Hugging Face's logo

lsnu
/
VLAarchtests

Robotics
TensorBoard
vision-language-action
bimanual-manipulation
rlbench
rgbd
Model card Files
xet
 Metrics Community
VLAarchtests / code /reveal_vla_bimanual /docs /upstream_pins.md
lsnu's picture
lsnu
Add files using upload-large-folder tool
6fa1956 verified
preview code
|
raw
history blame contribute delete
1.07 kB

Upstream Pins

Pinned on 2026-03-22 in /workspace/third_party.

Mandatory benchmark stack:

  • peract_bimanual: bb0232a6ba3fe116566e9568f0c7af980ed6703d
  • RLBench: 8af748c51287989294e00c9c670e3330a0e35ed5
  • PyRep: b8bd1d7a3182adcd570d001649c0849047ebf197
  • YARR: 6822ff78602c77878b27d4cfe759ce029c67bffb

Optional published baseline:

  • AnyBimanual: 76024e48b0e9489101459e85bc909c126ec581b4

Reveal-proxy stack candidate:

  • IsaacLab: v2.3.1 was cloned for inspection, but it targets Python 3.11 and Isaac Sim 5.x.
  • For the frozen project scope of Python 3.10 on Ubuntu 22.04, env B should stay on an Isaac Sim 4.5-compatible Isaac Lab release instead of the latest branch.

Notes:

  • peract_bimanual defaults to 6 cameras and older Python/Torch pins. This repo overrides camera selection and environment creation rather than running the upstream install scripts unchanged.
  • RLBench headless execution on this RunPod host will require an X server setup because the base image does not currently ship X, xvfb, or nvidia-xconfig.