Llama-3.1-405B-Instruct-xMADai-INT4 / quantize_config.json
onebitquantized's picture
Upload of AutoGPTQ quantized model
70b86ed verified
raw
history blame
313 Bytes
{
"bits": 4,
"group_size": 128,
"damp_percent": 0.025,
"desc_act": true,
"static_groups": false,
"sym": false,
"true_sequential": true,
"model_name_or_path": null,
"model_file_base_name": null,
"exponent_hinv": 4.0,
"shrink": 0.0625,
"quant_method": "gptq",
"checkpoint_format": "gptq"
}