Community fixes Jinja template for Qwen 3.8 reasoning_effort
A Reddit user's fix addresses official Qwen 3.8 chat template bugs that break disabling reasoning; the release adds prompt-steered `reasoning_effort` options `xhigh`, `medium`, and `low`.
1 source
Daily brief
Get tomorrow's AI brief in your inbox
More stories today
- Humyn Labs builds platform to capture human experience data for robotics
- Twin1 raises $20m seed to build AI digital twins for lawyers
- US trade talks fuel worry in Canada over tech and AI rules
- Quanta podcast probes AI reasoning with Melanie Mitchell
- AI data center opposition becomes bipartisan rallying cry before midterms