pythia-1.4B-deduped: benchmark results

Provider: EleutherAI. Released 2023-04-03. Access: Open.

Aggregate Rating 1129 ± 15, rank #10342 of 10916 rated models, from 79 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
MMLU-by-task - High School Statistics41.67Accuracy (%)60.7
MMLU-by-task - Global Facts34Accuracy (%)58.4
MMLU-by-task - College Computer Science35Accuracy (%)42
MMLU-by-task - College Mathematics30Accuracy (%)38.1
MMLU-by-task - Electrical Engineering32.41Accuracy (%)36.5
MMLU-by-task - Professional Accounting28.72Accuracy (%)35.6
MMLU-by-task - Abstract Algebra27Accuracy (%)34
MMLU-by-task - High School Computer Science33Accuracy (%)33.8
MMLU-by-task - Business Ethics34Accuracy (%)33.2
MMLU-by-task - College Biology29.86Accuracy (%)33.1
MMLU-by-task - High School Physics27.15Accuracy (%)30.2
MMLU-by-task - Elementary Mathematics25.93Accuracy (%)27.5

Interactive version: theaggregate.ai/model?slug=pythia-1-4b-deduped · How It Works · Data refreshed daily, snapshot 2026-11-08.