Thinking Machines Lab, founded by former OpenAI executives, unveils Tinker, a platform designed to simplify the process of fine-tuning large language models (LLMs) for wider accessibility. The platform automates much of the work involved in fine-tuning, allowing users to customize open-source AI models for specific tasks.
Thinking Machines Lab, a new venture founded by former OpenAI executives, is launching Tinker, a platform designed to simplify the process of fine-tuning large language models (LLMs). Mira Murati, co-founder and CEO of Thinking Machines, stated in an interview that Tinker aims to empower researchers and developers by making frontier AI capabilities more accessible.
The platform promises to automate much of the complex work involved in fine-tuning, allowing a wider range of users, from businesses to hobbyists, to customize open-source AI models for specific tasks. The team is betting that democratizing access to fine-tuning is the next major advancement in AI, and the launch is generating significant interest within the industry.\Tinker currently supports fine-tuning of two open-source models: Meta's Llama and Alibaba's Qwen. Users can utilize the Tinker API with minimal code to initiate fine-tuning via supervised learning, which involves adjusting the model using labeled datasets, or through reinforcement learning, a technique that uses feedback to improve model performance. The platform then allows users to download their fine-tuned models for deployment. This approach seeks to democratize access to powerful AI capabilities, moving beyond the current limitations of requiring significant resources and technical expertise. The team behind Tinker includes key figures who played pivotal roles in the creation of ChatGPT. Notably, the startup secured a substantial $2 billion in seed funding, indicating strong investor confidence and a valuation of $12 billion.\The industry is watching the launch closely due to the expertise and experience of the team. The platform aims to make cutting-edge AI research accessible to a broader audience. Tinker also offers a user-friendly experience compared to other fine-tuning tools. Beta testers have praised Tinker's combination of abstraction and tunability. While similar tools like VERL and SkyRL already exist, Tinker promises a more streamlined approach. A significant advantage lies in the platform's ability to simplify complex processes like reinforcement learning, enabling users to experiment and extract capabilities that were previously challenging to achieve. The team emphasizes their goal of providing full control over the training loop while abstracting away the complexities of distributed training. This control is coupled with a commitment to making frontier AI research accessible to a broader audience, which can help accelerate innovation. The company plans to offer the API for free initially but expects to transition to a paid model in the future. Tinker is positioned to be an essential tool for anyone looking to push the boundaries of AI research and development
Artificial Intelligence AI Fine-Tuning Large Language Models (Llms) Openai Reinforcement Learning
United States Latest News, United States Headlines
Similar News:You can also read news stories similar to this one that we have collected from other news sources.
Tinker Is Thinking Machines Lab's First AI Product, But It's Not The ChatGPT Rival Some ExpectedChris started blogging about tech by accident when he figured out his passion for consumer electronics, especially mobile devices, and telling stories could be intertwined.
Read more »
War machines: 10 mightiest military powers shaping global order and securityManpower, budgets, and cutting-edge weapons: the 10 most powerful militaries shaping global security in 2025.
Read more »
Mira Murati’s Thinking Machines Lab Unveils New AI Tinker Product Which Is Useful But Not A Big-Time BlockbusterFamed Mira Murati and her co-founded startup TML released their first-ever product, Tinker. I explain the problem that Tinker aims to solve. And what this means for AI.
Read more »
Padres Daily: The end hits hard; thinking change; ready for ABSBrief but heated exchange with umpires follows Game 3 loss, in which offense is again guilty of not doing enough
Read more »
6 Standing Moves That Eliminate Love Handles Faster Than Gym Machines After 50Your ultimate source for expert nutrition tips and health advice, covering wellness, healthy recipes, cooking hacks, food news, style trends and shopping.
Read more »
Scientists Team Up With Michelin Chefs to Recreate Ancient Yogurt—With AntsAccording to an old Bulgarian tradition, some ants are the perfect, all-natural yogurt-making machines.
Read more »