juniper-cascor-worker 0.1.0

Remote candidate training worker for the JuniperCascor neural network service

juniper-cascor-worker

Remote candidate training worker for the JuniperCascor cascade correlation neural network service.

Overview

This package enables distributed candidate training by connecting to a CasCor CandidateTrainingManager and processing training tasks on remote hardware.

Installation

pip install juniper-cascor-worker

Note: This package requires the JuniperCascor source code to be importable on the worker machine (the worker runs CasCor's training code locally).

CLI Usage

# Basic usage
juniper-cascor-worker --manager-host 192.168.1.100 --manager-port 50000 --workers 4

# With CasCor source path
juniper-cascor-worker --manager-host 192.168.1.100 --cascor-path /opt/juniper-cascor/src --workers 8

# Full options
juniper-cascor-worker \
    --manager-host 192.168.1.100 \
    --manager-port 50000 \
    --authkey my-secret-key \
    --workers 4 \
    --mp-context forkserver \
    --log-level INFO

Python API

from juniper_cascor_worker import CandidateTrainingWorker, WorkerConfig

config = WorkerConfig(
    manager_host="192.168.1.100",
    manager_port=50000,
    authkey="my-secret-key",
    num_workers=4,
)

with CandidateTrainingWorker(config) as worker:
    worker.start()
    # Workers process tasks from the remote queue
    input("Press Enter to stop...")
    worker.stop()

Environment Variables

Variable	Description	Default
CASCOR_MANAGER_HOST	Manager hostname	127.0.0.1
CASCOR_MANAGER_PORT	Manager port	50000
CASCOR_AUTHKEY	Authentication key	juniper
CASCOR_NUM_WORKERS	Worker count	1
CASCOR_MP_CONTEXT	Multiprocessing method	forkserver

Juniper Ecosystem

This package is part of the Juniper Cascade Correlation Neural Network Research Platform.

Package	Description	Install
juniper-data-client	Dataset service client	pip install juniper-data-client
juniper-cascor-client	Neural network service client	pip install juniper-cascor-client
juniper-cascor-worker	Distributed training worker (this package)	pip install juniper-cascor-worker

License

MIT License - see LICENSE for details.