Mawdistical/Draconic-Tease-70B
LoRA-style vs Steelskull/L3.3-Electra-R1-70b. 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 19 of 80 layers; norms, biases and embeddings are untouched.
Tensors changed: lm_head 0/1, Embeddings 0/1, Norms 0/161, MLP 57/240, Attention 76/320
2 derivatives checked
- Mawdistical/Draconic-Tease-70B-GGUF: Quantization
- TareksLab/Zootopia-L3.3-70B: Full fine-tune