Jackrong/Qwopus3.8-27B-Flash

LoRA-style vs Qwen/Qwen3.8-27B. Looks like a merged LoRA (or a fine-tune with frozen norms): only the weight matrices of down_proj, gate_proj, k_proj, o_proj, out_proj, q_proj, up_proj, v_proj changed, in 64 of 64 layers; norms, biases and embeddings are untouched.

Tensors changed: Norms 0/325, Attention 112/560, MLP 192/307, Embeddings 0/5, lm_head 0/1, other 0/1

38 derivatives checked

On the Hub · Downloads on Model Pulse