8000 Comparing openvla:main...moojink:main · openvla/openvla · GitHub
[go: up one dir, main page]
More Web Proxy on the site http://driver.im/
Skip to content
Permalink

Comparing changes

Choose two branches to see what’s changed or to start a new pull request. If you need to, you can also or learn more about diff comparisons.

Open a pull request

Create a new pull request by comparing changes across two branches. If you need to, you can also . Learn more about diff comparisons here.
base repository: openvla/openvla
Failed to load repositories. Confirm that selected base ref is valid, then try again.
Loading
base: main
Choose a base ref
...
head repository: moojink/openvla-oft
Failed to load repositories. Confirm that selected head ref is valid, then try again.
Loading
compare: main
Choose a head ref
  • 19 commits
  • 48 files changed
  • 1 contributor

Commits on Feb 28, 2025

  1. Public release: OpenVLA-OFT 

    First commit for paper: "Fine-Tuning Vision-Language-Action Models:
Optimizing Speed and Success" - Moo Jin Kim, Chelsea Finn, Percy Liang

Website: https://openvla-oft.github.io/
    @moojink
    moojink committed Feb 28, 2025
    Configuration menu
    Copy the full SHA
    3758245 View commit details
    Browse the repository at this point in the history

  2. Update README: Add arXiv link

    @moojink
    moojink authored Feb 28, 2025
    Configuration menu
    Copy the full SHA
    9bccb2a View commit details
    Browse the repository at this point in the history

Commits on Mar 1, 2025

  1. Update finetune.py: Sync model files on master process only 

    Prevents race conditions in multi-GPU training runs.
    @moojink
    moojink committed Mar 1, 2025
    Configuration menu
    Copy the full SHA
    40d6c89 View commit details
    Browse the repository at this point in the history

  2. Update ALOHA.md: Add ALOHA fine-tuning & eval instructions

    @moojink
    moojink committed Mar 1, 2025
    Configuration menu
    Copy the full SHA
    a45877f View commit details
    Browse the repository at this point in the history

Commits on Mar 7, 2025

  1. Update LIBERO.md: Add lfs to git clone command

    @moojink
    moojink committed Mar 7, 2025
    Configuration menu
    Copy the full SHA
    588f9f9 View commit details
    Browse the repository at this point in the history

  2. Update ALOHA eval instructions and VRAM usage note

    @moojink
    moojink committed Mar 7, 2025
    Configuration menu
    Copy the full SHA
    2fde5e9 View commit details
    Browse the repository at this point in the history

  3. Update openvla_utils.py: Add torch.inference_mode() to get_vla_action() 

    Also update ALOHA VRAM usage in README.md
    @moojink
    moojink committed Mar 7, 2025
    Configuration menu
    Copy the full SHA
    0b75f6f View commit details
    Browse the repository at this point in the history

  4. Update README: Update ALOHA VRAM usage to 18 GB 

    (Previous commit was supposed to do this)
    @moojink
    moojink committed Mar 7, 2025
    Configuration menu
    Copy the full SHA
    7f9efa6 View commit details
    Browse the repository at this point in the history

Commits on Mar 8, 2025

  1. Update aloha/constants.py: 25 Hz control, updated task configs

    @moojink
    moojink committed Mar 8, 2025
    Configuration menu
    Copy the full SHA
    d8bf499 View commit details
    Browse the repository at this point in the history

  2. Update ALOHA.md: Remove comment about Python 3.8.10 

    (Python 3.10 actually works for the client conda environment)
    @moojink
    moojink committed Mar 8, 2025
    Configuration menu
    Copy the full SHA
    b9d7702 View commit details
    Browse the repository at this point in the history

Commits on Mar 9, 2025

  1. Update BibTeX in README

    @moojink
    moojink authored Mar 9, 2025
    Configuration menu
    Copy the full SHA
    edc1f9e View commit details
    Browse the repository at this point in the history

Commits on Mar 26, 2025

  1. Update LIBERO.md: Remove lfs in modified_libero_rlds clone command

    @moojink
    moojink authored Mar 26, 2025
    Configuration menu
    Copy the full SHA
    1fe43cf View commit details
    Browse the repository at this point in the history

Commits on Apr 28, 2025

  1. Remove unnecessary usage of num_open_loop_steps in deploy functions

    @moojink
    moojink committed Apr 28, 2025
    Configuration menu
    Copy the full SHA
    72f94a8 View commit details
    Browse the repository at this point in the history

  2. Clean up unnecessary files and scripts from base repo

    @moojink
    moojink committed Apr 28, 2025
    Configuration menu
    Copy the full SHA
    d282364 View commit details
    Browse the repository at this point in the history

  3. Remove old base repo Bridge eval code for now 

    To avoid confusion; will add back later with OpenVLA-OFT integration
    @moojink
    moojink committed Apr 28, 2025
    Configuration menu
    Copy the full SHA
    f9bc894 View commit details
    Browse the repository at this point in the history

  4. Update names of saved rollout MP4s during eval

    @moojink
    moojink committed Apr 28, 2025
    Configuration menu
    Copy the full SHA
    43f33b9 View commit details
    Browse the repository at this point in the history

  5. Fix bug: diffusion inference with fewer steps at test time than train… 

    … time

This only affects evals with the OpenVLA-OFT diffusion variant when
the number of denoising steps at test time is different than the
amount specified during train time (e.g., fewer denoising steps for
faster inference)
    @moojink
    moojink committed Apr 28, 2025
    Configuration menu
    Copy the full SHA
    94d345f View commit details
    Browse the repository at this point in the history

  6. Add explicit lora_rank arg to deploy scripts 

    In case the user changes it from the default value (32) during
training; since the lora_rank at test time must match the value
used during train time
    @moojink
    moojink committed Apr 28, 2025
    Configuration menu
    Copy the full SHA
    ae22da4 View commit details
    Browse the repository at this point in the history

  7. Add OpenVLA-OFT policy trained on all 4 LIBERO suites combined 

    https://huggingface.co/moojink/openvla-7b-oft-finetuned-libero-spatial-object-goal-10
    @moojink
    moojink committed Apr 28, 2025
    Configuration menu
    Copy the full SHA
    213c5ff View commit details
    Browse the repository at this point in the history
Loading
0