Llama-3.2-1B-Instruct-NVFP4 / quantization_config.json
weiweiz1's picture
Upload folder using huggingface_hub
bb1f85e verified
raw
history blame contribute delete
301 Bytes
{
"bits": 4,
"group_size": 16,
"sym": true,
"data_type": "nv_fp4",
"seqlen": 512,
"batch_size": 4,
"iters": 20,
"autoround_version": "0.6.0",
"quant_method": "auto-round",
"packing_format": "nv_fp",
"scale_format": [
"e8m0"
],
"scale_calculation_mode": [
"even"
]
}