Vision-Language-Action models for end-to-end robotic control. SmolVLA, RDT2-FM action generation.
AI & ML interests
None defined yet.
Recent Activity
Organization Card
/// RobotFlowLabs
Foundation Models
for Real Robots
We optimize vision, language, and action models for real-time edge deployment. Every model here is compressed, benchmarked, and ready for production robotics.
16+
Models
58
ANIMA Modules
4.5x
Max Compression
ROS2
Native
Mission
The robotics community deserves production-ready models, not just research checkpoints. We take the best open foundation models — CLIP, SAM2, DINOv2, Qwen, Depth Anything — and make them actually deployable on the hardware robots use: Jetson Orin, industrial PCs, edge GPUs.
Every model is quantized (INT4/INT8), exported (ONNX/SafeTensors/TorchScript), and benchmarked on real hardware. No guesswork, no "should work in theory" — measured performance on real silicon.
Model Collections
Organized by capability for the ANIMA robotics stack.
Perception
ANIMA Vision
Segmentation, features, depth estimation, and visual grounding for robotic scene understanding.
SAM2 · DINOv2 · CLIP · SigLIP · Depth Anything
Reasoning
ANIMA Language
INT4 quantized language models for instruction following, planning, and robotic reasoning.
Qwen2.5-7B · SmolLM2-1.7B
Understanding
ANIMA VLM
Vision-language models for visual QA, scene description, and grounding language to observations.
Qwen2.5-VL-7B · Qwen2.5-VL-3B
Action
ANIMA VLA
Vision-Language-Action models for end-to-end robotic control and manipulation.
SmolVLA · RDT2-FM · FORGE Students
FORGE — Compression Pipeline
Our 4-stage pipeline takes any 7B+ VLA model down to <2GB for real-time edge deployment.
Teacher Labels
5 teachers, multi-GPU
▶
Distillation
KD + curriculum
▶
Compression
Prune + quantize
▶
Export
ONNX / TRT / MLX
Automated hyperparameter optimization via Optuna · 400+ trials across 4 GPUs · W&B experiment tracking
models 12
robotflowlabs/qwen3-vl-4b-instruct-int4
Image-Text-to-Text • 5B • Updated • 15
robotflowlabs/qwen3-vl-2b-instruct-int4
Image-Text-to-Text • 2B • Updated • 12
robotflowlabs/qwen2.5-7b-instruct-int4
Text Generation • 8B • Updated • 12
robotflowlabs/depth-anything-v2-small
Depth Estimation • Updated
robotflowlabs/depth-anything-v2-large
Depth Estimation • Updated
robotflowlabs/siglip-so400m-patch14-384-int8
Image Feature Extraction • Updated • 11
robotflowlabs/dinov2-large-int8
Feature Extraction • Updated • 11
robotflowlabs/sam2.1-hiera-tiny-int8
Image Segmentation • Updated • 12
robotflowlabs/sam2.1-hiera-small-int8
Image Segmentation • Updated • 11
robotflowlabs/sam2.1-hiera-large-int8
Image Segmentation • Updated • 12
datasets 0
None public yet