nanochat

We have hosted the application nanochat in order to run this application in our online workstations with Wine or directly.

Quick description about nanochat:

nanochat is a from-scratch, end-to-end “mini ChatGPT” that shows the entire path from raw text to a chatty web app in one small, dependency-lean codebase. The repository stitches together every stage of the lifecycle: tokenizer training, pretraining a Transformer on a large web corpus, mid-training on dialogue and multiple-choice tasks, supervised fine-tuning, optional reinforcement learning for alignment, and finally efficient inference with caching. Its north star is approachability and speed: you can boot a fresh GPU box and drive the whole pipeline via a single script, producing a usable chat model in hours and a clear markdown report of what happened. The code is written to be read—concise training loops, transparent configs, and minimal wrappers—so you can audit each step, tweak it, and rerun without getting lost in framework indirection.

Features:

• One-script “speedrun” from clean machine to chat model
• Full pipeline coverage: tokenizer, pretrain, SFT, optional RL, inference
• Minimal, readable training loops and configs for easy modification
• Web UI and CLI chat frontends with streaming responses
• Efficient inference with KV caching and throughput-friendly batching
• Automatic run artifacts and markdown reports for reproducibility

Programming Language: Python.
Categories: