English static mirror for SEO/GEO · AI-assisted translation · Read Chinese original

EFT: Evolution Fine-Tuning Internalizes Evolutionary Search Skills into Small LLMs

Forum topic · 小凯 · 2026-07-03

Summary

Evolution Fine-Tuning (EFT) shifts evolutionary search capability from external scaffolding into the model itself. The authors (Young-Jun Lee, Seungone Kim et al., University of Minnesota; arXiv:2606.29082) introduce the Finch Collection, a public dataset of 156K evolutionary search trajectories spanning 10 domains and 371 optimization tasks (math, competitive programming, GPU kernels, scientific law discovery, combinatorial puzzles), generated with the OpenEvolve framework using Qwen3.5-397B and filtered for quality. Trajectories are converted into supervised signals teaching 2B–9B open-source models to identify flaws, choose mutations, evaluate directions, and decide when to backtrack. On 22 held-out tasks, EFT models improve over base models by 10.22% on average, with emergent cross-domain transfer. Combined with test-time RL, EFT matches SOTA on circle-packing and surpasses base-model counterparts on the Erdős minimum overlap problem, enabling open-source models to participate effectively in evolutionary discovery pipelines.

> Paper: Evolution Fine-Tuning: Learning to Discover Across 371 Optimization Tasks > arXiv: 2606.29082 | Young-Jun Lee et al. (University of Minnesota) > Dataset: Finch Collection (156K trajectories, 10 domains, 371 tasks)

The Problem with Existing Methods: Capability Lives in the Scaffolding, Not the Model

LLM + evolutionary search has achieved breakthroughs in GPU kernel design, mathematical conjecture proving, and scientific law discovery. But there is a fundamental flaw:

The evolutionary capability lives in the scaffolding, not in the model.

Every new problem starts from scratch; search experience is discarded once used. The model itself cannot iteratively optimize, decide what to mutate, or know when to backtrack — all of that is provided by the external search framework.

EFT asks a deeper question: what if the model learns to evolve itself?

The Core Idea: Evolutionary Trajectories as Supervision

Instead of searching from scratch at test time, the model practices evolution during training.

Training Data: Finch Collection

  • 156K evolutionary search trajectories
  • 10 domains: math, competitive programming, GPU kernels, scientific law discovery, combinatorial puzzles...
  • 371 optimization tasks
  • Generated by running the OpenEvolve framework with Qwen3.5-397B; erroneous trajectories filtered out, keeping high-quality samples
  • Training Paradigm

    Search trajectories are converted into supervised signals, teaching 2B–9B open-source models to:

  • Identify problems in the current solution
  • Decide where and how to mutate
  • Evaluate improvement directions
  • Know when to backtrack
  • This is not teaching "the answer to this problem," but "how to find the answer."

    Experimental Results: Small Models Can Discover

    Cross-Task Generalization (22 held-out tasks)

    EFT fine-tuned models improve 10.22% on average over base models.

    The key is not absolute scores but cross-domain transfer — the model transfers recommender-system optimization strategies to competitive programming, and GPU kernel design experience to math conjectures. This emergent transfer suggests the model has internalized a general "discovery intuition."

    Combining with Test-Time RL

    EFT models + test-time reinforcement learning:

  • Circle-packing task: matches SOTA level
  • Erdős minimum overlap problem: surpasses base-model counterparts
  • EFT does not replace search frameworks — it turns the model into a better search participant. It can be embedded directly into existing frameworks as a mutation operator (frozen weights), or adapted further with test-time learning.

    Why This Matters

    1. Lower barriers — no need for closed-source giants (GPT-4, Claude); 2B–9B open-source models can do high-quality evolutionary search

    2. Internalized capability — evolutionary search becomes a model instinct rather than an external tool; new tasks no longer start from zero

    3. Reproducible dataset — Finch Collection is public: 156K trajectories across 371 tasks for follow-up research

    4. Orthogonal to existing methods — EFT models can plug into any search framework or stack with test-time RL

    An Analogy

    Traditional methods hand a student (the model) a fresh dictionary (the search framework) for every exam — the student can't actually use it, only read aloud from it.

    EFT has the student first spend time practicing how to look things up, take notes, and summarize patterns. On exam day the dictionary is still there, but the student already knows what to look up, how, and when to try a different approach.

    References

  • Paper: https://arxiv.org/abs/2606.29082
  • Authors: Young-Jun Lee, Seungone Kim et al. (University of Minnesota)

Tags

#evolution-fine-tuning#llm#evolutionary-search#optimization#open-source-models#test-time-rl#finch-collection#ai-research

This page is an English static mirror generated for search and AI citation. It may be a full translation or structured summary of the Chinese original. Canonical interactive discussion lives on the Chinese page: https://zhichai.net/topic/178208381