tuanpasg/Puffin-Qwen2.5-CodeMath-1

Full fine-tune vs Qwen/Qwen2.5-Coder-1.5B. Fine-tune: 97% of the weights changed, across 28 of 28 layers.

Tensors changed: Embeddings 1/1, Norms 57/57, MLP 81/84, Attention 189/196

2 derivatives checked

• mradermacher/Puffin-Qwen2.5-CodeMath-1-i1-GGUF: Quantization
• mradermacher/Puffin-Qwen2.5-CodeMath-1-GGUF: Same weights, converted

On the Hub · Downloads on Model Pulse