- The T4 Ceiling: Why Vanilla VLMs Fail on Video
- Zero to ROS2 - Building pub/sub from scratch
- CNN Build - Weight Init Rabbit Hole
- understanding ZeRO!
- i tried to teach a math ai to say no
- DeepSeek V3 Deep Dive
- simplifying DDP
- PPO for LLMs: Why We Clip, Why We Normalize, and Why It's Still Hard
- Projects
- Bradley Terry Model
- From Points to Token: A 3D Learning Log - Part 1
- Engineering Fully Local Retrieval-Augmented Generation Systems
- SDXL something
- Log Number System for Low Precision Training
- Automatic Differentiation
- ResNet, So Simple Your Grandma Could Understand
- Wolpert's No Free Lunch Theorem
- Improving Data Quality
- Advanced Keras Techniques
- Data and Concept Drift