iiinova — where your digital self keeps evolving

Sunny Qin

@sunnytqin.bsky.social

26 followers 2 following 0 posts

Machine Learning PhD @ Harvard

Posts Replies Media

Reposted by Sunny Qin

David Alvarez-Melis @dmelis.bsky.social · 11/04/2025

In our newest work (led by the amazing @sunnytqin.bsky.social , w/ @emalach.bsky.social, Samy Jelassi), we investigate a core question for LLMs: "𝑡𝑜 𝑏𝑎𝑐𝑘𝑡𝑟𝑎𝑐𝑘 𝑜𝑟 𝑛𝑜𝑡 𝑡𝑜 𝑏𝑎𝑐𝑘𝑡𝑟𝑎𝑐𝑘" in two prototypical logic-heavy puzzles: CountDown and Sudoku.

132

Reposted by Sunny Qin

Naomi Saphra @nsaphra.bsky.social · 25/02/2025

Ever looked at LLM skill emergence and thought 70B parameters was a magic number? Our new paper shows sudden breakthroughs are samples from bimodal performance distributions across seeds. Observed accuracy jumps abruptly while the underlying accuracy DISTRIBUTION changes slowly!

Distributional Scaling Laws for Emergent Capabilities
Rosie Zhao, Tian Qin, David Alvarez-Melis, Sham Kakade, Naomi Saphra
In this paper, we explore the nature of sudden breakthroughs in language model performance at scale, which stands in contrast to smooth improvements governed by scaling laws. While advocates of "emergence" view abrupt performance gains as capabilities unlocking at specific scales, others have suggested that they are produced by thresholding effects and alleviated by continuous metrics. We propose that breakthroughs are instead driven by continuous changes in the probability distribution of training outcomes, particularly when performance is bimodally distributed across random seeds. In synthetic length generalization tasks, we show that different random seeds can produce either highly linear or emergent scaling trends. We reveal that sharp breakthroughs in metrics are produced by underlying continuous changes in their distribution across seeds. Furthermore, we provide a case study of inverse scaling and show that even as the probability of a successful run declines, the average performance of a successful run continues to increase monotonically. We validate our distributional scaling framework on realistic settings by measuring MMLU performance in LLM populations. These insights emphasize the role of random variation in the effect of scale on LLM capabilities.

36615