Michael-Kozu/Deimos-R1
LoRA-style vs Qwen/Qwen3.5-4B. Looks like a merged LoRA (or a fine-tune with frozen norms): only the weight matrices of down_proj, gate_proj, in_proj_a, in_proj_b, in_proj_qkv, in_proj_z, k_proj, o_proj, out_proj, q_proj, up_proj, v_proj changed, in 32 of 32 layers; norms, biases and embeddings are untouched.
Tensors changed: Embeddings 0/4, MLP 96/196, Attention 152/320, Norms 0/203
2 derivatives checked
- mradermacher/Deimos-R1-i1-GGUF: Quantization
- mradermacher/Deimos-R1-GGUF: Same weights, converted