/opt/conda/envs/py310/bin/python -m mlc_llm gen_config /models/Meta-Llama-3-8B-Instruct --quantization q0f16 --conv-template llama-3 --output /models/mlc-delivery/hf/mlc-ai/Llama-3-8B-Instruct-q0f16-MLC
[2024-06-04 03:12:21] INFO auto_config.py:116: Found model configuration: /models/Meta-Llama-3-8B-Instruct/config.json
[2024-06-04 03:12:21] INFO auto_config.py:154: Found model type: llama. Use `--model-type` to override.
[2024-06-04 03:12:21] INFO llama_model.py:52: context_window_size not found in config.json. Falling back to max_position_embeddings (8192)
[2024-06-04 03:12:21] INFO llama_model.py:72: prefill_chunk_size defaults to 2048
[2024-06-04 03:12:21] INFO config.py:107: Overriding max_batch_size from 1 to 80
[2024-06-04 03:12:21] INFO gen_config.py:143: [generation_config.json] Setting bos_token_id: 128000
[2024-06-04 03:12:21] INFO gen_config.py:143: [generation_config.json] Setting eos_token_id: [128001, 128009]
[2024-06-04 03:12:21] INFO gen_config.py:143: [generation_config.json] Setting temperature: 0.6
[2024-06-04 03:12:21] INFO gen_config.py:143: [generation_config.json] Setting top_p: 0.9
[2024-06-04 03:12:21] INFO gen_config.py:157: Not found tokenizer config: /models/Meta-Llama-3-8B-Instruct/tokenizer.model
[2024-06-04 03:12:21] INFO gen_config.py:155: Found tokenizer config: /models/Meta-Llama-3-8B-Instruct/tokenizer.json. Copying to /models/mlc-delivery/hf/mlc-ai/Llama-3-8B-Instruct-q0f16-MLC/tokenizer.json
[2024-06-04 03:12:21] INFO gen_config.py:157: Not found tokenizer config: /models/Meta-Llama-3-8B-Instruct/vocab.json
[2024-06-04 03:12:21] INFO gen_config.py:157: Not found tokenizer config: /models/Meta-Llama-3-8B-Instruct/merges.txt
[2024-06-04 03:12:21] INFO gen_config.py:157: Not found tokenizer config: /models/Meta-Llama-3-8B-Instruct/added_tokens.json
[2024-06-04 03:12:21] INFO gen_config.py:155: Found tokenizer config: /models/Meta-Llama-3-8B-Instruct/tokenizer_config.json. Copying to /models/mlc-delivery/hf/mlc-ai/Llama-3-8B-Instruct-q0f16-MLC/tokenizer_config.json
[2024-06-04 03:12:21] INFO gen_config.py:216: Detected tokenizer info: {'token_postproc_method': 'byte_level', 'prepend_space_in_encode': False, 'strip_space_in_decode': False}
[2024-06-04 03:12:21] INFO gen_config.py:32: [System default] Setting pad_token_id: 0
[2024-06-04 03:12:21] INFO gen_config.py:32: [System default] Setting presence_penalty: 0.0
[2024-06-04 03:12:21] INFO gen_config.py:32: [System default] Setting frequency_penalty: 0.0
[2024-06-04 03:12:21] INFO gen_config.py:32: [System default] Setting repetition_penalty: 1.0
[2024-06-04 03:12:21] INFO gen_config.py:32: [System default] Setting mean_gen_len: 128
[2024-06-04 03:12:21] INFO gen_config.py:32: [System default] Setting max_gen_len: 512
[2024-06-04 03:12:21] INFO gen_config.py:32: [System default] Setting shift_fill_factor: 0.3
[2024-06-04 03:12:21] INFO gen_config.py:223: Dumping configuration file to: /models/mlc-delivery/hf/mlc-ai/Llama-3-8B-Instruct-q0f16-MLC/mlc-chat-config.json
/opt/conda/envs/py310/bin/python -m mlc_llm convert_weight /models/Meta-Llama-3-8B-Instruct --quantization q0f16 --output /models/mlc-delivery/hf/mlc-ai/Llama-3-8B-Instruct-q0f16-MLC
[2024-06-04 03:12:23] INFO auto_config.py:116: Found model configuration: /models/Meta-Llama-3-8B-Instruct/config.json
[2024-06-04 03:12:24] INFO auto_device.py:79: Found device: cuda:0
[2024-06-04 03:12:26] INFO auto_device.py:88: Not found device: rocm:0
[2024-06-04 03:12:27] INFO auto_device.py:88: Not found device: metal:0
[2024-06-04 03:12:29] INFO auto_device.py:79: Found device: vulkan:0
[2024-06-04 03:12:29] INFO auto_device.py:79: Found device: vulkan:1
[2024-06-04 03:12:29] INFO auto_device.py:79: Found device: vulkan:2
[2024-06-04 03:12:29] INFO auto_device.py:79: Found device: vulkan:3
[2024-06-04 03:12:30] INFO auto_device.py:88: Not found device: opencl:0
[2024-06-04 03:12:30] INFO auto_device.py:35: Using device: cuda:0
[2024-06-04 03:12:30] INFO auto_weight.py:71: Finding weights in: /models/Meta-Llama-3-8B-Instruct
[2024-06-04 03:12:30] INFO auto_weight.py:137: Not found Huggingface PyTorch
[2024-06-04 03:12:30] INFO auto_weight.py:144: Found source weight format: huggingface-safetensor. Source configuration: /models/Meta-Llama-3-8B-Instruct/model.safetensors.index.json
[2024-06-04 03:12:30] INFO auto_weight.py:107: Using source weight configuration: /models/Meta-Llama-3-8B-Instruct/model.safetensors.index.json. Use `--source` to override.
[2024-06-04 03:12:30] INFO auto_weight.py:111: Using source weight format: huggingface-safetensor. Use `--source-format` to override.
[2024-06-04 03:12:30] INFO auto_config.py:154: Found model type: llama. Use `--model-type` to override.
[2024-06-04 03:12:30] INFO llama_model.py:52: context_window_size not found in config.json. Falling back to max_position_embeddings (8192)
[2024-06-04 03:12:30] INFO llama_model.py:72: prefill_chunk_size defaults to 2048
Weight conversion with arguments:
  --config          /models/Meta-Llama-3-8B-Instruct/config.json
  --quantization    NoQuantize(name='q0f16', kind='no-quant', model_dtype='float16')
  --model-type      llama
  --device          cuda:0
  --source          /models/Meta-Llama-3-8B-Instruct/model.safetensors.index.json
  --source-format   huggingface-safetensor
  --output          /models/mlc-delivery/hf/mlc-ai/Llama-3-8B-Instruct-q0f16-MLC
Start storing to cache /models/mlc-delivery/hf/mlc-ai/Llama-3-8B-Instruct-q0f16-MLC

  0%|          | 0/195 [00:00<?, ?it/s]
                                       
[2024-06-04 03:12:32] INFO huggingface_loader.py:185: Loading HF parameters from: /models/Meta-Llama-3-8B-Instruct/model-00004-of-00004.safetensors

  0%|          | 0/195 [00:00<?, ?it/s]
                                       
[2024-06-04 03:12:36] INFO huggingface_loader.py:175: [Not quantized] Parameter: "lm_head.weight", shape: (128256, 4096), dtype: float16

  0%|          | 0/195 [00:03<?, ?it/s]
  1%|          | 1/195 [00:06<20:46,  6.43s/it]
                                               
[2024-06-04 03:12:39] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.31.input_layernorm.weight", shape: (4096,), dtype: float16

  1%|          | 1/195 [00:06<20:46,  6.43s/it]
  1%|          | 2/195 [00:06<08:44,  2.72s/it]
                                               
