biondizzle/deepseek-v4-quant
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
24 Commits 4 Branches 0 Tags
cbfc5a9afb27f2bc77ebe045c2e34a73bce0faf2
Commit Graph

2 Commits

Author SHA1 Message Date
biondizzle
cbfc5a9afb Update nvfp4_experts_only to use dequantized BF16 model 2026-05-07 16:34:37 +00:00
biondizzle
6008cf128d Add model_opt_nvfp4_experts_only.py 
Quantizes only MoE expert weights to NVFP4, leaving attention untouched.
Includes comments documenting all available NVFP4 strategies.
Copy to model_opt_nvfp4_<strategy>.py for each new strategy.
 2026-05-07 15:16:08 +00:00