super-decider / hf_quant_config.json
Eljaja's picture
super-decider NVFP4 weights, model card, licence and NOTICE
7f8b3f7
Raw History Blame Contribute Delete
463 Bytes
{
"producer": {
"name": "modelopt",
"version": "0.45.0.dev127+ga3a02ca5b.d20260608"
},
"quantization": {
"quant_algo": "NVFP4",
"kv_cache_quant_algo": "FP8",
"group_size": 16,
"exclude_modules": [
"lm_head",
"*embed_vision*",
"*mlp*",
"*router*",
"*self_attn*",
"*self_conditioning*",
"*vision_tower*"
]
}
}