A-Evolve-Training: Autonomous Post-Training of a 30B Model

Unknown Date

AI Summary

This work proposes the first fully automated, closed-loop post-training system for large language models, eliminating the need for manual intervention that traditionally spans weeks. The framework autonomously executes multiple optimization rounds for the 30B-parameter Nemotron model and successfully scales to 120B and 550B parameters. By integrating automated data curation, training recipe search, dynamic evaluation, and adaptive policy adjustment, the system detects and corrects misalignments between internal metrics and external performance, demonstrating emergent self-directed optimization capabilities beyond conventional methods. The resulting models achieve a score of 0.86 on the NVIDIA Nemotron-Reasoning Challenge—just below the human best of 0.87—and rank 8th among approximately 4,000 participants.

Abstract

Post-training a frontier model is normally weeks of human work: proposing data and recipe changes, launching runs, reading evals, deciding what to keep. We report an autonomous system that runs this loop with no human in the loop, post-training a 30B Nemotron across four rounds over multiple weeks. The autonomously produced model reaches a held-out score of 0.86 against the top human submission's 0.87 on the public NVIDIA Nemotron-Reasoning Challenge leaderboard, placing 8th of ~4000 at the time of writing. More striking than the number: the loop detected that its own dev metric had stopped tracking external performance on the weakest domain -- candidates drove dev to record highs without moving the external target -- and revised its own search policy, no longer maximizing dev but seeking interventions that lowered the now-misleading proxy while improving the external target. We treat this as direct, auditable evidence that a scaled autonomous loop can produce discovery, not only optimization: it detected that its measurement frame had become misleading and changed what counted as evidence. We take the operational view that any system worth the "recursive self-improvement" label must eventually perform end-to-end post-training of a frontier-class model; this is one datapoint of that bar being cleared. We do not claim a "first autonomous match" of human researchers. The claim we make is narrower and auditable: to our knowledge, this is the first publicly reported autonomous post-training run at this scale, where prior public autonomous-ML-research demonstrations sit at GPT-2-class (~124M) budgets. The same system also post-trains the 120B and 550B Nemotron; with no public human baseline there, this shows only that the loop closes at that scale, not that its output is competitive -- infrastructure evidence, with the effectiveness claim deferred until a comparable human anchor exists.

Problem

• autonomous post-training
• frontier model
• recursive self-improvement
• large language models
• automated machine learning

Innovation

• autonomous post-training
• recursive self-improvement
• proxy misalignment detection
• large language model
• automated machine learning