AI prompts.
Agent skills.

Generates PyTorch embeddings for graph edge features by mapping categorical strings to indices and concatenating learned embeddings, specifically handling device, net, and terminal attributes.

Generates PyTorch embeddings for categorical edge features in a Graph Neural Network by mapping string values to indices and concatenating learned embeddings according to a specific structure.

Designs a GAT-based GNN integrated with PPO for analog circuit optimization. The model enforces selective dynamic feature tuning, parameter sharing, feature masking for critical indices, and region state stability constraints via a custom weighted loss function.

Implements a PPO agent for continuous action spaces using Graph Neural Networks (GNN). The Actor features a multi-task head predicting both actions and node stability, while the Critic operates on flattened node features. Integrates dynamic stability loss and entropy regularization with Tanh acti...

Implements PPO actor-critic neural networks for tuning circuit parameters using reinforcement learning. Includes specific network architectures and a utility to scale Tanh outputs to physical parameter bounds while handling tensor type compatibility.

Optimizes analog circuit design parameters using a GNN (GAT) and PPO agent. Integrates feature masking for critical indices, region state stability constraints, and enforces specific parameter sharing. Separates graph connectivity (edge_index) from edge attributes (edge_features) and handles bipa...

Create a training loop that integrates learning rate scheduling (e.g., StepLR), logs metrics (loss, learning rate, validation accuracy), and follows production practices like checkpointing or early stopping (though the user didn't explicitly ask for checkpointing, the 'production level' and 'many...

PyTorch model training skill with custom training loops, gradient management, and GPU optimization.

Model interpretability and explainability using SHAP (SHapley Additive exPlanations). Use this skill when explaining machine learning model predictions, computing feature importance, generating SHAP plots (waterfall, beeswarm, bar, scatter, force, heatmap), debugging models, analyzing model bias ...

Configuration framework for complex applications (Hydra). Dynamic hierarchical configuration by composition and override via CLI, YAML, and structured configs. Use for ML experiment management, multi-environment deployment, hyperparameter sweeps, and reproducible research workflows. Integrates wi...

NVIDIA TensorRT — deep learning inference optimizer. FP16/INT8/INT4 quantization, kernel auto-tuning, layer fusion, and dynamic shapes. Max throughput on NVIDIA GPUs for production inference.

Fine-tune and serve Physical Intelligence OpenPI models (pi0, pi0-fast, pi0.5) using JAX or PyTorch backends for robot policy inference across ALOHA, DROID, and LIBERO environments. Use when adapting pi0 models to custom datasets, converting JAX checkpoints to PyTorch, running policy inference se...