HF RL Explorer

The AI table generation and data extraction steps apply a reasoningEffort override to all LLM models via the…

The AI table generation and data extraction steps apply a reasoningEffort override to all LLM models via the…: a task in LegoFlow-SWE (Harbor dataset). The current fallback logic that retries the request without reasoning effort after an error should be removed entirely. Instead, the steps should…

The task

The current fallback logic that retries the request without reasoning effort after an error should be removed entirely. Instead, the steps should only apply reasoning effort to models that actually support it.

Part of Lego-X/LegoFlow-SWE.