dfurman/CalmeRys-78B-Orpo-v0.1

LoRA-style vs MaziyarPanahi/calme-2.4-rys-78b. 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 86 of 86 layers; norms, biases and embeddings are untouched.

Tensors changed: Embeddings 0/1, Norms 0/173, MLP 258/258, Attention 344/602, lm_head 0/1

3 derivatives checked

On the Hub · Downloads on Model Pulse