redashes/Qwen3.8-27B-BF16-SSMFIX
LoRA-style vs Qwen/Qwen3.8-27B. Looks like a merged LoRA (or a fine-tune with frozen norms): only the weight matrices of conv1d changed, in 8 of 64 layers; norms, biases and embeddings are untouched.
Tensors changed: Norms 0/325, Attention 8/560, MLP 0/307, Embeddings 0/5, lm_head 0/1, other 0/1
2 derivatives checked
- Luis23333/Qwen3.8-27B-SSMFIX-UD-Q3_K_XL-GGUF: Quantization
- CreekHop/Qwen3.8-27B-SSMFIX-GGUF: Same weights, converted