variables

This table contains all the variables of all the datasets we have at Our World In Data. Note that this includes only the information about the variables like name and default display settings, not the data itself.

Data license: CC-BY

14 rows where datasetId = 5515 sorted by id descending

✎ View and edit SQL

This data as json, CSV (advanced)

Suggested facets: createdAt, display, createdAt (date), updatedAt (date)

id ▲ name unit description createdAt updatedAt code coverage timespan datasetId sourceId shortUnit display columnOrder originalMetadata grapherConfigAdmin shortName catalogPath dimensions schemaVersion processingLevel processingLog titlePublic titleVariant attributionShort attribution descriptionShort descriptionFromProducer descriptionKey descriptionProcessing licenses license grapherConfigETL type sort dataChecksum metadataChecksum
450415 Compute_sponsor_large     2022-07-20 14:19:28 2023-06-15 05:05:42       AI Input Trends Data – Sevilla et al. (2023) 5515 Sevilla et al. (2023) 21327   
{}
 0           1                                  
450123 Equivalent_training_time_hours     2022-07-09 10:01:46 2023-06-15 05:05:42       AI Input Trends Data – Sevilla et al. (2023) 5515 Sevilla et al. (2023) 21327   
{}
 0           1                                  
450122 Compute_sponsor_all     2022-07-09 09:48:24 2023-06-15 05:05:42       AI Input Trends Data – Sevilla et al. (2023) 5515 Sevilla et al. (2023) 21327   
{}
 0           1                                  
450118 Cost_training_computation   Cost of training computation was calculated as the <a href="https://ourworldindata.org/grapher/artificial-intelligence-training-computation" target=”_blank”>system's training computation</a> divided by a floating-point operation (FLOP)/$ value. The FLOP/$ value was calculated using one of two methods: 1. The value of FLOP/s per $ at the time of the system's publication (according to the <a href="https://epochai.org/blog/trends-in-gpu-price-performance" target=”_blank”>"Our data" trend line in Figure 1 here</a>). 2. Dividing the theoretical peak throughput (including "Tensor Core" performance) by the reported unit price of the hardware that was actually used for training. The authors expect that Method 2 is more accurate on average. If an estimate via Method 2 is available, they report that estimate; otherwise, they use Method 1. Additionally, the authors made the following assumptions for all systems, in order to convert theoretical peak FLOP/s per $ into realized FLOP/$: 1. Hardware utilization is 35% 2. The amount of GPU time the given hardware is used for in its lifetime is 2 years. 2022-07-09 09:28:43 2023-06-15 05:05:42       AI Input Trends Data – Sevilla et al. (2023) 5515 Sevilla et al. (2023) 21327   
{
    "unit": "real 2020 US$",
    "zeroDay": "2020-01-01",
    "shortUnit": "$",
    "yearIsDay": true,
    "includeInTable": true,
    "numDecimalPlaces": 0
}
 0           1                                  
450117 Training_data_gb     2022-07-09 09:28:43 2023-06-15 05:05:42       AI Input Trends Data – Sevilla et al. (2023) 5515 Sevilla et al. (2023) 21327   
{
    "includeInTable": true
}
 0           1                                  
450116 Inference_time_ms     2022-07-09 09:28:43 2023-06-15 05:05:42       AI Input Trends Data – Sevilla et al. (2023) 5515 Sevilla et al. (2023) 21327   
{}
 0           1                                  
450114 Inference_computation_flop     2022-07-09 09:28:43 2023-06-15 05:05:42       AI Input Trends Data – Sevilla et al. (2023) 5515 Sevilla et al. (2023) 21327   
{
    "includeInTable": true
}
 0           1                                  
450113 Training_datapoints     2022-07-09 09:28:43 2023-06-15 05:05:42       AI Input Trends Data – Sevilla et al. (2023) 5515 Sevilla et al. (2023) 21327   
{
    "unit": "data points",
    "zeroDay": "2020-01-01",
    "yearIsDay": true,
    "includeInTable": true
}
 0           1                                  
450112 Parameters     2022-07-09 09:28:43 2023-06-15 05:05:42       AI Input Trends Data – Sevilla et al. (2023) 5515 Sevilla et al. (2023) 21327   
{
    "name": "Parameters",
    "unit": "parameters",
    "zeroDay": "2020-01-01",
    "shortUnit": "",
    "yearIsDay": true,
    "includeInTable": true
}
 0           1                                  
412581 Researcher_affiliation   The affiliation of the research team building a particular notable AI system was classified according to the following: — Academia: 100% of researchers affiliated with academia — Collaboration, Academia-majority: 71–99% affiliated with academia — Collaboration: 30–70% affiliated with academia — Collaboration, Industry-majority: 71–99% affiliated with industry — Industry: 100% of researchers affiliated with industry This data corresponds to the "Organization Categorization" column in the primary source data spreadsheet. 2022-03-20 21:59:37 2023-06-15 05:05:42       AI Input Trends Data – Sevilla et al. (2023) 5515 Sevilla et al. (2023) 21327   
{
    "includeInTable": true
}
 0           1                                  
312259 Training_computation_petaflop   A petaFLOP is 10¹⁵ floating-point operations. 2022-02-15 16:45:06 2023-06-15 05:05:42       AI Input Trends Data – Sevilla et al. (2023) 5515 Sevilla et al. (2023) 21327   
{
    "name": "petaFLOP",
    "unit": "petaFLOP",
    "zeroDay": "2020-01-01",
    "shortUnit": "",
    "yearIsDay": true,
    "includeInTable": true
}
 0           1                                  
312258 Training_computation_petaflop_sec_day   A petaflop/s-day (pfs-d) consists of performing 10¹⁵ neural network operations per second for one day, for a total of about 10²⁰ operations. 2022-02-15 16:25:39 2023-06-15 05:05:42       AI Input Trends Data – Sevilla et al. (2023) 5515 Sevilla et al. (2023) 21327   
{
    "unit": "petaflop/s-days",
    "zeroDay": "2020-01-01",
    "shortUnit": "pfs-d",
    "yearIsDay": true,
    "includeInTable": true
}
 0           1                                  
312257 Training_computation_flop     2022-02-15 13:49:44 2023-06-15 05:05:42       AI Input Trends Data – Sevilla et al. (2023) 5515 Sevilla et al. (2023) 21327   
{
    "unit": "Floating-point operations",
    "zeroDay": "2020-01-01",
    "shortUnit": "FLOP",
    "yearIsDay": true,
    "includeInTable": true
}
 0           1                                  
312256 Domain     2022-02-15 13:49:44 2023-06-15 05:05:42       AI Input Trends Data – Sevilla et al. (2023) 5515 Sevilla et al. (2023) 21327   
{
    "includeInTable": true
}
 0           1                                  

