| --- |
| license: apache-2.0 |
| tags: |
| - robotics |
| - vla |
| - dreamzero |
| - lora |
| - arx_x5 |
| base_model: Luminis-Sim/DreamZero-AgiBot |
| --- |
| |
| # DreamZero — Epoch 4 checkpoint |
|
|
| LoRA fine-tune of [DreamZero-AgiBot](https://huggingface.co/Luminis-Sim/DreamZero-AgiBot) on the RoboDojo arx_x5 dataset (6-DOF dual-arm). |
| |
| - Checkpoint step: **35,000** (epoch 4) |
| - Action: relative for arm/head/waist, absolute for grippers |
| - Base model: DreamZero-AgiBot (7-DOF AgiBot G1 pretrain) |
| |
| ## Training loss |
| |
|  |
| |
| Red dashed lines mark epoch boundaries (step 8,750 / 17,500 / 26,250 / 35,000). |
| |
| | step | total | arm | ee (gripper) | aux | |
| |---:|---:|---:|---:|---:| |
| | 100 | 0.418 | 0.427 | 1.037 | 0.026 | |
| | 500 | 0.288 | 0.255 | 0.979 | 0.016 | |
| | 1,000 | 0.238 | 0.199 | 0.513 | 0.005 | |
| | 2,000 | 0.200 | 0.158 | 0.179 | 0.011 | |
| | 3,500 (warmup end) | 0.169 | 0.201 | 0.175 | 0.002 | |
| | 5,000 | 0.140 | 0.137 | 0.124 | 0.002 | |
| | **8,750 (epoch 1)** | 0.109 | 0.154 | 0.178 | 0.011 | |
| | 12,000 | 0.098 | 0.057 | 0.097 | 0.004 | |
| | **17,500 (epoch 2)** | 0.118 | 0.051 | 0.033 | 0.001 | |
| | 22,000 | 0.118 | 0.144 | 0.085 | 0.001 | |
| | **26,250 (epoch 3)** | 0.087 | 0.115 | 0.061 | 0.002 | |
| | 30,000 | 0.087 | 0.062 | 0.049 | 0.013 | |
| | **35,000 (epoch 4)** | **0.084** | **0.081** | **0.039** | **0.003** | |
| |
| Summary (step 0 → 35,000): |
| - Total loss: 0.42 → **0.08** (−80%) |
| - Arm loss: 0.43 → **0.08** (−81%) |
| - Gripper loss: 1.04 → **0.04** (−96%) |
| - Aux loss: 0.03 → **0.003** (live-channel mask working) |
| |
| ## Files |
| | File | Purpose | |
| |---|---| |
| | `model.safetensors` | LoRA weights (317 MB) | |
| | `experiment_cfg/conf.yaml` | Full Hydra training config (required at inference) | |
| | `config.json` | HF Trainer config | |
| | `trainer_state.json` | Per-step loss history | |
| | `loss_curves.png` | Loss visualization | |
|
|
| ## Bug fixes vs original DreamZero training |
|
|
| Includes the fixes shipped in [Luminis-Sim/XPolicyLab#31](https://github.com/Luminis-Sim/XPolicyLab/pull/31): |
| 1. J3-locked joint routing for 6-DOF arms onto AgiBot G1's 7-DOF slot layout |
| 2. Mask-aware action loss (live-channel detection) |
| 3. Gripper normalization q99 → min_max |
| 4. Per-modality (arm/ee/aux) loss logging |
| |
