Julia Gong

We're really excited to announce NetMind.AI’s ongoing partnership with Carnegie Mellon University's School of Computer Science! Through this collaboration, we already had the privilege of providing computing power for a groundbreaking study focused on optimising AI inference efficiency. Their groundbreaking study, (titled “An Empirical Analysis of Compute-Optimal Inference for Problem-Solving with Language Models,") explored ways to balance computational demands with performance, particularly in optimising language models (LLMs) for both efficiency and accuracy. 💡 So what was the key takeaway?  The findings imply that smaller models - when paired with sophisticated algorithms - can actually match the performance of their larger counterparts, paving the way for far more accessible and efficient AI, especially for end-device applications. We’re really proud that NetMind's dynamic, decentralised computing infrastructure played a crucial role in enabling this important research. This ongoing partnership also serves to highlight our commitment to pushing the envelope of AI research by aligning with top-tier academic institutions that push the boundaries of what's possible. So why partner with leading research institutes like Carnegie Mellon? Because we believe in the democratisation of AI. By supporting these amazing research institutions, we aim to make AI technologies beneficial for all - not just a select few.  We also want to take a moment to say a huge thank you to Prof. Graham Neubig and Zhiqing Sun for making this collaboration possible. Congrats on the publication of your research! 👏🏽 Curious to dive into the study? You can check it out right here: https://lnkd.in/gZXd47hH Stay tuned for more exciting and innovative research collaborations as we continue to push the frontiers of AI! #AIResearch #NetMind #CarnegieMellon 

4

1 Comment

Curious to learn how Large Languages models are being used to train empathy and negotiation skills? If so, check out this great work by Stanford Institute for Human-Centered Artificial Intelligence (HAI) Yang Diyi

3

This week in #TheBatch: 🤖 Waymo argues its cars are safer than those with human drivers 🌍 AI treaty supports innovation and human rights ⚖️ A method to balance datasets automatically Plus, Andrew Ng, discusses how most AI capabilities develop gradually, if you’re paying close attention. Read The Batch now: https://hubs.la/Q02PDh8k0

51

1 Comment

📢 Insight from Stanford HAI: Do Healthcare Algorithms Need to Be Generalizable? In a blog post by the Stanford Institute for Human-Centered Artificial Intelligence (HAI), Dr. Nigam Shah suggests that while generalization is key for AI reliability, it’s just as important to tailor models to the specific needs of hospitals and medical institutions. By considering local differences in how people interact with their healthcare providers, we can ensure that AI models are not just reliable but also effective in serving diverse communities. Want to know more? Dive into the full article here: 🔗 https://lnkd.in/dZe_SGDt #AI #Healthcare #MachineLearning #StanfordHAI #DigitalHealth #HealthTech #AIinHealthcare #DataScience #Innovation

7

Afforai has officially partnered with Semantic Scholar, an AI-powered research tool for scientific literature, based at the Allen Institute for AI (AI2). Using the Semantic Scholar API, researchers can supplement Afforai AI research assistant verified knowledge from a 128 million peer-reviewed research papers database. Afforai is now featured on the Semantic Scholar Gallery. This showcases innovative applications of the Semantic Scholar APIs, highlighting how partners like us are contributing to the mission of accelerating scientific discovery.  This collaboration marks a significant step forward in our commitment to enhancing the research capabilities of scholars worldwide. *Allen Institute for AI (AI2) is a non-profit research institute founded by the visionary Paul G. Allen, Semantic Scholar represents the forefront of AI research and engineering. AI2’s mission is to leverage artificial intelligence for the greater good, a goal that aligns perfectly with our own at Afforai.

20

4 Comments

Exciting developments in AI bias reduction! 🎉 Oregon State University doctoral student Slyman, in collaboration with Adobe researchers, has developed FairDeDup, an innovative and cost-effective training technique to reduce social biases in AI systems. Key points about FairDeDup: Utilizes pruning to eliminate redundant data, mitigating biases related to occupation, race, gender, age, geography, and culture. Enhances AI training accuracy and equity without relying on predefined fairness notions. Slyman unveiled the algorithm at the IEEE/CVF Conference on Computer Vision and Pattern Recognition in Seattle. This project also included valuable contributions from OSU assistant professor Stefan Lee and Adobe researchers Scott Cohen and Kushal Kafle. Let's celebrate this leap forward in creating fairer AI systems! 🚀 #AI #BiasReduction #Innovation #FairDeDup https://lnkd.in/eRK6Js7g

2

Join us on May 29, 2024, for a webinar covering deep learning frameworks on Argonne National Laboratory's Aurora exascale supercomputer. In this developer session, ALCF's Corey Adams will discuss deep learning frameworks optimized for Aurora, including TensorFlow, PyTorch, and the experimental release of JAX. Other topics include how to achieve optimal single device performance, as well as scaling up and scaling out models. The session will include examples and demonstrations that can run on Sunspot or Aurora. https://lnkd.in/gY2WMZ3x

