Stable Beluga vs Stable Beluga 2

Stable Beluga

Stability AI's instruction-tuned 65B LLM built on LLaMA

Stable Beluga is a 65-billion parameter language model from Stability AI, fine-tuned on LLaMA 65B using an internal Orca-style dataset to enhance instruction following. It excels at answering questions, completing writing tasks, and following complex instructions. Released under a non-commercial CC BY-NC-4.0 license, it targets researchers and developers building open-source LLM applications.

Stable Beluga 2

Stability AI's 70B instruction-following model based on Llama 2

Stable Beluga 2 is a 70-billion parameter language model from Stability AI, fine-tuned on Llama 2 70B using an Orca-style supervised dataset for improved instruction adherence. It supports system prompts and multi-turn conversation, making it suitable for chat applications and complex task completion. Available on Hugging Face under a non-commercial license, it targets researchers and enterprise AI teams.