anakin87/Phi-3.5-mini-ITA
LoRA-style vs microsoft/Phi-3.5-mini-instruct. 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 22 of 32 layers; norms, biases and embeddings are untouched.
Tensors changed: Embeddings 0/1, Norms 0/65, MLP 18/64, Attention 18/64, lm_head 0/1
2 derivatives checked
- mradermacher/Phi-3.5-mini-ITA-i1-GGUF: Quantization
- mradermacher/Phi-3.5-mini-ITA-GGUF: Same weights, converted