Cosmopedia
Benchmarks
| Task Name | Dataset Name | SOTA Result | Trend | |
|---|---|---|---|---|
| Language Modeling | Cosmopedia (val) | Perplexity8.937 | 11 | |
| Language Modeling | Cosmopedia v2 | Final Perplexity (PPL)6.49 | 5 | |
| Maximum-likelihood infilling | Cosmopedia 30 prompts | Mean Reward184 | 2 |
Showing 3 of 3 rows