---

name: modules-and-configuration description: "Configure Tensorforce networks, layers, memories, policies, objectives, optimizers, parameters, preprocessing, and config dictionaries." disable-model-invocation: true metadata: disco-role: operating license: Apache 2.0

Modules and Configuration

Use this sub-skill when a task asks how to express Tensorforce module specifications, choose network/layer/preprocessing/memory/objective/optimizer/policy/parameter components, translate JSON/dict configs, or debug unknown module type values.

Route by need

• Read module reference for registry names and where each module family is used.
• Read network and layer recipes for network='auto', layered networks, register/retrieve DAGs, preprocessing, and Keras-layer boundaries.
• Run scripts/inspect_module_registry.py in the user's Tensorforce environment to print installed registry names.
• Use troubleshooting for unknown module names, invalid shorthand, shape issues, TensorFlow Addons/Keras surprises, or debug/eager mode confusion.

Boundaries

This sub-skill owns configuration shape and component selection. Route agent lifecycle, action masks, and act/observe loops to agents-and-specifications; route environment state/action discovery to environments-and-interaction; route full training loops to runner-and-cli-workflows.

Minimal pattern

agent = dict(
    agent='ppo',
    network=[dict(type='dense', size=64, activation='tanh')],
    batch_size=10,
    update_frequency=2,
    learning_rate=3e-4,
    baseline=dict(type='auto', size=32, depth=1),
    state_preprocessing='linear_normalization',
    config=dict(device='CPU', seed=7),
)

Before a long run, instantiate the agent with the target environment and call agent.get_specification() or agent.get_architecture() to verify Tensorforce accepted the config.