zait-ai/toffee-3b-beta
LoRA-style (fewer layers) vs mistralai/Ministral-3-3B-Base-2512. zait-ai/toffee-3b-beta matches part of mistralai/Ministral-3-3B-Base-2512 (236 of its 458 tensors; the rest of mistralai/Ministral-3-3B-Base-2512 has no counterpart). On the shared part: 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 26 of 26 layers; norms, biases and embeddings are untouched.
Tensors changed: Embeddings 0/1, Norms 0/53, MLP 78/78, Attention 104/104
2 derivatives checked
- mradermacher/toffee-3b-beta-i1-GGUF: Quantization
- mradermacher/toffee-3b-beta-GGUF: Same weights, converted