Meta Llama

library active free

Meta's family of open-weight LLMs, including Llama 2 and Llama 3. Llama models can be downloaded and run locally or fine-tuned, making them the foundation for most open-source LLM development. Llama 3.1 405B rivals closed-source models.

Implements

Concepts this tool claims to implement:

  • Large Language Model primary

    Llama 3.1 (8B, 70B, 405B), Llama 3 (8B, 70B), Llama 2 (7B, 13B, 70B). 405B competitive with GPT-4.

  • Foundation Model primary

    Open weights enable fine-tuning and adaptation. Base models available for custom training.

  • Instruction Tuning secondary

    Llama 3.1 Instruct variants are instruction-tuned for chat and assistant use cases.

Integration Surfaces

  • Model weights (Hugging Face)
  • Llama.cpp
  • vLLM
  • TGI
  • Ollama

Details

Vendor
Meta
License
Llama Community License
Runs On
local, cloud
Used By
human, agent, system

Links

Notes

Llama democratized LLM development by releasing capable open-weight models. Most open-source LLM projects build on Llama. The 405B model proved that open models can match proprietary capabilities.

Related Tools