15 Graphs That Explain the State of AI in 2024
Google DeepMind Releases TxGemma: Open AI Models For Drug Discovery
In a move that could accelerate and reshape therapeutic development, Google DeepMind has launched TxGemma, a suite of open AI models designed to make drug discovery faster, smarter, and more cost-efficient. The announcement was made by Shekoofeh Azizi, Staff Research Scientist, on March 25, 2025.
What is TxGemma?TxGemma is a collection of lightweight, high-performing open models that are fine-tuned from DeepMind's Gemma 2 architecture. These models are built specifically to assist in therapeutic development, from identifying promising drug targets to predicting outcomes of clinical trials.
With over 7 million therapeutic examples used in training, TxGemma models excel in prediction and conversational analysis, potentially helping researchers cut down the years and billions traditionally spent in drug development.
The need for efficient drug development is greater than ever. Nearly 90% of drug candidates fail beyond Phase 1 trials, costing companies time, resources, and human capital. TxGemma offers a powerful solution by using large language models (LLMs) to better understand molecular properties, toxicity, blood-brain barrier crossing, binding affinities, and more.
In early benchmarks, the TxGemma 27B Predict model either outperformed or matched Google's previous state-of-the-art model, Tx-LLM, across 64 of 66 tasks. It even held its ground against specialized single-task models, outperforming them on 26 out of 66 tasks.
Who is it for and how can it be used?The TxGemma release includes three model sizes: 2B, 9B, and 27B, each available in two versions:
To support real-world application, Google has released Colab notebooks for developers to fine-tune TxGemma on their own data, using datasets like TrialBench. This enables more accurate, proprietary predictions tailored to specific drug development pipelines.
What's new with Agentic-Tx?TxGemma is also part of Agentic-Tx, an advanced agentic AI system built on Gemini 2.0 Pro. With access to 18 specialized tools, Agentic-Tx can handle multi-step reasoning using data from PubMed, Wikipedia, and various gene/protein databases.
Agentic-Tx has achieved state-of-the-art performance on benchmarks like ChemBench and Humanity's Last Exam, showcasing its ability to answer complex biology and chemistry questions.
Where to access it?TxGemma is now available on:
• Vertex AI Model Garden• Hugging Face
Researchers and developers are encouraged to try out the inference models, fine-tuning tutorials, and agent system demos. The goal is to foster an open, collaborative ecosystem for improving therapeutic AI research.
Qualcomm Acquires MovianAI To Bolster Its AI Capabilities
Qualcomm Technologies Inc. Today announced that it has acquired MovianAI, a Vietnamese artificial intelligence developer.
MovianAI is the generative AI arm of VinAI Application and Research JSC, a machine learning research lab. VinAI is a unit of Vingroup, one of Vietnam's largest conglomerates. The latter company operates hotels, manufactures cars and competes in several other markets.
"By bringing in high-caliber talent from VinAI, we are strengthening our ability to deliver cutting-edge AI solutions that will benefit a wide range of industries and consumers," said Jilei Hou, Qualcomm's senior vice president of engineering.
VinAI, MovianAI's parent organization, is led by former DeepMind research scientist Hung Bui. It has published multiple papers on generative AI since launching in 2019. VinAI has also uploaded dozens of open-source projects to GitHub.
About two years ago, VinAI researchers presented a new approach to building image generation models. The technique involves using an existing image generation model to train a smaller algorithm that can perform the same tasks using less hardware. According to VinAI, the technology reduces the need for training data and thereby eases development.
Besides new machine learning techniques, the lab has also developed several open-source AI models. One of its image generation algorithms combines three different neural network designs: the diffusion, Transformer and Mamba architectures. Mamba is a relatively new alternative to the Transformer architecture that can be more hardware-efficient in certain situations.
VinAI's research efforts also extend beyond image generation. It has developed a custom large language model, RecGPT-7B, that is optimized for recommendation generation tasks.
"We are ready to contribute to Qualcomm's mission of making breakthroughs in fundamental AI research and scale them across industries, including smartphones, PCs, software-defined vehicles, and more," said VinAI CEO Hung Bui, who is joining the chipmaker as part of the deal.
MovianAI's model development expertise could help Qualcomm enhance its suite of AI development tools. Those tools make it easier for companies to deploy machine learning applications on devices
Qualcomm provides a library of preconfigured AI models optimized to run on its silicon. Those algorithms remove the need for companies to train new algorithms from scratch, which speeds up development. For software teams that nevertheless wish to build custom models, Qualcomm provides tools that help optimize those models' performance.
The MovianAI deal comes less than a month after the chipmaker acquired Edge Impulse Inc., another AI development specialist. The latter company's software makes it easier to build AI models that can run efficiently on connected devices. Prior to the deal, Edge Impulse raised about $54 million in funding and built up an installed base of more than 170,000 developers.
Photo: Qualcomm Your vote of support is important to us and it helps us keep the content FREE. One click below supports our mission to provide free, deep, and relevant content. Join our community on YouTube Join the community that includes more than 15,000 #CubeAlumni experts, including Amazon.Com CEO Andy Jassy, Dell Technologies founder and CEO Michael Dell, Intel CEO Pat Gelsinger, and many more luminaries and experts."TheCUBE is an important partner to the industry. You guys really are a part of our events and we really appreciate you coming and I know people appreciate the content you create as well" – Andy Jassy
THANK YOU
Isomorphic Labs, Google's A.I. Drug Business, Raises $600 Million
Over the last 12 months, Google's efforts to use artificial intelligence to accelerate drug design have achieved breakthroughs in mimicking human biology and won its top scientists the Nobel Prize in Chemistry.
Now Isomorphic Labs, the division within the software giant meant to develop and commercialize the technology, is taking another big step: raising money from an outside investor.
Isomorphic announced on Monday that it had raised $600 million, led by Thrive Capital, the venture capital firm that has bet big on A.I. Companies including OpenAI. GV, Google's venture capital arm, and Alphabet, Google's parent company, also invested.
The announcement underscores Google's ambitions for Isomorphic, which was spun out of the company's DeepMind lab to focus on drugs discovery. It is built on software that DeepMind, a central intelligence lab in London, has developed. That includes AlphaFold, which can predict the structure of millions of proteins and more.
AlphaFold, which in its third iteration can now predict the complex behavior of DNA and RNA, has promised to slash the development time of new drugs. Such is its promise that Demis Hassabis, a co-founder of Isomorphic and DeepMind, and John M. Jumper, a DeepMind researcher, shared half of the Nobel in chemistry last year.
The goal, according to Mr. Hassabis, is to eventually conduct most of the drug discovery process via computers, rather than traditional labs that require biological materials, strict safety requirements — and lots of time.
We are having trouble retrieving the article content.
Please enable JavaScript in your browser settings.
Thank you for your patience while we verify access. If you are in Reader mode please exit and log into your Times account, or subscribe for all of The Times.
Thank you for your patience while we verify access.
Already a subscriber? Log in.
Want all of The Times? Subscribe.
Comments
Post a Comment