Daniel Olmedilla

New research tackles the fight against misinformation with AI-powered counter-narratives! A fascinating paper explores using Large Language Models (LLMs) to automatically evaluate the effectiveness of Counter Narratives (CNs) in combating online misinformation. The study reveals that LLMs, particularly chat-aligned models, excel at generating and evaluating CNs when fine-tuned with domain-specific data. This promising approach could revolutionize how we address misinformation and promote healthier online discourse. #misinformation #disinformation #cognitivesecurity #generativeAI #LLM #AI #techforgood https://lnkd.in/gHrVk9bs

11

We are not slowing down on Gen AI, come see why...and how it changes things, the nature of work. https://lnkd.in/gW3ettg3

2

Another demonstration of ScienceCast.org #AI enhancements of arXiv postings - free, simple interface https://lnkd.in/e_gXsF_9

3

1 Comment

Faster, more accurate, less data needed = BaseModel.AI

3

"How Meta trains large language models at scale" - Great article by Meta that provides a thorough overview of the infrastructure challenges and innovations involved in their efforts to train large language models at scale for generative AI. Some key highlights that stood out: 👉 They had to modify their Grand Teton platform that uses NVIDIA H100 GPUs on a very tight schedule, increasing the GPU TDP to 700W and moving to HBM3. This required mechanical and thermal design changes plus a validation cycle, all while staying within the constraints of their existing air-cooling infrastructure. Making significant hardware changes like this under time pressure and resource constraints is a really impressive feat. 👉 Meta decided to build two 24k GPU clusters simultaneously, one with RoCE and one with InfiniBand, to gain operational learnings and inform their future directions. Pursuing both paths in parallel at that scale is a significant undertaking, showing real commitment to thoroughly evaluating the tradeoffs. 👉 They mention observing various hardware failure modes during their scaling journey, such as GPUs "dropping off" the PCIe bus and uncorrectable memory errors. Monitoring, identifying "repeat offenders", and proactively mitigating these issues at scale required comprehensive telemetry and automation. 👉 The article notes that in the coming years Meta anticipate "handling even larger volumes of data, and dealing with longer distances and latencies." This suggests they can predict new networking and data management challenges as they continue to increase scale and potentially distribute training across larger geographical areas. #ai #generativeai #llms #enterpriseai #aitraining https://lnkd.in/eQsGmndf

5

1 Comment

*Mixture of Agents (MoA)* is an innovative way to boost Large Language Models’ performance by leveraging the strengths of multiple agents as a team. Layered Architecture: MoA arranges many LLMs in layered architecture, each layer composed of numerous individual agents(LLMs). Collaboration: These agents generate responses that are based on outputs from the previous layer’s agents. They repeatedly refine and enhance the last output. Diverse Insights: By integrating different capabilities and insights from various models, MoA achieves a more powerful combined model that is secure against diverse risks. Performance Boost: MoA outperforms standalone LLMs by a wide margin. For instance, it set a new benchmark with 65.1% on AlpacaEval 2.0 as against GPT-4 Omni at 57.5%, using only open-source models. In short, if we combine language modelers’ strengths then it results in increased efficiency and versatility. Here is a nice implementation of MOA by Together AI Github source: https://lnkd.in/dMCVRS3D #AI #LLM #LargeLanguageModels #MachineLearning

59

3 Comments

Good perspective on evolution of training costs of AI Models

26

3 Comments

Top 5 LLM Testing and Evaluation Tools: 🔭 Galileo: An end-to-end platform for GenAI evaluation, experimentation, and observability AgentOps.ai: Build reliable AI agents with monitoring, testing, and replay analytics. Toloka: A data-centric environment to support fast and scalable AI development with the help of human insight. Gantry: Powerful observability, analytics, and evaluation for your AI-powered products. RagaAI Inc: Fix AI issues in three easy steps - Detect, Diagnose and Fix. #AI