[2024-06-04 03:12:39] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.31.mlp.down_proj.weight", shape: (4096, 14336), dtype: float16

  1%|          | 2/195 [00:06<08:44,  2.72s/it]
  2%|▏         | 3/195 [00:07<05:29,  1.72s/it]
                                               
[2024-06-04 03:12:39] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.31.post_attention_layernorm.weight", shape: (4096,), dtype: float16

  2%|▏         | 3/195 [00:07<05:29,  1.72s/it]
                                               
[2024-06-04 03:12:39] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.norm.weight", shape: (4096,), dtype: float16

  2%|▏         | 3/195 [00:07<05:29,  1.72s/it]
                                               
[2024-06-04 03:12:39] INFO huggingface_loader.py:197: Unloading HF weight file: /models/Meta-Llama-3-8B-Instruct/model-00004-of-00004.safetensors

  2%|▏         | 3/195 [00:07<05:29,  1.72s/it]
                                               
[2024-06-04 03:12:39] INFO huggingface_loader.py:185: Loading HF parameters from: /models/Meta-Llama-3-8B-Instruct/model-00001-of-00004.safetensors

  2%|▏         | 3/195 [00:07<05:29,  1.72s/it]
                                               
[2024-06-04 03:12:45] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.embed_tokens.weight", shape: (128256, 4096), dtype: float16

  2%|▏         | 3/195 [00:13<05:29,  1.72s/it]
  3%|▎         | 6/195 [00:16<08:03,  2.56s/it]
                                               
[2024-06-04 03:12:48] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.0.input_layernorm.weight", shape: (4096,), dtype: float16

  3%|▎         | 6/195 [00:16<08:03,  2.56s/it]
                                               
[2024-06-04 03:12:49] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.0.mlp.down_proj.weight", shape: (4096, 14336), dtype: float16

  3%|▎         | 6/195 [00:16<08:03,  2.56s/it]
  4%|▍         | 8/195 [00:16<05:15,  1.69s/it]
                                               
[2024-06-04 03:12:49] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.0.mlp.gate_up_proj.weight", shape: (28672, 4096), dtype: float16

  4%|▍         | 8/195 [00:17<05:15,  1.69s/it]
  5%|▍         | 9/195 [00:17<04:50,  1.56s/it]
                                               
[2024-06-04 03:12:50] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.0.post_attention_layernorm.weight", shape: (4096,), dtype: float16

  5%|▍         | 9/195 [00:17<04:50,  1.56s/it]
                                               
[2024-06-04 03:12:50] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.0.self_attn.qkv_proj.weight", shape: (6144, 4096), dtype: float16

  5%|▍         | 9/195 [00:17<04:50,  1.56s/it]
  6%|▌         | 11/195 [00:18<03:06,  1.01s/it]
                                                
[2024-06-04 03:12:50] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.0.self_attn.o_proj.weight", shape: (4096, 4096), dtype: float16

  6%|▌         | 11/195 [00:18<03:06,  1.01s/it]
  6%|▌         | 12/195 [00:18<02:31,  1.20it/s]
                                                
[2024-06-04 03:12:50] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.1.input_layernorm.weight", shape: (4096,), dtype: float16

  6%|▌         | 12/195 [00:18<02:31,  1.20it/s]
                                                
[2024-06-04 03:12:51] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.1.mlp.down_proj.weight", shape: (4096, 14336), dtype: float16

  6%|▌         | 12/195 [00:18<02:31,  1.20it/s]
  7%|▋         | 14/195 [00:18<01:51,  1.63it/s]
                                                
[2024-06-04 03:12:51] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.1.mlp.gate_up_proj.weight", shape: (28672, 4096), dtype: float16

  7%|▋         | 14/195 [00:19<01:51,  1.63it/s]
  8%|▊         | 15/195 [00:19<02:09,  1.39it/s]
                                                
[2024-06-04 03:12:52] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.1.post_attention_layernorm.weight", shape: (4096,), dtype: float16

  8%|▊         | 15/195 [00:19<02:09,  1.39it/s]
                                                
[2024-06-04 03:12:52] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.1.self_attn.qkv_proj.weight", shape: (6144, 4096), dtype: float16

  8%|▊         | 15/195 [00:19<02:09,  1.39it/s]
  9%|▊         | 17/195 [00:20<01:27,  2.02it/s]
                                                
[2024-06-04 03:12:52] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.1.self_attn.o_proj.weight", shape: (4096, 4096), dtype: float16

  9%|▊         | 17/195 [00:20<01:27,  2.02it/s]
  9%|▉         | 18/195 [00:20<01:14,  2.37it/s]
                                                
[2024-06-04 03:12:52] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.2.input_layernorm.weight", shape: (4096,), dtype: float16

  9%|▉         | 18/195 [00:20<01:14,  2.37it/s]
                                                
[2024-06-04 03:12:53] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.2.mlp.down_proj.weight", shape: (4096, 14336), dtype: float16

  9%|▉         | 18/195 [00:20<01:14,  2.37it/s]
 10%|█         | 20/195 [00:20<01:03,  2.77it/s]
                                                
[2024-06-04 03:12:53] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.2.mlp.gate_up_proj.weight", shape: (28672, 4096), dtype: float16

 10%|█         | 20/195 [00:21<01:03,  2.77it/s]
 11%|█         | 21/195 [00:21<01:30,  1.93it/s]
                                                
[2024-06-04 03:12:54] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.2.post_attention_layernorm.weight", shape: (4096,), dtype: float16

 11%|█         | 21/195 [00:21<01:30,  1.93it/s]
                                                
[2024-06-04 03:12:54] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.2.self_attn.qkv_proj.weight", shape: (6144, 4096), dtype: float16

 11%|█         | 21/195 [00:21<01:30,  1.93it/s]
 12%|█▏        | 23/195 [00:22<01:03,  2.70it/s]
                                                
[2024-06-04 03:12:54] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.2.self_attn.o_proj.weight", shape: (4096, 4096), dtype: float16

 12%|█▏        | 23/195 [00:22<01:03,  2.70it/s]
 12%|█▏        | 24/195 [00:22<00:55,  3.08it/s]
                                                
[2024-06-04 03:12:54] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.3.input_layernorm.weight", shape: (4096,), dtype: float16

 12%|█▏        | 24/195 [00:22<00:55,  3.08it/s]
                                                
[2024-06-04 03:12:55] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.3.mlp.down_proj.weight", shape: (4096, 14336), dtype: float16

 12%|█▏        | 24/195 [00:22<00:55,  3.08it/s]
 13%|█▎        | 26/195 [00:22<00:50,  3.33it/s]
                                                
[2024-06-04 03:12:55] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.3.mlp.gate_up_proj.weight", shape: (28672, 4096), dtype: float16

 13%|█▎        | 26/195 [00:23<00:50,  3.33it/s]
 14%|█▍        | 27/195 [00:23<01:19,  2.12it/s]
                                                
[2024-06-04 03:12:56] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.3.post_attention_layernorm.weight", shape: (4096,), dtype: float16

 14%|█▍        | 27/195 [00:23<01:19,  2.12it/s]
                                                
[2024-06-04 03:12:56] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.3.self_attn.qkv_proj.weight", shape: (6144, 4096), dtype: float16

 14%|█▍        | 27/195 [00:23<01:19,  2.12it/s]
 15%|█▍        | 29/195 [00:24<00:56,  2.92it/s]
                                                
[2024-06-04 03:12:56] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.3.self_attn.o_proj.weight", shape: (4096, 4096), dtype: float16

 15%|█▍        | 29/195 [00:24<00:56,  2.92it/s]
 15%|█▌        | 30/195 [00:24<00:49,  3.31it/s]
                                                
