q-tuning-joint-pruning-efficient-training

【数据与研究】Dramatically reduce training data requirements (to 12.5% of original) while improving model performance using joint sample and token pruning guided by Error-Uncertainty plane diagnostics. Asymmetric pruning preserves calibration signals while removing redundant tokens from misconception examples.

独立 GitHub 快照 · 自动分类 · 以原作者文档为准