Understanding MXFP4 Quantization
Visualizer for MXFP4 quantization
Kapil Sharma
KV Caching Illustrated
KV Caching: Training vs Inference in Multi-Head Attention
GPT OSS - OpenAI Reference Implementation
Load OpenAI's reference implementation
GPT OSS - Inference Huggingface Model
Testing out the huggingface version for the gpt-oss-20b locally on consumer hardware
Transformers for Recommender Systems - Part 3
Regularization to improve duplication penalty loss
GPU Mode - Triton Internals Talk
How Triton Compiler Works Under the Hood!
Deep Dive into Triton Internals (Part 3)
Enough MLIR to be dangerous - how Triton uses MLIR passes to progressively lower IR
Transformers for Recommender Systems - Part 2
Improve model to reduce duplicates and improve training performance
Transformers for Recommender Systems - Part 1
Exploring a simple transformer model for sequence modelling in recommender systems
Deep Dive into Triton Internals (Part 2)
What happens when triton.compile is called in the frontend?