[2024-06-04 03:12:56] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.4.input_layernorm.weight", shape: (4096,), dtype: float16

 15%|█▌        | 30/195 [00:24<00:49,  3.31it/s]
                                                
[2024-06-04 03:12:57] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.4.mlp.down_proj.weight", shape: (4096, 14336), dtype: float16

 15%|█▌        | 30/195 [00:24<00:49,  3.31it/s]
 16%|█▋        | 32/195 [00:24<00:46,  3.49it/s]
                                                
[2024-06-04 03:12:57] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.4.mlp.gate_up_proj.weight", shape: (28672, 4096), dtype: float16

 16%|█▋        | 32/195 [00:25<00:46,  3.49it/s]
 17%|█▋        | 33/195 [00:25<01:14,  2.18it/s]
                                                
[2024-06-04 03:12:58] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.4.post_attention_layernorm.weight", shape: (4096,), dtype: float16

 17%|█▋        | 33/195 [00:25<01:14,  2.18it/s]
                                                
[2024-06-04 03:12:58] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.4.self_attn.qkv_proj.weight", shape: (6144, 4096), dtype: float16

 17%|█▋        | 33/195 [00:25<01:14,  2.18it/s]
 18%|█▊        | 35/195 [00:26<00:53,  2.99it/s]
                                                
[2024-06-04 03:12:58] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.4.self_attn.o_proj.weight", shape: (4096, 4096), dtype: float16

 18%|█▊        | 35/195 [00:26<00:53,  2.99it/s]
 18%|█▊        | 36/195 [00:26<00:47,  3.37it/s]
                                                
[2024-06-04 03:12:58] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.5.input_layernorm.weight", shape: (4096,), dtype: float16

 18%|█▊        | 36/195 [00:26<00:47,  3.37it/s]
                                                
[2024-06-04 03:12:59] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.5.mlp.down_proj.weight", shape: (4096, 14336), dtype: float16

 18%|█▊        | 36/195 [00:26<00:47,  3.37it/s]
 19%|█▉        | 38/195 [00:26<00:44,  3.51it/s]
                                                
[2024-06-04 03:13:00] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.5.mlp.gate_up_proj.weight", shape: (28672, 4096), dtype: float16

 19%|█▉        | 38/195 [00:27<00:44,  3.51it/s]
 20%|██        | 39/195 [00:27<01:12,  2.15it/s]
                                                
[2024-06-04 03:13:00] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.5.post_attention_layernorm.weight", shape: (4096,), dtype: float16

 20%|██        | 39/195 [00:27<01:12,  2.15it/s]
                                                
[2024-06-04 03:13:00] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.5.self_attn.qkv_proj.weight", shape: (6144, 4096), dtype: float16

 20%|██        | 39/195 [00:28<01:12,  2.15it/s]
 21%|██        | 41/195 [00:28<00:52,  2.94it/s]
                                                
[2024-06-04 03:13:00] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.5.self_attn.o_proj.weight", shape: (4096, 4096), dtype: float16

 21%|██        | 41/195 [00:28<00:52,  2.94it/s]
 22%|██▏       | 42/195 [00:28<00:45,  3.33it/s]
                                                
[2024-06-04 03:13:01] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.6.input_layernorm.weight", shape: (4096,), dtype: float16

 22%|██▏       | 42/195 [00:28<00:45,  3.33it/s]
                                                
[2024-06-04 03:13:01] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.6.mlp.down_proj.weight", shape: (4096, 14336), dtype: float16

 22%|██▏       | 42/195 [00:28<00:45,  3.33it/s]
 23%|██▎       | 44/195 [00:28<00:42,  3.52it/s]
                                                
[2024-06-04 03:13:02] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.6.mlp.gate_up_proj.weight", shape: (28672, 4096), dtype: float16

 23%|██▎       | 44/195 [00:29<00:42,  3.52it/s]
 23%|██▎       | 45/195 [00:30<01:09,  2.17it/s]
                                                
[2024-06-04 03:13:02] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.6.post_attention_layernorm.weight", shape: (4096,), dtype: float16

 23%|██▎       | 45/195 [00:30<01:09,  2.17it/s]
                                                
[2024-06-04 03:13:02] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.6.self_attn.qkv_proj.weight", shape: (6144, 4096), dtype: float16

 23%|██▎       | 45/195 [00:30<01:09,  2.17it/s]
 24%|██▍       | 47/195 [00:30<00:49,  2.99it/s]
                                                
[2024-06-04 03:13:03] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.6.self_attn.o_proj.weight", shape: (4096, 4096), dtype: float16

 24%|██▍       | 47/195 [00:30<00:49,  2.99it/s]
 25%|██▍       | 48/195 [00:30<00:43,  3.36it/s]
                                                
[2024-06-04 03:13:03] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.7.input_layernorm.weight", shape: (4096,), dtype: float16

 25%|██▍       | 48/195 [00:30<00:43,  3.36it/s]
                                                
[2024-06-04 03:13:03] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.7.mlp.down_proj.weight", shape: (4096, 14336), dtype: float16

 25%|██▍       | 48/195 [00:30<00:43,  3.36it/s]
 26%|██▌       | 50/195 [00:30<00:41,  3.46it/s]
                                                
[2024-06-04 03:13:04] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.7.mlp.gate_up_proj.weight", shape: (28672, 4096), dtype: float16

 26%|██▌       | 50/195 [00:31<00:41,  3.46it/s]
 26%|██▌       | 51/195 [00:32<01:06,  2.17it/s]
                                                
[2024-06-04 03:13:04] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.7.post_attention_layernorm.weight", shape: (4096,), dtype: float16

 26%|██▌       | 51/195 [00:32<01:06,  2.17it/s]
                                                
[2024-06-04 03:13:04] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.7.self_attn.qkv_proj.weight", shape: (6144, 4096), dtype: float16

 26%|██▌       | 51/195 [00:32<01:06,  2.17it/s]
 27%|██▋       | 53/195 [00:32<00:47,  2.98it/s]
                                                
[2024-06-04 03:13:05] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.7.self_attn.o_proj.weight", shape: (4096, 4096), dtype: float16

 27%|██▋       | 53/195 [00:32<00:47,  2.98it/s]
 28%|██▊       | 54/195 [00:32<00:41,  3.36it/s]
                                                
[2024-06-04 03:13:05] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.8.input_layernorm.weight", shape: (4096,), dtype: float16

 28%|██▊       | 54/195 [00:32<00:41,  3.36it/s]
                                                
[2024-06-04 03:13:05] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.8.mlp.down_proj.weight", shape: (4096, 14336), dtype: float16

 28%|██▊       | 54/195 [00:32<00:41,  3.36it/s]
 29%|██▊       | 56/195 [00:33<00:39,  3.53it/s]
                                                
[2024-06-04 03:13:06] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.8.mlp.gate_up_proj.weight", shape: (28672, 4096), dtype: float16

 29%|██▊       | 56/195 [00:33<00:39,  3.53it/s]
 29%|██▉       | 57/195 [00:34<01:02,  2.20it/s]
                                                
[2024-06-04 03:13:06] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.8.post_attention_layernorm.weight", shape: (4096,), dtype: float16

 29%|██▉       | 57/195 [00:34<01:02,  2.20it/s]
                                                
[2024-06-04 03:13:06] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.8.self_attn.qkv_proj.weight", shape: (6144, 4096), dtype: float16

 29%|██▉       | 57/195 [00:34<01:02,  2.20it/s]
 30%|███       | 59/195 [00:34<00:45,  3.02it/s]
                                                
[2024-06-04 03:13:07] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.8.self_attn.o_proj.weight", shape: (4096, 4096), dtype: float16

