MolmoAct2 is an open vision-language-action model for robot control. It builds on Molmo2-ER, an embodied-reasoning VLM backbone, and connects the autoregressive VLM to a flow-matching continuous action expert through per-layer KV (key-value) conditioning. This checkpoint is the post-trained, multi-embodiment MolmoAct2 model. It is intended as a foundation checkpoint for further robot fine-tuning rather than as a ready-to-run policy for a single deployment setting. Use this checkpoint for further fine-tuning on a target robot embodiment or benchmark. It contains the VLM and continuous action expert weights, plus normalization metadata for the post-training mixture in normstats.json. This…
Open-weight model · Robotics
molmoact2-droid-pour-water-5ep
by Sang Nguyen tsangb34/molmoact2-droid-pour-water-5ep
molmoact2-droid-pour-water-5ep is an open-weight model for robotics from Sang Nguyen, released under Apache License 2.0. It has 5.4B parameters. At 16-bit it needs about 13.1 GB of GPU memory, which fits on 1x MI300X from $1.85 an hour, at the lowest prices in the SAVRN Index.
MolmoAct2 is an open robotics foundation model from the Allen Institute for AI (Ai2) that maps camera images and language instructions to robot action chunks. The LeRobot implementation supports training and evaluation of the regular MolmoAct2 model.
Parameters5.4B
Context—
Weights40.0 GB
Licenseapache-2.0
AccessOpen weights
Monthly Downloads—
Runs On
What it takes to serve molmoact2-droid-pour-water-5ep (5.4B parameters): the memory its weights need at each precision, and the cheapest way to rent enough data-center GPUs to hold them.
| Precision | Weights | Memory needed | Cheapest setup | Per hour | Also fits |
|---|---|---|---|---|---|
| 16-bit | 10.9 GB | 13.1 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
| 8-bit | 5.4 GB | 6.5 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
| 4-bit | 2.7 GB | 3.3 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
Memory is the weights at that precision plus 20% for the runtime and a short context; a long context needs more. Prices are the lowest on-demand hourly rates in the SAVRN Index, read Oct 18, 2026.
molmoact2-droid-pour-water-5ep on every accelerator the SAVRN Index prices, at every precision
Model Card
By Sang Nguyen, published under apache-2.0, revision ad8eee671103.
MolmoAct2 is an open robotics foundation model from the Allen Institute for AI (Ai2) that maps camera images and language instructions to robot action chunks. The LeRobot implementation supports training and evaluation of the regular MolmoAct2 model. This policy has been trained and pushed to the Hub using LeRobot. Learn how to train and run it in the LeRobot molmoact2 guide, or browse the full documentation. The policy consumes these observation features and produces these action features. Inputs Outputs New to LeRobot? These guides cover the full workflow: - Install LeRobot — set up the lerobot package. - Hardware setup — assemble, wire, and calibrate your robot and cameras. - Record data…
Read Sang Nguyen's full model card
Model Card for molmoact2
MolmoAct2 is an open robotics foundation model from the Allen Institute for AI (Ai2) that maps camera images and language instructions to robot action chunks. The LeRobot implementation supports training and evaluation of the regular MolmoAct2 model.
This policy has been trained and pushed to the Hub using LeRobot.
Learn how to train and run it in the LeRobot molmoact2 guide, or browse the full documentation.
Model Details
- License: apache-2.0
- Robot type:
franka - Cameras:
wrist,left_image,right_image
Inputs & Outputs
The policy consumes these observation features and produces these action features.
Inputs
| Feature | Type | Shape |
|---|---|---|
observation.state |
STATE | (8,) |
observation.cartesian_position |
STATE | (6,) |
observation.images.wrist |
VISUAL | (3, 360, 640) |
observation.images.left_image |
VISUAL | (3, 360, 640) |
observation.images.right_image |
VISUAL | (3, 360, 640) |
Outputs
| Feature | Type | Shape |
|---|---|---|
action |
ACTION | (8,) |
Training Dataset
- Repository: local/droid-pour-water
- Episodes: 100
- Frames: 91891
- Frame rate: 30 FPS
- Task(s): "power water"
Training Configuration
| Setting | Value |
|---|---|
| Training steps | 7180 |
| Batch size | 64 |
| Optimizer | adamw |
| Learning rate | 1e-05 |
| Seed | 1000 |
| LeRobot version | 0.6.1 |
How to Get Started with the Model
New to LeRobot? These guides cover the full workflow:
- Install LeRobot — set up the
lerobotpackage. - Hardware setup — assemble, wire, and calibrate your robot and cameras.
- Record data & train a policy — the end-to-end imitation-learning walkthrough.
- CLI cheat-sheet — quick reference for the
lerobot-*commands.
