📢Introducing🔥#PyRCA🔥, an open-source Python ML Library for Root Cause Analysis (RCA) as a holistic framework to uncover complex time-series causal dependencies & automatically locate root causes in AIOps. Code: github.com/salesforce/PyR… Paper: arxiv.org/abs/2306.11417 (1/n)

Excited to unveil 🚀MOIRAI🚀 - our time series foundation model designed for universal forecasting, and 🚀LOTSA🚀, a largest collection of time series datasets with 27B observations.

Congratulations on the exciting announcement!

🚀 Exciting Update: Moirai is now open source! 🚀 Foundation models for time series forecasting are gaining momentum! If you’ve been eagerly awaiting the release of Moirai since our paper (Unified Training of Universal Time Series Forecasting Transformers

🔍 Excited for our “Engineering Energizers” series! Meet Sravanthi Konduru, Lead Member at Salesforce Engineering, innovating with our Warden AIOps platform. Discover how her team's AI integration reshapes Salesforce's production! #AIOps Read here 👇 sforce.co/49fbGFS

Happy to share our work has been accepted by ICML24. See you in Vienna!

Future-proof your predictions: Morai 1.1, our time-series foundation model for forecasting, achieves ~20% improvement in NMAE for low-frequency data. #TimeSeries #AIForecasting ⏲️ Code on Github: bit.ly/4d0810W ⏲️ Model on HugginFace: bit.ly/4d081hs ⏲️ Paper:

Thrilled to be attending ICML 2024 in Vienna! Excited to reconnect with old friends and make new connections. We will be giving an oral presentation on Moirai. Reach out if you're interested in discussing time-series foundation models.

It is very important to reduce KV cache memory consumption during long context inference. We introduce ThinK, a method that exploits substantial redundancy across the channel dimension of the KV cache. Our findings reveal that even KV caches optimized by eviction methods, such

Welcome to explore our KDD2024 paper, "Calibration of Time-Series Forecasting: Detecting and Adapting Context-Driven Distribution Shift", on Tuesday, August 27 11:00-13:00, at session "Time Series I", Room 121. ArXiv: lnkd.in/g99Zvc-y Github: lnkd.in/gwh9Wu54

(1/4) Foundation models are revolutionizing time series analysis—but their success depends on large, diverse, high-quality datasets, which poses a major challenge. Enter synthetic data, reshaping Time Series Foundation Models (TSFMs) & Time Series LLMs (TSLLMs). Our survey

🚀 We are happy to organize the BERT²S workshop NeurIPS Conference 2025 on Recent Advances in Time Series Foundation Models. 🌐 berts-workshop.github.io 📜Submit by August 22 🎓Speakers and panelists: Chenghao Liu Mingsheng Long Zoe Piran Danielle Maddix Ameet Talwalkar Qingsong Wen, PhD, Head of AI @ Squirrel AI

⏳ Time-series forecasting is essential for data-driven decision-making — yet it’s still a missing piece in AGI. 🚀 Meet Moirai 2.0 — our latest time-series foundation model, now back on top of the GIFT-Eval leaderboard. ⚡ Faster, 🎯 more accurate, and packed with upgrades. 📖

Ranking #1 on GIFTEval, time series foundational model Moirai 2.0 delivers 16% better accuracy, 44% faster inference, 96% smaller model size. Dig into the team's technical thread👇

It blows my mind that Moirai-2 hits 🔥5 Million🔥 downloads just one month after it's release. Time-series community deserves more attention!