#transformer-model

Kronos is the first open-source foundation model designed specifically for financial candlestick (K-line) sequences, trained on data from over 45 global exchanges. It utilizes a novel two-stage framework: a specialized tokenizer quantizes continuous, multi-dimensional K-line data into hierarchical discrete tokens, followed by pre-training a large autoregressive Transformer on these tokens. Kronos is built to handle the unique, high-noise characteristics of financial data, serving as a unified model for diverse quantitative tasks.