LLM-SRBENCH
Benchmarks
TaskDatasetSOTA resultTrendResults
LLM-SRBench (phys_osc)
8.999Best Reward
16
LLM-SRBench matsci
8.25Best Reward
16
LLM-SRBench chem_react
9Best Reward
16
LLM-SRBench bio_pop_growth
8.973Best Reward
16
LLM-SRBench Symbolic
34.4Term Recall
14
LLM-SRBench OOD (test)
0.325NMSE
14
LLM-SRBench ID (test)
0.4NMSE
14
LLM-SRBENCH LSR-Transform
0.067NMSE
13
LLM-SRBench 129-task synthetic (test)
35Chemistry 95% Acc (Tol=0.01)
11
LLM-SRBench LSR-Synth Biology
0.64NMSE
10
LLM-SRBench LSR-Synth Chemistry
0NMSE
10
LLM-SRBENCH LSR-Syn
0Chemistry Error
9
LLM-SRBench (official 239-problem split)
77Acc0.1 (%)
6
LLM-SRBench 129-task synthetic subset OOD (test)
32Chemistry Accuracy (Tol 0.01)
5
LLM-SRBench Overall
0.875Median R2
3
LLM-SRBench Transform
1.253Median R2
3
LLM-SRBench Synthetic
0.984Median R2
3