List your company

Assortment optimization on SUSHI Preference Data Set (test)

80Worst CR30 real data
2841.55568.5Feb 6, 2026

Evaluation results

MethodLinks
30 real dataAssortment Size (B)=52026.02800.19999999
DescriptionModel=GPT-4o, Few-shot=false, Assortment Size (B)=22026.02710.2959695
DescriptionModel=GPT-4o, Few-shot=false, Assortment Size (B)=52026.02710.2999998
15 real dataAssortment Size (B)=52026.02710.24989899
Popularity scoreAssortment Size (B)=52026.0270—10010098
SamplingModel=GPT-4o, Few-shot=false, Assortment Size (B)=22026.02690.23979789
SamplingModel=GPT-4o, Few-shot=false, Assortment Size (B)=52026.02690.23999898
PersonaModel=GPT-4o, Few-shot=false, Assortment Size (B)=22026.02690.24979888
PersonaModel=GPT-4o, Few-shot=false, Assortment Size (B)=52026.02690.2410010099
DescriptionModel=GPT-5-mini, Few-shot=false, Assortment Size (B)=52026.02690.191009999
DescriptionModel=Mistral Large 3, Few-shot=false, Assortment Size (B)=52026.02670.19999998
PersonaModel=GPT-4o, Few-shot=true, Assortment Size (B)=22026.02660.22979693
PersonaModel=GPT-4o, Few-shot=true, Assortment Size (B)=52026.02660.221009999
SamplingModel=GPT-4o, Few-shot=true, Assortment Size (B)=22026.02650.23979692
SamplingModel=GPT-4o, Few-shot=true, Assortment Size (B)=52026.02650.23989899
30 real dataModel=Baseline, Assortment size (B)=22026.02650.19969790
30 real dataModel=Baseline, Assortment size (B)=52026.02650.19989797
DescriptionModel=Gemini 3 Flash, Few-shot=true, Assortment Size (B)=52026.02620.21989898
DescriptionModel=GPT-4o-mini, Few-shot prompting=false, Assortment size (B)=22026.02620.19979589
DescriptionModel=GPT-4o-mini, Few-shot prompting=false, Assortment size (B)=52026.02620.19999996
5 real dataAssortment Size (B)=22026.02610.29949593
DescriptionModel=GPT-4o-mini, Few-shot prompting=true, Assortment size (B)=22026.02610.19959488
DescriptionModel=GPT-4o-mini, Few-shot prompting=true, Assortment size (B)=52026.02610.19979796
SamplingModel=GPT-5-mini, Few-shot=false, Assortment Size (B)=22026.02600.39798100
SamplingModel=Gemini 3 Flash, Few-shot=false, Assortment Size (B)=22026.02600.319797100
DescriptionModel=GPT-4o, Few-shot prompting=false, Assortment size (B)=22026.02580.2959376
DescriptionModel=GPT-4o, Few-shot prompting=false, Assortment size (B)=52026.02580.2969695
BatchModel=GPT-4o, Few-shot=false, Assortment Size (B)=22026.02570.34929187
BatchModel=GPT-4o, Few-shot=false, Assortment Size (B)=52026.02570.34979798
DescriptionModel=Mistral Large 2, Few-shot prompting=false, Assortment size (B)=22026.02570.19939179
DescriptionModel=Mistral Large 2, Few-shot prompting=false, Assortment size (B)=52026.02570.19959692
BatchModel=GPT-4o-mini, Few-shot prompting=false, Assortment size (B)=22026.02540.26918976
BatchModel=GPT-4o-mini, Few-shot prompting=false, Assortment size (B)=52026.02540.26929485
BatchModel=Gemini 1.5 Flash, Few-shot prompting=true, Assortment size (B)=22026.02540.25979281
BatchModel=Gemini 1.5 Flash, Few-shot prompting=true, Assortment size (B)=52026.02540.25989791
Popularity scoreModel=Baseline, Assortment size (B)=22026.0253—9610056
Popularity scoreModel=Baseline, Assortment size (B)=52026.0253—9910089
15 real dataModel=Baseline, Assortment size (B)=22026.02530.24989591
15 real dataModel=Baseline, Assortment size (B)=52026.02530.24949597
DescriptionModel=GPT-4o, Few-shot prompting=true, Assortment size (B)=22026.02520.19949284
DescriptionModel=GPT-4o, Few-shot prompting=true, Assortment size (B)=52026.02520.19959490
PersonaModel=Gemini 1.5 Flash, Few-shot prompting=false, Assortment size (B)=22026.02510.24939280
PersonaModel=Gemini 1.5 Flash, Few-shot prompting=false, Assortment size (B)=52026.02510.24969797
BatchModel=Gemini 1.5 Flash, Few-shot prompting=false, Assortment size (B)=22026.02510.25958875
BatchModel=Gemini 1.5 Flash, Few-shot prompting=false, Assortment size (B)=52026.02510.25999895
DescriptionModel=Gemini 1.5 Flash, Few-shot prompting=false, Assortment size (B)=22026.02510.2908875
DescriptionModel=Gemini 1.5 Flash, Few-shot prompting=false, Assortment size (B)=52026.02510.2999894
BatchModel=GPT-4o-mini, Few-shot prompting=true, Assortment size (B)=22026.02500.25908573
BatchModel=GPT-4o-mini, Few-shot prompting=true, Assortment size (B)=52026.02500.25929380
