RolexAlexander/cape-chemistry-tutor-gemma-3
Updated
Rolex Antoine Alexander
RolexAlexander
AI & ML interests
None yet
Recent Activity
updated
a model
23 days ago
RolexAlexander/cape-chemistry-tutor-gemma-3
updated
a model
25 days ago
RolexAlexander/cape-chemistry-gemma3-cpt-lora
published
a model
25 days ago
RolexAlexander/cape-chemistry-gemma3-cpt-lora
Organizations
None yet
updated
a
model
23 days ago
updated
a
model
25 days ago
published
a
model
25 days ago
liked
a
model
25 days ago
updated
a
model
27 days ago
published
a
model
27 days ago
updated
a
model
about 1 month ago
published
2
models
about 1 month ago
updated
2
models
4 months ago
published
2
models
4 months ago
updated
2
models
5 months ago
published
a
model
5 months ago
updated
a
model
6 months ago
published
a
model
6 months ago
reacted to
andito's
post with 🔥
6 months ago
Post
4065
🧠👁️ Can AI visualize solutions?
Humans often solve visual problems by sketching ideas in our minds. What if Vision-Language Models (VLMs) could do something similar, not by generating full images, but by using internal “mental sketches”?
That’s the idea behind Mirage, a new framework that empowers VLMs to reason using latent visual tokens. Instead of just thinking in words, Mirage mixes in abstract visual representations that help the model solve complex tasks.
These aren't photorealistic images. They're compact, internal representations optimized purely to support reasoning.
🔧 Mirage is trained in two phases:
1) Grounding: It learns to produce latent tokens anchored in real images.
2) Refinement: The model drops the images and learns to generate visual tokens on its own.
📈 And yes, it works!
On challenging benchmarks like Visual Spatial Planning, Jigsaw puzzles, and Spatial Attention Tasks, Mirage clearly outperforms GPT-4o and other strong baselines.
Smart sketches > empty words.
By mimicking the way humans visualize solutions, Mirage gives AI a new kind of imagination, one that’s faster, more efficient, and more human-like.
Kudos to the teams at UMass Amherst and MIT behind this exciting work.
Check the paper: Machine Mental Imagery: Empower Multimodal Reasoning with Latent Visual Tokens (2506.17218)
Humans often solve visual problems by sketching ideas in our minds. What if Vision-Language Models (VLMs) could do something similar, not by generating full images, but by using internal “mental sketches”?
That’s the idea behind Mirage, a new framework that empowers VLMs to reason using latent visual tokens. Instead of just thinking in words, Mirage mixes in abstract visual representations that help the model solve complex tasks.
These aren't photorealistic images. They're compact, internal representations optimized purely to support reasoning.
🔧 Mirage is trained in two phases:
1) Grounding: It learns to produce latent tokens anchored in real images.
2) Refinement: The model drops the images and learns to generate visual tokens on its own.
📈 And yes, it works!
On challenging benchmarks like Visual Spatial Planning, Jigsaw puzzles, and Spatial Attention Tasks, Mirage clearly outperforms GPT-4o and other strong baselines.
Smart sketches > empty words.
By mimicking the way humans visualize solutions, Mirage gives AI a new kind of imagination, one that’s faster, more efficient, and more human-like.
Kudos to the teams at UMass Amherst and MIT behind this exciting work.
Check the paper: Machine Mental Imagery: Empower Multimodal Reasoning with Latent Visual Tokens (2506.17218)
- 4 replies
published
a
model
7 months ago