Advanced export

JSON shape: default, array, newline-delimited, object

CSV options: 

CREATE TABLE "variables" (
	"id" INTEGER PRIMARY KEY AUTOINCREMENT,
	"name" VARCHAR(750) NULL  ,
	"unit" VARCHAR(255) NOT NULL  ,
	"description" TEXT NULL  ,
	"createdAt" DATETIME NOT NULL DEFAULT CURRENT_TIMESTAMP ,
	"updatedAt" DATETIME NULL  ,
	"code" VARCHAR(255) NULL  ,
	"coverage" VARCHAR(255) NOT NULL  ,
	"timespan" VARCHAR(255) NOT NULL  ,
	"datasetId" INTEGER NOT NULL  ,
	"sourceId" INTEGER NULL  ,
	"shortUnit" VARCHAR(255) NULL  ,
	"display" TEXT NOT NULL  ,
	"columnOrder" INTEGER NOT NULL DEFAULT '0' ,
	"originalMetadata" TEXT NULL  ,
	"grapherConfigAdmin" TEXT NULL  ,
	"shortName" VARCHAR(255) NULL  ,
	"catalogPath" VARCHAR(767) NULL  ,
	"dimensions" TEXT NULL  ,
	"schemaVersion" INTEGER NOT NULL DEFAULT '1' ,
	"processingLevel" VARCHAR(30) NULL  ,
	"processingLog" TEXT NULL  ,
	"titlePublic" VARCHAR(512) NULL  ,
	"titleVariant" VARCHAR(255) NULL  ,
	"attributionShort" VARCHAR(512) NULL  ,
	"attribution" TEXT NULL  ,
	"descriptionShort" TEXT NULL  ,
	"descriptionFromProducer" TEXT NULL  ,
	"descriptionKey" TEXT NULL  ,
	"descriptionProcessing" TEXT NULL  ,
	"licenses" TEXT NULL  ,
	"license" TEXT NULL  ,
	"grapherConfigETL" TEXT NULL  ,
	"type" TEXT NULL  ,
	"sort" TEXT NULL  ,
	"dataChecksum" VARCHAR(64) NULL  ,
	"metadataChecksum" VARCHAR(64) NULL,
	FOREIGN KEY("datasetId") REFERENCES "datasets" ("id") ON UPDATE RESTRICT ON DELETE RESTRICT,
	FOREIGN KEY("sourceId") REFERENCES "sources" ("id") ON UPDATE RESTRICT ON DELETE RESTRICT
);
CREATE UNIQUE INDEX "idx_catalogPath" ON "variables" ("catalogPath");
CREATE UNIQUE INDEX "unique_short_name_per_dataset" ON "variables" ("shortName", "datasetId");
CREATE UNIQUE INDEX "variables_code_fk_dst_id_7bde8c2a_uniq" ON "variables" ("code", "datasetId");
CREATE INDEX "variables_datasetId_50a98bfd_fk_datasets_id" ON "variables" ("datasetId");
CREATE UNIQUE INDEX "variables_name_fk_dst_id_f7453c33_uniq" ON "variables" ("name", "datasetId");
CREATE INDEX "variables_sourceId_31fce80a_fk_sources_id" ON "variables" ("sourceId");