Text8
Benchmarks
| Task Name | Dataset Name | SOTA Result | Trend | |
|---|---|---|---|---|
| Character-level Language Modeling | text8 (test) | BPC1.038 | 128 | |
| Language Modeling | Text8 (test) | BPC1.23 | 35 | |
| Language Modeling | text8 | BPC1.02 | 23 | |
| Language Modeling | Text8 word-level (test) | Perplexity99.9 | 19 | |
| Language Modeling | Text8 word-level (val) | Perplexity81.64 | 18 | |
| Character-level language modeling | text8 | BPC0.98 | 16 | |
| Character-Level Language Modeling | text8 (held-out 1M tokens) | BPC2.3 | 14 | |
| Character-level language modeling | text8 (dev) | BPC1.01 | 13 | |
| Language Modeling | text8 English (test) | BPC1.133 | 8 | |
| Character-Level Language Modeling | text8 (most recent 1M tokens) | BPC2.23 | 7 | |
| Character-level Language Modeling | text8 100M regime Backward stream | Backward BPC2.17 | 7 | |
| Character-level Language Modeling | text8 100M regime (Current stream split) | Current BPC2.12 | 7 | |
| Character-level Language Modeling | text8 100M regime (Forward split) | Forward BPC2.19 | 7 | |
| Text Generation | text8 | BPC1.45 | 2 | |
| Unconditional text generation | text8 unconditional | Metric- | 0 |
Showing 15 of 15 rows