Tolgay Atınç Uzun.

Tolgay Atinc Uzun et al.

Introduces a large-scale framework with more than 14,000 architectures and 750,000 structured training records spanning generative, evaluative, and deployment pipelines.

• Dataset
• AutoML
• Deployment

Tolgay Atinc Uzun, Dmitry Ignatov, Radu Timofte

Combines AST mutation, conditional code generation, and performance feedback to discover executable vision architectures with non-standard channel-width patterns.

• LLM
• Channel Search
• CIFAR-100

Roman Kochnev, Waleed Khalid, Tolgay Atinc Uzun et al.

Presents a self-improving AutoML framework that unifies architecture synthesis, hyperparameter optimization, accuracy prediction, retrieval-augmented generation, and reinforcement learning.

• AutoML
• LLM
• PyTorch

Tolgay Atinc Uzun, Radu Timofte, Dmitry Ignatov

Scales LLM-guided channel search to 2,000 candidates and studies accuracy, parameter efficiency, and architectural regularities across 462 verified CIFAR-100 evaluations.

• LLM
• NAS
• Computer Vision

Arash Torabi Goodarzi, Tolgay Atinc Uzun et al.

Introduces LEMUR, an open-source dataset and framework of PyTorch-based neural networks with unified templates, structured results, and Optuna-based hyperparameter optimization for seamless AutoML.

• Dataset
• AutoML
• PyTorch