{"version":1,"type":"story","url":"https://digestai.news/story/nvidia-releases-kumo-forecast-time-series-model-with-retrieval-and","json":"https://digestai.news/story/nvidia-releases-kumo-forecast-time-series-model-with-retrieval-and.json","markdown":"https://digestai.news/story/nvidia-releases-kumo-forecast-time-series-model-with-retrieval-and.md","slug":"nvidia-releases-kumo-forecast-time-series-model-with-retrieval-and","headline":"Nvidia releases Kumo Forecast time-series model with retrieval and interpretability","summary":"Nvidia has open‑sourced Kumo Forecast, a multivariate time‑series forecasting model that blends a pretrained Transformer encoder with a retrieval‑augmented forecasting path. The model builds on the MOMENT‑1‑large backbone, using an approximately 340‑million‑parameter encoder that remains frozen while a trainable forecasting head learns the target task.\n\nIn Domain‑Aware Representation and Retrieval (DARR) mode, Kumo Forecast pulls similar patterns from a separate historical DataFrame via a k‑nearest‑neighbor search and merges those signals with the direct forecast. An interpretability framework can then generate horizon‑specific lag and feature attributions, semantic‑flow diagnostics, and trajectory‑stability analyses without changing the underlying forecaster. The library offers a DataFrame‑first API, configurable horizons, and support for input windows of 256, 512, 1,024 or 2,048 observations.\n\nThe model code (NVIDIA/Kumo-TS) and pretrained weights (nvidia/Kumo-Forecast) are hosted on Hugging Face under an Apache 2.0 license, allowing developers to run inference out‑of‑the‑box or fine‑tune on their own CSV data.","keyPoints":["Kumo Forecast uses a 340‑million‑parameter MOMENT‑1‑large encoder plus a trainable forecasting head.","Domain‑Aware Representation and Retrieval (DARR) adds nearest‑neighbor pattern retrieval from a separate historical DataFrame.","The model and weights are released under Apache 2.0 on Hugging Face, supporting up to 2,048 observation windows."],"whyItMatters":"Combining pretrained representations with historical pattern retrieval and built‑in interpretability gives practitioners a ready‑to‑use, open‑source forecasting tool that can improve accuracy and transparency for multivariate time‑series tasks.","category":{"slug":"models","name":"Generative AI & Models","url":"https://digestai.news/category/models"},"entities":{"companies":["NVIDIA"],"models":["Kumo Forecast"],"people":[]},"firstPublishedAt":"2026-10-21T12:51:24Z","updatedAt":"2026-10-21T13:13:09Z","sourceCount":2,"hasPrimarySource":true,"sources":[{"outlet":"Hugging Face","title":"NVIDIA Kumo Forecast Combines Pretrained Time-Series Representations with Historical Pattern Retrieval and Interpretability","url":"https://huggingface.co/blog/nvidia/kumo-forecast","publishedAt":"2026-10-21T12:51:24Z","type":"primary","primary":true,"lead":true},{"outlet":"Unite.AI","title":"NVIDIA Announces Open Kumo Forecast Multivariate Time-Series Model","url":"https://unite.ai/nvidia-announces-open-kumo-forecast-multivariate-time-series-model","publishedAt":"2026-10-21T13:13:09Z","type":"press","primary":false,"lead":false}],"sourceNotes":null,"discussions":[],"thread":null,"cite":{"text":"Digest AI, \"Nvidia releases Kumo Forecast time-series model with retrieval and interpretability\", 21 October 2026, https://digestai.news/story/nvidia-releases-kumo-forecast-time-series-model-with-retrieval-and","publisher":"Digest AI","title":"Nvidia releases Kumo Forecast time-series model with retrieval and interpretability","datePublished":"2026-10-21T12:51:24Z","url":"https://digestai.news/story/nvidia-releases-kumo-forecast-time-series-model-with-retrieval-and"},"generatedBy":"Written by Digest AI's editorial model from the linked sources; the sources are the record.","license":"Headlines, digests and key points are written by Digest AI and may be quoted with a link to the story page. Linked articles belong to their publishers. Terms: https://digestai.news/terms#reuse"}