 30%|███       | 59/195 [00:34<00:45,  3.02it/s]
 31%|███       | 60/195 [00:34<00:39,  3.40it/s]
                                                
[2024-06-04 03:13:07] INFO huggingface_loader.py:197: Unloading HF weight file: /models/Meta-Llama-3-8B-Instruct/model-00001-of-00004.safetensors

 31%|███       | 60/195 [00:34<00:39,  3.40it/s]
                                                
[2024-06-04 03:13:07] INFO huggingface_loader.py:185: Loading HF parameters from: /models/Meta-Llama-3-8B-Instruct/model-00002-of-00004.safetensors

 31%|███       | 60/195 [00:35<00:39,  3.40it/s]
                                                
[2024-06-04 03:13:11] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.10.input_layernorm.weight", shape: (4096,), dtype: float16

 31%|███       | 60/195 [00:39<00:39,  3.40it/s]
 31%|███▏      | 61/195 [00:39<02:58,  1.34s/it]
                                                
[2024-06-04 03:13:12] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.10.mlp.down_proj.weight", shape: (4096, 14336), dtype: float16

 31%|███▏      | 61/195 [00:39<02:58,  1.34s/it]
 32%|███▏      | 62/195 [00:39<02:31,  1.14s/it]
                                                
[2024-06-04 03:13:12] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.10.mlp.gate_up_proj.weight", shape: (28672, 4096), dtype: float16

 32%|███▏      | 62/195 [00:40<02:31,  1.14s/it]
 32%|███▏      | 63/195 [00:40<02:29,  1.13s/it]
                                                
[2024-06-04 03:13:13] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.10.post_attention_layernorm.weight", shape: (4096,), dtype: float16

 32%|███▏      | 63/195 [00:40<02:29,  1.13s/it]
                                                
[2024-06-04 03:13:13] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.10.self_attn.qkv_proj.weight", shape: (6144, 4096), dtype: float16

 32%|███▏      | 63/195 [00:41<02:29,  1.13s/it]
 33%|███▎      | 65/195 [00:41<01:31,  1.42it/s]
                                                
[2024-06-04 03:13:13] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.10.self_attn.o_proj.weight", shape: (4096, 4096), dtype: float16

 33%|███▎      | 65/195 [00:41<01:31,  1.42it/s]
 34%|███▍      | 66/195 [00:41<01:14,  1.74it/s]
                                                
[2024-06-04 03:13:14] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.11.input_layernorm.weight", shape: (4096,), dtype: float16

 34%|███▍      | 66/195 [00:41<01:14,  1.74it/s]
                                                
[2024-06-04 03:13:14] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.11.mlp.down_proj.weight", shape: (4096, 14336), dtype: float16

 34%|███▍      | 66/195 [00:41<01:14,  1.74it/s]
 35%|███▍      | 68/195 [00:41<00:56,  2.23it/s]
                                                
[2024-06-04 03:13:15] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.11.mlp.gate_up_proj.weight", shape: (28672, 4096), dtype: float16

 35%|███▍      | 68/195 [00:42<00:56,  2.23it/s]
 35%|███▌      | 69/195 [00:42<01:14,  1.69it/s]
                                                
[2024-06-04 03:13:15] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.11.post_attention_layernorm.weight", shape: (4096,), dtype: float16

 35%|███▌      | 69/195 [00:42<01:14,  1.69it/s]
                                                
[2024-06-04 03:13:15] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.11.self_attn.qkv_proj.weight", shape: (6144, 4096), dtype: float16

 35%|███▌      | 69/195 [00:43<01:14,  1.69it/s]
 36%|███▋      | 71/195 [00:43<00:50,  2.44it/s]
                                                
[2024-06-04 03:13:15] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.11.self_attn.o_proj.weight", shape: (4096, 4096), dtype: float16

 36%|███▋      | 71/195 [00:43<00:50,  2.44it/s]
 37%|███▋      | 72/195 [00:43<00:43,  2.83it/s]
                                                
[2024-06-04 03:13:16] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.12.input_layernorm.weight", shape: (4096,), dtype: float16

 37%|███▋      | 72/195 [00:43<00:43,  2.83it/s]
                                                
[2024-06-04 03:13:16] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.12.mlp.down_proj.weight", shape: (4096, 14336), dtype: float16

 37%|███▋      | 72/195 [00:43<00:43,  2.83it/s]
 38%|███▊      | 74/195 [00:43<00:38,  3.16it/s]
                                                
[2024-06-04 03:13:17] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.12.mlp.gate_up_proj.weight", shape: (28672, 4096), dtype: float16

 38%|███▊      | 74/195 [00:44<00:38,  3.16it/s]
 38%|███▊      | 75/195 [00:44<00:57,  2.08it/s]
                                                
[2024-06-04 03:13:17] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.12.post_attention_layernorm.weight", shape: (4096,), dtype: float16

 38%|███▊      | 75/195 [00:44<00:57,  2.08it/s]
                                                
[2024-06-04 03:13:17] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.12.self_attn.qkv_proj.weight", shape: (6144, 4096), dtype: float16

 38%|███▊      | 75/195 [00:45<00:57,  2.08it/s]
 39%|███▉      | 77/195 [00:45<00:40,  2.90it/s]
                                                
[2024-06-04 03:13:17] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.12.self_attn.o_proj.weight", shape: (4096, 4096), dtype: float16

 39%|███▉      | 77/195 [00:45<00:40,  2.90it/s]
 40%|████      | 78/195 [00:45<00:35,  3.29it/s]
                                                
[2024-06-04 03:13:18] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.13.input_layernorm.weight", shape: (4096,), dtype: float16

 40%|████      | 78/195 [00:45<00:35,  3.29it/s]
                                                
[2024-06-04 03:13:18] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.13.mlp.down_proj.weight", shape: (4096, 14336), dtype: float16

 40%|████      | 78/195 [00:45<00:35,  3.29it/s]
 41%|████      | 80/195 [00:45<00:32,  3.50it/s]
                                                
[2024-06-04 03:13:19] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.13.mlp.gate_up_proj.weight", shape: (28672, 4096), dtype: float16

 41%|████      | 80/195 [00:46<00:32,  3.50it/s]
 42%|████▏     | 81/195 [00:46<00:51,  2.21it/s]
                                                
[2024-06-04 03:13:19] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.13.post_attention_layernorm.weight", shape: (4096,), dtype: float16

 42%|████▏     | 81/195 [00:46<00:51,  2.21it/s]
                                                
[2024-06-04 03:13:19] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.13.self_attn.qkv_proj.weight", shape: (6144, 4096), dtype: float16

 42%|████▏     | 81/195 [00:47<00:51,  2.21it/s]
 43%|████▎     | 83/195 [00:47<00:36,  3.06it/s]
                                                
[2024-06-04 03:13:19] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.13.self_attn.o_proj.weight", shape: (4096, 4096), dtype: float16

 43%|████▎     | 83/195 [00:47<00:36,  3.06it/s]
 43%|████▎     | 84/195 [00:47<00:32,  3.46it/s]
                                                
[2024-06-04 03:13:20] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.14.input_layernorm.weight", shape: (4096,), dtype: float16

 43%|████▎     | 84/195 [00:47<00:32,  3.46it/s]
                                                
[2024-06-04 03:13:20] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.14.mlp.down_proj.weight", shape: (4096, 14336), dtype: float16

 43%|████▎     | 84/195 [00:47<00:32,  3.46it/s]
 44%|████▍     | 86/195 [00:47<00:29,  3.64it/s]
                                                
[2024-06-04 03:13:20] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.14.mlp.gate_up_proj.weight", shape: (28672, 4096), dtype: float16

 44%|████▍     | 86/195 [00:48<00:29,  3.64it/s]
 45%|████▍     | 87/195 [00:48<00:47,  2.26it/s]
                                                
