[Feature Request] Open-source training / fine-tuning scripts for MiniCPM-RobotManip

We are trying to adapt MiniCPM-RobotManip to a new robot arm for a specific manipulation task. The inference side is clear — vla_infer.py exposes the embodiment_id interface and action_head.max_num_embodiments for multi-embodiment support. However, the current repository only includes inference code; no training or fine-tuning scripts are provided for RobotManip.

Questions

1. Fine-tuning scripts: Is there a plan to open-source the training/fine-tuning code for MiniCPM-RobotManip? Even a minimal script covering data loading, optimizer setup, and checkpoint saving would be very helpful.

2. New embodiment registration: What is the recommended workflow for adding a new embodiment_id beyond the ones already in the released weights? Does this require retraining from scratch, or is there a fine-tuning path that only updates the action head for the new embodiment?

3. Data format: Could you document or provide a sample of the training data format expected by RobotManip (state shape, action shape, image resolution, etc.)? The inference script implies STATE_DIM = 80 and action shape (30, 80), but the mapping from real robot joints to this 80-dim space is unclear.

4. Data quantity guidance: Are there any ablation results or rule-of-thumb recommendations for how many demonstration episodes are needed to fine-tune on a new embodiment/task?

Any pointers — even a rough timeline or a link to an internal script — would be greatly appreciated. Thank you for the great work on this project!

Start with vla_infer.py and its embodiment_id, action_head.max_num_embodiments, STATE_DIM=80, and (30, 80) action shape. A complete contribution would need to define training or fine-tuning entry points, new-embodiment registration, the expected data format, and guidance on demonstration quantity; no training files or tests are named in the issue.