← back to paper

Review history

arxiv: 2604.13977 · 2 revisions

How Can We Synthesize High-Quality Pretraining Data? A Systematic Study of Prompt Design, Generator Model, and Source Data

1. 2026-08-02 CONDITIONAL MODERATE v1.3.0-alltime-deepseek novelty 7.0
   135948 ms 33465 in 17012 out 2026-08-02T16:14:21.786207+00:00
2. 2026-05-10 UNVERDICTED LOW v0.9.0 novelty 6.0
   42356 ms 5545 in 1387 out 2026-05-10T13:21:42.591708+00:00