novastar112/pusht_96_norm4

96px PushT PPO成功轨迹数据集。这些轨迹是由1Hz PPO PushT求解器生成的，使用动作编解码器`norm4`，渲染的图像存储在`history[*].image`、`image_prev`和`image_next`中，分辨率为`96x96`，JPEG质量为`85`。数据集分为训练集和测试集，训练集包含500,000条记录，测试集包含200条记录，均为gzip压缩的JSONL格式。训练集和测试集的初始状态通过`init_state_hash`进行过滤以确保不重叠。`move`动作中的坐标使用归一化的[0, 1]坐标，并四舍五入到四位小数；执行时将它们解码回PushT 512x512物理坐标。渲染的图像为`96px`。

96px PushT PPO successful trajectory dataset. The trajectories are generated by a 1Hz PPO PushT solver with action codec `norm4`, rendered images in `history[*].image`, `image_prev`, and `image_next` at `96x96` and JPEG quality `85`. The dataset is split into train and test sets, with 500,000 records in the train set and 200 records in the test set, both in gzip-compressed JSONL format. Train/test initial states are filtered to be disjoint by `init_state_hash`. Coordinates in `move` actions use normalized [0, 1] coordinates rounded to exactly four decimal places; execution decodes them back to PushT 512x512 physical coordinates. Rendered images are `96px`.