Skip to main content

OnPremiseClassifier

class OnPremiseClassifier(, host, dataset_name='dataset', model='nicl-small', timeout_s=900, metadata=None, user=None, api_version=None, default_headers=None)​

Sklearn-style estimator proxying fit/predict calls to an on-premise NICL server.

This classifier connects to a local or self-hosted NICL (Neural In-Context Learning) server and provides a scikit-learn compatible interface for classification tasks. It’s designed for users who have deployed NICL on their own infrastructure.

  • Parameters:
    • host (str) – Base URL of the NICL server (e.g., “http://localhost:8000”). This is a required parameter.
    • dataset_name (str , default="dataset") – Name identifier for the dataset used in API requests.
    • model (str , default="nicl-small") – Model identifier/path to use for inference (e.g., “nicl-small”, “nicl-large”).
    • timeout_s (int , default=900) – Request timeout in seconds for API calls.
    • metadata (dict , optional) – Optional metadata dictionary to include with requests.
    • user (str , optional) – Optional user identifier for request tracking.
    • api_version (str , optional) – Optional API version string to send as ‘Nicl-Version’ header.
    • default_headers (dict , optional) – Optional default headers to include with every request (e.g., authentication).
  • Variables:
    • classes (ndarray of shape *(*n_classes , )) – The unique class labels found during fit.
    • X_train (ndarray of shape *(*n_samples , n_features )) – Training data stored during fit.
    • y_train (ndarray of shape *(*n_samples , )) – Training labels stored during fit.
    • last_response (dict , optional) – The last response received from the remote server (set after predict/predict_proba).

Examples​

>>> from neuralk import OnPremiseClassifier
>>> import numpy as np
>>>
>>> # Initialize the classifier with your NICL server URL
>>> clf = OnPremiseClassifier(
...     host="http://localhost:8000",
...     model="nicl-small",
...     timeout_s=300
... )
>>>
>>> # Generate some training data
>>> X_train = np.random.randn(100, 10).astype(np.float32)
>>> y_train = np.random.randint(0, 2, 100).astype(np.int64)
>>>
>>> # Fit the classifier (stores training data)
>>> clf.fit(X_train, y_train)
>>>
>>> # Make predictions
>>> X_test = np.random.randn(10, 10).astype(np.float32)
>>> predictions = clf.predict(X_test)
>>> probabilities = clf.predict_proba(X_test)

Notes​

  • The classifier requires numeric input data (numpy arrays with float32 dtype).
  • Training labels must be integers (int64 dtype).
  • The fit method only stores the training data; actual model training happens on the remote server during predict/predict_proba calls.
  • For advanced use cases, consider using the low-level helpers in neuralk.model.classify directly.

On this page