Directories
¶
| Path | Synopsis |
|---|---|
|
cmd
|
|
|
api
command
|
|
|
build-index
command
build-index pre-computes IVF or Grid indexes from references.json.gz.
|
build-index pre-computes IVF or Grid indexes from references.json.gz. |
|
partstats
command
|
|
|
rfval
command
|
|
|
train
command
Command train trains a decision tree from references.json.gz and exports it as a v2 GBDT binary (resources/gbdt.bin).
|
Command train trains a decision tree from references.json.gz and exports it as a v2 GBDT binary (resources/gbdt.bin). |
|
internal
|
|
|
dataset
Package dataset provides functionality for loading and managing the reference dataset.
|
Package dataset provides functionality for loading and managing the reference dataset. |
|
gbdt
Package gbdt provides fast inference using a pre-trained Gradient Boosted Decision Tree.
|
Package gbdt provides fast inference using a pre-trained Gradient Boosted Decision Tree. |
|
handler
Package handler provides HTTP request handlers for the fraud detection API.
|
Package handler provides HTTP request handlers for the fraud detection API. |
|
knn
Package knn implements k-nearest-neighbor search over 14-dimensional vectors.
|
Package knn implements k-nearest-neighbor search over 14-dimensional vectors. |
|
logreg
Package logreg implements a logistic regression model for binary classification.
|
Package logreg implements a logistic regression model for binary classification. |
|
middleware
Package middleware provides low-overhead request telemetry for fraudctl.
|
Package middleware provides low-overhead request telemetry for fraudctl. |
|
model
Package model provides data structures for the fraud detection API.
|
Package model provides data structures for the fraud detection API. |
|
rawhttp
Package rawhttp provides a minimal, zero-allocation HTTP/1.1 server optimized for the fraud detection API hot path.
|
Package rawhttp provides a minimal, zero-allocation HTTP/1.1 server optimized for the fraud detection API hot path. |
|
vectorizer
Package vectorizer provides functionality to convert transaction requests into 14-dimensional normalized vectors for fraud detection using KNN.
|
Package vectorizer provides functionality to convert transaction requests into 14-dimensional normalized vectors for fraud detection using KNN. |
Click to show internal directories.
Click to hide internal directories.