ | 142/199 [01:08<00:14, 3.92it/s] [2024-06-06 23:43:30] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.14.mlp.down_proj.q_scale[0m", shape: (3584, 592), dtype: float16 71%|ββββββββ | 142/199 [01:08<00:14, 3.92it/s] 73%|ββββββββ | 145/199 [01:08<00:11, 4.77it/s] [2024-06-06 23:43:30] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.14.post_attention_layernorm.weight[0m", shape: (3584,), dtype: float16 73%|ββββββββ | 145/199 [01:08<00:11, 4.77it/s] [2024-06-06 23:43:30] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.15.input_layernorm.weight[0m", shape: (3584,), dtype: float16 73%|ββββββββ | 145/199 [01:08<00:11, 4.77it/s] [2024-06-06 23:43:30] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.15.mlp.down_proj.q_weight[0m", shape: (3584, 2368), dtype: uint32 73%|ββββββββ | 145/199 [01:08<00:11, 4.77it/s] [2024-06-06 23:43:31] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.15.mlp.down_proj.q_scale[0m", shape: (3584, 592), dtype: float16 73%|ββββββββ | 145/199 [01:08<00:11, 4.77it/s] 74%|ββββββββ | 148/199 [01:08<00:08, 5.68it/s] [2024-06-06 23:43:32] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.15.mlp.gate_up_proj.q_weight[0m", shape: (37888, 448), dtype: uint32 74%|ββββββββ | 148/199 [01:10<00:08, 5.68it/s] [2024-06-06 23:43:32] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.15.mlp.gate_up_proj.q_scale[0m", shape: (37888, 112), dtype: float16 74%|ββββββββ | 148/199 [01:10<00:08, 5.68it/s] 75%|ββββββββ | 149/199 [01:10<00:17, 2.82it/s] [2024-06-06 23:43:32] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.15.post_attention_layernorm.weight[0m", shape: (3584,), dtype: float16 75%|ββββββββ | 149/199 [01:10<00:17, 2.82it/s] [2024-06-06 23:43:32] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.15.self_attn.c_attn.bias[0m", shape: (4608,), dtype: float16 75%|ββββββββ | 149/199 [01:10<00:17, 2.82it/s] [2024-06-06 23:43:32] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.15.self_attn.c_attn.q_weight[0m", shape: (4608, 448), dtype: uint32 75%|ββββββββ | 149/199 [01:10<00:17, 2.82it/s] [2024-06-06 23:43:32] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.15.self_attn.c_attn.q_scale[0m", shape: (4608, 112), dtype: float16 75%|ββββββββ | 149/199 [01:10<00:17, 2.82it/s] 76%|ββββββββ | 152/199 [01:10<00:11, 4.15it/s] [2024-06-06 23:43:32] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.15.self_attn.o_proj.q_weight[0m", shape: (3584, 448), dtype: uint32 76%|ββββββββ | 152/199 [01:10<00:11, 4.15it/s] [2024-06-06 23:43:32] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.15.self_attn.o_proj.q_scale[0m", shape: (3584, 112), dtype: float16 76%|ββββββββ | 152/199 [01:10<00:11, 4.15it/s] [2024-06-06 23:43:32] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.16.input_layernorm.weight[0m", shape: (3584,), dtype: float16 76%|ββββββββ | 152/199 [01:10<00:11, 4.15it/s] [2024-06-06 23:43:32] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.16.mlp.down_proj.q_weight[0m", shape: (3584, 2368), dtype: uint32 76%|ββββββββ | 152/199 [01:10<00:11, 4.15it/s] [2024-06-06 23:43:33] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.16.mlp.down_proj.q_scale[0m", shape: (3584, 592), dtype: float16 76%|ββββββββ | 152/199 [01:10<00:11, 4.15it/s] 78%|ββββββββ | 155/199 [01:10<00:09, 4.84it/s] [2024-06-06 23:43:34] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.16.mlp.gate_up_proj.q_weight[0m", shape: (37888, 448), dtype: uint32 78%|ββββββββ | 155/199 [01:12<00:09, 4.84it/s] [2024-06-06 23:43:34] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.16.mlp.gate_up_proj.q_scale[0m", shape: (37888, 112), dtype: float16 78%|ββββββββ | 155/199 [01:12<00:09, 4.84it/s] 78%|ββββββββ | 156/199 [01:12<00:17, 2.52it/s] [2024-06-06 