$ render --page ml-notes

ML Notes

Topics

Notes By Topic

Basics

Cv

NeRF

Let’s take a detailed journey into Neural Radiance Fields (NeRF)—both at the conceptual and mathematical level. NeRF was introduced as a way to represent a 3D scene continuously using a...

Open note →

Score Distillation Sampling

Let’s dive deep into Score Distillation Sampling (SDS), a key technique introduced (for example, in DreamFusion) to distill the rich, text-conditioned knowledge of a pre-trained diffusion model into another model—in...

Open note →

Handson

01 Cloud Deploy

Fine-tuning Flan-T5-XXL on AWS SageMaker can be quite expensive due to the model’s size and computational requirements. The cost depends on several factors, including: Instance Type: You’ll likely need a...

Open note →

Transformers

Attention

prompt - can you please explain below excerpt in details in the context of attention, encoder-decoder architechture, "To understand the flow in more details,let us consider the French translation example...

Open note →

Encoder-Decoder Architecture

Encoder: The encoder processes the input sequence (e.g., a sentence) one timestep at a time. At each timestep, it produces a hidden state (hidden variable) that captures information about the...

Open note →