May 4, 2022
On Tuesday, Facebook’s parent company, Meta announced plans to open up access to a large language model for AI research.
The company stated that restricting access to such models in the past had been “hindering progress on efforts to improve their robustness and mitigate known issues such as bias and toxicity.”
Therefore, making the “Open Pretrained Transformer (OPT-175B)” model available would help improve researchers’ ability to understand how large models work.
While Meta hopes to increase the diversity of voices defining the ethical considerations of such technologies, the company will however take steps to prevent misuse.
To prevent misuse, Meta will release the model under a noncommercial license to focus on research use cases.
Those that will have access to the model include academic researchers and people affiliated with government, civil society and academic organizations, and industry research laboratories.
“Large language models” are seen as natural processing systems which are trained on massive volumes of text, and are capable of answering reading comprehension questions or generating new text.
The sources for this piece include a story in Reuters.
