sliding-window-attention-adaptation

【开发与工程】Adapt full-attention language models to sliding window attention without expensive retraining. Combine five synergistic strategies (full decode, sink tokens, interleaved layers, chain-of-thought, fine-tuning) achieving 30-100% speedups while maintaining 90-100% accuracy.

独立 GitHub 快照 · 自动分类 · 以原作者文档为准