.eval_results
A structured evaluation results directory in Hugging Face model repositories, containing YAML or JSON files that record a model's scores on various benchmarks, metric names and evaluation timestamps, which the platform aggregates automatically to generate leaderboards.
Timeline (last 90 days)
Once a Hugging Face dataset is officially marked as an 'official benchmark', its page automatically generates a leaderboard, with data coming from the .eval_results files published by each model repository.
The Hugging Face model repository allows publishers to upload a .eval_results directory containing structured files (usually YAML or JSON) that record the model's scores on a given benchmark, metric names, evaluation time, and so on.
All Facts (2)
Once a Hugging Face dataset is officially marked as an 'official benchmark', its page automatically generates a leaderboard, with data coming from the .eval_results files published by each model repository.
50%UnverifiedThe Hugging Face model repository allows publishers to upload a .eval_results directory containing structured files (usually YAML or JSON) that record the model's scores on a given benchmark, metric names, evaluation time, and so on.
50%