variables: 736555

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

This data as json

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
736555 Top-5 accuracy - state of the art % The top-5 accuracy measure is used to assess how frequently a model's top five predictions include the correct answer from a list of 1000 options. Here's an example to illustrate what this benchmark tests: When an image classification model is presented with an image of an animal, it will assign probabilities to each possible label. Based on these probabilities, the model generates its top five predictions out of a total of 1000 animal labels. For instance, the model might output the following predictions as its top five guesses: * Cat * Dog * Elephant * Lion * Tiger Suppose the correct label for the image is "dog." If "dog" appears among the model's top five predictions, then the model's prediction is considered correct according to the top-5 accuracy metric. On the other hand, if the correct label is "giraffe" and "giraffe" is not included in the model's top five predictions, then the model's prediction would be considered incorrect based on the top-5 accuracy measure. To calculate the top-5 accuracy, researchers evaluate the model's performance on a large dataset with known labels. They compute the percentage of examples in the dataset where the correct label is present within the model's top five predictions out of the 1000 possible options. This measure provides a broader perspective on the model's performance by considering whether the correct answer is among its top guesses, even if it's not the model's absolute top prediction. 2023-07-03 14:54:57 2024-07-08 16:38:15       6103 29583 % 
{
    "name": "Top-5 accuracy",
    "unit": "%",
    "zeroDay": "2019-01-01",
    "shortUnit": "%",
    "yearIsDay": true,
    "numDecimalPlaces": 1
}
 0     papers_with_code_imagenet_top5_state_of_the_art grapher/artificial_intelligence/2023-06-14/papers_with_code_benchmarks_state_of_the_art/papers_with_code_benchmarks_state_of_the_art#papers_with_code_imagenet_top5_state_of_the_art   1                 
[]
         float 
[]
 62c1f0df7ed63b479161c02b89807098 4424413581606c35ff8203d92e7deb4e

Links from other tables