DeepSeek-R1-NVFP4-RTN / quantization_config.json
weiweiz1's picture
Upload folder using huggingface_hub
e5568e7 verified
raw
history blame contribute delete
283 Bytes
{
"bits": 4,
"group_size": 16,
"sym": true,
"data_type": "nv_fp4",
"batch_size": 4,
"iters": 0,
"autoround_version": "0.6.0",
"quant_method": "auto-round",
"packing_format": "nv_fp",
"scale_format": [
"e8m0"
],
"scale_calculation_mode": [
"even"
]
}