This repository serves as a curated list of "awesome" resources related to huge AI models, primarily focusing on Large Language Models (LLMs) and Vision Models. It's a valuable reference for researchers, engineers, and practitioners tracking the rapid advancements and open-source contributions in the field of large-scale AI.

How It Works

The collection is organized into categories such as Language Models, Vision Models, Reinforcement Learning, Speech, and supporting frameworks for training and inference. Each entry typically includes the model's name, developer, release date, parameter count, training data size, architecture, and licensing information. The emphasis is on highlighting open-source models and providing links to relevant papers or repositories.

Quick Start & Requirements

This is a curated list, not a runnable project. To use specific models, refer to their individual project pages or papers linked within the repository.

Highlighted Details

• Comprehensive catalog of LLMs and Vision Models, detailing parameters, training data, and architectures.
• Tracks the evolution of models from early transformers (BERT, GPT-2) to massive models (GPT-4, PaLM, Switch Transformer).
• Includes extensive lists of distributed training frameworks (PyTorch, TensorFlow, JAX ecosystems) and inference tools.
• Features surveys and key papers that shaped the field of large-scale AI.

Maintenance & Community

The repository is maintained by zhengzangw. Updates reflect the fast-paced nature of LLM development, with entries dated up to June 2023.

Licensing & Compatibility

The repository itself is a list and does not have a specific license. Individual models listed have varying licenses, with many open-source models released under permissive licenses like Apache 2.0. However, some prominent models are closed-source.

Limitations & Caveats

The information is a snapshot as of June 2023 and may not include the very latest models or developments. Some details for older models might be less precise or incomplete.