Synthetic dataset can train linear probes on huge vision models better than real images. Demonstrated by MIT researchers
🟢 Why it matters and usefull?
1️⃣ Take a frozen base model (DINO, CLIP, etc.)
2️⃣ Observe the gradients it produces on real images
3️⃣ Generate synthetic images so that the gradients match
4️⃣ Train a linear classifier - and it works better than training ining ng on thng on the original data
Why this is useful?
— works across models (generated for DINO → works great on CLIP)
— especially strong on fine-grained classifications, where micro-details matter
— helps to see what the model is really looking at: serious correlations, similar clusters, embedding space structure
This changes the understanding of data.
Before:
"You need to collect millions of images."
Now:
"You need to correctly generate dozens."
•••••••••••••••••••••••••••••••••••••
🤖 Data Science, ML & Big Data with @DataXplore