Nous Capybara 7B

About

The Nous Capybara 7B is a large language model from NousResearch, built on the Llama architecture with 7 billion parameters. Known for its conversational proficiency, it excels in multi-turn dialogues and complex summarization. It was trained on a dataset of approximately 20,000 conversational examples using a novel synthesis technique called Amplify-Instruct. Capable of recalling information up to late 2022, the model processes up to 4096 tokens per interaction, making it ideal for chatbots and virtual assistants, although it faces limitations in data size and awareness of recent developments.

Capabilities