MaziyarPanahi/calme-2.1-phi3.5-4b
Partial fine-tune vs microsoft/Phi-3.5-mini-instruct. Partial change: 72% of the weights changed (attention, mlp, norms), in 31 of 32 layers.
Tensors changed: Embeddings 0/1, Norms 9/65, MLP 54/64, Attention 45/64, lm_head 0/1
5 derivatives checked
- mradermacher/calme-2.1-phi3.5-4b-i1-GGUF: Quantization
- bartowski/calme-2.1-phi3.5-4b-GGUF: Same weights, converted
- mradermacher/calme-2.1-phi3.5-4b-GGUF: Same weights, converted
- MaziyarPanahi/calme-2.1-phi3.5-4b-GGUF: Quantization
- bunnycore/Phi-3.5-Mini-Hyper: Full fine-tune