List your company

Soft constrained reward optimization on XSum on Gemma-7B

0.147Average Soft Constrained RewardMLP-CSR
-0.005880.033810.07350.11319May 14, 2026

Evaluation results

MethodLinks
MLP-CSRPrompt set size (K)=50, Per-arm budget (b)=82026.050.147
MLP-CSRPrompt set size (K)=100, Per-arm budget (b)=102026.050.147
MLP-CSRPrompt set size (K)=100, Per-arm budget (b)=82026.050.145
CSRPrompt set size (K)=30, Per-arm budget (b)=82026.050.144
CSRPrompt set size (K)=30, Per-arm budget (b)=102026.050.143
CSRPrompt set size (K)=50, Per-arm budget (b)=52026.050.143
MLP-CSRPrompt set size (K)=50, Per-arm budget (b)=32026.050.143
CSRPrompt set size (K)=100, Per-arm budget (b)=102026.050.143
MLP-CSRPrompt set size (K)=30, Per-arm budget (b)=102026.050.142
MLP-CSRPrompt set size (K)=50, Per-arm budget (b)=52026.050.142
MLP-CSRPrompt set size (K)=50, Per-arm budget (b)=102026.050.142
CSRPrompt set size (K)=100, Per-arm budget (b)=82026.050.142
MLP-CSRPrompt set size (K)=100, Per-arm budget (b)=52026.050.142
CSRPrompt set size (K)=50, Per-arm budget (b)=82026.050.141
CSRPrompt set size (K)=100, Per-arm budget (b)=52026.050.141
MLP-CSRPrompt set size (K)=30, Per-arm budget (b)=82026.050.14
CSRPrompt set size (K)=50, Per-arm budget (b)=102026.050.14
MLP-CSRPrompt set size (K)=30, Per-arm budget (b)=52026.050.139
MLP-CSRPrompt set size (K)=100, Per-arm budget (b)=32026.050.139
CSRPrompt set size (K)=30, Per-arm budget (b)=52026.050.137
CSRPrompt set size (K)=100, Per-arm budget (b)=32026.050.134
MLP-CSRPrompt set size (K)=30, Per-arm budget (b)=32026.050.123
CSRPrompt set size (K)=50, Per-arm budget (b)=32026.050.122
CSRPrompt set size (K)=30, Per-arm budget (b)=32026.050.117
UniformPrompt set size (K)=100, Per-arm budget (b)=52026.050.066
UniformPrompt set size (K)=100, Per-arm budget (b)=102026.050.044
UniformPrompt set size (K)=30, Per-arm budget (b)=102026.050.037
UniformPrompt set size (K)=50, Per-arm budget (b)=82026.050.037
UniformPrompt set size (K)=100, Per-arm budget (b)=82026.050.037
UniformPrompt set size (K)=50, Per-arm budget (b)=102026.050.036
UniformPrompt set size (K)=30, Per-arm budget (b)=82026.050.03
UniformPrompt set size (K)=50, Per-arm budget (b)=32026.050.021
UniformPrompt set size (K)=50, Per-arm budget (b)=52026.050.021
UniformPrompt set size (K)=100, Per-arm budget (b)=32026.050.021
UniformPrompt set size (K)=30, Per-arm budget (b)=32026.050.015
UniformPrompt set size (K)=30, Per-arm budget (b)=52026.050