NVIDIA Releases Kumo Forecast for Time-Series Forecasting With Retrieval and Interpretability

According to a Hugging Face blog post published Oct. 21, 2026, the model, formerly known as NV-Tesseract Forecasting, is built on the MOMENT-1-large backbone and uses an approximately 340-million-parameter base encoder plus a trainable forecasting head. The library provides a DataFrame-first API, configurable forecast horizons, optional historical context for retrieval, and an interpretability framework that produces horizon-specific explanations without modifying the underlying forecaster. Model code is listed as NVIDIA/Kumo-TS and model weights as nvidia/Kumo-Forecast.

Kumo Forecast draws on two information paths. In its direct forecasting path, a MOMENT-based Transformer and a trained forecasting head predict future values; for horizons beyond the model's native capability, autoregressive extension is supported. In Domain-Aware Representation and Retrieval (DARR) mode, the system also retrieves related patterns from historical data using a nearest-neighbor approach and combines that information with the direct forecast. The input window supplies recent observations for the forecast, while a separate historical DataFrame can supply additional context for pattern retrieval. Developers can adjust how strongly the final prediction relies on each pathway through the library's configuration options, and the SDK accepts historical context through a separate context_df argument.

The model combines four main components: a pretrained temporal encoder, a trainable forecasting head, a historical-pattern retrieval mechanism and an interpretability framework. Encoder and embedder weights are kept frozen while the forecasting head is trained, separating the pretrained representation model from the task-specific prediction layer. The interpretability framework is model-agnostic and, when enabled, produces an explanation bundle containing lag and feature attributions per forecast horizon, semantic-flow diagnostics, trajectory-stability diagnostics and optional embedding Integrated Gradients.

Data must be numeric, temporal data. Supported input sequence lengths include 256, 512, 1024 and 2048 observations. Inputs consist of a timestamp column and one or more numeric value columns, represented as a pandas DataFrame or a CSV/JSON input. When using DARR mode, the dimensions of the context must be similar to the context of the input.

NVIDIA also said the Factory Operations (FOX) Blueprint includes Early Access to the Predictive Maintenance module, which leverages Kumo Anomaly and Kumo Forecast to analyze real-time data and predict future anomalies. The company pointed to a GTC San Jose 2026 session for further details.

For application-specific evaluation, NVIDIA recommends comparing direct and retrieval-enhanced configurations on the same held-out forecasting windows and validating forecast quality across the intended horizons and operating conditions, consistent with its requirement for use-case-specific testing. The Kumo-TS repository also provides an example for fine-tuning Kumo Forecast on a user-supplied CSV.