StarCoderBase-1B: benchmark results
Provider: BigCode. Access: Open.
Aggregate Rating 1070 ± 44, rank #10906 of 10916 rated models, from 9 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard v1 - TruthfulQA MC2 | 45.79 | MC2 (%) (0-shot) | 32.2 |
| Open LLM Leaderboard v1 - GSM8K | 0.91 | Accuracy (%) (5-shot) | 14.6 |
| Open LLM Leaderboard v1 - MMLU | 26.67 | Accuracy (%) (5-shot) | 13.4 |
| EvalPlus (HumanEval+ & MBPP+) | 12.2 | Pass@1 avg (%) | 6.9 |
| Open LLM Leaderboard v1 - HellaSwag | 34.31 | Normalized accuracy (%) (10-shot) | 6.9 |
| Open LLM Leaderboard v1 - WinoGrande | 49.96 | Accuracy (%) (5-shot) | 2.7 |
| Open LLM Leaderboard v1 - ARC Challenge | 22.7 | Normalized accuracy (%) (25-shot) | 1.5 |
| CanItEdit (Descriptive) | 4.9 | pass@1 (%; share of the 105 hand-written Python editing prob | 0 |
| CanItEdit (Lazy) | 5.48 | pass@1 (%; share of the 105 hand-written Python editing prob | 0 |
Interactive version: theaggregate.ai/model?slug=starcoderbase-1b · How It Works · Data refreshed daily, snapshot 2026-11-08.