Fetching from the wire…
01
02
03
04
05
06
07
08
Research2026-07-29 · source-backed
Bits and Memories measures verbatim reproduction of training data rather than membership inference, using Pythia models with known-memorized sequence sets across five precision levels down to four bits. Verbatim memorization drops faster than perplexity at every precision and size under two unrelated quantization algorithms. But at the largest model studied, four-bit quantization still reproduces most memorized sequences while giving up a few percent of capability, and the surviving memorized fraction grows with model size.
Each link below shares sources, entities, or timing with this story.
Shared entity: Bits / Same source domain / Shared topic / Earlier coverage / Tension
Both cover Bits; reported by the same outlet (arxiv.org); overlapping topics (bits, model).
Same source domain / Shared topic / Tension
Reported by the same outlet (arxiv.org); overlapping topics (data, faster, model); pushes against this story (against).
Shared entity: Quantization / Shared topic / Earlier coverage
Both cover Quantization; overlapping topics (drop, model); earlier Quantization coverage from 2026-06-07.
Same source domain / Shared topic / Tension
Reported by the same outlet (arxiv.org); overlapping topics (capability, data); pushes against this story (but).
Reported by the same outlet (arxiv.org); overlapping topics (down, faster); pushes against this story (but).
Reported by the same outlet (arxiv.org); overlapping topics (data, model); pushes against this story (against).
Reported by the same outlet (arxiv.org); overlapping topics (capability, model); pushes against this story (against).
Same source domain / Shared topic / Downstream implication
Reported by the same outlet (arxiv.org); overlapping topics (capability, data); traces where this leads (downstream).