Auto-correlation Inequality Minimization on ThirdAutoCorrIneq
3.1586Best ScoreInitial
Evaluation results
| Method | Links | ||
|---|---|---|---|
| InitialModel=ProRL-1.5B-v2, Step=02025.11 | 3.1586 | — | |
| InitialModel=Distill-Qwen3-8B, Step=02025.11 | 3.1586 | — | |
| ThetaEvolve (w/o RL early)Model=ProRL-1.5B-v2, Step=2002025.11 | 1.6155 | 1.6831 | |
| ThetaEvolve (w/o RL late)Model=ProRL-1.5B-v2, Step=6002025.11 | 1.6123 | 1.6766 | |
| ThetaEvolve (w/ RL)Model=ProRL-1.5B-v2, Step=2002025.11 | 1.6053 | 1.6412 | |
| ThetaEvolve (w/o RL early)Model=Distill-Qwen3-8B, Step=652025.11 | 1.5084 | 1.5498 | |
| ThetaEvolve (w/o RL late)Model=Distill-Qwen3-8B, Step=1002025.11 | 1.5084 | 1.5491 | |
| ThetaEvolve (w/ RL)Model=Distill-Qwen3-8B, Step=652025.11 | 1.493 | 1.521 | |
| GEPALLM Backbone=GPT-52026.02 | 1.4758 | 1.4768 | |
| ShinkaLLM Backbone=GPT-52026.02 | 1.4614 | 1.4785 | |
| OpenEvolveLLM Backbone=GPT-52026.02 | 1.461 | 1.4671 | |
| EvoXLLM Backbone=GPT-52026.02 | 1.4609 | 1.4714 | |
| OpenEvolveLLM Backbone=Gemini-3.0-Pro2026.02 | 1.46 | 1.4609 | |
| GEPALLM Backbone=Gemini-3.0-Pro2026.02 | 1.4598 | 1.4678 | |
| Human2026.02 | 1.4581 | — | |
| ShinkaLLM Backbone=Gemini-3.0-Pro2026.02 | 1.4578 | 1.4595 | |
| EvoXLLM Backbone=Gemini-3.0-Pro2026.02 | 1.4558 | 1.4589 | |
| AlphaEvolve2026.02 | 1.4557 | — |