WAM-Policy / README.md
dontKnow23456's picture codex's picture SteveZeyuZhang's picture codex's picture
Release snapshot
bb81548
|
Raw History Blame Contribute Delete
10 kB
---
license: apache-2.0
pipeline_tag: robotics
tags:
- robotics
- world-action-model
- quadruped
- loco-manipulation
- on-policy-distillation
- lora
- mujoco
- libero
datasets:
- AIGeeksGroup/QLM-Bench
base_model:
- robbyant/lingbot-va-base
- robbyant/lingbot-va-posttrain-libero-long
---
# WAM-Policy: World Action Models with On-Policy Distillation for Quadrupedal Loco-Manipulation
Weight repository for WAM-Policy checkpoints and prompt embeddings.
> **WAM-Policy: World Action Models with On-Policy Distillation for Quadrupedal Loco-Manipulation**
>
> [Junlin Long](https://github.com/JalenLoong)\*, [Xu Deng](https://github.com/Ethan-Xu-Deng)\*, [Yiran Wang](https://scholar.google.com/citations?user=eg_nBF0AAAAJ&hl=en), [Yue Yang](https://au.linkedin.com/in/yue-yang-982714286)†, [Yang Zhao](https://yangyangkiki.github.io/)†, and [Zeyu Zhang](https://steve-zeyu-zhang.github.io/)‡
>
> \*Equal contribution. †Corresponding authors. ‡Project lead.
>
> #### [Paper](https://arxiv.org/abs/2611.24011) | [Website](https://aigeeksgroup.github.io/WAM-Policy/) | [WAM-Policy code](https://github.com/AIGeeksGroup/WAM-Policy) | [QLM-Bench code](https://github.com/AIGeeksGroup/WAM-Policy/tree/qlm-bench) | [QLM-Bench dataset](https://huggingface.co/datasets/AIGeeksGroup/QLM-Bench)
## ✏️ Citation
If you find our work helpful for your research, please consider citing our paper:
```bibtex
@article{long2026wampolicy,
title={WAM-Policy: World Action Models with On-Policy Distillation for Quadrupedal Loco-Manipulation},
author={Long, Junlin and Deng, Xu and Wang, Yiran and Yang, Yue and Zhao, Yang and Zhang, Zeyu},
journal={arXiv preprint arXiv:2611.24011},
year={2026}
}
```
---
## 🏃 Introduction
Arm-free quadrupeds can perform loco-manipulation, walking on three legs while manipulating with a front leg, yet no existing policy for it follows language instructions. World action models (WAMs) have emerged as effective language-conditioned manipulation policies, but are pretrained for robot arms and sample slowly. WAM-Policy turns one WAM into a quadrupedal loco-manipulation policy via two stages. The first, Cross-Morphology Policy Adaptation and Acceleration, adapts the arm-centric WAM to quadrupeds and distills the transferred policy into a student with one video and two action steps by on-policy self-distillation (OPSD). This student, however, is bounded by its teacher, which only imitates demonstrations. The second stage, Task-Specific Improvement and Merging, therefore tunes the teacher by reinforcement learning (RL) and merges the slow specialists into one few-step student by multi-teacher on-policy distillation (M-OPD). On QLM-Bench, ten simulated tasks with 1,000 demonstrations synthesized without teleoperation, the transferred policy reaches 77.0%, and the OPSD student keeps 76.7% while sampling 14 times faster. Task-specific RL and M-OPD lift it to 85.0%. We also deploy the OPSD student on a real Unitree Go2 Air. WAM-Policy thus achieves a stronger quadrupedal policy at a fraction of the inference cost.
![The OPSD student on the real robot (top) and in simulation (bottom)](./readme_assets/teaser.png)
- **Cross-morphology adaptation and acceleration.** Stage 1 replaces only the visual and action interfaces of a WAM, fine-tunes it on quadruped demonstrations, and accelerates it by a few-step form of on-policy self-distillation (OPSD), in which the same network with more sampling steps labels the student's own samples. The transferred policy reaches 77.0%, and the OPSD student keeps 76.7% (69.3% without distillation) while sampling 14 times faster.
- **Task-specific improvement and merging.** Stage 2 tunes specialist teachers by per-task reinforcement learning and merges them by multi-teacher on-policy distillation (M-OPD), which routes each context to the teacher of its task. On pushing a box and opening a door, the specialists succeed in every episode, against 47% and 53% for the OPSD teacher, and the M-OPD student reaches 90% and 87%, raising the ten-task success from 76.7% to 85.0%, above the OPSD teacher (77.0%).
- **QLM-Bench**, ten simulated loco-manipulation tasks with 1,000 language-annotated demonstrations synthesized without teleoperation, and evaluation in simulation and on a real Unitree Go2 Air, where the OPSD student runs without real-robot data.
### Method
![Stage 1: Cross-Morphology Policy Adaptation and Acceleration](./readme_assets/transfer.png)
![On-Policy Self-Distillation (Stage 1.2, top) and Stage 2 (bottom)](./readme_assets/opd.png)
### QLM-Bench
See the [QLM-Bench code README](https://github.com/AIGeeksGroup/WAM-Policy/tree/qlm-bench) for data preparation and evaluation setup,
and the [QLM-Bench dataset card](https://huggingface.co/datasets/AIGeeksGroup/QLM-Bench) for downloads, formats and splits.