[2024-06-04 03:13:21] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.14.post_attention_layernorm.weight", shape: (4096,), dtype: float16

 45%|████▍     | 87/195 [00:48<00:47,  2.26it/s]
                                                
[2024-06-04 03:13:21] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.14.self_attn.qkv_proj.weight", shape: (6144, 4096), dtype: float16

 45%|████▍     | 87/195 [00:49<00:47,  2.26it/s]
 46%|████▌     | 89/195 [00:49<00:33,  3.12it/s]
                                                
[2024-06-04 03:13:21] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.14.self_attn.o_proj.weight", shape: (4096, 4096), dtype: float16

 46%|████▌     | 89/195 [00:49<00:33,  3.12it/s]
 46%|████▌     | 90/195 [00:49<00:29,  3.52it/s]
                                                
[2024-06-04 03:13:21] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.15.input_layernorm.weight", shape: (4096,), dtype: float16

 46%|████▌     | 90/195 [00:49<00:29,  3.52it/s]
                                                
[2024-06-04 03:13:22] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.15.mlp.down_proj.weight", shape: (4096, 14336), dtype: float16

 46%|████▌     | 90/195 [00:49<00:29,  3.52it/s]
 47%|████▋     | 92/195 [00:49<00:27,  3.68it/s]
                                                
[2024-06-04 03:13:22] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.15.mlp.gate_up_proj.weight", shape: (28672, 4096), dtype: float16

 47%|████▋     | 92/195 [00:50<00:27,  3.68it/s]
 48%|████▊     | 93/195 [00:50<00:44,  2.28it/s]
                                                
[2024-06-04 03:13:23] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.15.post_attention_layernorm.weight", shape: (4096,), dtype: float16

 48%|████▊     | 93/195 [00:50<00:44,  2.28it/s]
                                                
[2024-06-04 03:13:23] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.15.self_attn.qkv_proj.weight", shape: (6144, 4096), dtype: float16

 48%|████▊     | 93/195 [00:50<00:44,  2.28it/s]
 49%|████▊     | 95/195 [00:51<00:31,  3.14it/s]
                                                
[2024-06-04 03:13:23] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.15.self_attn.o_proj.weight", shape: (4096, 4096), dtype: float16

 49%|████▊     | 95/195 [00:51<00:31,  3.14it/s]
 49%|████▉     | 96/195 [00:51<00:27,  3.54it/s]
                                                
[2024-06-04 03:13:23] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.16.input_layernorm.weight", shape: (4096,), dtype: float16

 49%|████▉     | 96/195 [00:51<00:27,  3.54it/s]
                                                
[2024-06-04 03:13:24] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.16.mlp.down_proj.weight", shape: (4096, 14336), dtype: float16

 49%|████▉     | 96/195 [00:51<00:27,  3.54it/s]
 50%|█████     | 98/195 [00:51<00:26,  3.70it/s]
                                                
[2024-06-04 03:13:24] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.16.mlp.gate_up_proj.weight", shape: (28672, 4096), dtype: float16

 50%|█████     | 98/195 [00:52<00:26,  3.70it/s]
 51%|█████     | 99/195 [00:52<00:41,  2.29it/s]
                                                
[2024-06-04 03:13:25] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.16.post_attention_layernorm.weight", shape: (4096,), dtype: float16

 51%|█████     | 99/195 [00:52<00:41,  2.29it/s]
                                                
[2024-06-04 03:13:25] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.16.self_attn.qkv_proj.weight", shape: (6144, 4096), dtype: float16

 51%|█████     | 99/195 [00:52<00:41,  2.29it/s]
 52%|█████▏    | 101/195 [00:53<00:29,  3.16it/s]
                                                 
[2024-06-04 03:13:25] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.16.self_attn.o_proj.weight", shape: (4096, 4096), dtype: float16

 52%|█████▏    | 101/195 [00:53<00:29,  3.16it/s]
 52%|█████▏    | 102/195 [00:53<00:26,  3.57it/s]
                                                 
[2024-06-04 03:13:25] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.17.input_layernorm.weight", shape: (4096,), dtype: float16

 52%|█████▏    | 102/195 [00:53<00:26,  3.57it/s]
                                                 
[2024-06-04 03:13:26] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.17.mlp.down_proj.weight", shape: (4096, 14336), dtype: float16

 52%|█████▏    | 102/195 [00:53<00:26,  3.57it/s]
 53%|█████▎    | 104/195 [00:53<00:24,  3.71it/s]
                                                 
[2024-06-04 03:13:26] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.17.mlp.gate_up_proj.weight", shape: (28672, 4096), dtype: float16

 53%|█████▎    | 104/195 [00:54<00:24,  3.71it/s]
 54%|█████▍    | 105/195 [00:54<00:39,  2.29it/s]
                                                 
[2024-06-04 03:13:27] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.17.post_attention_layernorm.weight", shape: (4096,), dtype: float16

 54%|█████▍    | 105/195 [00:54<00:39,  2.29it/s]
                                                 
[2024-06-04 03:13:27] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.17.self_attn.qkv_proj.weight", shape: (6144, 4096), dtype: float16

 54%|█████▍    | 105/195 [00:54<00:39,  2.29it/s]
 55%|█████▍    | 107/195 [00:54<00:27,  3.15it/s]
                                                 
[2024-06-04 03:13:27] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.17.self_attn.o_proj.weight", shape: (4096, 4096), dtype: float16

 55%|█████▍    | 107/195 [00:55<00:27,  3.15it/s]
 55%|█████▌    | 108/195 [00:55<00:24,  3.57it/s]
                                                 
[2024-06-04 03:13:27] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.18.input_layernorm.weight", shape: (4096,), dtype: float16

 55%|█████▌    | 108/195 [00:55<00:24,  3.57it/s]
                                                 
[2024-06-04 03:13:28] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.18.mlp.down_proj.weight", shape: (4096, 14336), dtype: float16

 55%|█████▌    | 108/195 [00:55<00:24,  3.57it/s]
 56%|█████▋    | 110/195 [00:55<00:22,  3.71it/s]
                                                 
[2024-06-04 03:13:28] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.18.mlp.gate_up_proj.weight", shape: (28672, 4096), dtype: float16

 56%|█████▋    | 110/195 [00:56<00:22,  3.71it/s]
 57%|█████▋    | 111/195 [00:56<00:36,  2.29it/s]
                                                 
[2024-06-04 03:13:29] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.18.post_attention_layernorm.weight", shape: (4096,), dtype: float16

 57%|█████▋    | 111/195 [00:56<00:36,  2.29it/s]
                                                 
[2024-06-04 03:13:29] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.18.self_attn.qkv_proj.weight", shape: (6144, 4096), dtype: float16

 57%|█████▋    | 111/195 [00:56<00:36,  2.29it/s]
 58%|█████▊    | 113/195 [00:56<00:26,  3.15it/s]
                                                 
[2024-06-04 03:13:29] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.18.self_attn.o_proj.weight", shape: (4096, 4096), dtype: float16

 58%|█████▊    | 113/195 [00:56<00:26,  3.15it/s]
 58%|█████▊    | 114/195 [00:57<00:22,  3.56it/s]
                                                 
[2024-06-04 03:13:29] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.19.input_layernorm.weight", shape: (4096,), dtype: float16

 58%|█████▊    | 114/195 [00:57<00:22,  3.56it/s]
                                                 
[2024-06-04 03:13:29] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.19.mlp.down_proj.weight", shape: (4096, 14336), dtype: float16

 58%|█████▊    | 114/195 [00:57<00:22,  3.56it/s]
 59%|█████▉    | 116/195 [00:57<00:21,  3.71it/s]
                                                 
