UnfilteredAI/Dan-Qwen3.5-4B
Full fine-tune vs Qwen/Qwen3.5-4B. Fine-tune: 93% of the weights changed, across 32 of 32 layers.
Tensors changed: Embeddings 1/4, MLP 96/196, Attention 181/320, Norms 63/203
4 derivatives checked
- mradermacher/Qwen3.5-4B-Unfiltered-GGUF: Same weights, converted
- mradermacher/Qwen3.5-4B-Unfiltered-i1-GGUF: Quantization
- mradermacher/Dan-Qwen3.5-4B-GGUF: Same weights, converted
- mradermacher/Dan-Qwen3.5-4B-i1-GGUF: Quantization
1 other repos hold byte-for-byte identical weight files.