Hands-On Build Guide: Paged-Attention KV Cache Manager with Attention-Score-Aware Eviction
A practical implementation of paged attention with score‑aware eviction, ready to extend for LLM serving workloads.
Cv
From-Scratch LoRA Fine‑Tuning Trainer with Manual Gradient Checkpointing and Adapter Merging in Pure PyTorch
A hands‑on guide to implementing LoRA fine‑tuning with manual gradient checkpointing and adapter merging in pure PyTorch, ready to run and extend.
Build a Pure‑Python Transformer Encoder with RoPE and Learned KV Cache
A step‑by‑step tutorial and runnable code for a minimal transformer encoder that uses RoPE and a mutable KV cache, perfect for a portfolio CV.
Pure‑Python Product‑Quantized IVF Index for Fast ANN Search
A step‑by‑step guide to building a pure‑Python product‑quantized IVF index for fast approximate nearest‑neighbor search, complete with runnable code, benchmarks, and extension paths for production use.
Build a LoRA Module from Scratch in Pure PyTorch
Learn to implement LoRA adapters in pure PyTorch, from theory to a runnable prototype, and how to extend it for production‑grade model adaptation.