datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
PhysicsNeMo-Datacenter-CFD
Overview
This asset provides the dataset and reference OpenFOAM configuration to be used for the Datacenter CFD use case here.
The dataset contains normalized OpenFOAM simulations for various configurations of a typical hot aisle in a datacenter. The dataset includes variation for the number of IT racks and hot-aisle geometry parameters like length, height and width of the hot aisle. These configurations are solved with OpenFOAM assuming maximum flow rate and rack exit temperature… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/PhysicsNeMo-Datacenter-CFD.Datacenter-watchdatacenter-tracesim3-datacenter-olmoearth-segmentation
OlmoEarth v1.2-ready datacenter energy dataset
Derived cook 20260830T191500Z from pinned source 1f3ee5f334dd53940d088a89db57aed62cc925d5.
The branch uses no train/validation/test split. Every accepted row is tagged all.
Accepted imagery rows: 1417
Quality-unavailable rows: 0
Release mode: partial with documented quality-unavailable exclusions
Clearest-available quality fallbacks: 1
Sentinel-2 L2A: uint16 [12,1,128,128]
Band order: B02,B03,B04,B08,B05,B06,B07,B8A,B11,B12,B01,B09… See the full description on the dataset page: https://huggingface.co/datasets/sarkarghya/im3-datacenter-olmoearth-segmentation.hyperscale-datacenter-segmentation-naip
Hyperscale Data Center Segmentation (NAIP)
Hand-digitized training data for detecting hyperscale data center
footprints in aerial imagery, with a trained baseline model.
Contents
path
what it is
datacenters.geojson
190 hand-digitized data center footprint polygons (QGIS; named facilities, e.g. vantage_0)
chips/
757 NAIP aerial chips, 256×256 px, 4-band RGBN, 0.6 m resolution (GeoTIFF, georeferenced) — labeled facilities plus surrounding negatives… See the full description on the dataset page: https://huggingface.co/datasets/rbhughes/hyperscale-datacenter-segmentation-naip.ifuse-final-datacenter-traces-20260624
I-Fuse Final Datacenter Traces
This dataset contains the final Scarab/DynamoRIO memtrace bundles for the
I-Fuse datacenter workloads.
Workloads included:
bc
bfs
chemcrow
cc
cd
dfs
langchain_web
mongodb
mysql
pagerank
postgres
rag_haystack
sssp_ego_fb
swe_agent
tc
toolformer
The original datacenter workload directories contain the Scarab trace structure,
including:
traces_simp/trace/
traces_simp/bin/
fingerprint/
simpoints/
trace_clustering_info.json
The five newly collected… See the full description on the dataset page: https://huggingface.co/datasets/harry1332/ifuse-final-datacenter-traces-20260624.datacenter-tracesdatacenters
Cloud Datacenter Locations
This dataset tracks the locations of datacenters operated by large cloud providers (AWS, Azure, GCP) with each cloud provider having their own json file.
It based on my research collected from publicly available information.
Accuracy of locations
The granularity of each entry's location is recorded explicitly (see geoPrecision below).
Cloud providers will provide region or zone as options while procuring their services, these refer to a… See the full description on the dataset page: https://huggingface.co/datasets/GridGrapher/datacenters.datacenter-thermal-performance-coherence-risk-v0.1What this repo is for
Detect when heat and performance start to decouple.
Focus
• hotspot temperature vs throttling
• cooling margin vs p95 latency stability
• fan headroom vs sustained load
• early drift before alarms and shutdowns
Why it matters
Many outages are preceded by thermal drift.
Operators see it as “random latency.”
This dataset turns it into an early warning signal.
wue-african-datacenters
Water Efficiency Dataset for African Data Centers
Description
This dataset accompanies the paper "A Water Efficiency Dataset for African Data Centers" presented at COMPASS '25 (ACM SIGCAS/SIGCHI Conference on Computing and Sustainable Societies, July 2025, Toronto, Canada).
It provides nation-level estimates of water usage efficiency (WUE) for data centers across 46 African countries spanning five distinct climate regions. The dataset covers:
Direct (onsite) water… See the full description on the dataset page: https://huggingface.co/datasets/masterlion/wue-african-datacenters.les-datacenters-finances-par-la-banque-des-territoires
Les Datacenters financés par la Banque des Territoires
[!NOTE]
Ce jeu de données Hugging Face est vide. Cette carte sert seulement à référencer le jeu de données Les Datacenters financés par la Banque des Territoires qui est disponible à l'adresse https://www.data.gouv.fr/datasets/60e4f0325ecc1ba0a0de38ef
Description
Le jeu de données mis à disposition présente le parc des Centres de données financés par la Direction des Investissements de la Banque des Territoires, en… See the full description on the dataset page: https://huggingface.co/datasets/french-open-data/les-datacenters-finances-par-la-banque-des-territoires.datacenter-network-routing-coherence-risk-v0.1What this repo is for
Detect routing instability before outages.
Focus
• packet loss vs route flapping
• jitter vs path stability
• link utilization vs reroutes
• early latency drift
Why it matters
Most outages show network symptoms before full failure.
This dataset turns those symptoms into early warnings.
datacenter-node-failure-cascade-risk-v0.1What this repo is for
Detect cascading node failure risk before full cluster outage.
Focus
• replication margin vs load
• failover speed
• service dependency density
• node failure clustering
Why it matters
Most major outages begin with small local failures.
The cascade forms when redundancy, load shift, and recovery fall out of alignment.
This dataset lets operators detect that drift early.
datacenter-power-load-coherence-risk-v0.1What this repo is for
Detect power instability before outages.
Focus
• rack load vs PSU margin
• UPS buffer vs spike frequency
• voltage variance vs reset events
• sustained load vs safe envelope
Why it matters
Power issues often look random.
They usually follow slow margin erosion.
This dataset turns that erosion into a visible signal.
datacenter-water-cooling-demand-coherence-risk-v0.1What this repo is for
Detect when datacenter compute heat
outpaces cooling water capacity.
Flags
drought stress and discharge limits under high heat load
curtailment events under rising demand
heat and water draw decouple from throttling
abnormal overdraw at low heat load
datacenter-job-queue-resource-coherence-risk-v0.1What this repo is for
Detect scheduler pressure before customer-visible incidents.
Focus
• queue depth vs wait time
• utilization vs SLO breaches
• preemption vs job failure
• fragmentation where capacity exists but cannot be used
Why it matters
Operators often see queue problems too late.
This dataset spots the drift earlier and links it to real failure signals.
Datacenter_traineco-datacenter
eco-datacenter
Data center design within ethical principles of material sourcing, energy consumption, data privacy, ownership and transparency.
1. conceptual-submodule-expansions
1.1 mycelial-compute-fabric
Expanding upon the eco-datacenter, this is a distributed computing protocol modeled after fungal mycelium networks. It dynamically routes inference tasks to nodes powered by surplus renewable energy, optimizing for planetary metabolic balance rather than… See the full description on the dataset page: https://huggingface.co/datasets/ronniross/eco-datacenter.oaks-datacenterdatacenter_qadatacenter_qa