Google DeepMind·released 2023-05-171 source
PaLM 2-L 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 86.1 on TriviaQA (self-reported)·0 of 4 independently reproduced
| Benchmark | Score | Evidence status | Measured |
|---|---|---|---|
| TriviaQA | 86.1accuracy (%) | self-reported· optimizedT1 | 2023-05-17PaLM 2 Technical Report |
| HellaSwag | 82.4accuracy (%) | self-reported· optimizedT1 | 2023-05-17PaLM 2 Technical Report |
| Winogrande | 66accuracy (%) | self-reported· optimizedT1 | 2023-05-17PaLM 2 Technical Report |
| ARC AI2 | 58.93accuracy (%) | unverified· optimizedT1 | 2023-05-17The Falcon Series of Open Language Models |
Claims drawn from cited facts, not live model generation.
This model was developed by Google DeepMind and released in 2023.
2 cited facts
A 4-benchmark record is thin; treat it as a sample of behavior under specific harnesses, too narrow to profile the model. Leading none of them places it with the field, a context that calibrates how to read its scores without capping what it can do. These scores are self-reported, meaning they are claimed by the vendor and have not been independently confirmed.
3 cited facts
With an average gap of 15.01 points to the leader and zero top scores on benchmarks, the model is well behind the frontier.
2 cited facts
1 facts cross-checked across data sources: 1 single-source.
Source facts, citations, and refresh stamp for this record.
Sources: epoch_benchmarks
PaLM 2-L is an AI model developed by Google DeepMind, released 2023-05-17. Its benchmark record below tags every score as independently reproduced, self-reported, or unverified.
PaLM 2-L has recorded scores on 4 benchmarks, each shown with its evidence status.
PaLM 2-L has recorded scores on 4 benchmarks — Winogrande, TriviaQA, ARC AI2, HellaSwag. 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.
PaLM 2-L has 4 tracked claims: 3 self-reported, 1 unverified.