Download hf_quant_config.json from stepfun-ai/Step-5-Preview-NVFP4: direct link, hf CLI and curl.
- Browser
- Download file 12.4 kB
-
https://huggingface.co/stepfun-ai/Step-5-Preview-NVFP4/resolve/main/hf_quant_config.json
- Command line
-
hf download hf://stepfun-ai/Step-5-Preview-NVFP4/hf_quant_config.json
-
curl -L -o hf_quant_config.json https://huggingface.co/stepfun-ai/Step-5-Preview-NVFP4/resolve/main/hf_quant_config.json
12.4 kB
| { | |
| "producer": { | |
| "name": "modelopt", | |
| "version": "0.47.0" | |
| }, | |
| "quantization": { | |
| "quant_algo": "NVFP4", | |
| "kv_cache_quant_algo": null, | |
| "group_size": 16, | |
| "exclude_modules": [ | |
| "lm_head", | |
| "model.embed_tokens", | |
| "model.norm", | |
| "*layers.0.self_attn*", | |
| "*layers.0.mlp*", | |
| "*layers.1.self_attn*", | |
| "*layers.1.mlp*", | |
| "*layers.2.self_attn*", | |
| "*layers.2.mlp*", | |
| "*layers.3.self_attn*", | |
| "*layers.3.share_expert*", | |
| "*layers.3.moe.share_expert*", | |
| "*layers.3.moe.gate*", | |
| "*layers.4.self_attn*", | |
| "*layers.4.share_expert*", | |
| "*layers.4.moe.share_expert*", | |
| "*layers.4.moe.gate*", | |
| "*layers.5.self_attn*", | |
| "*layers.5.share_expert*", | |
| "*layers.5.moe.share_expert*", | |
| "*layers.5.moe.gate*", | |
| "*layers.6.self_attn*", | |
| "*layers.6.share_expert*", | |
| "*layers.6.moe.share_expert*", | |
| "*layers.6.moe.gate*", | |
| "*layers.7.self_attn*", | |
| "*layers.7.share_expert*", | |
| "*layers.7.moe.share_expert*", | |
| "*layers.7.moe.gate*", | |
| "*layers.8.self_attn*", | |
| "*layers.8.share_expert*", | |
| "*layers.8.moe.share_expert*", | |
| "*layers.8.moe.gate*", | |
| "*layers.9.self_attn*", | |
| "*layers.9.share_expert*", | |
| "*layers.9.moe.share_expert*", | |
| "*layers.9.moe.gate*", | |
| "*layers.10.self_attn*", | |
| "*layers.10.share_expert*", | |
| "*layers.10.moe.share_expert*", | |
| "*layers.10.moe.gate*", | |
| "*layers.11.self_attn*", | |
| "*layers.11.share_expert*", | |
| "*layers.11.moe.share_expert*", | |
| "*layers.11.moe.gate*", | |
| "*layers.12.self_attn*", | |
| "*layers.12.share_expert*", | |
| "*layers.12.moe.share_expert*", | |
| "*layers.12.moe.gate*", | |
| "*layers.13.self_attn*", | |
| "*layers.13.share_expert*", | |
| "*layers.13.moe.share_expert*", | |
| "*layers.13.moe.gate*", | |
| "*layers.14.self_attn*", | |
| "*layers.14.share_expert*", | |
| "*layers.14.moe.share_expert*", | |
| "*layers.14.moe.gate*", | |
| "*layers.15.self_attn*", | |
| "*layers.15.share_expert*", | |
| "*layers.15.moe.share_expert*", | |
| "*layers.15.moe.gate*", | |
| "*layers.16.self_attn*", | |
| "*layers.16.share_expert*", | |
| "*layers.16.moe.share_expert*", | |
| "*layers.16.moe.gate*", | |
| "*layers.17.self_attn*", | |
| "*layers.17.share_expert*", | |
| "*layers.17.moe.share_expert*", | |
| "*layers.17.moe.gate*", | |
| "*layers.18.self_attn*", | |
| "*layers.18.share_expert*", | |
| "*layers.18.moe.share_expert*", | |
| "*layers.18.moe.gate*", | |
| "*layers.19.self_attn*", | |
| "*layers.19.share_expert*", | |
| "*layers.19.moe.share_expert*", | |
| "*layers.19.moe.gate*", | |
| "*layers.20.self_attn*", | |
| "*layers.20.share_expert*", | |
| "*layers.20.moe.share_expert*", | |
