In this article
What's changed
@react-native-rag/executorchruns on react-native-executorch 0.11.ExecuTorchLLMloads the model through the new asynccreateLLMRunner, fits the conversation to the model'smaxContextLen, and passes the tokenizer's BOS and EOS tokens to the chat template (#27 by @msluszniak)- The adapter accepts
react-native-workletsup to 0.13 (#27 by @msluszniak)
Breaking changes
@react-native-rag/executorchrequiresreact-native-executorch^0.11.0 andreact-native-blob-util0.24.11 or newer. Its API is unchanged.
Full changelog: v0.10.0...v0.11.0
Release card