Gemini 1.5
Google DeepMind — Gemini
partial detail 7 method fields published
Evaluation
Long-sequence perplexity (NLL) diagnostic
Negative log-likelihood (NLL) of tokens at different positions over long documents (up to 1M tokens) and shuffled code repositories (up to 10M tokens); power-law fit
How it was run
Open a release to see the setup, scoring, and other details that the lab published. Only fields the lab actually disclosed are shown.