![The ten tasks of QLM-Bench](./readme_assets/qlm_bench_tasks.png)
## Checkpoints
| Checkpoint | Folder | Weights | NFE (video / action) |
|---|---|---|---|
| Stage 1.1 teacher | `stage1_1_teacher/` | Full transformer in `transformer/` (~18.9 GiB) | 25 / 50 |
| Stage 1.2 OPSD student (S300 EMA) | `stage1_2_opsd_student/` | `student_adapter.pt` | 1 / 2 |
| Stage 2.1 specialists | `stage2_1_specialists/{push_box,open_room_door}/` | One `adapter.pt` per task | 25 / 50 |
| Stage 2.2 M-OPD student (S300 EMA) | `stage2_2_mopd_student/` | `student_adapter.pt` | 1 / 2 |
| Arm OPSD student (S300 EMA) | `supplementary/arm_libero_opsd_student/` | `student_adapter.pt` | 1 / 2 |
Quadruped adapters load unmerged on the Stage 1.1 teacher. The 1/2-NFE baseline without distillation uses the same
teacher weights. Arm adapters use the LingBot-VA LIBERO-Long base.
The [checkpoint manifest](./paper_checkpoints.json) lists file paths, sizes and SHA-256 digests. Each checkpoint
directory includes loader metadata (`wam_policy.json` for quadruped and `arm_libero.json` for LIBERO).
Keep `prompt_embeddings/<task>/embedding.pt` with its `reference.json` for each quadruped task.
See the [QLM-Bench dataset card](https://huggingface.co/datasets/AIGeeksGroup/QLM-Bench) for the task instructions.
## Ablation checkpoints
| Embodiment | Variant | Updates | Folder |
|---|---|---|---|
| Quadruped | Teacher cache only | 300, 2,000 | `ablations/quadruped/teacher_cache_only/{s300_ema,s2000_ema}/` |
| Quadruped | Streaming only | 300 | `ablations/quadruped/streaming_only/s300_ema/` |
| Quadruped | OPSD | 900 | `ablations/quadruped/opsd/s900_ema/` |
| Quadruped | RGB + proprioception | 5,000 | `ablations/quadruped/rgb_proprioception/s5000/` |
| Arm (LIBERO) | Teacher cache only | 300, 2,000 | `ablations/arm_libero/teacher_cache_only/{s300_ema,s2000_ema}/` |
| Arm (LIBERO) | Streaming only | 300 | `ablations/arm_libero/streaming_only/s300_ema/` |
| Arm (LIBERO) | Collected once | 300 | `ablations/arm_libero/collected_once/s300_ema/` |
| Arm (LIBERO) | SFT only | 692 | `ablations/arm_libero/sft_only/s692_ema/` |
| Arm (LIBERO) | OPSD | 900 | `ablations/arm_libero/opsd/s900_ema/` |
The ablation adapters use EMA weights. The RGB + proprioception checkpoint is a full transformer evaluated at
25/50 NFE. See Tab. 6 in the [code README's reproduction table](https://github.com/AIGeeksGroup/WAM-Policy#reproducing-the-paper) for the evaluation entry point.
To download one ablation, select its folder along with any base model and prompts it requires.
```bash
huggingface-cli download AIGeeksGroup/WAM-Policy --revision main \
--include "ablations/quadruped/teacher_cache_only/s300_ema/*"
```
## Upstream models
| Needed for | Repository | Revision |
|---|---|---|
| VAE and text encoder config for quadruped inference | [robbyant/lingbot-va-base](https://huggingface.co/robbyant/lingbot-va-base) | `68b7bc1b35da6ddc67ea94c4ceb58d768fbb3f9c` |
| Base model of the arm student | [robbyant/lingbot-va-posttrain-libero-long](https://huggingface.co/robbyant/lingbot-va-posttrain-libero-long) | `0e89d1e753019988aba484e8da2dc0810e264d9f` |
## Usage
Download only what you need, pinned to this branch, and the LingBot-VA base (VAE and text-encoder config):
```bash
pip install -U "huggingface_hub[cli]"
ROOT=$(huggingface-cli download AIGeeksGroup/WAM-Policy --revision main --quiet \
--include "stage1_1_teacher/*" "stage2_2_mopd_student/*" "prompt_embeddings/*")
LINGBOT_VA_BASE=$(huggingface-cli download robbyant/lingbot-va-base \
--revision 68b7bc1b35da6ddc67ea94c4ceb58d768fbb3f9c --quiet)
```
Serve a policy over qlm-live-v1 (two 640×360 RGB views + instruction → a chunk of 16 9-D commands). This needs the
WAM-Policy code and the `qlm_bench.live` package from QLM-Bench (`pip install -e "<QLM-Bench>[live]"`, or put
`<QLM-Bench>/src` on `PYTHONPATH`):
```bash
python -m wam_policy.inference.serve --policy mopd_student \
--teacher "$ROOT/stage1_1_teacher" --adapter "$ROOT/stage2_2_mopd_student" \
--base-model "$LINGBOT_VA_BASE" --prompt-root "$ROOT/prompt_embeddings" --port 8862
```
`--policy` accepts `teacher`, `specialist`, `opsd_student` and `mopd_student`. The default sampler comes from
`wam_policy.json`. Use the separate [LIBERO evaluation guide](https://github.com/AIGeeksGroup/WAM-Policy/blob/main/supplementary/arm_libero/README.md#evaluation)
for arm checkpoints.
Use the WAM-Policy [experiment configs](https://github.com/AIGeeksGroup/WAM-Policy/tree/main/configs) for training
and the [reproduction guide](https://github.com/AIGeeksGroup/WAM-Policy#reproducing-the-paper) for the Standard100
evaluation settings.
## License
The checkpoints in this repository are released under the [Apache License 2.0](https://www.apache.org/licenses/LICENSE-2.0).
The Stage 1.1 teacher adapts the LingBot-VA base with a 9-D action head and fine-tuning on QLM-Bench demonstrations, and the OPSD student, specialist teachers, and M-OPD student are LoRA adapters on that teacher.
The arm student is a LoRA adapter on the LIBERO-Long base, and the original weights are available from the repositories in [Upstream models](#upstream-models).