# NVIDIA — NVIDIA Kumo Forecast Combines Pretrained Time-Series Representations with Historical Pattern Retrieval and Interpretability

- Company: NVIDIA (nvidia.com)
- Announced: 2026-10-21T12:51:24+00:00
- Category: new-model
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://huggingface.co/blog/nvidia/kumo-forecast
- Record: https://forck.live/items/25911-nvidia-kumo-forecast-combines-pretrained-time-series-representations-with
- Open weights: yes
- Models affected: NVIDIA Kumo Forecast, NV-Tesseract Forecasting, MOMENT-1-large, Kumo Anomaly
- License: Apache 2.0

NVIDIA has introduced Kumo Forecast, a multivariate time-series forecasting model that pairs a pretrained Transformer backbone with retrieval of patterns from historical data. 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. Its Domain-Aware Representation and Retrieval mode adds k-nearest-neighbor retrieval from historical context, which the SDK accepts through a separate context_df argument, and developers can adjust how strongly the final prediction relies on each pathway. The library offers a DataFrame-first API, configurable forecast horizons, and autoregressive extension for horizons beyond the model's native capability. An interpretability framework produces lag and feature attributions per forecast horizon, semantic-flow and trajectory-stability diagnostics, and optional embedding Integrated Gradients without changing the underlying forecaster. Kumo Forecast is available under the Apache 2.0 license, with model code at NVIDIA/Kumo-TS and weights at nvidia/Kumo-Forecast. Inputs must be numeric temporal data with sequence lengths of 256, 512, 1024 or 2048 observations, and in DARR mode the context dimensions must be similar to those of the input.

## Evidence

Verbatim from https://huggingface.co/blog/nvidia/kumo-forecast:

> NVIDIA Kumo Forecast (formerly known as NV-Tesseract Forecasting) is a multi-variate time-series forecasting model that combines a pretrained Transformer backbone with Domain-Aware Representation and Retrieval (DARR), a context-enhanced forecasting mode that blends direct model predictions with patterns retrieved from historical data.

---

Record: https://forck.live/items/25911-nvidia-kumo-forecast-combines-pretrained-time-series-representations-with
Catalogue: https://forck.live/llms.txt
Current issue: https://forck.live/feed.md