Home ยป Introducing Stable LM 2: A Compact Linguistic Model by Stability AI, Outperforming Other SLM Models in Rigorous Testing

Introducing Stable LM 2: A Compact Linguistic Model by Stability AI, Outperforming Other SLM Models in Rigorous Testing

Stability AI, the company behind Stable Diffusion image generation models, has introduced a new version of their language model called Stable LM 2. This updated version comes as an improvement over the initial Stable LM model, which was launched in April 2023.

One interesting aspect of Stable LM 2 is its reduced size, now comprising only 1.6 billion parameters. Compared to the initial version with 3 billion parameters, it falls into the category of small language models (SLM). The advantage of this smaller size is that it can run efficiently on low-spec hardware, offering greater accessibility compared to larger language models (LLM).

Stability AI claims that the performance of Stable LM 2 surpasses other small language models in the market, including Microsoft Phi-1.5 (1.3B) and Phi-2 (2.7B), TinyLlama 1.1B, and Falcon 1B. It outperforms models with less than 2 billion parameters and even competes with Phi-2 in certain test sets.

As for availability, Stable LM 2 can be used for free if it is not for commercial purposes. However, for commercial usage, users are required to subscribe to the Stability AI Membership at a monthly price of $20.

TLDR: Stability AI has launched Stable LM 2, an updated language model with a smaller size of 1.6 billion parameters. It outperforms other small language models in the market and offers free usage for non-commercial purposes. For commercial usage, a Stability AI Membership subscription at $20 per month is necessary.

More Reading

Post navigation

Leave a Comment

Leave a Reply

Your email address will not be published. Required fields are marked *

[Whispers] AI Stability in Crisis: Financial Troubles, Inability to Pay GPU Costs, Employee Turnover

AI Unleashed: Crafting 3D Object Videos with Multifaceted Perspectives Using Image Inputs

Revolutionary AI Unveils Exquisite Stable Audio Model with Unmatched Velocity