Patch release moving to v0.10.4-libs, whose XNNPACK backend fixes PReLU models returning NaN and Kokoro TTS being killed on iPhone, plus LLM and install fixes. LLMRunner.generate no longer takes echo; the prompt is never echoed.
What's Changed
- fix(libs): stop shipping backends the app opted out of by @msluszniak in #1466
- fix(llm): keep every terminal token out of the chat response by @msluszniak in #1486
- fix(llm): chunk prefill at the bound the graph declares by @msluszniak in #1491
- fix(llm): pin the model load mode instead of inheriting upstream defaults by @msluszniak in #1492
- fix(install): make the native lib download work on Windows by @msluszniak in #1494
Full Changelog: v0.10.2...v0.10.3