Articles
Tech and related writings
What does a real dialogue with LLMs look like?
A future beyond discrete, turn-based LLM conversations.
LLMs as Complex Sequence Learners
A look at how LLMs learn, and what this says about their limitations.
Late Chunking on long documents
Late-chunking might be the closest we've ever been to solving retrieval with text embeddings models. This post gets it working on very long documents.
A Working Taxonomy for LLM Agents
This blog offers an initial, evolving taxonomy for LLM agents.
Fine-Tuning Embedding Models with Unsloth
Fine-tuning a BERT embeddings model with QLoRA using unsloth and Sentence Transformers.
Tips for LLM prompts
A PyTorch SLERP implementation
Merging an arbitrary number of Binary Trees
Complex Rayleigh Weight Initializations
Normalizing spectrograms for Deep Learning
