CATT Benchmark Exposes Gaps in Training Data Attribution
A new CATT benchmark, from an undergraduate team advised by davidbau, reveals that existing training data attribution methods often fail to find the exact documents a model learned from, even when those documents are known.
2026-10-30 ~ 2026-10-31 · 2 related posts
- CATT benchmark shows attribution methods fail to trace model answers to training data — cindy_x_wu · 2026-10-30
- CATT benchmark probes why training data attribution fails even with known ground truth — TuhinChakr · 2026-10-31