Alibaba-AAIG/YuFeng-XGuard-Reason-8B

Partial fine-tune vs Qwen/Qwen3-8B. Partial change: 42% of the weights changed (attention, mlp, norms), in 18 of 36 layers.

Tensors changed: Embeddings 0/1, Norms 72/145, MLP 54/108, Attention 72/144, lm_head 0/1

1 derivatives checked

• mradermacher/YuFeng-XGuard-Reason-8B-GGUF: Same weights, converted

On the Hub · Downloads on Model Pulse