snorkelai.sdk.client.gts
Ground truth related functions. Ground truth is the set of labels that are considered to be the “true” labels for a dataset. Ground truth is used to evaluate the performance of a model, calculate labeling function performance, and power analysis and auto-suggest tools. Ground truth labels are associated with a single model node in an application.
Functions
add_ground_truth(node[, x_uids, labels, df, ...]) | This function registers ground truth to the Snorkel Flow platform. |
align_external_ground_truth(node_uid, ...[, ...]) | Note: This SDK is only necessary when you use a non-dataframe format with sf.add_ground_truth for sequence tagging applications. |
create_ground_truth_version(node, name[, ...]) | Create a new ground truth version for a node. |
delete_ground_truth_version(node, gt_version_uid) | Delete a specified ground truth version for a node. |
get_document_ground_truth(node[, context_uids]) | Get document-level ground truths as Pandas DataFrame. |
get_ground_truth(node[, split, start_date, ...]) | Get ground truth from the Snorkel Flow platform. |
get_inferred_document_ground_truth_from_span_ground_truth(...) | Available only for text extraction and entity classification tasks. |
get_span_level_ground_truth_conflicts(node) | Retrieve a list of span-level ground truth conflicts. |
list_ground_truth_versions(node) | Fetch a list of all ground truth versions for a node. |
load_ground_truth_version(node, gt_version_uid) | Load a specified ground truth version for a node. |