[2024-06-04 03:13:31] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.19.mlp.gate_up_proj.weight", shape: (28672, 4096), dtype: float16

 59%|█████▉    | 116/195 [00:58<00:21,  3.71it/s]
 60%|██████    | 117/195 [00:59<00:44,  1.76it/s]
                                                 
[2024-06-04 03:13:31] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.19.post_attention_layernorm.weight", shape: (4096,), dtype: float16

 60%|██████    | 117/195 [00:59<00:44,  1.76it/s]
                                                 
[2024-06-04 03:13:32] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.19.self_attn.qkv_proj.weight", shape: (6144, 4096), dtype: float16

 60%|██████    | 117/195 [00:59<00:44,  1.76it/s]
 61%|██████    | 119/195 [00:59<00:30,  2.48it/s]
                                                 
[2024-06-04 03:13:32] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.19.self_attn.o_proj.weight", shape: (4096, 4096), dtype: float16

 61%|██████    | 119/195 [00:59<00:30,  2.48it/s]
 62%|██████▏   | 120/195 [00:59<00:26,  2.85it/s]
                                                 
[2024-06-04 03:13:32] INFO huggingface_loader.py:185: Loading HF parameters from: /models/Meta-Llama-3-8B-Instruct/model-00003-of-00004.safetensors

 62%|██████▏   | 120/195 [00:59<00:26,  2.85it/s]
                                                 
[2024-06-04 03:13:37] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.20.mlp.gate_up_proj.weight", shape: (28672, 4096), dtype: float16

 62%|██████▏   | 120/195 [01:04<00:26,  2.85it/s]
 62%|██████▏   | 121/195 [01:05<01:56,  1.58s/it]
                                                 
[2024-06-04 03:13:38] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.20.self_attn.qkv_proj.weight", shape: (6144, 4096), dtype: float16

 62%|██████▏   | 121/195 [01:05<01:56,  1.58s/it]
 63%|██████▎   | 122/195 [01:05<01:31,  1.25s/it]
                                                 
[2024-06-04 03:13:38] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.20.self_attn.o_proj.weight", shape: (4096, 4096), dtype: float16

 63%|██████▎   | 122/195 [01:05<01:31,  1.25s/it]
 63%|██████▎   | 123/195 [01:05<01:09,  1.04it/s]
                                                 
[2024-06-04 03:13:38] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.9.input_layernorm.weight", shape: (4096,), dtype: float16

 63%|██████▎   | 123/195 [01:05<01:09,  1.04it/s]
                                                 
[2024-06-04 03:13:38] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.9.mlp.down_proj.weight", shape: (4096, 14336), dtype: float16

 63%|██████▎   | 123/195 [01:05<01:09,  1.04it/s]
 64%|██████▍   | 125/195 [01:06<00:46,  1.51it/s]
                                                 
[2024-06-04 03:13:39] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.9.mlp.gate_up_proj.weight", shape: (28672, 4096), dtype: float16

 64%|██████▍   | 125/195 [01:06<00:46,  1.51it/s]
 65%|██████▍   | 126/195 [01:07<00:54,  1.26it/s]
                                                 
[2024-06-04 03:13:40] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.9.post_attention_layernorm.weight", shape: (4096,), dtype: float16

 65%|██████▍   | 126/195 [01:07<00:54,  1.26it/s]
                                                 
[2024-06-04 03:13:40] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.9.self_attn.qkv_proj.weight", shape: (6144, 4096), dtype: float16

 65%|██████▍   | 126/195 [01:07<00:54,  1.26it/s]
 66%|██████▌   | 128/195 [01:07<00:35,  1.89it/s]
                                                 
[2024-06-04 03:13:40] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.9.self_attn.o_proj.weight", shape: (4096, 4096), dtype: float16

 66%|██████▌   | 128/195 [01:07<00:35,  1.89it/s]
 66%|██████▌   | 129/195 [01:07<00:29,  2.24it/s]
                                                 
[2024-06-04 03:13:40] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.20.input_layernorm.weight", shape: (4096,), dtype: float16

 66%|██████▌   | 129/195 [01:07<00:29,  2.24it/s]
                                                 
[2024-06-04 03:13:40] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.20.mlp.down_proj.weight", shape: (4096, 14336), dtype: float16

 66%|██████▌   | 129/195 [01:08<00:29,  2.24it/s]
 67%|██████▋   | 131/195 [01:08<00:24,  2.65it/s]
                                                 
[2024-06-04 03:13:41] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.20.post_attention_layernorm.weight", shape: (4096,), dtype: float16

 67%|██████▋   | 131/195 [01:08<00:24,  2.65it/s]
                                                 
[2024-06-04 03:13:41] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.21.input_layernorm.weight", shape: (4096,), dtype: float16

 67%|██████▋   | 131/195 [01:08<00:24,  2.65it/s]
                                                 
[2024-06-04 03:13:41] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.21.mlp.down_proj.weight", shape: (4096, 14336), dtype: float16

 67%|██████▋   | 131/195 [01:08<00:24,  2.65it/s]
 69%|██████▊   | 134/195 [01:08<00:17,  3.48it/s]
                                                 
[2024-06-04 03:13:42] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.21.mlp.gate_up_proj.weight", shape: (28672, 4096), dtype: float16

 69%|██████▊   | 134/195 [01:09<00:17,  3.48it/s]
 69%|██████▉   | 135/195 [01:10<00:26,  2.30it/s]
                                                 
[2024-06-04 03:13:42] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.21.post_attention_layernorm.weight", shape: (4096,), dtype: float16

 69%|██████▉   | 135/195 [01:10<00:26,  2.30it/s]
                                                 
[2024-06-04 03:13:42] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.21.self_attn.qkv_proj.weight", shape: (6144, 4096), dtype: float16

 69%|██████▉   | 135/195 [01:10<00:26,  2.30it/s]
 70%|███████   | 137/195 [01:10<00:19,  3.01it/s]
                                                 
[2024-06-04 03:13:43] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.21.self_attn.o_proj.weight", shape: (4096, 4096), dtype: float16

 70%|███████   | 137/195 [01:10<00:19,  3.01it/s]
 71%|███████   | 138/195 [01:10<00:17,  3.35it/s]
                                                 
[2024-06-04 03:13:43] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.22.input_layernorm.weight", shape: (4096,), dtype: float16

 71%|███████   | 138/195 [01:10<00:17,  3.35it/s]
                                                 
[2024-06-04 03:13:43] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.22.mlp.down_proj.weight", shape: (4096, 14336), dtype: float16

 71%|███████   | 138/195 [01:10<00:17,  3.35it/s]
 72%|███████▏  | 140/195 [01:11<00:16,  3.42it/s]
                                                 
[2024-06-04 03:13:44] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.22.mlp.gate_up_proj.weight", shape: (28672, 4096), dtype: float16

 72%|███████▏  | 140/195 [01:11<00:16,  3.42it/s]
 72%|███████▏  | 141/195 [01:12<00:24,  2.17it/s]
                                                 
[2024-06-04 03:13:44] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.22.post_attention_layernorm.weight", shape: (4096,), dtype: float16

 72%|███████▏  | 141/195 [01:12<00:24,  2.17it/s]
                                                 
[2024-06-04 03:13:44] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.22.self_attn.qkv_proj.weight", shape: (6144, 4096), dtype: float16

 72%|███████▏  | 141/195 [01:12<00:24,  2.17it/s]
 73%|███████▎  | 143/195 [01:12<00:17,  2.95it/s]
                                                 
[2024-06-04 03:13:45] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.22.self_attn.o_proj.weight", shape: (4096, 4096), dtype: float16