| "*layers.20.moe.gate*", | |
| "*layers.21.self_attn*", | |
| "*layers.21.share_expert*", | |
| "*layers.21.moe.share_expert*", | |
| "*layers.21.moe.gate*", | |
| "*layers.22.self_attn*", | |
| "*layers.22.share_expert*", | |
| "*layers.22.moe.share_expert*", | |
| "*layers.22.moe.gate*", | |
| "*layers.23.self_attn*", | |
| "*layers.23.share_expert*", | |
| "*layers.23.moe.share_expert*", | |
| "*layers.23.moe.gate*", | |
| "*layers.24.self_attn*", | |
| "*layers.24.share_expert*", | |
| "*layers.24.moe.share_expert*", | |
| "*layers.24.moe.gate*", | |
| "*layers.25.self_attn*", | |
| "*layers.25.share_expert*", | |
| "*layers.25.moe.share_expert*", | |
| "*layers.25.moe.gate*", | |
| "*layers.26.self_attn*", | |
| "*layers.26.share_expert*", | |
| "*layers.26.moe.share_expert*", | |
| "*layers.26.moe.gate*", | |
| "*layers.27.self_attn*", | |
| "*layers.27.share_expert*", | |
| "*layers.27.moe.share_expert*", | |
| "*layers.27.moe.gate*", | |
| "*layers.28.self_attn*", | |
| "*layers.28.share_expert*", | |
| "*layers.28.moe.share_expert*", | |
| "*layers.28.moe.gate*", | |
| "*layers.29.self_attn*", | |
| "*layers.29.share_expert*", | |
| "*layers.29.moe.share_expert*", | |
| "*layers.29.moe.gate*", | |
| "*layers.30.self_attn*", | |
| "*layers.30.share_expert*", | |
| "*layers.30.moe.share_expert*", | |
| "*layers.30.moe.gate*", | |
| "*layers.31.self_attn*", | |
| "*layers.31.share_expert*", | |
| "*layers.31.moe.share_expert*", | |
| "*layers.31.moe.gate*", | |
| "*layers.32.self_attn*", | |
| "*layers.32.share_expert*", | |
| "*layers.32.moe.share_expert*", | |
| "*layers.32.moe.gate*", | |
| "*layers.33.self_attn*", | |
| "*layers.33.share_expert*", | |
| "*layers.33.moe.share_expert*", | |
| "*layers.33.moe.gate*", | |
| "*layers.34.self_attn*", | |
| "*layers.34.share_expert*", | |
| "*layers.34.moe.share_expert*", | |
| "*layers.34.moe.gate*", | |
| "*layers.35.self_attn*", | |
| "*layers.35.share_expert*", | |
| "*layers.35.moe.share_expert*", | |
| "*layers.35.moe.gate*", | |
| "*layers.36.self_attn*", | |
| "*layers.36.share_expert*", | |
| "*layers.36.moe.share_expert*", | |
| "*layers.36.moe.gate*", | |
| "*layers.37.self_attn*", | |
| "*layers.37.share_expert*", | |
| "*layers.37.moe.share_expert*", | |
| "*layers.37.moe.gate*", | |
| "*layers.38.self_attn*", | |
| "*layers.38.share_expert*", | |
| "*layers.38.moe.share_expert*", | |
| "*layers.38.moe.gate*", | |
| "*layers.39.self_attn*", | |
| "*layers.39.share_expert*", | |
| "*layers.39.moe.share_expert*", | |
| "*layers.39.moe.gate*", | |
| "*layers.40.self_attn*", | |
| "*layers.40.share_expert*", | |
| "*layers.40.moe.share_expert*", | |
| "*layers.40.moe.gate*", | |
| "*layers.41.self_attn*", | |
| "*layers.41.share_expert*", | |
| "*layers.41.moe.share_expert*", | |
| "*layers.41.moe.gate*", | |
| "*layers.42.self_attn*", | |
| "*layers.42.share_expert*", | |
| "*layers.42.moe.share_expert*", | |
| "*layers.42.moe.gate*", | |
| "*layers.43.self_attn*", | |
| "*layers.43.share_expert*", | |
| "*layers.43.moe.share_expert*", | |
| "*layers.43.moe.gate*", | |
| "*layers.44.self_attn*", | |
| "*layers.44.share_expert*", | |
| "*layers.44.moe.share_expert*", | |
