Timestral

A curated map of modern time-series forecasting architectures, covering classical baselines, linear models, transformers, diffusion models, and pretrained forecasters. It explains the mechanisms and trade-offs of representative approaches and provides compact, runnable PyTorch implementations, with a short README for each architecture.

Kronos is an open-source family of decoder-only foundation models for financial-market K-line sequences, developed by the repository authors and distributed with pretrained models through the Hugging Face Hub. A specialized tokenizer quantizes multidimensional OHLCV market data into hierarchical discrete tokens, which an autoregressive Transformer models for forecasting and other quantitative tasks. The Python predictor accepts historical market data and timestamps, applies preprocessing and normalization, generates probabilistic forecasts for future open, high, low, close, volume, and amount values, and reverses the normalization. It supports single-series and GPU-parallel batch prediction, with configurable sampling temperature, nucleus-sampling probability, and forecast-path count. The repository also provides a Qlib-based example pipeline for data preparation, tokenizer and predictor fine-tuning, and simple top-K backtesting; it explicitly describes this example as a demonstration rather than a production-ready trading system. The project is licensed under the MIT License.

Mitra-v2 Regressor is a tabular foundation model from AutoGluon for regression tasks. It is pretrained entirely on synthetic datasets sampled from mixed random-regressor priors, including a Hybrid SCM prior, and uses in-context learning with a 12-layer 2D Transformer that attends across rows and columns. Regression is formulated as classification over 1,000 target bins: the model predicts a distribution for each row, whose mean provides the point prediction and whose probabilities can also produce quantiles and probabilistic metrics such as CRPS. Using the accompanying mitra-finetune package, the model is fine-tuned for 50 steps and bagged across eight copies on a CUDA GPU. It accepts tabular features and regression targets and returns point predictions or full predictive distributions. The weights are distributed through Hugging Face under the Apache-2.0 License; the package is required because standard AutoGluon expects the scalar head used by Mitra-v1 rather than Mitra-v2's distributional head.

Mitra-v2 Classifier is a tabular foundation model from AutoGluon and Amazon for classification tasks. It is pretrained entirely on synthetic datasets sampled from random classifier priors, including a Hybrid SCM prior, rather than on real-world datasets. The model uses a 12-layer 2D Transformer with attention across both rows and columns and incorporates in-context learning. Its recommended recipe fine-tunes 50 steps on a user's table and bags eight model copies for prediction; it can be run through the mitra-finetune package or used as a Mitra model in AutoGluon Tabular. The repository describes it as the second generation of Mitra, with a longer context, more features, and an improved optimizer than Mitra-v1. The weights are distributed on Hugging Face under the Apache-2.0 license and require a CUDA GPU for the documented fine-tuning and eight-fold bagging recipe.

mitra-finetune is a Python package for fine-tuning and running inference with second-generation Mitra v2 tabular foundation-model checkpoints. It supports classification and regression through the `MitraFinetune` API, including `fit`, `predict`, `predict_proba`, and distributional regression outputs. Each fit performs a 50-step full fine-tuning run through AutoGluon's eight-fold bagging wrapper. The package uses capped, optionally class-balanced in-context support for large tables, narrows wide tables with top-K feature selection or truncated-SVD projection, and uses hierarchical label decomposition for classification problems exceeding the checkpoint's native 10-class head. Regression is represented as classification over 1,000 target bins, with predictions returned as point estimates or predictive distributions. The package requires Python 3.11–3.13, a CUDA GPU, AutoGluon with the Mitra extra, and the TabArena execution wrapper. It uses checkpoints hosted on Hugging Face, including `autogluon/mitra-classifier-2` and `autogluon/mitra-regressor-2`; FlashAttention is optional. The project is distributed from its Hugging Face repository and can be installed from a cloned checkout.