The short version to run and train this policy:
Run the policy on your robot
lerobot-rollout \
--strategy.type=base \
--robot.type=franka \
--robot.port=<your_robot_port> \
--robot.cameras="{ <camera_1>: {type: opencv, index_or_path: <index_or_path>, width: 640, height: 480, fps: 30}, <camera_2>: {type: opencv, index_or_path: <index_or_path>, width: 640, height: 480, fps: 30}}" \
--policy.path=tsangb34/molmoact2-droid-pour-water-5ep \
--task="power water" \
--duration=60
Replace the remaining <...> placeholders with your own values: --robot.port and the camera names/indices are specific to your machine, and the camera names must match the observation keys this policy was trained on.
When --strategy.type=base is used the script doesn't record the episodes. Skipping duration will make the policy run indefinitely. For more information look at rollout documentation.
Train your own policy
lerobot-train \
--dataset.repo_id=${HF_USER}/<dataset> \
--policy.type=molmoact2 \
--output_dir=outputs/train/<policy_repo_id> \
--job_name=lerobot_training \
--policy.device=cuda \
--policy.repo_id=${HF_USER}/<policy_repo_id> \
--wandb.enable=true
Writes checkpoints to outputs/train/<policy_repo_id>/checkpoints/.
Evaluation
No evaluation results have been provided for this policy yet.
Citation
If you use this policy, please cite the method linked in the description above, along with LeRobot:
@misc{cadene2024lerobot,
author = {Cadene, Remi and Alibert, Simon and Soare, Alexander and Gallouedec, Quentin and Zouitine, Adil and Palma, Steven and Kooijmans, Pepijn and Aractingi, Michel and Shukor, Mustafa and Aubakirova, Dana and Russi, Martino and Capuano, Francesco and Pascal, Caroline and Choghari, Jade and Moss, Jess and Wolf, Thomas},
title = {LeRobot: State-of-the-art Machine Learning for Real-World Robotics in Pytorch},
howpublished = "\url{https://github.com/huggingface/lerobot}",
year = {2024}
}
Identity and Version
- Repository
- tsangb34/molmoact2-droid-pour-water-5ep
- Publisher
- Sang Nguyen
- Task
- Robotics
- Modality
- Control
- Library
- lerobot
- Parameters
- 5.4B parameters
- Languages
- Not stated by the source
- Revision
- ad8eee671103ee2f20320a9e7eb1e4ba130aa54c
- First published
- 2026-10-13
- Last updated
- 2026-10-13
Files and Weights
21 files, 40.0 GB in total. The weights are 8 files totalling 40.0 GB in safetensors.
Weights8 files · 40.0 GB
Configuration11 files · 57.9 KB
Documentation1 file · 5.8 KB
Repository1 file · 1.5 KB
Every file
| File | Type | Size | SHA-256 |
|---|---|---|---|
| checkpoints/007180/pretrained_model/model.safetensors | Weights | 10.9 GB | 4d3ca6d791d4 |
| checkpoints/007180/pretrained_model/policy_postprocessor_step_1_molmoact2_masked_unnormalizer.safetensors | Weights | 10.2 KB | abd3a93a2618 |
| checkpoints/007180/pretrained_model/policy_preprocessor_step_3_molmoact2_masked_normalizer.safetensors | Weights | 10.2 KB | cba100b8bbf3 |
| checkpoints/007180/training_state/optimizer_state.safetensors | Weights | 18.3 GB | 012b81b782a7 |
| checkpoints/007180/training_state/rng_state.safetensors | Weights | 15.7 KB | 8004406a417c |
| model.safetensors | Weights | 10.9 GB | 4d3ca6d791d4 |
| policy_postprocessor_step_1_molmoact2_masked_unnormalizer.safetensors | Weights | 10.2 KB | abd3a93a2618 |
| policy_preprocessor_step_3_molmoact2_masked_normalizer.safetensors | Weights | 10.2 KB | cba100b8bbf3 |
| checkpoints/007180/pretrained_model/config.json | Configuration | 4.3 KB | — |
| checkpoints/007180/pretrained_model/policy_postprocessor.json | Configuration | 914 B | — |
| checkpoints/007180/pretrained_model/policy_preprocessor.json | Configuration | 2.8 KB | — |
| checkpoints/007180/pretrained_model/train_config.json | Configuration | 8.8 KB | — |
| checkpoints/007180/training_state/optimizer_param_groups.json | Configuration | 23.8 KB | — |
| checkpoints/007180/training_state/scheduler_state.json | Configuration | 385 B | — |
| checkpoints/007180/training_state/training_step.json | Configuration | 66 B | — |
| config.json | Configuration | 4.3 KB | — |
| policy_postprocessor.json | Configuration | 914 B | — |
| policy_preprocessor.json | Configuration | 2.8 KB | — |
| train_config.json | Configuration | 8.8 KB | — |
| README.md | Documentation | 5.8 KB | — |
| .gitattributes | Repository | 1.5 KB | — |
License and Download
- License
- apache-2.0
- Access
- Open weights, no gate
- Download size
- 40.0 GB
Download from Sang Nguyen
Released by Sang Nguyen through its official repository on Hugging Face. Read the license.