SamplingModel=Mistral Large 3, Few-shot=false, Assortment Size (B)=22026.02470.28979388
SamplingModel=GPT-4o-mini, Few-shot prompting=true, Assortment size (B)=22026.02470.26929595
SamplingModel=GPT-4o-mini, Few-shot prompting=true, Assortment size (B)=52026.02470.26979696
PersonaModel=Mistral Large 2, Few-shot prompting=false, Assortment size (B)=22026.02470.23908765
PersonaModel=Mistral Large 2, Few-shot prompting=false, Assortment size (B)=52026.02470.23979792
RandomAssortment Size (B)=22026.02460.23858583
PersonaModel=Mistral Large 2, Few-shot prompting=true, Assortment size (B)=22026.02450.24878676
PersonaModel=Mistral Large 2, Few-shot prompting=true, Assortment size (B)=52026.02450.24979793
DescriptionModel=Mistral Large 2, Few-shot prompting=true, Assortment size (B)=22026.02450.24828172
DescriptionModel=Mistral Large 2, Few-shot prompting=true, Assortment size (B)=52026.02450.24888887
PersonaModel=GPT-4o, Few-shot prompting=false, Assortment size (B)=22026.02440.249698100
PersonaModel=GPT-4o, Few-shot prompting=false, Assortment size (B)=52026.02440.24999896
SamplingModel=GPT-4o-mini, Few-shot prompting=false, Assortment size (B)=22026.02440.39697100
SamplingModel=GPT-4o-mini, Few-shot prompting=false, Assortment size (B)=52026.02440.3868993
PersonaModel=GPT-4o, Few-shot prompting=true, Assortment size (B)=22026.02430.22979182
PersonaModel=GPT-4o, Few-shot prompting=true, Assortment size (B)=52026.02430.22989794
SamplingModel=Gemini 1.5 Flash, Few-shot prompting=false, Assortment size (B)=22026.02430.31969487
SamplingModel=Gemini 1.5 Flash, Few-shot prompting=false, Assortment size (B)=52026.02430.31868995
SamplingModel=GPT-4o, Few-shot prompting=false, Assortment size (B)=22026.02420.23969174
SamplingModel=GPT-4o, Few-shot prompting=false, Assortment size (B)=52026.02420.23939295
PersonaModel=GPT-4o-mini, Few-shot prompting=true, Assortment size (B)=22026.02410.22868378
PersonaModel=GPT-4o-mini, Few-shot prompting=true, Assortment size (B)=52026.02410.22989696
DescriptionModel=Gemini 1.5 Flash, Few-shot prompting=true, Assortment size (B)=22026.02400.21787873
DescriptionModel=Gemini 1.5 Flash, Few-shot prompting=true, Assortment size (B)=52026.02400.21949490
5 real dataModel=Baseline, Assortment size (B)=22026.02400.29919183
5 real dataModel=Baseline, Assortment size (B)=52026.02400.29869090
SamplingModel=GPT-4o, Few-shot prompting=true, Assortment size (B)=22026.02390.23979082
SamplingModel=GPT-4o, Few-shot prompting=true, Assortment size (B)=52026.02390.23949495
PersonaModel=GPT-4o-mini, Few-shot prompting=false, Assortment size (B)=22026.02380.22697673
PersonaModel=GPT-4o-mini, Few-shot prompting=false, Assortment size (B)=52026.02380.22969597
BatchModel=GPT-4o, Few-shot prompting=false, Assortment size (B)=22026.02370.34837871
BatchModel=GPT-4o, Few-shot prompting=false, Assortment size (B)=52026.02370.34899191
PersonaModel=Gemini 1.5 Flash, Few-shot prompting=true, Assortment size (B)=22026.02370.24717673
PersonaModel=Gemini 1.5 Flash, Few-shot prompting=true, Assortment size (B)=52026.02370.24858586
BatchModel=GPT-4o, Few-shot prompting=true, Assortment size (B)=22026.02350.34757769
BatchModel=GPT-4o, Few-shot prompting=true, Assortment size (B)=52026.02350.34848782
SamplingModel=Mistral Large 2, Few-shot prompting=true, Assortment size (B)=22026.02350.27767772
SamplingModel=Mistral Large 2, Few-shot prompting=true, Assortment size (B)=52026.02350.27929188
BatchModel=Mistral Large 2, Few-shot prompting=false, Assortment size (B)=22026.02350.34817968
BatchModel=Mistral Large 2, Few-shot prompting=false, Assortment size (B)=52026.02350.34878883
SamplingModel=Gemini 1.5 Flash, Few-shot prompting=true, Assortment size (B)=22026.02340.28707471
SamplingModel=Gemini 1.5 Flash, Few-shot prompting=true, Assortment size (B)=52026.02340.28828078
SamplingModel=Mistral Large 2, Few-shot prompting=false, Assortment size (B)=22026.02340.28968371
SamplingModel=Mistral Large 2, Few-shot prompting=false, Assortment size (B)=52026.02340.28979797
RandomModel=Baseline, Assortment size (B)=22026.02310.23716555
RandomModel=Baseline, Assortment size (B)=52026.02310.23838276
BatchModel=Mistral Large 2, Few-shot prompting=true, Assortment size (B)=22026.02300.35697367
BatchModel=Mistral Large 2, Few-shot prompting=true, Assortment size (B)=52026.02300.35798580