reports/rlbench_debug_common23_pushbox_ep1/rollout_eval.md

RLBench Rollout Eval

• Checkpoint: /workspace/VLAarchtests_hf_rlbench/artifacts/outputs/rlbench_current/rlbench_subset3_backbone_only_clip_current_common23/checkpoint_best.pt
• Plan requested: False
• Plan applied: False
• Support-mode conditioning: True
• Task conditioning: True
• Geometry enabled: True
• World-model mode: checkpoint_default
• Mean success: 0.000

Per-task

• bimanual_push_box: mean_success=0.000, returns=[0.0]