23:43:34] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.16.post_attention_layernorm.weight[0m", shape: (3584,), dtype: float16 78%|ββββββββ | 156/199 [01:12<00:17, 2.52it/s] [2024-06-06 23:43:34] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.16.self_attn.c_attn.bias[0m", shape: (4608,), dtype: float16 78%|ββββββββ | 156/199 [01:12<00:17, 2.52it/s] [2024-06-06 23:43:34] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.16.self_attn.c_attn.q_weight[0m", shape: (4608, 448), dtype: uint32 78%|ββββββββ | 156/199 [01:12<00:17, 2.52it/s] [2024-06-06 23:43:34] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.16.self_attn.c_attn.q_scale[0m", shape: (4608, 112), dtype: float16 78%|ββββββββ | 156/199 [01:12<00:17, 2.52it/s] 80%|ββββββββ | 159/199 [01:12<00:10, 3.79it/s] [2024-06-06 23:43:34] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.16.self_attn.o_proj.q_weight[0m", shape: (3584, 448), dtype: uint32 80%|ββββββββ | 159/199 [01:12<00:10, 3.79it/s] [2024-06-06 23:43:34] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.16.self_attn.o_proj.q_scale[0m", shape: (3584, 112), dtype: float16 80%|ββββββββ | 159/199 [01:12<00:10, 3.79it/s] [2024-06-06 23:43:34] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.17.input_layernorm.weight[0m", shape: (3584,), dtype: float16 80%|ββββββββ | 159/199 [01:12<00:10, 3.79it/s] [2024-06-06 23:43:35] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.17.mlp.down_proj.q_weight[0m", shape: (3584, 2368), dtype: uint32 80%|ββββββββ | 159/199 [01:12<00:10, 3.79it/s] [2024-06-06 23:43:35] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.17.mlp.down_proj.q_scale[0m", shape: (3584, 592), dtype: float16 80%|ββββββββ | 159/199 [01:12<00:10, 3.79it/s] 81%|βββββββββ | 162/199 [01:12<00:08, 4.60it/s] [2024-06-06 23:43:36] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.17.mlp.gate_up_proj.q_weight[0m", shape: (37888, 448), dtype: uint32 81%|βββββββββ | 162/199 [01:13<00:08, 4.60it/s] [2024-06-06 23:43:36] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.17.mlp.gate_up_proj.q_scale[0m", shape: (37888, 112), dtype: float16 81%|βββββββββ | 162/199 [01:14<00:08, 4.60it/s] [2024-06-06 23:43:36] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.17.post_attention_layernorm.weight[0m", shape: (3584,), dtype: float16 81%|βββββββββ | 162/199 [01:14<00:08, 4.60it/s] 82%|βββββββββ | 164/199 [01:14<00:11, 3.14it/s] [2024-06-06 23:43:36] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.17.self_attn.c_attn.bias[0m", shape: (4608,), dtype: float16 82%|βββββββββ | 164/199 [01:14<00:11, 3.14it/s] [2024-06-06 23:43:36] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.17.self_attn.c_attn.q_weight[0m", shape: (4608, 448), dtype: uint32 82%|βββββββββ | 164/199 [01:14<00:11, 3.14it/s] [2024-06-06 23:43:36] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.17.self_attn.c_attn.q_scale[0m", shape: (4608, 112), dtype: float16 82%|βββββββββ | 164/199 [01:14<00:11, 3.14it/s] 83%|βββββββββ | 166/199 [01:14<00:08, 3.96it/s] [2024-06-06 23:43:36] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.17.self_attn.o_proj.q_weight[0m", shape: (3584, 448), dtype: uint32 83%|βββββββββ | 166/199 [01:14<00:08, 3.96it/s] [2024-06-06 23:43:36] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.17.self_attn.o_proj.q_scale[0m", shape: (3584, 112), dtype: float16 83%|βββββββββ | 166/199 [01:14<00:08, 3.96it/s] [2024-06-06 23:43:36] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.18.input_layernorm.weight[0m", shape: (3584,), dtype: float16 83%|βββββββββ | 166/199 [01:14<00:08, 3.96it/s] [2024-06-06 23:43:36] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.18.mlp.down_proj.q_weight[0m", shape: (3584, 2368), dtype: uint32 83%|βββββββββ | 166/199 [01:14<00:08, 3.96it/s] [2024-06-06 23:43:36] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.18.mlp.down_proj.q_scale[0m", shape: (3584, 592), dtype: float16 83%|βββββββββ | 166/199 [01:14<00:08, 3.96it/s] 85%|βββββββββ | 169/199 [01:14<00:06, 4.77it/s] [2024-06-06 23:43:37] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.18.mlp.gate_up_proj.q_weight[0m", shape: (37888, 448), dtype: uint32 85%|βββββββββ | 169/199 [01:15<00:06, 4.77it/s] [2024-06-06 23:43:38] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.18.mlp.gate_up_proj.q_scale[0m", shape: (37888, 112), dtype: float16 85%|βββββββββ | 169/199 [01:15<00:06, 4.77it/s] 85%|βββββββββ | 170/199 [01:15<00:10, 2.81it/s] [2024-06-06 23:43:38] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.18.post_attention_layernorm.weight[0m", shape: (3584,), dtype: float16 85%|βββββββββ | 170/199 [01:15<00:10, 2.81it/s] [2024-06-06 23:43:38] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.18.self_attn.c_attn.bias[0m", shape: (4608,), dtype: float16 85%|βββββββββ | 170/199 [01:15<00:10, 2.81it/s] [2024-06-06 23:43:38] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.18.self_attn.c_attn.q_weight[0m", shape: (4608, 448), dtype: uint32 85%|βββββββββ | 170/199 [01:16<00:10, 2.81it/s] [2024-06-06 23:43:38] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.18.self_attn.c_attn.q_scale[0m", shape: (4608, 112), dtype: float16 85%|βββββββββ | 170/199 [01:16<00:10, 2.81it/s] 87%|βββββββββ | 173/199 [01:16<00:06, 4.23it/s] [2024-06-06 23:43:38] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.18.self_attn.o_proj.q_weight[0m", shape: (3584, 448), dtype: uint32 87%|βββββββββ | 173/199 [01:16<00:06, 4.23it/s] [2024-06-06 23:43:38] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.18.self_attn.o_proj.q_scale[0m", shape: (3584, 112), dtype: float16 87%|βββββββββ | 173/199 [01:16<00:06, 4.23it/s] [2024-06-06 23:43:38] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.19.input_layernorm.weight[0m", shape: (3584,), dtype: float16 87%|βββββββββ | 173/199 [01:16<00:06, 4.23it/s] [2024-06-06 23:43:38] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.19.mlp.down_proj.q_weight[0m", shape: (3584, 2368), dtype: uint32 87%|βββββββββ | 173/199 [01:16<00:06, 4.23it/s] [2024-06-06 23:43:38] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.19.mlp.down_proj.q_scale[0m", shape: (3584, 592), dtype: float16 87%|βββββββββ | 173/199 [01:16<00:06, 4.23it/s] 88%|βββββββββ | 176/199 [01:16<00:04, 4.89it/s] [2024-06-06 23:43:39] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.19.mlp.gate_up_proj.q_weight[0m", shape: (37888, 448), dtype: uint32 88%|βββββββββ | 176/199 [01:17<00:04, 4.89it/s] [2024-06-06 23:43:39] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.19.mlp.gate_up_proj.q_scale[0m", shape: (37888, 112), dtype: float16 88%|βββββββββ | 176/199 [01:17<00:04, 4.89it/s] 89%|βββββββββ | 177/199 [01:17<00:07, 2.85it/s] [2024-06-06 23:43:39] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.19.post_attention_layernorm.weight[0m", shape: (3584,), dtype: float16 89%|βββββββββ | 177/199 [01:17<00:07, 2.85it/s] [2024-06-06 23:43:39] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.19.self_attn.c_attn.bias[0m", shape: (4608,), dtype: float16 89%|βββββββββ | 177/199 [01:17<00:07, 2.85it/s] [2024-06-06 23:43:39] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.19.self_attn.c_attn.q_weight[0m", shape: (4608, 448), dtype: uint32 89%|βββββββββ | 177/199 [01:17<00:07, 2.85it/s] [2024-06-06 23:43:40] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.19.self_attn.c_attn.q_scale[0m", shape: (4608, 112), dtype: float16 89%|βββββββββ | 177/199 [01:17<00:07, 2.85it/s] 90%|βββββββββ | 180/199 [01:17<00:04, 