 73%|███████▎  | 143/195 [01:12<00:17,  2.95it/s]
 74%|███████▍  | 144/195 [01:12<00:15,  3.32it/s]
                                                 
[2024-06-04 03:13:45] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.23.input_layernorm.weight", shape: (4096,), dtype: float16

 74%|███████▍  | 144/195 [01:12<00:15,  3.32it/s]
                                                 
[2024-06-04 03:13:45] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.23.mlp.down_proj.weight", shape: (4096, 14336), dtype: float16

 74%|███████▍  | 144/195 [01:12<00:15,  3.32it/s]
 75%|███████▍  | 146/195 [01:13<00:14,  3.48it/s]
                                                 
[2024-06-04 03:13:46] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.23.mlp.gate_up_proj.weight", shape: (28672, 4096), dtype: float16

 75%|███████▍  | 146/195 [01:13<00:14,  3.48it/s]
 75%|███████▌  | 147/195 [01:14<00:22,  2.17it/s]
                                                 
[2024-06-04 03:13:46] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.23.post_attention_layernorm.weight", shape: (4096,), dtype: float16

 75%|███████▌  | 147/195 [01:14<00:22,  2.17it/s]
                                                 
[2024-06-04 03:13:47] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.23.self_attn.qkv_proj.weight", shape: (6144, 4096), dtype: float16

 75%|███████▌  | 147/195 [01:14<00:22,  2.17it/s]
 76%|███████▋  | 149/195 [01:14<00:15,  2.98it/s]
                                                 
[2024-06-04 03:13:47] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.23.self_attn.o_proj.weight", shape: (4096, 4096), dtype: float16

 76%|███████▋  | 149/195 [01:14<00:15,  2.98it/s]
 77%|███████▋  | 150/195 [01:14<00:13,  3.36it/s]
                                                 
[2024-06-04 03:13:47] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.24.input_layernorm.weight", shape: (4096,), dtype: float16

 77%|███████▋  | 150/195 [01:14<00:13,  3.36it/s]
                                                 
[2024-06-04 03:13:47] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.24.mlp.down_proj.weight", shape: (4096, 14336), dtype: float16

 77%|███████▋  | 150/195 [01:14<00:13,  3.36it/s]
 78%|███████▊  | 152/195 [01:15<00:12,  3.54it/s]
                                                 
[2024-06-04 03:13:49] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.24.mlp.gate_up_proj.weight", shape: (28672, 4096), dtype: float16

 78%|███████▊  | 152/195 [01:16<00:12,  3.54it/s]
 78%|███████▊  | 153/195 [01:17<00:25,  1.62it/s]
                                                 
[2024-06-04 03:13:49] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.24.post_attention_layernorm.weight", shape: (4096,), dtype: float16

 78%|███████▊  | 153/195 [01:17<00:25,  1.62it/s]
                                                 
[2024-06-04 03:13:49] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.24.self_attn.qkv_proj.weight", shape: (6144, 4096), dtype: float16

 78%|███████▊  | 153/195 [01:17<00:25,  1.62it/s]
 79%|███████▉  | 155/195 [01:17<00:17,  2.31it/s]
                                                 
[2024-06-04 03:13:50] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.24.self_attn.o_proj.weight", shape: (4096, 4096), dtype: float16

 79%|███████▉  | 155/195 [01:17<00:17,  2.31it/s]
 80%|████████  | 156/195 [01:17<00:14,  2.67it/s]
                                                 
[2024-06-04 03:13:50] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.25.input_layernorm.weight", shape: (4096,), dtype: float16

 80%|████████  | 156/195 [01:17<00:14,  2.67it/s]
                                                 
[2024-06-04 03:13:50] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.25.mlp.down_proj.weight", shape: (4096, 14336), dtype: float16

 80%|████████  | 156/195 [01:17<00:14,  2.67it/s]
 81%|████████  | 158/195 [01:17<00:12,  2.99it/s]
                                                 
[2024-06-04 03:13:51] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.25.mlp.gate_up_proj.weight", shape: (28672, 4096), dtype: float16

 81%|████████  | 158/195 [01:18<00:12,  2.99it/s]
 82%|████████▏ | 159/195 [01:19<00:20,  1.80it/s]
                                                 
[2024-06-04 03:13:52] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.25.post_attention_layernorm.weight", shape: (4096,), dtype: float16

 82%|████████▏ | 159/195 [01:19<00:20,  1.80it/s]
                                                 
[2024-06-04 03:13:52] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.25.self_attn.qkv_proj.weight", shape: (6144, 4096), dtype: float16

 82%|████████▏ | 159/195 [01:19<00:20,  1.80it/s]
 83%|████████▎ | 161/195 [01:19<00:13,  2.52it/s]
                                                 
[2024-06-04 03:13:52] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.25.self_attn.o_proj.weight", shape: (4096, 4096), dtype: float16

 83%|████████▎ | 161/195 [01:19<00:13,  2.52it/s]
 83%|████████▎ | 162/195 [01:19<00:11,  2.90it/s]
                                                 
[2024-06-04 03:13:52] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.26.input_layernorm.weight", shape: (4096,), dtype: float16

 83%|████████▎ | 162/195 [01:19<00:11,  2.90it/s]
                                                 
[2024-06-04 03:13:52] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.26.mlp.down_proj.weight", shape: (4096, 14336), dtype: float16

 83%|████████▎ | 162/195 [01:19<00:11,  2.90it/s]
 84%|████████▍ | 164/195 [01:20<00:09,  3.18it/s]
                                                 
[2024-06-04 03:13:53] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.26.mlp.gate_up_proj.weight", shape: (28672, 4096), dtype: float16

 84%|████████▍ | 164/195 [01:20<00:09,  3.18it/s]
 85%|████████▍ | 165/195 [01:21<00:15,  1.91it/s]
                                                 
[2024-06-04 03:13:54] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.26.post_attention_layernorm.weight", shape: (4096,), dtype: float16

 85%|████████▍ | 165/195 [01:21<00:15,  1.91it/s]
                                                 
[2024-06-04 03:13:54] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.26.self_attn.qkv_proj.weight", shape: (6144, 4096), dtype: float16

 85%|████████▍ | 165/195 [01:21<00:15,  1.91it/s]
 86%|████████▌ | 167/195 [01:21<00:10,  2.66it/s]
                                                 
[2024-06-04 03:13:54] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.26.self_attn.o_proj.weight", shape: (4096, 4096), dtype: float16

 86%|████████▌ | 167/195 [01:21<00:10,  2.66it/s]
 86%|████████▌ | 168/195 [01:22<00:08,  3.04it/s]
                                                 
[2024-06-04 03:13:54] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.27.input_layernorm.weight", shape: (4096,), dtype: float16

 86%|████████▌ | 168/195 [01:22<00:08,  3.04it/s]
                                                 
[2024-06-04 03:13:54] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.27.mlp.down_proj.weight", shape: (4096, 14336), dtype: float16

 86%|████████▌ | 168/195 [01:22<00:08,  3.04it/s]
 87%|████████▋ | 170/195 [01:22<00:07,  3.31it/s]
                                                 
[2024-06-04 03:13:56] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.27.mlp.gate_up_proj.weight", shape: (28672, 4096), dtype: float16

 87%|████████▋ | 170/195 [01:23<00:07,  3.31it/s]
 88%|████████▊ | 171/195 [01:24<00:15,  1.59it/s]
                                                 
[2024-06-04 03:13:57] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.27.post_attention_layernorm.weight", shape: (4096,), dtype: float16

 88%|████████▊ | 171/195 [01:24<00:15,  1.59it/s]
                                                 
[2024-06-04 03:13:57] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.27.self_attn.qkv_proj.weight", shape: (6144, 4096), dtype: float16

