{"@attributes":{"version":"2.0"},"channel":{"title":"Pranav Jadhav","description":"Personal website and blog for Pranav Jadhav.","link":"https:\/\/pranavjadhav.com\/","item":[{"title":"Finetuning LLMs for ReAct","link":"https:\/\/pranavjadhav.com\/blog\/finetuning-llms-for-react\/","guid":"https:\/\/pranavjadhav.com\/blog\/finetuning-llms-for-react\/","description":"Unleashing the power of finetuning to improve multi-hop question-answering ability in LLMs."},{"title":"GPT-2 from scratch in MLX","link":"https:\/\/pranavjadhav.com\/blog\/gpt-2-from-scratch-in-mlx\/","guid":"https:\/\/pranavjadhav.com\/blog\/gpt-2-from-scratch-in-mlx\/","description":"A from-scratch GPT-2 implementation optimized for Apple silicon using MLX."},{"title":"Music Transformer","link":"https:\/\/pranavjadhav.com\/blog\/music-transformer\/","guid":"https:\/\/pranavjadhav.com\/blog\/music-transformer\/","description":"Training a transformer to generate MIDI piano compositions."},{"title":"Optimizing a MatMul with CuTe C++ (Part 1)","link":"https:\/\/pranavjadhav.com\/blog\/optimizing-a-matmul-with-cute\/","guid":"https:\/\/pranavjadhav.com\/blog\/optimizing-a-matmul-with-cute\/","description":"Implementing CUDA SGEMM optimizations with NVIDIA CuTe layouts and tensors."}]}}