YangyiYY/rl-value-eval download history
YangyiYY/rl-value-eval is a text generation dataset on the Hugging Face Hub. In the last 30 days it was downloaded 27 times (3 in the last 7 days), and 93 times in total. It ranks #297,781 among datasets by monthly downloads.
rl-value-eval Held-out evaluation set (774 prompts) for the Block-3 confidence / correctness-estimation comparison. Pairs with YangyiYY/rl-value-confidence-train (training) and the nct-ppo actor/critic models. Each row is in the verl RL schema so the shared grader scores it directly.