Writings
Technical notes, deep dives, and short essays. Full archive on bearblog.
- REFRAG: Recursive Fragmentation for Efficient Retrieval-Augmented Decoding(RAG)17 Sep 2025
- Self-Supervised Learning (SSL) Deep Dive(Representation Learning)14 Sep 2025
- Mixture of Experts (MoE): Theory and Implementation(Architectures)19 Aug 2025
- KV Caching(Inference)18 Aug 2025
- Prefix Tuning(Fine-tuning)18 Aug 2025
- LoRA and QLoRA: Efficient Fine-Tuning for Large Language Models(Fine-tuning)03 Aug 2025
- Towards Autonomous Preference Formation in AI: When Does “Changing One's Mind” Become Meaningful?(Essay)03 Aug 2025
- Pairwise Ranking Problem: ML Interview(Notes)19 Jul 2025