Writing
Notes and deep dives on NLP, interpretability, and agentic systems.
Notion Chain-of-Thought Reasoning without Prompting An exploration of implicit reasoning capabilities in large language models without explicit chain-of-thought prompting. Notion Direct Logit Attribution A deep dive into interpretability methods for understanding model predictions through logit attribution. Medium BayLearn Machine Learning Symposium 2024 Key takeaways and highlights from the BayLearn 2024 symposium. Medium Multi-label Classification of User Utterances from Natural Language A technical walkthrough of building multi-label classification systems for user intent understanding.