Satori-reasoning/Satori-SWE-SFT-32B
Full fine-tune vs Qwen/Qwen2.5-Coder-32B-Instruct. Fine-tune: 97% of the weights changed, across 64 of 64 layers.
Tensors changed: Embeddings 0/1, Norms 125/129, MLP 191/192, Attention 448/448, lm_head 1/1
3 derivatives checked
- mradermacher/Satori-SWE-SFT-32B-i1-GGUF: Quantization
- mradermacher/Satori-SWE-SFT-32B-GGUF: Quantization
- Satori-reasoning/Satori-SWE-RL-32B: Full fine-tune