Update evaluation results with benchmark scores

#4
by pwxcdct - opened
Owner

Summary

This PR updates the model card with proper evaluation results from the Open LLM Leaderboard.

Changes:

  • Added model-index metadata with benchmark scores
  • Replaced exaggerated performance claims with factual benchmark data
  • Added evaluation table showing standard benchmark results

Benchmark Scores Added:

Benchmark Score
MMLU (5-shot) 46.87
ARC-Challenge (25-shot) 53.07
HellaSwag (10-shot) 78.59
TruthfulQA (0-shot) 41.75
Winogrande (5-shot) 73.76
GSM8K (5-shot) 16.05

Source: Open LLM Leaderboard evaluation of meta-llama/Llama-2-7b-hf

pwxcdct changed discussion status to closed

Sign up or log in to comment