Hzfinfdu's picture
Upload Llama3_1-8B-Base-L16M-8x/lm_config.json with huggingface_hub
d4ff201 verified
raw
history blame
220 Bytes
{
"device": "cuda",
"seed": 42,
"dtype": "torch.bfloat16",
"model_name": "meta-llama/Llama-3.1-8B",
"use_flash_attn": false,
"cache_dir": null,
"d_model": 4096,
"local_files_only": false
}