OLMo 2 7B Instruct

Permissive

Allen Institute for AI · OLMo · Released 2024-11

7BText

A fully open model in the truest sense: weights, training data, and training code are all published, not just the checkpoint.

Strengths

  • +Apache 2.0 license
  • +Full training data and recipe transparency, unusual among open-weight models
  • +Competitive with similarly-sized Llama/Qwen releases

Limitations

  • -Smaller ecosystem of fine-tunes and quantizations than Llama/Qwen
  • -4K-32K context depending on stage, smaller than some 2024+ peers

License

Apache 2.0

use commercially with attribution niceties

Hardware

runs on a good consumer GPU (or Apple Silicon) with quantization

Links

Stats

allenai/OLMo-2-1124-7B-Instruct— downloads·— likes

via Hugging Face

Related models