kaggle-api-expert
Expert agent for Kaggle API authentication, dataset management, and running Kaggle notebooks on Texas Tech HPCC. Specializes in connecting Jupyter notebooks to Kaggle API and submitting to code competitions. Always checks VPN connection first before HPCC operations.
pytorch-geometric
Library for Graph Neural Networks (GNNs). Covers MessagePassing layers, modular aggregation schemes, and handling large graphs via mini-batching with disjoint graph representation. (pyg, messagepassing, gnn, gcn, gat, edge_index, knn_graph, global_mean_pool)
training-improvements-v245
Training improvements: LR warmup, validation intervals, reward weights. Trigger when: (1) training unstable in early epochs, (2) need more validation visibility, (3) model too conservative.
export-wizard
Master coordinator for model export. Guides the user through selecting the right format and initiating the process.
pytorch-lightning
High-level training framework for PyTorch that abstracts boilerplate while maintaining flexibility. Includes the Trainer, LightningModule, and support for multi-GPU scaling and reproducibility. (lightning, pytorch-lightning, lightningmodule, trainer, callback, ddp, fast_dev_run, seed_everything)
credit-model-validation-banking
Автоматизация процесса валидации моделей кредитного риска в банковской сфере. Используется для полного цикла валидации - от загрузки pickle модели и анализа данных до генерации детального отчета с метриками (AUC, Gini, Recall, Precision, F1, KS, PSI, CSI), визуализациями и соответствием регуляторным требованиям Казахстана.
arxiv-learn
INTERNAL MODULE - Use `arxiv learn` instead. This module provides the learn pipeline implementation for the arxiv skill.
few-shot-learning-finance
Use when implementing models that learn from minimal data or need to adapt to new market regimes rapidly. Covers episodic learning, context sets, support and query sequences, zero-shot vs few-shot learning, meta-learning for finance, transfer learning across assets and regimes, and quick adaptation to market changes.
describe-image
Uses a local model to describe something about an image
rl-foundations
Master RL theory - MDPs, value functions, Bellman equations, value/policy iteration, TD
marker-engine-rl
Vertieft den Marker-Engine-Skill um SFT/RL-Feinabstimmung mit LeanDeep 4.0; lädt Marker aus Supabase/ZIP und lernt eine Policy zur präzisen, kontextualisierten Marker-Anwendung bei strikter Bottom-up-Logik.
ai-engineering-skill
Practical guide for building production ML systems based on Chip Huyen's AI Engineering book. Use when users ask about model evaluation, deployment strategies, monitoring, data pipelines, feature engineering, cost optimization, or MLOps. Covers metrics, A/B testing, serving patterns, drift detection, and production best practices.
training-resilience
Fix PPO training early-stop issues. Trigger when: (1) impossible drawdown values (>100%), (2) training stops too early, (3) need adaptive recovery instead of hard stop.
trl
This skill should be used when users want to train or fine-tune language models using TRL (Transformer Reinforcement Learning) on Hugging Face Jobs infrastructure. Covers SFT, DPO, GRPO and reward modeling training methods, plus GGUF conversion for local deployment. Includes guidance on the TRL Jobs package, UV scripts with PEP 723 format, dataset preparation and validation, hardware selection, cost estimation, Trackio monitoring, Hub authentication, and model persistence. Should be invoked for tasks involving cloud GPU training, GGUF conversion, or when users mention training on Hugging Face Jobs without local GPU setup.
training-mlps
A skill for defining and training Multi-Layer Perceptrons (MLPs) using Flax NNX.
reward-shaping-engineering
Master reward function design - potential-based shaping, hacking patterns, validation
recommendation-ml
ML recommendation system development with collaborative filtering (Matrix Factorization), content-based filtering, and hybrid approaches. Use when building recommendation models, implementing Feast feature stores, setting up MLflow model registry, handling cold-start problems for new users/products, implementing diversity with MMR algorithm, or adding exploration with Thompson Sampling/epsilon-greedy bandits.
model-explainability-and-interpretability
Techniques and tools for understanding how machine learning models make decisions and explaining those decisions to stakeholders.
model-bias-and-fairness
Identifying, measuring, and mitigating algorithmic bias to ensure equitable outcomes in AI systems.
x-trend-architecture
Use when implementing X-Trend or attention-based trading models. Covers LSTM encoders, cross-attention, self-attention, sequence representations, entity embeddings, Variable Selection Networks, encoder-decoder patterns, Deep Momentum Networks, and interpretable predictions for trend-following strategies.
game-scoring
Use when working with candidate scoring, confidence calculation, softmax aggregation, or guess decision logic. Load for understanding how candidates are ranked, when the system decides to guess, and how semantic + geographic scores combine. Covers temperature tuning, entropy thresholds, and margin logic.
stan-fundamentals
Foundational knowledge for writing Stan 2.37 models including program structure, type system, distributions, and best practices. Use when creating or reviewing Stan models.
developing-flax-models
A comprehensive guide for developing, training, and managing neural networks using Flax NNX. Use when defining models, managing state, or writing training loops.