--- title: NVIDIA Kumo Relational Outperforms Supervised Machine... description: NVIDIA's Kumo Relational is a pre-trained foundation model that predicts outcomes directly on relational databases without task-specific training. It reads up... canonical: https://daily.dev/posts/nvidia-kumo-relational-outperforms-supervised-machine-learning-on-relational-data-no-training-requi-ykf01wckv twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: NVIDIA Kumo Relational Outperforms Supervised Machine Learning on Relational Data, No Training Required | daily.dev og:description: NVIDIA's Kumo Relational is a pre-trained foundation model that predicts outcomes directly on relational databases without task-specific training. It reads up... og:url: https://daily.dev/posts/nvidia-kumo-relational-outperforms-supervised-machine-learning-on-relational-data-no-training-requi-ykf01wckv og:image: https://api.daily.dev/og/posts/YKF01WcKV.png og:image:alt: NVIDIA Kumo Relational Outperforms Supervised Machine Learning on Relational Data, No Training Required og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # NVIDIA Kumo Relational Outperforms Supervised Machine Learning on Relational Data, No Training Required **[Hugging Face](https://daily.dev/sources/huggingface)** · 11 min read · 0 upvotes · 0 comments ## Summary NVIDIA's Kumo Relational is a pre-trained foundation model that predicts outcomes directly on relational databases without task-specific training. It reads up to 10,000 historical examples as context and makes predictions in a single forward pass, skipping feature engineering and table flattening. Across 41 tasks on 15 databases, it beats supervised relational deep learning models and gradient-boosted tree pipelines. On SAP SALT, it surpasses AutoGluon ensembles by 8% in mean reciprocal rank, rising to 16% after fine-tuning. On the RelArena-α leaderboard, it sits on the accuracy-efficiency Pareto front, running about 430 times faster than the highest-scoring competing entry at similar accuracy. It scales to databases with 500B+ rows and is available via NVIDIA's open-source structured-data-models library or as an NVIDIA NIM. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### What is NVIDIA Kumo Relational and how does it make predictions without training? Kumo Relational is a pre-trained relational foundation model that predicts directly on enterprise databases using in-context learning. Given a predictive query, it gathers up to 10,000 historical examples as context and produces predictions like churn, demand, or fraud in a single forward pass, requiring no task-specific training, feature engineering, or table flattening. _developers evaluating foundation models for database-driven predictions can track releases like this on daily.dev_ ### How does Kumo Relational perform against AutoGluon and XGBoost on relational data benchmarks? On the SAP SALT enterprise benchmark, Kumo Relational surpasses large AutoGluon ensembles by 8% in mean reciprocal rank out of the box, rising to 16% after fine-tuning. It also beats XGBoost on expert-engineered features by 7 MRR points out of the box and 13 MRR points after fine-tuning, and outperforms gradient-boosted trees by nearly 4 AUROC points on RelBench V1 classification tasks. _teams weighing a switch from XGBoost pipelines can compare benchmark results like these on daily.dev_ ### How large of a database can NVIDIA's structured-data-models library handle for relational predictions? The library's custom graph engine handles databases with over 500 billion rows via memory-mapped I/O at speeds up to 5 GB/s. The model is available through the open-source structured-data-models library, which downloads weights from the Hugging Face Hub, or can be deployed as a separate NVIDIA NIM service. _engineers scaling predictive workloads to massive databases can follow tooling updates like this on daily.dev_ ## Similar posts on daily.dev - [NVIDIA Kumo Tabular Sets a New Accuracy-Efficiency Frontier for Tabular Prediction](https://daily.dev/posts/nvidia-kumo-tabular-sets-a-new-accuracy-efficiency-frontier-for-tabular-prediction-avcjvvwsc) · Hugging Face · 0 upvotes · 0 comments - [NVIDIA Kumo Forecast Combines Pretrained Time-Series Representations with Historical Pattern Retrieval and Interpretability](https://daily.dev/posts/nvidia-kumo-forecast-combines-pretrained-time-series-representations-with-historical-pattern-retriev-h7yttslab) · Hugging Face · 2 upvotes · 0 comments - [Kumo’s new foundation model replaces months of data science