VLM-R1: A Stable and Generalizable R1-style Large Vision-Language Model
Paper
•
2504.07615
•
Published
•
32
•
2
Tony Zhao
tianchez
AI & ML interests
Multimodal Agent, Generative AI
Recent Activity
commented on
a paper
about 2 months ago
VLM-R1: A Stable and Generalizable R1-style Large Vision-Language Model
updated
a model
about 2 months ago
omlab/VLM-R1-Qwen2.5VL-3B-Math-0305
updated
a model
about 2 months ago
omlab/Qwen2.5VL-3B-VLM-R1-REC-500steps
Organizations
tianchez's activity
commented
a
paper
about 2 months ago
updated
3
models
about 2 months ago
updated
a
collection
about 2 months ago
updated
a
Space
about 2 months ago
upvoted
a
paper
about 2 months ago
upvoted
2
articles
3 months ago
Article
Improving Object Detection through Reinforcement Learning with VLM-R1
By
and 5 others
•
•
2
Article
Trials, Errors, and Breakthroughs: Our Rocky Road to OVD SOTA with Reinforcement Learning
By
and 5 others
•
•
1We now share our latest insights in the blog here.
https://om-ai-lab.github.io/index.html
liked
2
Spaces
3 months ago
published
a
Space
3 months ago
upvoted
a
collection
3 months ago
Apply for community grant: Personal project (gpu)
#3 opened 3 months ago
by
tianchez
replied to
their
post
4 months ago
looks very cool!
reacted to
their
post with 👍
4 months ago
Post
4438
Introducing VLM-R1!
GRPO has helped DeepSeek R1 to learn reasoning. Can it also help VLMs perform stronger for general computer vision tasks?
The answer is YES and it generalizes better than SFT. We trained Qwen 2.5 VL 3B on RefCOCO (a visual grounding task) and eval on RefCOCO Val and RefGTA (an OOD task).
https://github.com/om-ai-lab/VLM-R1
GRPO has helped DeepSeek R1 to learn reasoning. Can it also help VLMs perform stronger for general computer vision tasks?
The answer is YES and it generalizes better than SFT. We trained Qwen 2.5 VL 3B on RefCOCO (a visual grounding task) and eval on RefCOCO Val and RefGTA (an OOD task).
https://github.com/om-ai-lab/VLM-R1
-
- 3 replies
Fixes 500 error for some users
1
#1 opened 4 months ago
by
Tonic
reacted to
their
post with ❤️
4 months ago
Post
4438
Introducing VLM-R1!
GRPO has helped DeepSeek R1 to learn reasoning. Can it also help VLMs perform stronger for general computer vision tasks?
The answer is YES and it generalizes better than SFT. We trained Qwen 2.5 VL 3B on RefCOCO (a visual grounding task) and eval on RefCOCO Val and RefGTA (an OOD task).
https://github.com/om-ai-lab/VLM-R1
GRPO has helped DeepSeek R1 to learn reasoning. Can it also help VLMs perform stronger for general computer vision tasks?
The answer is YES and it generalizes better than SFT. We trained Qwen 2.5 VL 3B on RefCOCO (a visual grounding task) and eval on RefCOCO Val and RefGTA (an OOD task).
https://github.com/om-ai-lab/VLM-R1
-
-
-
- 3 replies