42

Our Chief Data Scientist Roberto Pieraccini shares an in-depth article on the top 3 things AI Scientists contribute to in today's world of AI democratization. These three things are: 1) Through years of training and specialization, scientists have deep phenomenological knowledge and can apply their knowledge it to build models in their respective areas such as speech, linguistics, reasoning, etc. 2) Machine learning (algorithms) - AI scientists are trained in algorithms (maths) and know how to train/infer to get closed/open form solutions, know how to generate/analyze loss functions, a knowledge that is applicable irrespective of the type of algorithms. 3) Deep understanding of empirical analysis and experimentation: As scientists are trained to form & test hypothesis and benchmark algorithms, this is the key requirement in industry when there are already 149 new foundation models released in 2023 (AI index report) alone. Being able to quickly create the right metrics and benchmark on internal datasets is key to success in applied AI industry. Hope you enjoy this article!

15

🚀 Transform AI Conversations: Join Our Exclusive vCon Use Cases Zoominar on May 14th! 🗓️ Attention trailblazers in the world of AI and communication! 🌟 Following the resounding success of our "Introduction to vCon" Zoominar, Thomas McCarthy-Howe and I are thrilled to invite you to Part II of our groundbreaking series: "vCon Use Cases: Revolutionizing AI Conversations". 🎉 As a pending Internet Engineering Task Force (IETF) RFC, virtualized conversations (vCons) are set to become a critical building block for AI task managers, revolutionizing the way we store, analyze, and leverage customer dialogues in AI applications. 📈 Join us on May 14th at 1 PM Eastern US for an interactive online Zoominar, where we'll explore the transformative use cases of vCons across various domains: 🌟 Customer Experience: Conversational CRM Listening to Customers Mass Customization 🌟 Efficiency: Compliance and Consent Generative AI 🌟 Observability and Transparency: Robot Food Enterprise Search Agent Governance 🌟 Business Transformation: New Revenue Business Improvement Throughout the Zoominar, you'll have ample opportunities to ask questions, share your insights, and engage in thought-provoking discussions with our expert speakers and fellow attendees. 💬 🔥 Limited spots available! Secure your place now by registering here: https://lnkd.in/ec5q5ErV 🔥 If you missed our previous Zoominar, "Introduction to vCon," don't worry! Simply send an email to jeff@pulver.com, and we'll provide you with exclusive access to the recording. 📩 Seize this opportunity to explore the transformative potential of vCons and stay at the forefront of AI and communication innovation. Let's unlock new possibilities together! 🔓 #vCon #ArtificialIntelligence #CX

22

Interesting article about humongous foundational LLMs potentially reaching the ceiling in terms of improving model quality. I believe that smaller, domain specific models are the hew frontier that might offer better ROI in terms of costs and model quality for a specific domain/use case. This should be especially true for enterprise use cases, when companies possess significant proprietary data corpuses. #artificialintelligence #largelanguagemodels #airevolution

52

8 Comments

Just announced by #OpenAI -- read more here: https://lnkd.in/gVsBPRBx Data Analysis is one place OpenAI already really shines thanks to the ADA feature -- can't wait to ee what this adds. #GenAI #AIforBusiness #DataAnalysis

2

How ethical is this really? Shutterstock just announced another deal with yet another company to “ethically” use its image & video data. But there are so many questions here and very little transparency… Maybe Jon Oringer can help? The big question on my mind is, Why would a stock photographer/videographer/contributor knowingly support the use of their creative work to build products that will literally replace the need for them to do these very same jobs? As a stock photographers, why would you allow the use of your images for fractions of a penny to enable giant tech companies to create a product that literally replaces your job forever? Why would any photographer support this? Shutterstock reportedly made over 104 MILLION dollars in dataset sales related to AI product development. (link in comments) How much of that made it back to their contributors who supplied all of this data? Is the amount of money contributors get remotely enough for them to make a living of residuals? On a more granular level, Did all these contributors explicitly opt-in for the use of their stock images to be used to train Gen-AI products that replace their job? And how much money did contributors actually get on the 104 MIL? 1%? 2%? 5%? 10%? (its 15% on VectorField Co-Operative!) Expanding on these questions a bit, 1.     Was this “ethical” deal with shutterstock & databricks only include data contributed post to the addition of the ability to opt out that was added in February 2023?   2.     Wouldn’t opt in be way more ethical then opt out as a rule of thumb? (VectorField Co-Operative is opt in only)   3.     Are there any mechanisms to transparently validate any of these B2B transactions? Or do all of the contributors just need to take Shutterstock's word for it? Are these mechanisms Cryptographically secure? Offer Zerotrust? (These features are baked into the design of VectorField Co-Operative)   4.     If Data was used prior to 02/23 were terms negotiated with each contributor for the % & cost of their data?     So is this ethical at all? Ethical for businesses & corporations? Maybe? Does it pass the legal threshold of fair use for contributions submitted before 02/23? The US legal system will decide. On a basic human, person to person level does this seem fair? Ethical? What do you think?      

