Symbolic Regression on LSR-Synth Material Science
0NMSEPiT-PO
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| PiT-POBackbone=Llama-3.1-8B-Instruct2026.02 | 0 | 12 | 92 | — | |
| LLM-SRBackbone=Llama-3.1-8B-Instruct2026.02 | 0 | 4.1 | 88.12 | — | |
| PySR+Optimization Strategy=SAGE-Fit (Inner-loop)2026.05 | 0 | 16 | 41.67 | — | |
| PySROptimization Strategy=Post-hoc Refinement2026.05 | 0 | 4 | 11.76 | — | |
| LLM-SR+Backbone=Qwen2.5-7B, Optimization Strategy=SAGE-Fit (Inner-loop)2026.05 | 0 | 8 | 15.91 | — | |
| LLM-SRBackbone=Qwen2.5-7B, Optimization Strategy=Post-hoc Refinement2026.05 | 0 | 4 | 11.36 | — | |
| LaSRBackbone=Llama-3.1-8B-Instruct2026.02 | 0.0001 | 8.21 | 64.22 | — | |
| Deliberate EvolutionBackbone=Qwen3-4B-Instruct-25072026.06 | 0.0001 | — | — | 56 | |
| Deliberate EvolutionBackbone=Llama3.1-8B-Instruct2026.06 | 0.0003 | — | — | 64 | |
| LaSR+Backbone=Qwen2.5-7B, Optimization Strategy=SAGE-Fit (Inner-loop)2026.05 | 0.0006 | 4 | 8 | — | |
| LASRBackbone=Qwen3-4B-Instruct-25072026.06 | 0.0006 | — | — | 8 | |
| LASRBackbone=Llama3.1-8B-Instruct2026.06 | 0.0009 | — | — | 32 | |
| uDSR+Optimization Strategy=SAGE-Fit (Inner-loop)2026.05 | 0.0011 | 0 | 15.79 | — | |
| uDSROptimization Strategy=Post-hoc Refinement2026.05 | 0.0012 | 0 | 13.04 | — | |
| LLMDirectBackbone=Qwen3-4B-Instruct-25072026.06 | 0.0014 | — | — | 52 | |
| LaSRBackbone=Qwen2.5-7B, Optimization Strategy=Post-hoc Refinement2026.05 | 0.0015 | 0 | 4 | — | |
| LLM-SRBackbone=Qwen3-4B-Instruct-25072026.06 | 0.0036 | — | — | 44 | |
| SGABackbone=Qwen3-4B-Instruct-25072026.06 | 0.0102 | — | — | 16 | |
| SGABackbone=Llama-3.1-8B-Instruct2026.02 | 0.0435 | 0 | 12.12 | — | |
| SGABackbone=Llama3.1-8B-Instruct2026.06 | 0.0435 | — | — | 12 | |
| LLMDirectBackbone=Llama3.1-8B-Instruct2026.06 | 0.0816 | — | — | 24 | |
| Direct PromptingBackbone=Llama-3.1-8B-Instruct2026.02 | 0.0826 | 0 | 0 | — | |
| LLM-SRBackbone=Llama3.1-8B-Instruct2026.06 | 0.216 | — | — | 60 |