Language Modeling on LightEval Benchmark Suite
13.77Macro AverageDCLM
Evaluation results
| Method | Links | |
|---|---|---|
| DCLMDataset Type=Non-synthetic, Training Steps=10K2026.04 | 13.77 | |
| Nemotron-HQ-SynthDataset Type=Synthetic, Training Steps=10K2026.04 | 13.54 | |
| REWIREDataset Type=Synthetic, Training Steps=10K2026.04 | 13.49 | |
| Ultra-FineWebDataset Type=Non-synthetic, Training Steps=10K2026.04 | 13 | |
| FineWeb-HQDataset Type=Non-synthetic, Training Steps=10K2026.04 | 11.82 | |
| CosmopediaDataset Type=Synthetic, Training Steps=10K2026.04 | 10.33 | |
| SYNTHDataset Type=Synthetic, Training Steps=10K2026.04 | 10.03 | |
| FineWeb-LQDataset Type=Non-synthetic, Training Steps=10K2026.04 | 8.83 |