samankeon.com#llm
- 2026
- RoPE: The Phase Geometry Behind Long-Context Transformers
- 2025
- From Greedy to Nucleus, How LLMs Choose the Next Token
- The Command Line is Learning to Talk
- Breaking the Quadratic Barrier
- Attention Isn't All You Need. You Also Need a Memory Budget
- From Word2Vec to BERT: The Evolution of Language Embeddings
- From Brute Force to Finesse: The Story of Parameter-Efficient Fine-Tuning (PEFT)
- RAG: Giving LLMs an External Brain
- How MoE Powers Modern LLMs
- Continuous Batching: The Secret Sauce of High-Throughput LLM Inference