bunnycore/Qwen3-4B-Mini-Merge

Full fine-tune vs Bingguang/FunReason-MT. Fine-tune: 100% of the weights changed, across 36 of 36 layers.

Tensors changed: lm_head 1/1, MLP 108/108, Norms 144/145, Attention 144/144, Embeddings 1/1

2 derivatives checked

• mradermacher/Qwen3-4B-Mini-Merge-i1-GGUF: Quantization
• mradermacher/Qwen3-4B-Mini-Merge-GGUF: Same weights, converted

On the Hub · Downloads on Model Pulse