SWE-rebench
A SWE dataset used for fine-tuning coding models, whose data quality can be improved through filtering with quality assessment methods.
Timeline (last 90 days)
Fine-tuning a coding model on SWE-rebench data filtered with ExLabel, the model solved 66.2% of problems on SWE-bench Verified, 14.5% higher than the model trained on data filtered with the best baseline.
The authors conducted a stored-memory audit of two reproducible traces (OpenHands/SWE-rebench and PlasmaFAIR/sdf-xarray, totaling 185 messages), ultimately classifying 64 note entries.
All Facts (2)
Fine-tuning a coding model on SWE-rebench data filtered with ExLabel, the model solved 66.2% of problems on SWE-bench Verified, 14.5% higher than the model trained on data filtered with the best baseline.
50%UnverifiedThe authors conducted a stored-memory audit of two reproducible traces (OpenHands/SWE-rebench and PlasmaFAIR/sdf-xarray, totaling 185 messages), ultimately classifying 64 note entries.
50%