Views
No views yet
| Paper row | Table 1: OpenVLA-OFT Keep 2 Action (extreme action compression) |
| Dropped blocks | Action head (MLPResNet): reduce hidden dimension 4096 -> 256. Vision and language backbone untouched. |
| Recovery training | LoRA rank 32 (merged), batch size 16, 100K steps, lr 5e-4 |
| LIBERO success rate | Spatial 89.2 / Object 99.2 / Goal 76.6 / Long 90.8 / Avg 89.0 |
config.json (text_config.drop_attn_list / drop_mlp_list), so evaluation
reconstructs the pruned graph automatically. Requires the VLADrop modified
transformers (transformers-openvla-oft-dropped) from https://github.com/s1ghhh/VLADrop1python -m experiments.robot.libero.run_libero_eval_random_dropped \
2 --pretrained_checkpoint <this_repo_local_path> \
3 --task_suite_name libero_spatial \
4 --num_trials_per_task 50 --center_crop Trueaction_head--*.pt and proprio_projector--*.pt (L1 regression head, 2 input images + proprio).1@article{sun2026vladrop,
2 title={Drop-Then-Recovery: How Redundant Are Vision-Language-Action Models?},
3 author={Sun, Guoheng and Feng, Kaixi and He, Shwai and Gong, Xiaochuan and He, Yexiao and Wang, Ziyao and Shen, Zheyu and Ye, Wanghao and Kompella, Ramana Rao and Liu, Gaowen and Li, Ang},
4 journal={arXiv preprint arXiv:2606.27755},
5 year={2026}
6}