Atlas

Models

← All models

Gemma 2 27B

Google · 2024-06-27 · 27.2B parameters

Gemma 2 27B is Google DeepMind's open-weight, text-to-text, decoder-only pre-trained checkpoint in the Gemma 2 family, released on June 27, 2024. Trained from scratch on 13 trillion primarily English tokens, it alternates 4,096-token local and 8,192-token global attention layers with grouped-query attention; Google designed full-precision inference for a single A100 80GB or H100 GPU or a Google Cloud TPU host.

Benchmark scores

BenchmarkScore
BBH (Open LLM Leaderboard v2)56.4
Capability113.8
Global PIQA - Non-Parallel (Log-Likelihood Accuracy)63.4
Global PIQA - Parallel (Log-Likelihood Accuracy)30.7
GPQA (Open LLM Leaderboard v2)35.1
IFEval (Open LLM Leaderboard v2)24.8
MATH Level 5 (Open LLM Leaderboard v2)16.6
MMLU-Pro (Open LLM Leaderboard v2)43.7
MuSR (Open LLM Leaderboard v2)44.0
Loading Atlas data…