7

On this episode of #YourAIInjection, I had a great chat with professors Karen Jacobs, EdD and Alyson Stover about the potential of AI in occupational therapy. We explored how AI is helping tackle challenges like practitioner shortages and long wait times, using tools like computer vision and augmented reality to gather objective data and personalize therapy. Tune in now to explore the transformative role of AI in healthcare, and the future of rehabilitative sciences: https://lnkd.in/gKthYM5Z

9

1 Comment

Researchers at UC Berkeley Unveil a Novel Interpretation of the U-Net Architecture Through the Lens of Generative Hierarchical Models https://lnkd.in/g__ww994 Artificial intelligence and machine learning are fields focused on creating algorithms to enable machines to understand data, make decisions, and solve problems. Researchers in this domain seek to design models that can process vast amounts of information efficiently and accurately, a crucial aspect in advancing automation and predictive analysis. This focus on the efficiency and precision of AI systems remains a central challenge, particularly as the complexity and size of datasets continue to grow. AI researchers encounter significant progress in improving mixing models for high performance without compromising accuracy. With data sets expanding in size and complexity, the computational cost associated with training and running these models is a critical concern. The goal is to create models that can...

1

Mixed feelings about AI in healthcare? You're not alone. While AI shows promise in medical imaging and preventive care, 60% of Americans are uneasy with AI in their medical care. 🏥💬 🔗 Check out our most recent blog post to learn how UCLA Health and Collibra are pioneering responsible AI governance in healthcare: https://lnkd.in/gWgjBwbD

20

1 Comment

Mixture-of-Agents as an Event-Driven System 🤖☎️ This paper by Wang et al. shows you how to ensemble smaller LLMs to create a system that can outperform state-of-the-art larger models. We’ve implemented this paper in a fully async, event-driven workflow system thanks to Ravi Theja Desetty - treat each “small LLM” as an event-driven step that can process incoming events and respond to it, independently and in parallel. ✅ Take full advantage of processing an entire batch of requests  ✅ Cleaner, readable code LlamaPack: https://lnkd.in/gRuqQGk2 Learn more about workflows: https://lnkd.in/gqWwz7Xs

311

9 Comments

🔍 New Research Alert from CDS! Shi Feng, a Postdoc Researcher at CDS, alongside Arjun P. and CDS Associate Professor of Linguistics and Data Science Sam Bowman, have unveiled findings on a novel AI bias. Their study, titled “LLM Evaluators Recognize and Favor Their Own Generations,” reveals that large language models like GPT-4 often prefer their own text. This self-preference bias could affect everything from content moderation to AI evaluations, and has the potential to interfere with human decision-making. Read more about their pioneering work: https://lnkd.in/eS4F3Zkw #DataScience

17

1 Comment

If you're in the Bay Area, join Prashanth Rao at GitHub HQ for the AICamp meetup on June 4th! In our talk there, we'll be exploring some of the ways in which knowledge graphs are used in conjunction with LLMs and graph databases to power RAG applications that provide insights from structured or unstructured data. We'll also go over the basics of graph databases and what Kùzu brings to the table. RSVP at the link below, and share with your friends and colleagues who are interested in learning more about graphs and their intersection with #rag! https://lnkd.in/eKQdFD8E

19

LongRoPE is making it possible to extend language model context windows, including for the Microsoft Phi-3 family of SLMs, while maintaining performance. Learn about the work, featured at #ICML2024, with podcast guest and Senior Researcher Li Lyna Zhang. https://msft.it/6043lbpeH

55

This week on The Changelog: Carol Lee, PhD! 👏 She joins Jerod Santo & Adam Stacoviak to share her research on code review anxiety. We dive deep into her recent research paper “Understanding and Effectively Mitigating Code Review Anxiety”. We get into all the nooks and crannies of this topic — common code review myths, strategies for coping, the need for awareness and self-reflection, the value of exposure and practice to build confidence, the importance of team dynamics, respect, empathy, and connection, and more. This show is jam-packed with goodies for everyone… #softwareengineering #github #git #collaboration #brainscience #anxiety #research #opensource #podcast

9

2 Comments

This is exciting news from the Allen Institute for AI at the University of Washington! They have unveiled OpenScholar, a new AI system for scientific discovery. OpenScholar utilizes a fine-tuned, retrieval-augmented LLM constructed from a vast data repository of 45 million open-access scientific papers. What sets this system apart is its active retrieval system, which extracts pertinent papers, synthesizes findings, and continually enhances results through natural language feedback and the assimilation of additional data. Here are the reasons for my enthusiasm: - The outcomes are verifiably factual, supported by citations. - The LLM is developed from a meticulously curated data store of high-quality, peer-reviewed papers. - The entire system has been made open source, fostering collaboration and innovation. - With an 8 billion parameter model, OpenScholar is not only efficient to construct and operate, but also more cost-effective and sustainable compared to larger language models like ChatGPT or Claude. This advancement signifies better results, promotes open access, and demonstrates a commitment to sustainability and cost-efficiency. It's a prime example of #ResponsibleAI in action! Read more about OpenScholar's success in scientific research: [VentureBeat Article](https://lnkd.in/eJd7h9Xe)

5

2 Comments