Built From
- Trained on (disclosed) local/droid-pour-water
Memory Requirements
| Precision | Weights in memory |
|---|---|
| As published | 40.0 GB |
| 16-bit | 10.9 GB |
| 8-bit | 5.4 GB |
| 4-bit | 2.7 GB |
Weights only, from the published parameter count; the key-value cache and runtime add to this.
Questions About molmoact2-droid-pour-water-5ep
How much GPU memory does molmoact2-droid-pour-water-5ep need?
About 13.1 GB at 16-bit and 3.3 GB at 4-bit: the weights (5.4B parameters) plus a working margin. A long context needs more.
What is the cheapest GPU to run molmoact2-droid-pour-water-5ep on?
At 16-bit, 1x MI300X from $1.85 an hour; at 4-bit, 1x MI300X from $1.85 an hour, at the lowest on-demand prices the SAVRN Index lists.
Can I use molmoact2-droid-pour-water-5ep commercially?
Yes. molmoact2-droid-pour-water-5ep is released under Apache License 2.0. The Apache License 2.0 is a permissive open-source license. It permits commercial use, modification and redistribution. It requires keeping the license and copyright notices and any NOTICE file, stating significant changes, and it includes an express patent grant from contributors.
Similar Models
MolmoAct2 is an open vision-language-action model for robot control. It builds on Molmo2-ER and attaches a flow-matching continuous action expert that conditions on the VLM key-value cache through a per-layer connection. This checkpoint is fine-tuned on the SO-100/101 mixture with absolute joint-pose control and annotated language instructions. It is intended for both further fine-tuning and SO-100/101 policy inference. Use this checkpoint for SO-100/101 inference or for further fine-tuning. Dataset normalization metadata is stored in normstats.json. pass normtag="so100so101molmoact2" at inference time. Continuous action prediction is the intended and recommended inference mode. Discrete…
MolmoAct2 is an open vision-language-action model for robot control. It builds on Molmo2-ER and attaches a flow-matching continuous action expert that conditions on the VLM key-value cache through a per-layer connection. This checkpoint is fine-tuned on the full LIBERO training mixture, combining Spatial, Object, Goal, and Long suites. It is intended for both further fine-tuning and LIBERO policy inference. Use this checkpoint for LIBERO inference or for further fine-tuning. Dataset normalization metadata is stored in normstats.json. pass normtag="libero" at inference time. Continuous action prediction is the intended and recommended inference mode. Discrete action prediction is exposed for…
MolmoAct2 is an open vision-language-action model for robot control. It builds on Molmo2-ER and attaches a flow-matching continuous action expert that conditions on the VLM key-value cache through a per-layer connection. This checkpoint is fine-tuned on the bimanual YAM mixture with absolute joint-pose control and annotated language instructions. It is intended for both further fine-tuning and bimanual YAM policy inference. Use this checkpoint for bimanual YAM inference or for further fine-tuning. Dataset normalization metadata is stored in normstats.json. pass normtag="yamdualmolmoact2" at inference time. Continuous action prediction is the intended and recommended inference mode. Discrete…
Arm b0fft (group baseline) of the MolmoAct2 baseline pair in experimental/lerobotpolicymolmoact2/modal/ (alpha-robotics, branch policytestfanqi), run with the same protocol as the π0.5 grid in experimental/lerobotpolicypi05/. The only test is the real robot: submit fanqi-robo/molmoact2busyboxpushgreenbuttonb0fft (or a step repo) on https://huggingface.co/spaces/armnet/armnet-eval (embodiment lerobot/so-101, task pushgreenbutton, 20 rollouts, variation seed 42). Native horizon: execute the full 30-step chunk (nactionsteps=30).
MolmoAct2 fine-tuned with LeRobot (lerobot commit c13d79e6e350c589babc5b7c5cf5ef05cb0fa58e), exported from /home/sakakibara/bachelorthesis/outputs/train/molmoact2yamspellceateceasyftcontinuousgpu220261008232437/checkpoints/008000/pretrainedmodel (trainer step 8000). LoRA adapters are merged; weights are float32. normstats.json holds the normalization statistics of the fine-tuning run under yamdualmolmoact2. lerobotcheckpoint/ holds the LeRobot policy config, training config, and processor files (no weights); exportmanifest.json records the source checkpoint hash and the export verification. Loading with dtype=torch.bfloat16 rounds the merged weights and loses much of the fine-tuning update.