Meta AI·released 2023-02-241 source
LLaMA-65B 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 86 on TriviaQA (unverified)·0 of 10 independently reproduced
| Benchmark | Score | Evidence status | Measured |
|---|---|---|---|
| TriviaQA | 86accuracy (%) | unverified· optimizedT1 | 2023-02-24Llama 2: Open Foundation and Fine-Tuned Chat Models |
| HellaSwag | 78.93accuracy (%) | unverified· optimizedT1 | 2023-02-24LLaMA: Open and Efficient Foundation Language Models |
| LAMBADA | 77.7accuracy (%) | self-reported· optimizedT1 | 2023-02-24Qwen Technical Report |
| PIQA | 65.6accuracy (%) | self-reported· optimizedT1 | 2023-02-24Qwen Technical Report |
| ARC AI2 | 59.33accuracy (%) | self-reported· optimizedT1 | 2023-02-24Qwen Technical Report |
| GSM8K | 54.4accuracy (%) | unverified· optimizedT1 | 2023-02-24Stanford HELM |
| Winogrande | 54accuracy (%) | unverified· optimizedT1 | 2023-02-24LLaMA: Open and Efficient Foundation Language Models |
| MMLU | 51.2accuracy (%) | unverified· optimizedT1 | 2023-02-24Llama 2: Open Foundation and Fine-Tuned Chat Models |
| OpenBookQA | 46.93accuracy (%) | unverified· optimizedT1 | 2023-02-24Llama 2: Open Foundation and Fine-Tuned Chat Models |
| BBH | 44.53accuracy (%) | self-reported· optimizedT1 | 2023-02-24Qwen Technical Report |
Claims drawn from cited facts, not live model generation.
This model was developed by Meta AI and released in 2023 as a large, 65-billion-parameter system. That scale implies substantial headroom for complex tasks, but also considerable computational expense.
4 cited facts
This model is tracked on 10 benchmarks and currently holds no top score on any of them. These scores are self-reported by the vendor and have not been independently confirmed, so treat them as claims rather than verified results.
3 cited facts
With an average gap of 23.95 points behind the leader, this model trails by a wide margin and sits well behind the front of the pack. It holds the top score on none of the benchmarks, so there is no 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
LLaMA-65B is an AI model developed by Meta AI, released 2023-02-24. Its benchmark record below tags every score as independently reproduced, self-reported, or unverified.
LLaMA-65B has recorded scores on 10 benchmarks, each shown with its evidence status.
LLaMA-65B 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.
LLaMA-65B has 10 tracked claims: 4 self-reported, 6 unverified.