Llama-Guard-3-8B-Q8-GPTQ / quantize_config.json
ktoprakucar's picture
AutoGPTQ model for meta-llama/Llama-Guard-3-8B: 8bits, gr128, desc_act=True
6a9ced6 verified
raw
history blame contribute delete
265 Bytes
{
"bits": 8,
"group_size": 128,
"damp_percent": 0.01,
"desc_act": true,
"static_groups": false,
"sym": true,
"true_sequential": true,
"model_name_or_path": null,
"model_file_base_name": null,
"is_marlin_format": false,
"quant_method": "gptq"
}