bigatuna/Qwen3.5-9b-Sushi-Coder-RL
LoRA-style vs bigatuna/Qwen3.5-9b-Sushi-Coder. Looks like a merged LoRA (or a fine-tune with frozen norms): only the weight matrices of q_proj, v_proj changed, in 8 of 32 layers; norms, biases and embeddings are untouched.
Tensors changed: Norms 0/215, lm_head 0/1, Embeddings 0/4, Attention 16/332, MLP 0/208
2 derivatives checked
- bigatuna/Qwen3.5-9b-Sushi-Coder-RL-GGUF: Quantization
- bigatuna/Qwen3.5-9b-Sushi-Coder-RL-MLX: Quantization