Chronos: Pretrained Models for Time Series Forecasting

This is a Google colab based notebook that allows you to test Amazon's Chronos Time Series Forecasting model against some demo data, or your own.

A pure C implementation of the Chronos-Bolt pretrained time series forecasting model.

Chronos is an open-source, institutional-grade time series forecasting engine for production. Bayesian uncertainty, multi-model ensembles, configurable pipelines, and compliance-ready audit trails with full data lineage and diagnostics—built on a rigorous modular architecture for deterministic, reproducible runs.

End-to-end time series forecasting system using Hugging Face Chronos and Amazon SageMaker. Includes training, deployment, and Streamlit interface.

Chrono - Forecasting Linguistic patterns to the future and beyond

AI-powered e-commerce competitor analysis with price forecasting (0.38% MAPE) and sentiment analysis using Chronos, Prophet, and Llama 3.3 70B

This project focuses on process monitoring using the NASA Bearing Dataset, leveraging the Chronos-Forecasting library to predict bearing data and evaluate prediction accuracy through the Mean Absolute Error (MAE) metric.

Time series forecasting with the Chronos model from Amazon

Chronos Forecasting

Time Series Forecasting with Chronos Models

Making time series forecasting with Amazon's Foundation Model Chronos simple and accessible.

Reproducible fev benchmark for US GDP forecasting across baseline, classical, ML, and Chronos-2 models.

Time-series AutoGluon Forecaster for the Forecasting problem category as per Ready Tensor specifications.

Sample set of notebooks that explore causal inference with DoWhy & Time-series forecasting with Chronos/AutoGluonTS

Integrating DeepSeek R1 7B and Chronos Time Series Bolt Tiny to better forecast zero-shot and uncertain events

This project revolutionizes energy forecasting by moving away from traditional statistical models (ARIMA) and data-hungry Deep Learning (LSTMs) to Foundation Models (Amazon Chronos, Google TimesFM).

This project presents a comprehensive comparison between Chronos AI (Amazon's T5-based model) and ARIMA (Auto-Regressive Integrated Moving Average) for forecasting airline passenger numbers.

Chronos forecasting framework with covariate integration, incremental training with automatic rollback, checkpoint-based resumable training for large date ranges, and cost-optimized AWS orchestration ($0 idle cost). Built for proprietary time series data with configuration-driven architecture and state-managed concurrent run prevention.

Chronos Engine 3.5 is a next-gen “what-if” simulation platform combining Monte Carlo forecasting, narrative generation, FRED macro data, Unsplash imagery, and browser-native TTS. Built with React/Next.js and Supabase, it delivers immersive alternate-world modeling and analyst-grade insights.

amazon-chronos-forecasting-test

Prediction/Forecasting using Chronos and Chronos_Bolt

Production forecasting using Amazon Chronos

POC for Forecasting with Chronos

Chronos Network Forecasting (and benchmarking) Project

Trying out Chronos foundational forecasting model.

Time series forecasting using Amazon Chronos models

AI-Powered Portfolio Forecasting System using Chronos-2

Testing Amazon Science's Chronos 2 forecasting model

Forecasting sales using Prophet and experimenting with Chronos