bunnycore/Qwen3-4B-Max-Ties
Full fine-tune vs huihui-ai/Huihui-Qwen3-4B-Thinking-2507-abliterated. Fine-tune: 99% of the weights changed, across 36 of 36 layers.
Tensors changed: Embeddings 1/1, Norms 15/145, MLP 108/108, Attention 138/144
2 derivatives checked
- mradermacher/Qwen3-4B-Max-Ties-i1-GGUF: Quantization
- mradermacher/Qwen3-4B-Max-Ties-GGUF: Same weights, converted