 88%|████████▊ | 171/195 [01:24<00:15,  1.59it/s]
 89%|████████▊ | 173/195 [01:24<00:09,  2.27it/s]
                                                 
[2024-06-04 03:13:57] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.27.self_attn.o_proj.weight", shape: (4096, 4096), dtype: float16

 89%|████████▊ | 173/195 [01:24<00:09,  2.27it/s]
 89%|████████▉ | 174/195 [01:24<00:08,  2.60it/s]
                                                 
[2024-06-04 03:13:57] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.28.input_layernorm.weight", shape: (4096,), dtype: float16

 89%|████████▉ | 174/195 [01:24<00:08,  2.60it/s]
                                                 
[2024-06-04 03:13:57] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.28.mlp.down_proj.weight", shape: (4096, 14336), dtype: float16

 89%|████████▉ | 174/195 [01:25<00:08,  2.60it/s]
 90%|█████████ | 176/195 [01:25<00:06,  2.98it/s]
                                                 
[2024-06-04 03:13:59] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.28.mlp.gate_up_proj.weight", shape: (28672, 4096), dtype: float16

 90%|█████████ | 176/195 [01:27<00:06,  2.98it/s]
 91%|█████████ | 177/195 [01:27<00:14,  1.24it/s]
                                                 
[2024-06-04 03:14:00] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.28.post_attention_layernorm.weight", shape: (4096,), dtype: float16

 91%|█████████ | 177/195 [01:27<00:14,  1.24it/s]
                                                 
[2024-06-04 03:14:00] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.28.self_attn.qkv_proj.weight", shape: (6144, 4096), dtype: float16

 91%|█████████ | 177/195 [01:27<00:14,  1.24it/s]
 92%|█████████▏| 179/195 [01:28<00:08,  1.82it/s]
                                                 
[2024-06-04 03:14:00] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.28.self_attn.o_proj.weight", shape: (4096, 4096), dtype: float16

 92%|█████████▏| 179/195 [01:28<00:08,  1.82it/s]
 92%|█████████▏| 180/195 [01:28<00:07,  2.14it/s]
                                                 
[2024-06-04 03:14:00] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.29.input_layernorm.weight", shape: (4096,), dtype: float16

 92%|█████████▏| 180/195 [01:28<00:07,  2.14it/s]
                                                 
[2024-06-04 03:14:01] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.29.mlp.down_proj.weight", shape: (4096, 14336), dtype: float16

 92%|█████████▏| 180/195 [01:28<00:07,  2.14it/s]
 93%|█████████▎| 182/195 [01:28<00:05,  2.55it/s]
                                                 
[2024-06-04 03:14:02] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.29.mlp.gate_up_proj.weight", shape: (28672, 4096), dtype: float16

 93%|█████████▎| 182/195 [01:30<00:05,  2.55it/s]
 94%|█████████▍| 183/195 [01:30<00:08,  1.42it/s]
                                                 
[2024-06-04 03:14:03] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.29.post_attention_layernorm.weight", shape: (4096,), dtype: float16

 94%|█████████▍| 183/195 [01:30<00:08,  1.42it/s]
                                                 
[2024-06-04 03:14:03] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.29.self_attn.qkv_proj.weight", shape: (6144, 4096), dtype: float16

 94%|█████████▍| 183/195 [01:30<00:08,  1.42it/s]
 95%|█████████▍| 185/195 [01:30<00:04,  2.04it/s]
                                                 
[2024-06-04 03:14:03] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.29.self_attn.o_proj.weight", shape: (4096, 4096), dtype: float16

 95%|█████████▍| 185/195 [01:31<00:04,  2.04it/s]
 95%|█████████▌| 186/195 [01:31<00:03,  2.38it/s]
                                                 
[2024-06-04 03:14:03] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.30.input_layernorm.weight", shape: (4096,), dtype: float16

 95%|█████████▌| 186/195 [01:31<00:03,  2.38it/s]
                                                 
[2024-06-04 03:14:04] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.30.mlp.down_proj.weight", shape: (4096, 14336), dtype: float16

 95%|█████████▌| 186/195 [01:31<00:03,  2.38it/s]
 96%|█████████▋| 188/195 [01:31<00:02,  2.78it/s]
                                                 
[2024-06-04 03:14:06] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.30.mlp.gate_up_proj.weight", shape: (28672, 4096), dtype: float16

 96%|█████████▋| 188/195 [01:33<00:02,  2.78it/s]
 97%|█████████▋| 189/195 [01:34<00:04,  1.25it/s]
                                                 
[2024-06-04 03:14:06] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.30.post_attention_layernorm.weight", shape: (4096,), dtype: float16

 97%|█████████▋| 189/195 [01:34<00:04,  1.25it/s]
                                                 
[2024-06-04 03:14:06] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.30.self_attn.qkv_proj.weight", shape: (6144, 4096), dtype: float16

 97%|█████████▋| 189/195 [01:34<00:04,  1.25it/s]
 98%|█████████▊| 191/195 [01:34<00:02,  1.82it/s]
                                                 
[2024-06-04 03:14:07] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.30.self_attn.o_proj.weight", shape: (4096, 4096), dtype: float16

 98%|█████████▊| 191/195 [01:34<00:02,  1.82it/s]
 98%|█████████▊| 192/195 [01:34<00:01,  2.15it/s]
                                                 
[2024-06-04 03:14:07] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.31.mlp.gate_up_proj.weight", shape: (28672, 4096), dtype: float16

 98%|█████████▊| 192/195 [01:35<00:01,  2.15it/s]
 99%|█████████▉| 193/195 [01:35<00:01,  1.45it/s]
                                                 
[2024-06-04 03:14:08] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.31.self_attn.qkv_proj.weight", shape: (6144, 4096), dtype: float16

 99%|█████████▉| 193/195 [01:36<00:01,  1.45it/s]
 99%|█████████▉| 194/195 [01:36<00:00,  1.72it/s]
                                                 
[2024-06-04 03:14:08] INFO huggingface_loader.py:175: [Not quantized] Parameter: "model.layers.31.self_attn.o_proj.weight", shape: (4096, 4096), dtype: float16

 99%|█████████▉| 194/195 [01:36<00:00,  1.72it/s]
100%|██████████| 195/195 [01:36<00:00,  2.12it/s]
100%|██████████| 195/195 [01:36<00:00,  2.02it/s]
[2024-06-04 03:14:09] INFO huggingface_loader.py:197: Unloading HF weight file: /models/Meta-Llama-3-8B-Instruct/model-00002-of-00004.safetensors
[2024-06-04 03:14:09] INFO huggingface_loader.py:197: Unloading HF weight file: /models/Meta-Llama-3-8B-Instruct/model-00003-of-00004.safetensors
[2024-06-04 03:14:10] INFO stats.py:77: Time usage: HF loading: 15.816 sec; Pre-quantization mapping: 36.574 sec; Quantization: 0.000 sec
[2024-06-04 03:14:10] INFO stats.py:91: RAM usage: Peak RAM: 18.469 GB. Total bytes loaded from disk: 29.915 GB
[2024-06-04 03:14:10] INFO convert_weight.py:155: Parameter size after quantization: 14.958 GB
[2024-06-04 03:14:10] INFO convert_weight.py:160: Total parameters: 8,030,261,248
[2024-06-04 03:14:10] INFO convert_weight.py:161: Bits per parameter: 16.000
[2024-06-04 03:14:10] INFO convert_weight.py:166: Saved to directory: /models/mlc-delivery/hf/mlc-ai/Llama-3-8B-Instruct-q0f16-MLC

All finished, 131 total shards committed, record saved to /models/mlc-delivery/hf/mlc-ai/Llama-3-8B-Instruct-q0f16-MLC/ndarray-cache.json