oakmindai/Qwen3.8-Flash-Next-NVFP4-FTW

Same weights, repacked (fewer layers) vs nvidia/Qwen3.8-Flash-Next-NVFP4. oakmindai/Qwen3.8-Flash-Next-NVFP4-FTW matches part of nvidia/Qwen3.8-Flash-Next-NVFP4 (3101 of its 299545 tensors; the rest of nvidia/Qwen3.8-Flash-Next-NVFP4 has no counterpart). On the shared part: Same weights in different files (re-saved, re-sharded or new metadata).

Tensors changed: Embeddings 0/2, other 0/3, Norms 0/8, Attention 0/8, MLP 0/28

On the Hub · Downloads on Model Pulse