8

5 Comments

AI is reshaping landscapes, altering industries and transforming our lives and work. With its rapid advancement, it's also causing disruption — for better or worse — in every field imaginable. While it promises efficiency and growth, AI also brings challenges and uncertainties that professionals and businesses must navigate. What can one do to pivot if AI is disrupting their industry? Authority Magazine turned to Reggie Townsend for the answers: http://2.sas.com/6041iPQUL. #TrustworthyAI #ArtificialIntelligence

33

Always a pleasure to read about strides in AI! Google DeepMind's latest research shows the impressive learning capabilities of LLMs, particularly in 'many-shot' In-Context Learning (ICL). This method simplifies task adaptation, enhancing AI accessibility. Reinforced and unsupervised ICL techniques show potential in reducing reliance on human-generated data and may help reduce the need for complicated RAG retrieval pipelines or laborious fine tuning of models. Right now, the spotlight is on scalability as a key factor in ensuring seamless integration and heightened efficiency. Looking forward to the impactful innovation generated by this research!

14

1 Comment

It’s finally here! I’m thrilled to introduce you to Llama 3, our latest generation of models that is setting a new standard for state-of-the art performance and efficiency for openly available LLMs. Key highlights: • 8B and 70B parameter openly available pre-trained and fine-tuned models • Trained on more than 15T tokens, 7x+ larger than Llama 2's dataset! • State-of-the-art performance across industry benchmarks • New capabilities, including enhanced reasoning and coding • 3x more efficient training than Llama 2 • New trust and safety tools with Llama Guard 2, Code Shield, and CyberSec Eval 2. • Training underway on a leading-edge 400B+ parameter model  We’re also releasing a new version of Meta AI powered by Llama 3. We’re making it available in 13 new countries, integrating enhanced search capabilities and adding new ways to experience it on the web at meta.ai. Our latest updates to image generation on Meta AI is state of the art, allowing you to create, animate and share stunning images with a simple text prompt in seconds. We think Meta AI is the world’s most intelligent AI assistant and we can't wait for you to try it and tell us what you think. Read more about it here: https://lnkd.in/eB5yzi8R

2,389

80 Comments

1. IMO, agent patterns are one of the most interesting ways to extend LLMs. memary is based on leveraging a graph database (neo4j) for long-term memory (to get around the limited memory inherent in even the largest context windows in LLMs) and augmenting it with other tools like web search and Computer Vision. https://lnkd.in/gfFcBqJx Looks interesting. 2. More about how consolidation effects the entire healthcare industry: https://lnkd.in/gXJKqNJB The Change hack/ransomware attack not only affected claims, service, refills, etc. but now CMS & NCQA are delaying the entire HEDIS reporting process (used for Medicare/Medicaid quality ratings) due to the impact. 3. NARA bans ChatGPT use agency-wide. https://lnkd.in/gW8T94JU Federal Government agencies see this as a risk, including their information leaking and being used to train models (which may then output the wrong information or show PII/PHI other proprietary information). 4. Oh great, now I can't even be angry anymore :) https://lnkd.in/gcrN9mjx

1

1 Comment