X-VLA is a soft-prompted, flow-matching Vision-Language-Action framework that treats each robot or hardware setup as a "task" encoded with a small set of learnable Soft Prompt embeddings, letting a single model reconcile diverse robot morphologies, sensors, and action spaces. This policy has been trained and pushed to the Hub using LeRobot. Learn how to train and run it in the LeRobot xvla 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…
xvla_sharpie_holder_v1 is an open-weight model for robotics from Parsh Gandhi, released under Apache License 2.0. It has 880M parameters. At 16-bit it needs about 2.1 GB of GPU memory, which fits on 1x MI300X from $1.85 an hour, at the lowest prices in the SAVRN Index.
X-VLA is a soft-prompted, flow-matching Vision-Language-Action framework that treats each robot or hardware setup as a "task" encoded with a small set of learnable Soft Prompt embeddings, letting a single model reconcile diverse robot morphologies, sensors…
Parameters880M
Context—
Weights1.8 GB
Licenseapache-2.0
AccessOpen weights
Monthly Downloads—
Runs On
What it takes to serve xvla_sharpie_holder_v1 (880M 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 | 1.8 GB | 2.1 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
| 8-bit | 0.9 GB | 1.1 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
| 4-bit | 0.4 GB | 0.5 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.
xvla_sharpie_holder_v1 on every accelerator the SAVRN Index prices, at every precision
Model Card
By Parsh Gandhi, published under apache-2.0, revision a4ad01cab241.
X-VLA is a soft-prompted, flow-matching Vision-Language-Action framework that treats each robot or hardware setup as a "task" encoded with a small set of learnable Soft Prompt embeddings, letting a single model reconcile diverse robot morphologies, sensors, and action spaces. This policy has been trained and pushed to the Hub using LeRobot. Learn how to train and run it in the LeRobot xvla 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…
Read Parsh Gandhi's full model card
Model Card for xvla
X-VLA is a soft-prompted, flow-matching Vision-Language-Action framework that treats each robot or hardware setup as a "task" encoded with a small set of learnable Soft Prompt embeddings, letting a single model reconcile diverse robot morphologies, sensors, and action spaces.
This policy has been trained and pushed to the Hub using LeRobot.
Learn how to train and run it in the LeRobot xvla guide, or browse the full documentation.
Model Details
- License: apache-2.0
- Fine-tuned from: lerobot/xvla-base
- Robot type:
so_follower - Cameras:
front,wrist
Inputs & Outputs
The policy consumes these observation features and produces these action features.
Inputs
| Feature | Type | Shape |
|---|---|---|
observation.images.image |
VISUAL | (3, 256, 256) |
observation.images.image2 |
VISUAL | (3, 256, 256) |
observation.state |
STATE | (8,) |
observation.images.image3 |
VISUAL | (3, 224, 224) |
observation.images.empty_camera_0 |
VISUAL | (3, 224, 224) |
Outputs
| Feature | Type | Shape |
|---|---|---|
action |
ACTION | (6,) |
Training Dataset
- Repository: ParshG/so101_sharpie_holder_20261006_112937
- Episodes: 50
- Frames: 35964
- Frame rate: 30 FPS
- Task(s): "Put the Sharpie in the pen holder"
Training Configuration
| Setting | Value |
|---|---|
| Training steps | 40000 |
| Batch size | 8 |
| Optimizer | xvla-adamw |
| Learning rate | 0.0001 |
| Seed | 1000 |
| LeRobot version | 0.6.2 |
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=so_follower \
--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=ParshG/xvla_sharpie_holder_v1 \
--task="Put the Sharpie in the pen holder" \
--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
Fine-tune your own from the same base model, lerobot/xvla-base:
lerobot-train \
--dataset.repo_id=${HF_USER}/<dataset> \
--policy.path=lerobot/xvla-base \
--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}
}
Configuration
- Hidden size
- 1,024
Identity and Version
- Repository
- ParshG/xvla_sharpie_holder_v1
- Publisher
- Parsh Gandhi
- Task
- Robotics
- Modality
- Control
- Library
- lerobot
- Parameters
- 880M parameters
- Languages
- Not stated by the source
- Revision
- a4ad01cab2410bea7cd6719cbdef9802e7f6b9fa
- First published
- 2026-10-08
- Last updated
- 2026-10-08
Files and Weights
11 files, 1.8 GB in total. The weights are 3 files totalling 1.8 GB in safetensors.
