Stability AI launches open-source ChatGPT rival
Artificial intelligence startup Stability AI recently launched StableLM, its first open-source language model.
Stability AI is the company behind the image generation model Stable Diffusion, whit it released last year.
The company said the suite of StableLM language models could generate text and code for powering a range of downstream applications.
“They demonstrate how small and efficient models can deliver high performance with appropriate training,” Stability AI said.
StableLM is trained on a new experimental dataset built on The Pile, which consists of Internet-scraped text from websites such as Wikipedia, PubMed, and StackExchange.
However, Stability AI has roughly tripled The Pile to 1.5 trillion tokens of content.
It said the richness of the expanded dataset offered “surprisingly” high performance in conversational and coding tasks, despite its small size of 3–7 billion parameters.
For reference, OpenAI’s GPT-3 language model has 175 billion parameters, making it much more performance-intensive.
Stability AI plans to roll out models with 15 billion and 65 billion parameters.
The screenshots below show examples of responses given by the current release of the StableLM language model.
Some have warned that companies invested in such models could tweak code with an inherent bias to help boost their profits.
Tech billionaire Elon Musk is among several big names interested in AI development who are concerned about the impact of generative AI models developed behind closed doors.
Stability AI said its release of StableLM demonstrated a commitment to transparent AI technology.
“Researchers can ‘look under the hood’ to verify performance, work on interpretability techniques, identify potential risks, and help develop safeguards,” Stability AI said.
“Organizations across the public and private sectors can adapt these open-source models for their own applications without sharing their sensitive data or giving up control of their AI capabilities.”
In addition to making the model more transparent, StableLM was designed to be broadly accessible and run locally, instead of relying on powerful hardware in a data centre acting as the model’s brains.
“Using these models, developers can build independent applications compatible with widely-available hardware instead of relying on proprietary services from one or two companies,” said Stability AI.
“In this way, the economic benefits of AI are shared by a broad community of users and developers.”
“Open, fine-grained access to our models allows the broad research and academic community to develop interpretability and safety techniques beyond what is possible with closed models.”
Finally, Stability AI hopes to be supportive of users rather than replacing them, like some experts believe AI will do.
“We are focused on efficient, specialized, and practical AI performance — not a quest for god-like intelligence.”
“We develop tools that help everyday people and everyday firms use AI to unlock creativity, boost their productivity, and open up new economic opportunities.”
For developers who want to give it a test run, StableLM’s code is available in a GitHub repository that was uploaded last week.
Now read: Real-world study shows generative AI tools made workers 14% more productive
Loading ...