mesolitica/Malaysian-Qwen2.5-7B-Reasoning-SFT
Full fine-tune vs mesolitica/Malaysian-Qwen2.5-7B-Instruct. Fine-tune: 100% of the weights changed, across 28 of 28 layers.
Tensors changed: Embeddings 1/1, Norms 55/57, MLP 84/84, Attention 196/196, lm_head 1/1
3 derivatives checked
- mradermacher/Malaysian-Qwen2.5-7B-Reasoning-SFT-i1-GGUF: Quantization
- mradermacher/Malaysian-Qwen2.5-7B-Reasoning-SFT-GGUF: Same weights, converted
- mesolitica/Malaysian-Qwen2.5-7B-Dialect-Reasoning-GRPO: Full fine-tune