What problem does it solve? Training and deploying neural networks typically requires managing GPU infrastructure, configuring distributed environments, and writing boilerplate code. This Skill lets you train, deploy, and manage neural networks through Flow Nexus MCP tools running in E2B sandboxes, without provisioning your own hardware. ## Core Features & Use Cases - Single-Node Training: Train feedforward, LSTM, GAN, autoencoder, and transformer architectures with configurable layers, optimizers, and training tiers from nano to large. - Distributed Training Clusters: Initialize clusters with mesh, ring, star, or hierarchical topologies, deploy worker and parameter-server nodes, and run federated learning where data stays on local nodes. - Template Marketplace: Browse, deploy, publish, and rate pre-trained model templates for NLP, vision, time series, and anomaly detection tasks. - Use Case: Train an LSTM time-series forecaster on the medium tier, monitor its training status, benchmark inference latency, then publish the model as a reusable marketplace template. ## Quick Start Ask the AI to train a feedforward classifier with two hidden layers on the small tier using the Flow Nexus neural training tool, then run predictions on the resulting model.
