Novaciano/Pyromancer-3.2-1B
Partial fine-tune vs Novaciano/qp-3.2-1B. Partial change: 83% of the weights changed (attention, embeddings, lm_head, mlp, norms), in 16 of 16 layers.
Tensors changed: Embeddings 1/1, lm_head 1/1, Norms 32/33, MLP 35/48, Attention 53/64
2 derivatives checked
- mradermacher/Pyromancer-3.2-1B-i1-GGUF: Quantization
- mradermacher/Pyromancer-3.2-1B-GGUF: Same weights, converted