SWE-bench/SWE-agent-LM-32B
Full fine-tune vs Qwen/Qwen2.5-Coder-32B-Instruct. Fine-tune: 98% of the weights changed, across 64 of 64 layers.
Tensors changed: Embeddings 0/1, Norms 126/129, MLP 192/192, Attention 448/448, lm_head 1/1
4 derivatives checked
- featherless-ai-quants/SWE-bench-SWE-agent-LM-32B-GGUF: Quantization
- DevQuasar/SWE-bench.SWE-agent-LM-32B-GGUF: Same weights, converted
- tensorblock/SWE-bench_SWE-agent-LM-32B-GGUF: Quantization
- Kebob/SWE-agent-LM-32B-GGUF: Quantization
1 other repos hold byte-for-byte identical weight files.