README.md

---
license: apache-2.0
language:
- en
library_name: Kumo-TS
---
# Kumo-Anomaly Overview

## Description: <br>
NVIDIA Kumo-Anomaly provides anomaly detection functionality. Rather than relying only on transformers, it introduces diffusion modeling, stabilized through curriculum learning, and pairs it with adaptive thresholding methods in a model purpose-built for anomaly detection. Together, these elements address some of the most challenging issues in the field: noisy, high-dimensional signals that drift over time and contain rare, irregular events.

### License/Terms of Use: <br>

Governing Terms: **Model:** The model is provided under the [Apache License, Version 2.0](https://www.apache.org/licenses/LICENSE-2.0.html). 

### Deployment Geography:
Global <br>

### Use Case: <br>
Companies, organizations, research hubs looking to do anomaly detection on temporal data. <br>

### Release Date:  <br>
29th June 2026 via https://huggingface.co/nvidia/Kumo-Anomaly

## Reference(s):
<a href="https://dl.acm.org/doi/full/10.1145/3787120.3787130">Segmented Confidence Sequences and Multi-Scale Adaptive Confidence Segments for Anomaly Detection in Nonstationary Time Series</a> </br>
<a href="https://arxiv.org/abs/2307.00754">ImDiffusion</a> </br>

## Model Architecture: <br>
**Architecture Type:** Diffusion based transformer <br>

**Network Architecture:** ResNet34 <br>

**Number of model parameters:** 8 million <br>

The Kumo-Anomaly is a diffusion-based model for time-series imputation and anomaly detection. <br>

## Input: <br>
**Input Type(s):** Tabular numeric <br>
**Input Format(s):** Tabular Pandas DataFrame or CSV/JSON <br>
**Input Parameters:** Two-Dimensional (2D) <br>
**Other Properties Related to Input:** Pre-Processing Needed <br>
Anomaly detection: Contains timestamp column and one or more numeric value columns.<br />

## Output: <br>
**Output Type(s):** Tabular numeric <br>
**Output Format:** Tabular Pandas DataFrame <br>
**Output Parameters:** Two-Dimensional (2D) <br>
**Other Properties Related to Output:** Post-Processing Needed <br>
Anomaly detection: Contains timestamp, value, and anomaly label columns.<br />

Our AI models are designed and/or optimized to run on NVIDIA GPU-accelerated systems. By leveraging NVIDIA's hardware (e.g. GPU cores) and software frameworks (e.g., CUDA libraries), the model achieves faster training and inference times compared to CPU-only solutions. <br>

## Software Integration: <br>
**Runtime Engine(s):** <br>
PyTorch <br>

**Supported Hardware Microarchitecture Compatibility:** <br>
**NVIDIA Ampere** <br>
**NVIDIA Hopper** <br>

**[Preferred/Supported] Operating System(s):** <br>
Linux<br>

The integration of foundation and fine-tuned models into AI systems requires additional testing using use-case-specific data to ensure safe and effective deployment. Following the V-model methodology, iterative testing and validation at both unit and system levels are essential to mitigate risks, meet technical and functional requirements, and ensure compliance with safety and ethical standards before deployment.

## Model Version(s):
Kumo-Anomaly<br>


# Training & Testing Datasets: <br>
80/20 split per dataset. <br />

**Data Modality:** Other: Numeric time series

**Training Data Size:** 3 million data points

### <a href="https://thedatumorg.github.io/TSB-AD/">TSB-AD-M</a><br />
**Data Collection Method by dataset**  
Hybrid: Automatic/Sensors, Human, Synthetic<br />
**Labeling Method by dataset**  
Hybrid: Human, Automatic, Synthetic<br/>
**Properties:** The TSB-AD-M benchmark consists of **17 public multivariate datasets** containing **198 curated multivariate time series** spanning diverse application domains. The datasets include both point and sequence anomalies with varying dimensionalities, anomaly ratios, anomaly lengths, and time-series lengths, providing a comprehensive benchmark for multivariate time-series anomaly detection.
<br>

# Evaluation Datasets: <br>

### <a href="https://www.kaggle.com/datasets/arbazkhan971/anomaly-detection">Detecting Anomalies in Wafer Manufacturing</a><br />
**Data Collection Method by dataset**  
Automatic/Sensors<br />
**Labeling Method by dataset**  
Human, Automatic<br/>
**Properties:** High-dimensional tabular anomaly detection dataset collected from semiconductor wafer manufacturing. Each sample represents a single wafer described by **1,558 anonymized process features** and a binary anomaly label indicating normal or anomalous production.
<br>

### <a href="https://archive.ics.uci.edu/dataset/156/calit2+building+people+counts">CalIt2 Building People Counts</a><br />
**Data Collection Method by dataset**  
Automatic/Sensors<br />
**Labeling Method by dataset**  
Human<br/>
**Properties:** Time-series dataset containing people counts collected from sensors at the main entrance of the Calit2 building at the University of California, Irvine. Measurements were recorded every 30 minutes over approximately 15 weeks, producing 48 observations per day.
<br>

### <a href="https://www.kaggle.com/datasets/ziya07/network-traffic-anomaly-detection-dataset">Network Traffic</a><br />
**Data Collection Method by dataset**  
Synthetic, Automatic/Sensors<br />
**Labeling Method by dataset**  
Automatic, Human<br/>
**Properties:** Labeled network traffic dataset for anomaly detection containing normal and malicious traffic records represented by multiple network features. The dataset is intended for evaluating supervised anomaly detection methods on network intrusion scenarios.
<br>

### <a href="https://www.kaggle.com/datasets/inIT-OWL/genesis-demonstrator-data-for-machine-learning">Genesis Demonstrator</a><br />
**Data Collection Method by dataset**  
Automatic/Sensors<br />
**Labeling Method by dataset**  
Human<br/>
**Properties:** Multivariate industrial time-series dataset collected from the Genesis portable pick-and-place demonstrator developed as part of the OPAK and IMPROVE projects. The dataset contains continuous and discrete sensor signals together with timestamps and labeled anomalous operating conditions.
<br>

### <a href="https://www.kaggle.com/datasets/jorekai/anomaly-detection-falling-people-events?select=data">Falling People</a><br />
**Data Collection Method by dataset**  
Human, Automatic/Sensors<br />
**Labeling Method by dataset**  
Human<br/>
**Properties:** This data set was used during a thesis to develop safer smart environments. The origin is a care independent smart home environment to detect the falling of elderly people.
<br>

## Inference:
**Engine:** PyTorch, Transformer Engine <br>
**Test Hardware:** <br>
* A100 (8 GPUs; each is 74 GB)<br>
* H100 (8 GPUs; each is 80 GB)

## Ethical Considerations:
NVIDIA believes Trustworthy AI is a shared responsibility and we have established policies and practices to enable development for a wide array of AI applications.  When downloaded or used in accordance with our terms of service, developers should work with their internal model team to ensure this model meets requirements for the relevant industry and use case and addresses unforeseen product misuse.

Please report model quality, risk, security vulnerabilities or NVIDIA AI Concerns [here](https://www.nvidia.com/en-us/support/submit-security-vulnerability/).