serag-ai/SAMF
LoRA-style (fewer layers) vs microsoft/Phi-3-mini-4k-instruct. microsoft/Phi-3-mini-4k-instruct matches part of serag-ai/SAMF (193 of its 380 tensors; the rest of serag-ai/SAMF 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_up_proj, o_proj, qkv_proj changed, in 32 of 32 layers; norms, biases and embeddings are untouched.
Tensors changed: Norms 0/65, MLP 64/64, Attention 64/64