redashes/Qwen3.8-27B-BF16-SSMFIX-apostate
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, down_proj, o_proj, out_proj changed, in 46 of 64 layers; norms, biases and embeddings are untouched.
Tensors changed: Norms 0/325, Attention 20/560, MLP 33/307, Embeddings 0/5, lm_head 0/1, other 0/1
2 derivatives checked
- redashes/Qwen3.8-27B-SSMFIX-apostate-GSQ-RCO-IQ3_S: Quantization
- CreekHop/Qwen3.8-27B-SSMFIX-Apostate-GGUF: Same weights, converted