MultihopSpatial: Multi-hop Compositional Spatial Reasoning Benchmark for Vision-Language Model
AI & ML interests
Visual Intelligence, Pretrained Vision-and-Language Model, Embodied AI, Collaborative Agents, Vision Task(Object Detection, Segmentation)
Recent Activity
Organization Card
🔥 We are the Visual Intelligence Research Section in the Superintelligence Creative Research Laboratory, Electronics and Telecommunications Research Institute, Daejeon, South Korea
| 🚨 Safe LLaVA/Safe QWen/Safe Gemma : AI Safety-tuned Vision Language Model |
|---|
 |
| 🐨 KOALA : text-to-image generation | 🌋 Ko-LLaVA : Korean Vision-Language Model |
|---|---|
| (feat. Knowledge Distillation based Stable Diffusion XL) | (feat. Korean Large Language and Vision Assistant) |
 |
 |
models 16
etri-vilab/MultiHopSpatial-Qwen3-VL-4B-Instruct
Image-Text-to-Text • 4B • Updated • 20
etri-vilab/SafeLLaVA-7B
Image-Text-to-Text • 7B • Updated • 24 • 3
etri-vilab/SafeLLaVA-13B
Image-Text-to-Text • 13B • Updated • 10 • 3
etri-vilab/SafeQwen2.5-VL-32B
Image-Text-to-Text • 33B • Updated • 11 • 3
etri-vilab/SafeQwen2.5-VL-7B
Image-Text-to-Text • 8B • Updated • 57 • 3
etri-vilab/SafeGem-27B
Image-Text-to-Text • 27B • Updated • 16 • 3
etri-vilab/SafeGem-12B
Image-Text-to-Text • 12B • Updated • 20 • 3
etri-vilab/koala-lightning-1.7b
Text-to-Image • Updated • 1 • 2
etri-vilab/koala-lightning-1b
Text-to-Image • Updated • 15 • 9
etri-vilab/koala-lightning-700m
Text-to-Image • Updated • 970 • 9