passdrill
AI & LLM Engineering · Evaluation, Testing & Safety · Card 004/010 medium

A research team reports that their new model scores surprisingly high on a widely-used public benchmark, higher than models with far more parameters. Before trusting this result, an independent reviewer wants to rule out one specific and common threat to benchmark validity: that some of the benchmark's test questions, or close paraphrases of them, appeared in the model's pretraining data. What is this threat called, and what is one concrete technique used to check for it?

  1. This threat is called catastrophic forgetting; it is checked by fine-tuning the model on an older task and confirming that performance on that older task does not drop
  2. This threat is called benchmark (data) contamination; one concrete detection technique is searching the pretraining corpus for high n-gram overlap with the benchmark's test examples, or checking for embedded 'canary string' markers placed in the benchmark specifically to detect this
  3. This threat is called mode collapse; it is checked by sampling many outputs from the model at high temperature and confirming that they remain diverse
  4. This threat is called reward hacking; it is checked by inspecting the model's reward model training data for label errors
Next card → Shuffle