4.27it/s] [2024-06-06 23:43:40] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.19.self_attn.o_proj.q_weight[0m", shape: (3584, 448), dtype: uint32 90%|βββββββββ | 180/199 [01:17<00:04, 4.27it/s] [2024-06-06 23:43:40] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.19.self_attn.o_proj.q_scale[0m", shape: (3584, 112), dtype: float16 90%|βββββββββ | 180/199 [01:17<00:04, 4.27it/s] [2024-06-06 23:43:40] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.20.input_layernorm.weight[0m", shape: (3584,), dtype: float16 90%|βββββββββ | 180/199 [01:17<00:04, 4.27it/s] [2024-06-06 23:43:40] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.20.mlp.down_proj.q_weight[0m", shape: (3584, 2368), dtype: uint32 90%|βββββββββ | 180/199 [01:18<00:04, 4.27it/s] [2024-06-06 23:43:40] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.20.mlp.down_proj.q_scale[0m", shape: (3584, 592), dtype: float16 90%|βββββββββ | 180/199 [01:18<00:04, 4.27it/s] 92%|ββββββββββ| 183/199 [01:18<00:03, 5.05it/s] [2024-06-06 23:43:41] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.20.mlp.gate_up_proj.q_weight[0m", shape: (37888, 448), dtype: uint32 92%|ββββββββββ| 183/199 [01:19<00:03, 5.05it/s] [2024-06-06 23:43:41] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.20.mlp.gate_up_proj.q_scale[0m", shape: (37888, 112), dtype: float16 92%|ββββββββββ| 183/199 [01:19<00:03, 5.05it/s] [2024-06-06 23:43:41] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.20.post_attention_layernorm.weight[0m", shape: (3584,), dtype: float16 92%|ββββββββββ| 183/199 [01:19<00:03, 5.05it/s] 93%|ββββββββββ| 185/199 [01:19<00:03, 3.56it/s] [2024-06-06 23:43:41] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.20.self_attn.c_attn.bias[0m", shape: (4608,), dtype: float16 93%|ββββββββββ| 185/199 [01:19<00:03, 3.56it/s] [2024-06-06 23:43:41] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.20.self_attn.c_attn.q_weight[0m", shape: (4608, 448), dtype: uint32 93%|ββββββββββ| 185/199 [01:19<00:03, 3.56it/s] [2024-06-06 23:43:41] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.20.self_attn.c_attn.q_scale[0m", shape: (4608, 112), dtype: float16 93%|ββββββββββ| 185/199 [01:19<00:03, 3.56it/s] 94%|ββββββββββ| 187/199 [01:19<00:02, 4.45it/s] [2024-06-06 23:43:41] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.20.self_attn.o_proj.q_weight[0m", shape: (3584, 448), dtype: uint32 94%|ββββββββββ| 187/199 [01:19<00:02, 4.45it/s] [2024-06-06 23:43:41] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.20.self_attn.o_proj.q_scale[0m", shape: (3584, 112), dtype: float16 94%|ββββββββββ| 187/199 [01:19<00:02, 4.45it/s] [2024-06-06 23:43:41] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.21.input_layernorm.weight[0m", shape: (3584,), dtype: float16 94%|ββββββββββ| 187/199 [01:19<00:02, 4.45it/s] [2024-06-06 23:43:41] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.21.mlp.down_proj.q_weight[0m", shape: (3584, 2368), dtype: uint32 94%|ββββββββββ| 187/199 [01:19<00:02, 4.45it/s] [2024-06-06 23:43:42] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.21.mlp.down_proj.q_scale[0m", shape: (3584, 592), dtype: float16 94%|ββββββββββ| 187/199 [01:19<00:02, 4.45it/s] 95%|ββββββββββ| 190/199 [01:19<00:01, 5.24it/s] [2024-06-06 23:43:42] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.21.mlp.gate_up_proj.q_weight[0m", shape: (37888, 448), dtype: uint32 95%|ββββββββββ| 190/199 [01:20<00:01, 5.24it/s] [2024-06-06 23:43:43] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.21.mlp.gate_up_proj.q_scale[0m", shape: (37888, 112), dtype: float16 95%|ββββββββββ| 190/199 [01:20<00:01, 5.24it/s] 96%|ββββββββββ| 191/199 [01:20<00:02, 3.01it/s] [2024-06-06 