mfielding92/littlemonster-reasoning-12B-QKVO-heretic-HF
LoRA-style vs p-e-w/gemma-3-12b-it-heretic. Looks like a merged LoRA (or a fine-tune with frozen norms): only the weight matrices of down_proj, o_proj, q_proj, v_proj changed, in 48 of 48 layers; norms, biases and embeddings are untouched.
Tensors changed: Embeddings 0/4, Norms 0/400, MLP 34/252, Attention 144/408, other 0/1
2 derivatives checked
- mradermacher/littlemonster-reasoning-12B-QKVO-heretic-HF-i1-GGUF: Quantization (fewer layers)
- mradermacher/littlemonster-reasoning-12B-QKVO-heretic-HF-GGUF: Quantization (fewer layers)