Open Models

We hope that the information on recently released AI models will be useful for education and research purposes.

presentation

Meta's LLaMA 3.2 (Released: 2024)

  • Types: Lightweight Language model (1B & 3B) and Multimodal model (11B & 90B)
  • Key Features: LLaMA is designed to enhance language understanding with improved efficiency and scalability. It focuses on NLP tasks like text generation, summarization, and translation. The model is optimized for accessibility in academic and research settings.
  • https://www.llama.com/

Pangu-α (Released: September 2024)

  • Types: Language model (MAX 200B)
  • Key Features: Developed by Huawei, Pangu-α is a transformer-based model optimized for a wide range of NLP tasks, including text generation, question-answering, and content summarization. It is engineered for performance at scale in real-world applications.
  • https://github.com/Huawei

OpenAI EvalAI (Released: October 2024)

  • Types: Evaluation Frameworks
  • Key Features: EvalAI is an evaluation framework designed to assess AI models' performance across a variety of real-world tasks. It helps in evaluating models for tasks such as text generation, reasoning, and decision-making.
  • https://eval.ai/