Want to create your own ChatGPT? Start with StableLM
- April 25, 2023
- 0
StableLM lets you get started with two rich models via GitHub if you agree to the prerequisites. If you want to get started with a Large Language Model
StableLM lets you get started with two rich models via GitHub if you agree to the prerequisites. If you want to get started with a Large Language Model
StableLM lets you get started with two rich models via GitHub if you agree to the prerequisites.
If you want to get started with a Large Language Model (LLM) like ChatGPT, you can now consult two models for free on GitHub. Stability AI has provided two models of its StableLM. The smallest model contains 3 billion parameters, the larger version 7 billion. In the long term, Stability AI also wants to put models with 15 billion and 65 billion parameters online so that developers can get started.
With ChatGPT, the manufacturer positions itself as an open source alternative to OpenAI. The latter, while having āopenā in its name, is anything but that these days. Stability AI believes that LLMs will be the backbone of our digital economy. “Everyone should have a voice in this,” says a blog post in which the model is made public.
StableLM claims that it achieves the same performance as GPT-3, although the latter was trained with 175 billion parameters. A smaller data set makes an LLM more efficient and easier to run locally on smartphones and tablets, for example.
Developers looking to get started with StableLM should review the rules surrounding all projects. The Company makes the LLM available under a Creative Commons BY-SA 4.0 license. In short, in projects you must give credit to the original creator and grant the same license.
Stability AI extracted training data from a new experimental dataset based on the open-source The Pile, but three times larger. Ars Technica started working with StableLM and found that the 7 billion parameter model was more responsive and robust than GPT-3. The release used an Alpaca Method-based model for dialogues.
StableLM is new, but Stability AI is not. The company provided funding to enable an open-source deployment of Stable Diffusion. This open tool can generate images, similar to OpenAI’s Dall-E, but allowed developers to refine everything further. What followed was a lightning-fast development of AI imagery. Stability AI was discredited for training its model on illegally sourced images, including those from Getty Images.
Source: IT Daily
As an experienced journalist and author, Mary has been reporting on the latest news and trends for over 5 years. With a passion for uncovering the stories behind the headlines, Mary has earned a reputation as a trusted voice in the world of journalism. Her writing style is insightful, engaging and thought-provoking, as she takes a deep dive into the most pressing issues of our time.