BigCode·released 2024-02-201 source
StarCoder 2 15B benchmark scores: 4 benchmarks tracked. 0% of its scores are independently reproduced — the rest are self-reported or unverified.
Leads on 0 of 4 benchmarks·best result 57.7 on GSM8K (self-reported)·0 of 4 independently reproduced
| Benchmark | Score | Evidence status | Measured |
|---|---|---|---|
| GSM8K | 57.7accuracy (%) | self-reported· optimizedT1 | 2024-02-20Qwen2.5-Coder Technical Report |
| MMLU | 52.13accuracy (%) | self-reported· optimizedT1 | 2024-02-20Qwen2.5-Coder Technical Report |
| ARC AI2 | 29.6accuracy (%) | self-reported· optimizedT1 | 2024-02-20Qwen2.5-Coder Technical Report |
| Winogrande | 28.6accuracy (%) | self-reported· optimizedT1 | 2024-02-20Qwen2.5-Coder Technical Report |
Claims drawn from cited facts, not live model generation.
This model, developed by BigCode, is a 15-billion-parameter mid-size model released in 2024. Its scale balances efficiency and cost against capability: it offers more headroom than compact models while avoiding the expense of frontier-scale systems.
3 cited facts
This model is tracked across four benchmarks. It currently holds no top score on any of them. These scores are self-reported, so they should be treated as claims rather than independently verified results.
3 cited facts
Based on the disclosed harness, this model trails the state-of-the-art leader by an average of 45.06 points, a wide gap that places it well behind the front-of-pack. It holds the top score on none of the tracked benchmarks, indicating no genuine category leadership yet.
2 cited facts
2 facts cross-checked across data sources: 2 single-source.
Source facts, citations, and refresh stamp for this record.
Sources: curated_param_counts, epoch_benchmarks
StarCoder 2 15B is an AI model developed by BigCode, released 2024-02-20. Its benchmark record below tags every score as independently reproduced, self-reported, or unverified.
StarCoder 2 15B has recorded scores on 4 benchmarks, each shown with its evidence status.
StarCoder 2 15B has recorded scores on 4 benchmarks — Winogrande, MMLU, ARC AI2, GSM8K. The full table above shows each score with its evidence status.
0 of 4 recorded scores (0%) are independently reproduced rather than self-reported by the lab.
StarCoder 2 15B has 4 tracked claims: 4 self-reported.