Rank Alignment Correlation on BigCodeBench Hard Set - Instruct
0.952Spearman RhoConv-to-Bench
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Conv-to-BenchEvaluator=Gemini-2.5-Flash, Configuration=Instructions-Only2026.04 | 0.952 | 0.001 | 0.857 | 0.002 | |
| Arena-Hard-Auto (Code)Evaluator=GPT-5-Mini2026.04 | 0.952 | 0.001 | 0.857 | 0.002 | |
| Arena-Hard-Auto (Code)Evaluator=Gemini-2.5-Flash2026.04 | 0.905 | 0.002 | 0.786 | 0.006 | |
| Conv-to-BenchEvaluator=GPT-5-Mini, Configuration=Instructions-Only2026.04 | 0.905 | 0.002 | 0.786 | 0.006 |