huginnfork/Qwen3.8-Flash-Next-NVFP4-Abliterated

Partial fine-tune vs local-inference-lab/Qwen3.8-Flash-Next-NVFP4. Partial change: 11% of the weights changed (attention, embeddings, mlp, other), in 48 of 48 layers.

Tensors changed: Attention 241/608, MLP 125/1561, Embeddings 1/266, lm_head 0/1, Norms 0/302, other 1/8

On the Hub · Downloads on Model Pulse