- Developer IDs normalized to kebab-case (WizardLM→wizard-lm, CohereForAI→cohere-for-ai, etc.)
- source_data URL now points to actual leaderboard CSV
- metric_id added to all MetricConfig entries
- Folder structure updated from alpaca_eval_v1/v2 to "AlpacaEval 1.0/2.0"
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
add AlpacaEval 1.0 and 2.0 leaderboard data (324 models)
Converts AlpacaEval v1 (102 models) and v2 (222 models) leaderboard CSVs
from https://github.com/tatsu-lab/alpaca_eval into EEE schema v0.2.2.
Metrics per model:
- Win Rate (raw, with bootstrap standard error)
- Length-Controlled Win Rate (v2 only)
- Discrete Win Rate
- Average Response Length
Submitted as part of the ACL 2026 Every Eval Ever shared task (Track 1).
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
