Llama 2 7B / NQ / Never Retrieve

1 benchmarks · 1 observations · 1 sources

Scores at a glance

1 benchmarks · large dot: Llama 2 7B / NQ / Never Retrieve; small dots: closest models where reported

PopQA ↗21.43%

Llama-2 7B / NQ / Never Retrieve · PopQA · exact-match accuracy · Adapt-LLM · PopQA test set ↗

Closest models by score

Similar positions on shared benchmark leaderboards. Smaller gaps are closer.

Llama-2 7B / SQuAD / Never Retrieve →Meta

20.0 percentile-point gap

1 of 1 shared benchmark comparisons

Llama-2 7B / NQ / Always Retrieve →Meta

20.0 percentile-point gap

1 of 1 shared benchmark comparisons

Llama-2 7B / SQuAD / Always Retrieve →Meta

40.0 percentile-point gap

1 of 1 shared benchmark comparisons

How similarity is measured

Mean absolute gap in percentile positions on shared benchmark settings; lower is closer. At least 60% overlap is required, with a minimum of three when available. Independent evaluations are used when there are at least three; otherwise published report comparisons are used. Repeated model observations within one setting are averaged before ranking. Writing-style traits and composite indices are excluded. This describes similarity, not an overall quality ranking.

Filter comparisons to this model →

Sources

Every plotted observation links to its evaluation. Reported configurations, benchmark versions and task subsets can differ; similarity describes shared scores, not training lift.

When to Retrieve: Teaching LLMs to Utilize Information Retrieval Effectively — published result table ↗
Labruna, Tiziano, Campos, Jon Ander, Azkune, Gorka · 2024

Names used in source tables

Llama-2 7B / NQ / Never Retrieve