team-suzuki/

hle-extract-qwen3235ba22b-20250815

	HLE Extract: Qwen3-235B-A22B Evaluation Results (2025-08-15)




	
	Dataset Description

This dataset contains the complete Human-Level Evaluation (HLE) benchmark with detailed evaluation results from the Qwen/Qwen3-235B-A22B model. It merges the original team-suzuki/hle-extract dataset with comprehensive model responses and human judgments.

Total Questions: 120 (complete HLE dataset) Evaluated Questions: 103 (85.8%) Unevaluated Questions: 17 (14.2%)… See the full description on the dataset page: https://huggingface.co/datasets/team-suzuki/hle-extract-qwen3235ba22b-20250815.