List your company

ResearchFormerly WizWand

Recipe Generation

Benchmarks

DatasetSOTA methodMetricTrendResultsLast updated

LLaVA-v1.5-7b-FT

8.4SacreBLEU

12

Jun 4, 2026

FoodKG 50k (test)

0.405BLEU-1

11

Feb 26, 2026

Recipe1M (test)

Real (Ground Truth)

80.33Success Rate

6

Feb 26, 2026

Food.com (test)

0.289BLEU-1

5

Feb 26, 2026

Allrecipes (test)

4.97PPL (All)

4

Feb 26, 2026

Allrecipes (val)

5.02PPL (All)

4

Feb 26, 2026

Recipe Gen (test)

89.7Structural Accuracy

3

Feb 26, 2026

Recipe Corpus original (test)

Typed Language Model

9.67Perplexity

3

Feb 26, 2026

Recipe Corpus modified type (test)

17.62Perplexity

1

Feb 26, 2026