Thinking machine lab, A heavy -funded startup conducted by prominent researchers from OpenAI has released its first product – a tool called Tinker, which automatically automatically automatically made the Custom Frontier AI models.
“We believe [Tinker] Researchers and developers will help to be empowered to experiment with models and make border capacity more accessible to all people, “said Mira Murai, Cafouder and CEO of Thinking Machines in an interview with Wired before this announcement.
Large companies and academic labs are already fine-tune open source AI models to create new varieties that are favorable for specific tasks, such as solving mathematics problems, drafting legal agreements or answering the question of treatment.
Generally, this job is involved in acquiring and managing the clusters of GPUs and using various software equipment and ensuring that large size training is stable and efficient. Tinker has promised more business, researchers and even hobbies to automatically allow their own AI models to be subtle.
Basically, the team is betting that helping people’s subtle-melted frontier models will be the biggest thing of AI. And there is reason to believe that they can be right. Researchers have helmed the thinking machine lab who played a key role in creating the ChatGP. And, comparing similar tools in the market, Tinker is more powerful and user -friendly, according to the beta examiners I talked to.
Murarati says that the thought machine labs hope to eradicate the work involved in the world’s most powerful AI models and make it possible for more people to explore the outer limits of AI. “We are creating a border capacity to be accessible to everyone else,” he said, and it is full game-changing, “he says. “There are a lot of smart people and we need as smart people as possible to study border AI.”
Tinker is currently allowing users to subtle two open source models: Meta Lama and Alibaba Queen. Users can write a few lines of codes to tap the Tinker API and start fine-tuning through supervised education, which means that the model is growing with a labeled data or adjusting to reinforced education, positive or negative reaction based on their results. Then users can download their fine tune model and run it wherever they want.
The AI industry is looking closely at the launch of some part because of the team’s calliba.