discourse-ai/evals/lib
Sam ab5edae121
FIX: make AI helper more robust (#1484)
* FIX: make AI helper more robust

- If JSON is broken for structured output then lean on a more forgiving parser
- Gemini 2.5 flash does not support temp, support opting out
- Evals for assistant were broken, fix interface
- Add some missing LLMs
- Translator was not mapped correctly to the feature - fix that
- Don't mix XML in prompt for translator

* lint

* correct logic

* simplify code

* implement best effort json parsing direct in the structured output object
2025-07-04 14:47:11 +10:00
..
prompts DEV: Support multiple tests per eval and followups per test (#1199) 2025-03-18 11:42:05 +08:00
boot.rb FIX: hardcoded require for evals (#1137) 2025-02-19 11:56:52 -08:00
cli.rb FEATURE: PDF support for rag pipeline (#1118) 2025-02-14 12:15:07 +11:00
eval.rb FIX: make AI helper more robust (#1484) 2025-07-04 14:47:11 +10:00
llm.rb DEV: Extract configs to a yml file and allow local config (#1142) 2025-02-24 16:22:19 +11:00
runner.rb DEV: Allow prompt-type evals to take in several prompts and messages (#1190) 2025-03-14 15:46:22 +11:00