tyrtleli/thinking-benchmark-deepseek-v4-pro download history

tyrtleli/thinking-benchmark-deepseek-v4-pro is a question answering dataset on the Hugging Face Hub. In the last 30 days it was downloaded 28 times (8 in the last 7 days), and 111 times in total. It ranks #289,218 among datasets by monthly downloads.

Thinking Benchmark — DeepSeek V4 Pro k=32 subset Five problems targeting DeepSeek V4 Pro's capability edge (25–75% success on the k=8 calibration pass; overall accuracy 75.4%). Selected with a preference for low censoring (few null/incomplete attempts) plus cross-model overlap for pai

Open tyrtleli/thinking-benchmark-deepseek-v4-pro on Hugging Face