From 3de7d2359a5aef7878d621d43f2361b3e5cab976 Mon Sep 17 00:00:00 2001 From: Maxime Ellerbach Date: Thu, 21 May 2026 15:37:37 +0200 Subject: [PATCH] fixing doc defaults args Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com> Signed-off-by: Maxime Ellerbach --- docs/source/policy_vla_jepa_README.md | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/docs/source/policy_vla_jepa_README.md b/docs/source/policy_vla_jepa_README.md index 977b68879..67f0d2015 100644 --- a/docs/source/policy_vla_jepa_README.md +++ b/docs/source/policy_vla_jepa_README.md @@ -84,12 +84,12 @@ Key parameters in `VLAJEPAConfig`: | Parameter | Default | Description | | ------------------------- | ------- | -------------------------------------------------------------- | -| `chunk_size` | 16 | Number of actions predicted per inference call | -| `n_action_steps` | 16 | Steps executed from the predicted chunk before re-planning | -| `num_video_frames` | 16 | Video clip length fed to the world model | +| `chunk_size` | 7 | Number of actions predicted per inference call | +| `n_action_steps` | 7 | Steps executed from the predicted chunk before re-planning | +| `num_video_frames` | 8 | Video clip length fed to the world model | | `enable_world_model` | `True` | Whether to load and train the V-JEPA2 predictor | | `world_model_loss_weight` | 0.1 | Weight of the JEPA prediction loss relative to the action loss | -| `num_inference_timesteps` | 10 | Euler integration steps for action denoising | +| `num_inference_timesteps` | 4 | Euler integration steps for action denoising | | `freeze_qwen` | `False` | Freeze the Qwen3-VL backbone and only train the action head | ---