TinyChat
LiveA 561M-parameter LLM trained from scratch for ~$95
561M
Parameters
~$95
Training Cost
65K tokens
Vocabulary
2048
Context Window
Overview
A language model built from scratch — custom BPE tokenizer, GPT architecture with RoPE and Multi-Query Attention, trained on ~38B tokens from FineWeb-EDU, then fine-tuned for conversation. Deployed on Modal serverless GPU with a Next.js frontend.
Screenshots
Tech Stack
PyTorchModal (T4 GPU)Next.jsTailwind CSSSSE Streaming