What's Changed
- Bump llama-cpp-python to 0.2.38 by @oobabooga in #5420
- Quadratic sampling by @kalomaze in #5403
- Remove non-HF ExLlamaV2 loader by @oobabooga in #5431
- Fix the n_batch slider by @BadisG in #5436
- Split by rows instead of layers for llama.cpp multi-gpu by @Ph0rk0z in #5435
- Improve ChatML template by @BadisG in #5411
- Truncate long chat completions inputs by @oobabooga in #5439
- Add custom sampler order support by @oobabooga in #5443
- Merge dev branch by @oobabooga in #5452
- Merge dev branch by @oobabooga in #5453
Full Changelog: snapshot-2024-02-04...snapshot-2024-02-11