List your company

ResearchFormerly WizWand

BookCorpus

Benchmarks

TaskDatasetSOTA resultTrendResults

Language Modeling

BookCorpus (test)

10.83PPL

18

Conditional Text Generation

BookCorpus

90.4BLEU-1

9

Language Modeling

BookCorpus (val)

17.81Perplexity

7

BookCorpus

90.58BLEU

5

LLM-as-judge evaluation

BookCorpus

2.4Diversity Rank

3

Open-ended text generation

BookCorpus Story Project Gutenberg

0.91MAUVE

2