engineering with plain-English queries](https://daily.dev/posts/kumo-s-new-foundation-model-replaces-months-of-data-science-engineering-with-plain-english-queries-xrcvjrwyq) · The New Stack · 0 upvotes · 0 comments - [NVIDIA Open Models Advance Europe’s Sovereign Intelligence Across Industries](https://daily.dev/posts/nvidia-open-models-advance-europe-s-sovereign-intelligence-across-industries-5kjyoxlix) · NVIDIA · 0 upvotes · 0 comments --- Tags: [#ai](https://daily.dev/tags/ai), [#machine-learning](https://daily.dev/tags/machine-learning), [#nvidia](https://daily.dev/tags/nvidia) [View this post on daily.dev](https://daily.dev/posts/nvidia-kumo-relational-outperforms-supervised-machine-learning-on-relational-data-no-training-requi-ykf01wckv) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"NVIDIA Kumo Relational Outperforms Supervised Machine Learning on Relational Data, No Training Required","url":"https://daily.dev/posts/nvidia-kumo-relational-outperforms-supervised-machine-learning-on-relational-data-no-training-requi-ykf01wckv","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/nvidia-kumo-relational-outperforms-supervised-machine-learning-on-relational-data-no-training-requi-ykf01wckv"},"datePublished":"2026-10-21T18:27:20.030Z","dateModified":"2026-10-21T19:05:02.856Z","description":"NVIDIA's Kumo Relational is a pre-trained foundation model that predicts outcomes directly on relational databases without task-specific training. It reads up...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/98eca42093b72130063d1ef40fda8424?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/98eca42093b72130063d1ef40fda8424?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"Hugging Face","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"Hugging Face","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/f1f55c67d81a4330acf5b90b26b0c8e1","url":"https://daily.dev/sources/huggingface"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/nvidia-kumo-relational-outperforms-supervised-machine-learning-on-relational-data-no-training-requi-ykf01wckv","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"ai,machine-learning,nvidia","timeRequired":"PT11M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Hugging Face","item":"https://daily.dev/sources/huggingface"},{"@type":"ListItem","position":3,"name":"NVIDIA Kumo Relational Outperforms Supervised Machine Learning on Relational Data, No Training Required"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/nvidia-kumo-relational-outperforms-supervised-machine-learning-on-relational-data-no-training-requi-ykf01wckv#faq","mainEntity":[{"@type":"Question","name":"What is NVIDIA Kumo Relational and how does it make predictions without training?","acceptedAnswer":{"@type":"Answer","text":"Kumo Relational is a pre-trained relational foundation model that predicts directly on enterprise databases using in-context learning. Given a predictive query, it gathers up to 10,000 historical examples as context and produces predictions like churn, demand, or fraud in a single forward pass, requiring no task-specific training, feature engineering, or table flattening. developers evaluating foundation models for database-driven predictions can track releases like this on daily.dev"}},{"@type":"Question","name":"How does Kumo Relational perform against AutoGluon and XGBoost on relational data benchmarks?","acceptedAnswer":{"@type":"Answer","text":"On the SAP SALT enterprise benchmark, Kumo Relational surpasses large AutoGluon ensembles by 8% in mean reciprocal rank out of the box, rising to 16% after fine-tuning. It also beats XGBoost on expert-engineered features by 7 MRR points out of the box and 13 MRR points after fine-tuning, and outperforms gradient-boosted trees by nearly 4 AUROC points on RelBench V1 classification tasks. teams weighing a switch from XGBoost pipelines can compare benchmark results like these on daily.dev"}},{"@type":"Question","name":"How large of a database can NVIDIA's structured-data-models library handle for relational predictions?","acceptedAnswer":{"@type":"Answer","text":"The library's custom graph engine handles databases with over 500 billion rows via memory-mapped I/O at speeds up to 5 GB/s. The model is available through the open-source structured-data-models library, which downloads weights from the Hugging Face Hub, or can be deployed as a separate NVIDIA NIM service. engineers scaling predictive workloads to massive databases can follow tooling updates like this on daily.dev"}}]} ```