Auto71%Harbor: pendingKnowledge & Reasoning

language-modeling-recommendation

This is the Big-Bench version of our language-based movie recommendation dataset

How to run

ltb · language-modeling-recommendation

$ npm i -g lettokenburn-cli
$ ltb bench info language-modeling-recommendation

External harness. No Harbor adapter has been generated for this benchmark yet. `ltb bench info` prints the known links.

Runs happen on your machine with your own subscription; only the score summary is pushed. CLI docs

1 sourced entry

		Harness	Source	Confidence		Link
GPT-2	48.8	N/A	Independent	medium	Dec 8, 2021	arxiv.org

Every number links to where it was published. Low-confidence rows are muted and need per-cell verification.

This is the Big-Bench version of our language-based movie recommendation dataset

https://github.com/google/BIG-bench/tree/main/bigbench/benchmark_tasks/movie_recommendation

GPT-2 has a 48.8% accuracy, chance is 25%.

Auto71%

Auto: Extracted automatically and reviewed by an LLM; some fields may be missing.

Missing:

Confirmed by 1 source channel. Fields are extracted automatically and can be wrong.

Apache-2.0redistributable

Citation. Cite “Zero-Shot Recommendation as Language Modeling” (2021), arXiv:2112.04184.

Index record

Review status: auto published

Extraction confidence: 80%

Updated: Oct 3, 2026