microsoft/Fara1.5-4B
Full fine-tune vs Qwen/Qwen3.5-4B. Fine-tune: 100% of the weights changed, across 32 of 32 layers.
Tensors changed: Embeddings 4/4, MLP 196/196, Attention 299/320, Norms 163/203
6 derivatives checked
- bartowski/Fara1.5-4B-GGUF: Same weights, converted
- webmp3/Sakura-Fara1.5-4B-GGUF: Quantization
- prithivMLmods/Fara1.5-4B-GGUF: Same weights, converted
- runanywhere/Fara1.5-4B-GGUF: Same weights, converted
- mlx-community/Fara1.5-4B-OptiQ-4bit: Quantization
- DevQuasar/microsoft.Fara1.5-4B-GGUF: Quantization
1 other repos hold byte-for-byte identical weight files.