--- datasets: - Marcus2112/minipile_density-proportioned language: - en base_model: - EleutherAI/pythia-1.4b-deduped --- | Benchmark | Measure | | 1.4B MiniPile | 1.4B Density | Percentage Difference in Means | | ---------------- | ---------- | --- | -------------------------- | ------------------------------ | ------------------------------ | | ARC-Challenge | acc | ↑ | **0.1903 ± 0.0115** | 0.1852 ± 0.0114 | -2.6800 | | MMLU | acc | ↑ | 0.2295 ± 0.0035 | 0.2295 ± 0.0035 | 0.0000 | | HellaSwag | acc | ↑ | 0.2579 ± 0.0044 | **0.2589 ± 0.0044** | 0.3877 | | WinoGrande | acc | ↑ | **0.5185 ± 0.0140** | 0.5043 ± 0.0141 | -2.7387 | | Lambada (OpenAI) | acc | ↑ | 0.0000 ± 0.0000 | 0.0000 ± 0.0000 | - | | Lambada (OpenAI) | perplexity | ↓ | 1564928.5258 ± 118691.4565 | **1420846.8323 ± 106563.1327** | -9.2069 | | Lambada (Std) | acc | ↑ | 0.0000 ± 0.0000 | 0.0000 ± 0.0000 | - | | Lambada (Std) | perplexity | ↓ | 8848600.9409 ± 745031.8900 | **7916035.3527 ± 664805.9178** | -10.5391 | | BLiMP | acc | ↑ | **0.5483 ± 0.0017** | 0.5422 ± 0.0017 | -1.1125 | | ARC-Easy | acc | ↑ | **0.2715 ± 0.0091** | 0.2698 ± 0.0091 | -0.6262 |