Posted by EditorDavid from the opening-day dept.
“From the beginning, we have believed that the right way to build these AI models is with open licenses,” says the Open Model Initiative. SD Times quotes them as saying that open licenses “allow creatives and businesses to build on each other’s work, facilitate research, and create new products and services without restrictive licensing constraints.”
Phoronix explains the community initiative “came about over the summer to help advance open-source AI models while now is becoming part of the Linux Foundation to further their cause.” As part of the Linux Foundation, the OMI will be working to establish a governance framework and working groups, create shared standards to enhance model interoperability and metadata practices, develop a transparent dataset for training and captioning, complete an alpha test model for targeted red teaming, and release an alpha version of a new model with fine-tuning scripts before the end of 2024.
The group was established “in response to a number of recent decisions by creators of popular open-source models to alter their licensing terms,” reports Silicon Angle: The creators highlighted the recent licensing change announced by Stability AI Ltd., regarding its popular image-generation model Stable Diffusion 3 (SD3). That model had previously been entirely free and open, but the changes introduced a monthly fee structure and imposed limitations on its usage. Stability AI was also criticized for the lack of clarity around its licensing terms, but it isn’t the only company to have introduced licensing restrictions on previously free software. The OMI intends to eliminate all barriers to enterprise adoption by focusing on training and developing AI models with “irrevocable open licenses without deletion clauses or recurring costs for access,” the Linux Foundation said.
InfoWorld also notes “the unavailability of source code and the license restrictions from LLM providers such as Meta, Mistral and Anthropic, who put caveats in the usage policies of their ‘open source’ models.” Meta, for instance, does provide the rights to use Llama models royalty free without any license, but does not provide the source code, according to [strategic research firm] Everest Group’s AI practice leader Suseel Menon. “Meta also adds a clause: ‘If, on the Meta Llama 3, monthly active users of the products or services is greater than 700 million monthly active users, you must request a license from Meta.’ This clause, combined with the unavailability of the source code, raises the question if the term open source should apply to Llama’s family of models,” Menon explained….
The OMI’s objectives and vision received mixed reactions from analysts. While Amalgam Insights’ chief analyst Hyoun Park believes that the OMI will lead to the development of more predictable and consistent standards for open source models, so that these models can potentially work with each other more easily, Everest Group’s Malik believes that the OMI may not be able to stand before the might of vendors such as Meta and Anthropic. “Developing LLMs is highly compute intensive and has cost big tech giants and start-ups billions in capital expenditure to achieve the scale they currently have with their open-source and proprietary LLMs,” Malik said, adding that this could be a major challenge for community-based LLMs.
The AI practice leader also pointed out that previous attempts at a community-based LLM have not garnered much adoption, as models developed by larger entities tend to perform better on most metrics… However, Malik said that the OMI might be able to find appropriate niches within the content development space (2D/3D image generation, adaptation, visual design, editing, etc.) as it begins to build its models… One of the other use cases for the OMI’s community LLMs is to see their use as small language models (SLMs), which can offer specific functionality at high effectiveness or functionality that is restricted to unique applications or use cases, analysts said. Currently, the OMI’s GitHub page has three repositories, all under the Apache 2.0 license.
Everyone can be taught to sculpt: Michelangelo would have had to be taught how not to. So it is with the great programmers.
Working…
