Model training and evaluation

Landing page for notes on:

• evaluation methods (offline, online, red-team)
• reliability and calibration for factual outputs
• multi-step agent evaluation (tool-use regressions, state carryover)

Notes

• Fluency Is Not Factuality
• Theory of mind in LLMs — what benchmarks test (and what they don’t)
• Orders of intentionality and recursive mindreading in LLM evaluation