tvall43/Qwen3.6-14B-A3B-FableVibes
LoRA-style vs tvall43/Qwen3.6-35B-A3B-heretic. 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, q_proj, up_proj, v_proj changed, in 40 of 40 layers; norms, biases and embeddings are untouched.
Tensors changed: lm_head 0/1, Embeddings 0/1, Norms 0/131, Attention 40/280, MLP 120/160
3 derivatives checked
- tvall43/Qwen3.6-14B-A3B-FableVibes-GGUF: Quantization
- khanh2023/Qwen3.6-14B-A3B-FableVibes-mlx-q4: Quantization
- khanh2023/Qwen3.6-14B-A3B-FableVibes-mlx-q6: Quantization