oasst-pythia-12B-flash-attn-5000-steps: benchmark results

Provider: Other. Access: Open.

Aggregate Rating 1147 ± 17, rank #10012 of 10916 rated models, from 68 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
MMLU-by-task - Abstract Algebra34Accuracy (%)82
MMLU-by-task - College Physics29.41Accuracy (%)71.1
MMLU-by-task - High School Physics31.13Accuracy (%)56.1
MMLU-by-task - Moral Scenarios25.47Accuracy (%)48.6
MMLU-by-task - Global Facts32Accuracy (%)44.3
MMLU-by-task - Elementary Mathematics27.25Accuracy (%)41.4
MMLU-by-task - ARC Challenge42.24Accuracy (%)37.2
MMLU-by-task - HellaSwag52.71Accuracy (%)35.1
MMLU-by-task - High School Mathematics25.93Accuracy (%)34.1
MMLU-by-task - High School US History29.41Accuracy (%)33
MMLU-by-task - Clinical Knowledge31.7Accuracy (%)32.7
MMLU-by-task - Astronomy30.92Accuracy (%)32.4

Interactive version: theaggregate.ai/model?slug=oasst-pythia-12b-flash-attn-5000-steps · How It Works · Data refreshed daily, snapshot 2026-11-08.