aaroncaozj/pi05_co-sft_blockpap_front

Checkpoints and Training Settings

👉 Normalize Statistics Files Stored in aaroncaozj/pi05_norm_stats_collection/BlockPAP-v0_Mix

Checkpoint Folder	norm_stats	Training Stride
/global_step_5000	norm_stats_unaligned.json	1
/global_step_15000	norm_stats_unaligned.json	1
/stride2_global_step_15000	norm_stats.json	2

Notes

• These models are trained using only front camera input.
• Using a stride when reading the dataset and constructing action labels with a larger timestep gap leads to more significant movement per action and faster task completion.
• Using different norm_stats.json results in different data distributions for the model to handle, so evaluation should use the same file as training.