Thinking of thinking machines, A strongly funded startup co -founded by OpenAi eminent researchers, revealed its first product – a tool called Tinker which automates the creation of personalized border AI models.
“We believe [Tinker] Will help allow researchers and developers to experiment with models and make border capacities much more accessible to all people, “said Mira Murati, co -founder and CEO of Thinking Machines, in an interview with Wired before the announcement.
Large companies and academic laboratories already refine the Open Source models to create new variants that are optimized for specific tasks, such as solving mathematical problems, drafting legal agreements or the answer to medical questions.
As a rule, this work consists in acquiring and managing GPU clusters and using various software tools to ensure that large -scale training races are stable and effective. Tinker promises to allow more companies, researchers and even lovers to settle their own AI models by automating a large part of this work.
Essentially, the team bet that helping people refine border models will be the next great thing in AI. And there are reasons to believe that they might be right. Thinking Machines Lab is led by researchers who have played a fundamental role in creating Chatgpt. And, compared to similar tools on the market, Tinker is more powerful and friendly, according to the beta testers with whom I spoke.
Murati says that Thinking Machines Lab hopes to demystify the work involved in setting the most powerful AI models in the world and allow more people to explore the external limits of the AI. “We do what is also a border capacity accessible to all, and that completely changes the situation,” she says. “There are a ton of intelligent people, and we need as many intelligent people as possible to carry out research on AI.”
Tinker currently allows users to adjust two open source models: Meta’s Llama and Qwen d’Alibaba. Users can write a few lines of code to press the API Tinker and start adjusting a supervised learning, which means adjusting the model with labeled data or by learning to strengthen, an increasingly popular method to regulate models by giving them positive or negative feedback depending on their outings. Users can then download their refined model and execute it wherever they want.
The AI industry looks closely at the launch – partly due to the caliber of the team behind.
