Change8

b10707

📦 llama-cppView on GitHub →
🐛 1 fixes🔧 2 symbols

Summary

Optimized kv-cells sequence scan for improved generation performance on longer contexts. This release also provides pre-compiled binaries for various platforms and hardware configurations.

🐛 Bug Fixes

  • Optimized kv-cells sequence scan to stop once all sequences for a cell are seen, improving generation speed for longer contexts. This change affects the n-gram path via the get_prev_tokens function.

Affected Symbols