Major Changes
- Updated to llama.cpp
815a2a5915f22ce6a760c676389c5dfe8535c08f(b10221) by @martindevans in #1414
Bug Fixes
- Implement Sequence ID pooling in BatchedExecutor to prevent native SeqMax overflow and crashes by @zsogitbe in #1386
- Disable GGML_F16C GGML_BMI2 for noavx buid (Fixes #1407) by @butameron in #1408
- Allow arbitrary ChatSession message interleaving by @YoMosa2009 in #1417
- Fix null DryRun output by @martindevans in #1419
Other Changes
- Remove
LLama.Experimentalby @martindevans in #1421 - Remove SemanticKernel and KernelMemory by @martindevans in #1420
New Contributors
- @butameron made their first contribution in #1408
- @YoMosa2009 made their first contribution in #1417
Full Changelog: v0.27.0...v0.29.0
Contributors
martindevans, butameron, and 2 other contributors