Ömer Yüksel


Articles tagged with benchmarking

Moby-Dick revisited

Retrying the earlier Moby Dick experiment with a more varied selection, including few frontier models.

Published:
Reconstructing Moby-Dick with large language models

An experiment testing whether LLMs can reproduce Moby Dick verbatim with mixed results.

Published:
The color memory test: a quick filter for language model quality

A simple yet effective test for weeding out unreliable LLMs using eye and hair color descriptions.