Models

getModels() lists the local PhEye models the deployment has in PHEYE_MODELS_DIRECTORY, ordered by name, and getModel(name) reads one. A policy uses a model by giving its name as a ph-eye filter's phEyeConfiguration.modelPath, and the model then runs inside Philter with no PhEye service. Use the list, for example, to offer only the models a deployment has when building a policy.

import ai.philterd.philter.model.Model;

for (Model model : client.getModels()) {
    System.out.println(model.getName() + " (" + model.getSizeBytes() + " bytes)");
}

Model model = client.getModel("pii-en-xsmall");

Both require Philter 4.0.0 or later and the policies:read scope. The models are the deployment's, the same for every user, and the list is empty when it has none. A key without the scope is a ClientException with status 403, and a name with no model one with status 404.

Getter	Value
getName()	The name a policy gives as modelPath.
getEncoder()	The encoder the model was trained from, or null when its configuration does not say.
getMaxLength()	The most words of a text the model reads; later words are not examined. null when its configuration does not say.
getMaxWidth()	The most words one detected entity can span. null when its configuration does not say.
getSizeBytes()	The combined size, in bytes, of its gliner_config.json, tokenizer.json, and model.onnx.

Philter does not return a model's location on disk. For how models are deployed and named, see Philter's models documentation.