MosaicML·released 2023-05-051 source
MPT-7B benchmark scores: 10 benchmarks tracked. 0% of its scores are independently reproduced — the rest are self-reported or unverified.
Leads on 0 of 10 benchmarks·best result 70 on LAMBADA (self-reported)·0 of 10 independently reproduced
| Benchmark | Score | Evidence status | Measured |
|---|---|---|---|
| LAMBADA | 70accuracy (%) | self-reported· optimizedT1 | 2023-05-05Qwen Technical Report |
| HellaSwag | 68.53accuracy (%) | self-reported· optimizedT1 | 2023-05-05Qwen Technical Report |
| TriviaQA | 61.6accuracy (%) | unverified· optimizedT1 | 2023-05-05Llama 2: Open Foundation and Fine-Tuned Chat Models |
| PIQA | 61.2accuracy (%) | self-reported· optimizedT1 | 2023-05-05Qwen Technical Report |
| Winogrande | 37.2accuracy (%) | unverified· optimizedT1 | 2023-05-05Llama 2: Open Foundation and Fine-Tuned Chat Models |
| OpenBookQA | 35.2accuracy (%) | unverified· optimizedT1 | 2023-05-05Llama 2: Open Foundation and Fine-Tuned Chat Models |
| ARC AI2 | 23.47accuracy (%) | self-reported· optimizedT1 | 2023-05-05Qwen Technical Report |
| BBH | 14.13accuracy (%) | unverified· optimizedT1 | 2023-05-05Baichuan 2: Open Large-scale Language Models |
| GSM8K | 9.1accuracy (%) | unverified· optimizedT1 | 2023-05-05Baichuan 2: Open Large-scale Language Models |
| MMLU | 7.73accuracy (%) | self-reported· optimizedT1 | 2023-05-05Qwen Technical Report |
Claims drawn from cited facts, not live model generation.
This model was developed by MosaicML and introduces a 7-billion-parameter, mid-size design in 2023. That scale balances modest computational expense with enough capacity for strong general-purpose performance.
4 cited facts
This model is scored on 10 tracked benchmarks, but holds no current top score on any of them. Because the scores are self-reported, they should be read as vendor claims rather than independently verified results.
3 cited facts
On the disclosed harness, this model trails the leader by an average of 46.99 points, a wide gap that places it well behind the front of the pack. It holds the top score on none of the tracked benchmarks, so there is no category leadership to report.
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
MPT-7B is an AI model developed by MosaicML, released 2023-05-05. Its benchmark record below tags every score as independently reproduced, self-reported, or unverified.
MPT-7B has recorded scores on 10 benchmarks, each shown with its evidence status.
MPT-7B has recorded scores on 10 benchmarks — PIQA, OpenBookQA, Winogrande, TriviaQA, MMLU, ARC AI2, and 4 more. The full table above shows each score with its evidence status.
0 of 10 recorded scores (0%) are independently reproduced rather than self-reported by the lab.
MPT-7B has 10 tracked claims: 5 self-reported, 5 unverified.