23:43:43] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.21.post_attention_layernorm.weight[0m", shape: (3584,), dtype: float16 96%|ββββββββββ| 191/199 [01:20<00:02, 3.01it/s] [2024-06-06 23:43:43] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.21.self_attn.c_attn.bias[0m", shape: (4608,), dtype: float16 96%|ββββββββββ| 191/199 [01:20<00:02, 3.01it/s] [2024-06-06 23:43:43] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.21.self_attn.c_attn.q_weight[0m", shape: (4608, 448), dtype: uint32 96%|ββββββββββ| 191/199 [01:21<00:02, 3.01it/s] [2024-06-06 23:43:43] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.21.self_attn.c_attn.q_scale[0m", shape: (4608, 112), dtype: float16 96%|ββββββββββ| 191/199 [01:21<00:02, 3.01it/s] 97%|ββββββββββ| 194/199 [01:21<00:01, 4.51it/s] [2024-06-06 23:43:43] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.21.self_attn.o_proj.q_weight[0m", shape: (3584, 448), dtype: uint32 97%|ββββββββββ| 194/199 [01:21<00:01, 4.51it/s] [2024-06-06 23:43:43] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.21.self_attn.o_proj.q_scale[0m", shape: (3584, 112), dtype: float16 97%|ββββββββββ| 194/199 [01:21<00:01, 4.51it/s] [2024-06-06 23:43:44] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.22.mlp.gate_up_proj.q_weight[0m", shape: (37888, 448), dtype: uint32 97%|ββββββββββ| 194/199 [01:22<00:01, 4.51it/s] [2024-06-06 23:43:44] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.22.mlp.gate_up_proj.q_scale[0m", shape: (37888, 112), dtype: float16 97%|ββββββββββ| 194/199 [01:22<00:01, 4.51it/s] 98%|ββββββββββ| 196/199 [01:22<00:01, 2.95it/s] [2024-06-06 23:43:44] INFO huggingface_loader.py:175: [Not quantized] Parameter: "[1mmodel.layers.22.self_attn.c_attn.bias[0m", shape: (4608,), dtype: float16 98%|ββββββββββ| 196/199 [01:22<00:01, 2.95it/s] [2024-06-06 23:43:44] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.22.self_attn.c_attn.q_weight[0m", shape: (4608, 448), dtype: uint32 98%|ββββββββββ| 196/199 [01:22<00:01, 2.95it/s] [2024-06-06 23:43:44] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.22.self_attn.c_attn.q_scale[0m", shape: (4608, 112), dtype: float16 98%|ββββββββββ| 196/199 [01:22<00:01, 2.95it/s] 99%|ββββββββββ| 198/199 [01:22<00:00, 3.78it/s] [2024-06-06 23:43:44] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.22.self_attn.o_proj.q_weight[0m", shape: (3584, 448), dtype: uint32 99%|ββββββββββ| 198/199 [01:22<00:00, 3.78it/s] [2024-06-06 23:43:44] INFO huggingface_loader.py:167: [Quantized] Parameter: "[1mmodel.layers.22.self_attn.o_proj.q_scale[0m", shape: (3584, 112), dtype: float16 99%|ββββββββββ| 198/199 [01:22<00:00, 3.78it/s] 100%|ββββββββββ| 199/199 [01:22<00:00, 2.41it/s] [2024-06-06 23:43:44] INFO huggingface_loader.py:197: Unloading HF weight file: /models/Qwen2-7B-Instruct/model-00002-of-00004.safetensors [2024-06-06 23:43:45] INFO huggingface_loader.py:197: Unloading HF weight file: /models/Qwen2-7B-Instruct/model-00003-of-00004.safetensors [2024-06-06 23:43:45] INFO stats.py:77: [92mTime usage[0m: HF loading: 17.289 sec; Pre-quantization mapping: 48.744 sec; Quantization: 3.234 sec [2024-06-06 23:43:45] INFO stats.py:91: [92mRAM usage[0m: Peak RAM: 14.397 GB. Total bytes loaded from disk: 28.370 GB [2024-06-06 23:43:45] INFO convert_weight.py:155: [92mParameter size[0m after quantization: 3.990 GB [2024-06-06 23:43:45] INFO convert_weight.py:160: [92mTotal parameters[0m: 7,615,616,512 [2024-06-06 23:43:45] INFO convert_weight.py:161: [92mBits per parameter[0m: 4.501 [2024-06-06 23:43:45] INFO convert_weight.py:166: Saved to directory: [1m/models/mlc-delivery/hf/mlc-ai/Qwen2-7B-Instruct-q4f16_1-MLC[0m All finished, 88 total shards committed, record saved to /models/mlc-delivery/hf/mlc-ai/Qwen2-7B-Instruct-q4f16_1-MLC/ndarray-cache.json |