---
license: apache-2.0
base_model: lerobot/smolvla_robotwin
tags:
- robotics
- vla
- smolvla
- lerobot
- robotwin
- imitation-learning
- multi-instruction
pipeline_tag: robotics
library_name: lerobot
---

# SmolVLA RoboTwin `place_container_plate` (50 ep, MULTI-instruction)

SmolVLA policy fine-tuned on 50 demonstration episodes of the **`place_container_plate`** task from **RoboTwin 2.0** (`demo_clean` config), with **per-episode random language instructions** sampled from RoboTwin's 100 instruction variations (seed=42 for reproducibility).

This is the **multi-instruction** counterpart to [`arrow-hf/smolvla-robotwin-place-container-plate-50ep`](https://huggingface.co/arrow-hf/smolvla-robotwin-place-container-plate-50ep) (which uses a single fixed instruction).

## Task

- **Robot**: Agilex dual-arm, end-effector control (16D state, 16D action)
- **Cameras**: 3 RGB streams — `dual_cam_global`, `cam_wrist_65`, `cam_wrist_75` (240×320, D435)
- **Control rate**: ~30 Hz (LeRobot metadata is 10 Hz; underlying RoboTwin sim ~30 Hz, used consistently for train/eval)
- **Instructions**: 50 unique sentences (one per episode), examples:
  - "Use the left arm to place the object into the basket"
  - "Pick the item up and drop it into the woven basket"
  - "Move the object from the table into the basket"

## Training

| Config | Value |
|---|---|
| Base checkpoint | `lerobot/smolvla_robotwin` |
| Training data | 50 RoboTwin demonstrations, **50 unique instructions** |
| Batch size | 32 |
| Steps | 6000 (~10-25 epochs) |
| Optimizer | AdamW, lr=1e-4 |
| Scheduler | Cosine, warmup=300, decay=6000 |
| Chunk size | 50 |

## Evaluation: Single vs Multi-Instruction Comparison

Evaluated in RoboTwin 2.0 simulator (`demo_clean` config), 10 episodes, `max_steps=400`, `action_chunk_exec=50`, **single fixed eval instruction** `"place the container on the plate"` (fair comparison).

| Variant | Eval setting | Success rate |
|---|---|---|
| Single-instruction training | Fixed `"place the container on the plate"` | **9/10 (90%)** |
| **Multi-instruction training (this model)** | Fixed `"place the container on the plate"` | **7/10 (70%)** |

The multi-instruction model trades some single-instruction performance for the ability to follow varied language commands. For tasks where instruction diversity helps (held-out new instructions), this trade-off may pay off.

## Usage

```python
from lerobot.policies.smolvla import SmolVLAPolicy

policy = SmolVLAPolicy.from_pretrained("arrow-hf/smolvla-robotwin-place-container-plate-50ep-multi")
```

See [LeRobot documentation](https://huggingface.co/docs/lerobot) for inference setup.

## Citation

Built on [SmolVLA](https://huggingface.co/lerobot/smolvla_base) and [SmolVLA-RoboTwin](https://huggingface.co/lerobot/smolvla_robotwin) pretrained base, fine-tuned on data collected from [RoboTwin 2.0](https://github.com/TianxingChen/RoboTwin).