| "*layers.44.moe.gate*", | |
| "*layers.45.self_attn*", | |
| "*layers.45.share_expert*", | |
| "*layers.45.moe.share_expert*", | |
| "*layers.45.moe.gate*", | |
| "*layers.46.self_attn*", | |
| "*layers.46.share_expert*", | |
| "*layers.46.moe.share_expert*", | |
| "*layers.46.moe.gate*", | |
| "*layers.47.self_attn*", | |
| "*layers.47.share_expert*", | |
| "*layers.47.moe.share_expert*", | |
| "*layers.47.moe.gate*", | |
| "*layers.48.self_attn*", | |
| "*layers.48.share_expert*", | |
| "*layers.48.moe.share_expert*", | |
| "*layers.48.moe.gate*", | |
| "*layers.49.self_attn*", | |
| "*layers.49.share_expert*", | |
| "*layers.49.moe.share_expert*", | |
| "*layers.49.moe.gate*", | |
| "*layers.50.self_attn*", | |
| "*layers.50.share_expert*", | |
| "*layers.50.moe.share_expert*", | |
| "*layers.50.moe.gate*", | |
| "*layers.51.self_attn*", | |
| "*layers.51.share_expert*", | |
| "*layers.51.moe.share_expert*", | |
| "*layers.51.moe.gate*", | |
| "*layers.52.self_attn*", | |
| "*layers.52.share_expert*", | |
| "*layers.52.moe.share_expert*", | |
| "*layers.52.moe.gate*", | |
| "*layers.53.self_attn*", | |
| "*layers.53.share_expert*", | |
| "*layers.53.moe.share_expert*", | |
| "*layers.53.moe.gate*", | |
| "*layers.54.self_attn*", | |
| "*layers.54.share_expert*", | |
| "*layers.54.moe.share_expert*", | |
| "*layers.54.moe.gate*", | |
| "*layers.55.self_attn*", | |
| "*layers.55.share_expert*", | |
| "*layers.55.moe.share_expert*", | |
| "*layers.55.moe.gate*", | |
| "*layers.56.self_attn*", | |
| "*layers.56.share_expert*", | |
| "*layers.56.moe.share_expert*", | |
| "*layers.56.moe.gate*", | |
| "*layers.57.self_attn*", | |
| "*layers.57.share_expert*", | |
| "*layers.57.moe.share_expert*", | |
| "*layers.57.moe.gate*", | |
| "*layers.58.self_attn*", | |
| "*layers.58.share_expert*", | |
| "*layers.58.moe.share_expert*", | |
| "*layers.58.moe.gate*", | |
| "*layers.59.self_attn*", | |
| "*layers.59.share_expert*", | |
| "*layers.59.moe.share_expert*", | |
| "*layers.59.moe.gate*", | |
| "*layers.60.self_attn*", | |
| "*layers.60.share_expert*", | |
| "*layers.60.moe.share_expert*", | |
| "*layers.60.moe.gate*", | |
| "*layers.61.self_attn*", | |
| "*layers.61.share_expert*", | |
| "*layers.61.moe.share_expert*", | |
| "*layers.61.moe.gate*", | |
| "*layers.62.self_attn*", | |
| "*layers.62.share_expert*", | |
| "*layers.62.moe.share_expert*", | |
| "*layers.62.moe.gate*", | |
| "*layers.63.self_attn*", | |
| "*layers.63.share_expert*", | |
| "*layers.63.moe.share_expert*", | |
| "*layers.63.moe.gate*", | |
| "*layers.64.self_attn*", | |
| "*layers.64.share_expert*", | |
| "*layers.64.moe.share_expert*", | |
| "*layers.64.moe.gate*", | |
| "*layers.65.self_attn*", | |
| "*layers.65.share_expert*", | |
| "*layers.65.moe.share_expert*", | |
| "*layers.65.moe.gate*", | |
| "*layers.66.self_attn*", | |
| "*layers.66.share_expert*", | |
| "*layers.66.moe.share_expert*", | |
| "*layers.66.moe.gate*", | |
| "*layers.67.self_attn*", | |
| "*layers.67.share_expert*", | |
| "*layers.67.moe.share_expert*", | |
| "*layers.67.moe.gate*", | |
| "*layers.68.self_attn*", | |
| "*layers.68.share_expert*", | |
| "*layers.68.moe.share_expert*", | |
| "*layers.68.moe.gate*", | |
