-
RoPE to NoPE and Back Again: Hybrid Attention for Long-Context LLMs
A Manim-based explainer of RoPE (Rotary Position Embedding) vs NoPE (No Position Embedding) in long-context LLMs.
-
Visualizing DBMSolver: Training-Free Diffusion Bridge Sampler
A Manim-based explainer of DBMSolver, a training-free diffusion bridge sampler for image-to-image translation.
-
Visualizing Deformable DETR Attention
Visualizations of multi-scale deformable attention mechanism in DETR.
-
Getting an ML Research Internship
How to apply for and get an ML Research Internship as an Indian undergrad