Weights3 files · 1.8 GB
Configuration4 files · 22.3 KB
Tokenizer2 files · 3.6 MB
Documentation1 file · 6.3 KB
Repository1 file · 1.5 KB
Every file
| File | Type | Size | SHA-256 |
|---|---|---|---|
| model.safetensors | Weights | 1.8 GB | 2c3c4a18a302 |
| policy_postprocessor_step_0_unnormalizer_processor.safetensors | Weights | 7.6 KB | 9b5072e9d669 |
| policy_preprocessor_step_7_normalizer_processor.safetensors | Weights | 7.6 KB | 3f98ef132d08 |
| config.json | Configuration | 6.0 KB | — |
| policy_postprocessor.json | Configuration | 660 B | — |
| policy_preprocessor.json | Configuration | 2.5 KB | — |
| train_config.json | Configuration | 13.1 KB | — |
| README.md | Documentation | 6.3 KB | — |
| .gitattributes | Repository | 1.5 KB | — |
| tokenizer/tokenizer.json | Tokenizer | 3.6 MB | — |
| tokenizer/tokenizer_config.json | Tokenizer | 360 B | — |
License and Download
- License
- apache-2.0
- Access
- Open weights, no gate
- Download size
- 1.8 GB
Download from Parsh Gandhi
Released by Parsh Gandhi through its official repository on Hugging Face. Read the license.
Built From
- Derived from lerobot/xvla-base
- Described by arXiv:2510.10274
- Trained on (disclosed) ParshG/so101_sharpie_holder_20261006_112937
Memory Requirements
| Precision | Weights in memory |
|---|---|
| As published | 1.8 GB |
| 16-bit | 1.8 GB |
| 8-bit | 0.9 GB |
| 4-bit | 0.4 GB |
Weights only, from the published parameter count; the key-value cache and runtime add to this.
Questions About xvla_sharpie_holder_v1
How much GPU memory does xvla_sharpie_holder_v1 need?
About 2.1 GB at 16-bit and 0.5 GB at 4-bit: the weights (880M parameters) plus a working margin. A long context needs more.
What is the cheapest GPU to run xvla_sharpie_holder_v1 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 xvla_sharpie_holder_v1 commercially?
Yes. xvla_sharpie_holder_v1 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
Paper: Zheng et al., 2025, “X-VLA: Soft-Prompted Transformer as Scalable Cross-Embodiment Vision-Language-Action Model” (arXiv:2510.10274) Successful generalist Vision-Language-Action (VLA) models rely on effective training across diverse robotic platforms with large-scale, cross-embodiment, heterogeneous datasets. To facilitate and leverage the heterogeneity in rich robotic data sources, X-VLA introduces a Soft Prompt approach with minimally added parameters: we infuse prompt-learning concepts into cross-embodiment robot learning, introducing separate sets of learnable embeddings for each distinct embodiment. These embodiment-specific prompts empower VLA models to exploit cross-embodiment…
X-VLA is a Vision-Language-Action foundation model that uses soft prompts to handle cross-embodiment and cross-domain robot control within a unified Transformer architecture. Original paper: X-VLA: Soft-Prompted Transformer as Scalable Cross-Embodiment Vision-Language-Action Model For full installation details (including optional video dependencies such as ffmpeg for torchcodec), see the official documentation: https://huggingface.co/docs/lerobot/installation If you’re training / fine-tuning, you typically call forward(...) to get a loss and then: - -policy.chunksize=... - -policy.nactionsteps=... - -policy.maxactiontokens=... - -policy.gradientcheckpointing=true You can use the record…
Paper: Zheng et al., 2025, “X-VLA: Soft-Prompted Transformer as Scalable Cross-Embodiment Vision-Language-Action Model” (arXiv:2510.10274) Successful generalist Vision-Language-Action (VLA) models rely on effective training across diverse robotic platforms with large-scale, cross-embodiment, heterogeneous datasets. To facilitate and leverage the heterogeneity in rich robotic data sources, X-VLA introduces a Soft Prompt approach with minimally added parameters: we infuse prompt-learning concepts into cross-embodiment robot learning, introducing separate sets of learnable embeddings for each distinct embodiment. These embodiment-specific prompts empower VLA models to exploit cross-embodiment…
A vectorv1.1 network, Robotensor's two-arm Behavior Prompting Policy, submitted to Attune's Vector competition (Bittensor netuid 58). Its image encoders start from the Behavior Prompting Policy authors' own LIBERO policy (austinpatel/libero, MIT; Patel, Pekarek, Castro Hernandez and Song, 2026); the rest was trained on RoboTwin expert demonstrations of two Franka arms.
SmolVLA is a compact, efficient vision-language-action model that achieves competitive performance at reduced computational costs and can be deployed on consumer-grade hardware. This policy has been trained and pushed to the Hub using LeRobot. See the full documentation at LeRobot Docs. For a complete walkthrough, see the training guide. Below is the short version on how to train and run inference/eval: Writes checkpoints to outputs/train/ /checkpoints/. Prefix the dataset repo with eval\ and supply --policy.path pointing to a local or hub checkpoint.