| "*layers.69.self_attn*", | |
| "*layers.69.share_expert*", | |
| "*layers.69.moe.share_expert*", | |
| "*layers.69.moe.gate*", | |
| "*layers.70.self_attn*", | |
| "*layers.70.share_expert*", | |
| "*layers.70.moe.share_expert*", | |
| "*layers.70.moe.gate*", | |
| "*layers.71.self_attn*", | |
| "*layers.71.share_expert*", | |
| "*layers.71.moe.share_expert*", | |
| "*layers.71.moe.gate*", | |
| "*layers.72.self_attn*", | |
| "*layers.72.share_expert*", | |
| "*layers.72.moe.share_expert*", | |
| "*layers.72.moe.gate*", | |
| "*layers.73.self_attn*", | |
| "*layers.73.share_expert*", | |
| "*layers.73.moe.share_expert*", | |
| "*layers.73.moe.gate*", | |
| "*layers.74.self_attn*", | |
| "*layers.74.share_expert*", | |
| "*layers.74.moe.share_expert*", | |
| "*layers.74.moe.gate*", | |
| "*layers.75.self_attn*", | |
| "*layers.75.share_expert*", | |
| "*layers.75.moe.share_expert*", | |
| "*layers.75.moe.gate*", | |
| "*layers.76.self_attn*", | |
| "*layers.76.share_expert*", | |
| "*layers.76.moe.share_expert*", | |
| "*layers.76.moe.gate*", | |
| "*layers.77.self_attn*", | |
| "*layers.77.share_expert*", | |
| "*layers.77.moe.share_expert*", | |
| "*layers.77.moe.gate*", | |
| "*layers.78.self_attn*", | |
| "*layers.78.share_expert*", | |
| "*layers.78.moe.share_expert*", | |
| "*layers.78.moe.gate*", | |
| "*layers.79.self_attn*", | |
| "*layers.79.share_expert*", | |
| "*layers.79.moe.share_expert*", | |
| "*layers.79.moe.gate*", | |
| "*layers.80.self_attn*", | |
| "*layers.80.share_expert*", | |
| "*layers.80.moe.share_expert*", | |
| "*layers.80.moe.gate*", | |
| "*layers.81.self_attn*", | |
| "*layers.81.share_expert*", | |
| "*layers.81.moe.share_expert*", | |
| "*layers.81.moe.gate*", | |
| "*layers.82.self_attn*", | |
| "*layers.82.share_expert*", | |
| "*layers.82.moe.share_expert*", | |
| "*layers.82.moe.gate*", | |
| "*layers.83.self_attn*", | |
| "*layers.83.share_expert*", | |
| "*layers.83.moe.share_expert*", | |
| "*layers.83.moe.gate*", | |
| "*layers.84.self_attn*", | |
| "*layers.84.share_expert*", | |
| "*layers.84.moe.share_expert*", | |
| "*layers.84.moe.gate*", | |
| "*layers.85.self_attn*", | |
| "*layers.85.share_expert*", | |
| "*layers.85.moe.share_expert*", | |
| "*layers.85.moe.gate*", | |
| "*layers.86.self_attn*", | |
| "*layers.86.share_expert*", | |
| "*layers.86.moe.share_expert*", | |
| "*layers.86.moe.gate*", | |
| "*layers.87.self_attn*", | |
| "*layers.87.share_expert*", | |
| "*layers.87.moe.share_expert*", | |
| "*layers.87.moe.gate*", | |
| "*layers.88.self_attn*", | |
| "*layers.88.share_expert*", | |
| "*layers.88.moe.share_expert*", | |
| "*layers.88.moe.gate*", | |
| "*layers.88.moe.experts*", | |
| "*layers.89.self_attn*", | |
| "*layers.89.share_expert*", | |
| "*layers.89.moe.share_expert*", | |
| "*layers.89.moe.gate*", | |
| "*layers.89.moe.experts*", | |
| "*layers.90.self_attn*", | |
| "*layers.90.share_expert*", | |
| "*layers.90.moe.share_expert*", | |
| "*layers.90.moe.gate*", | |
| "*layers.90.moe.experts*", | |
| "*layers.91.self_attn*", | |
| "*layers.91.mlp*", | |
| "*layers.92.*", | |
| "*layers.93.*", | |
| "*layers.94.*", | |
| "vision_model*", | |
| "vit_large_projector*" | |
| ] | |
| } | |
| } | |