Joe Wyer

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

The key hurdle in deploying AI at scale is bridging the gap in trust, security, and compliance of LLM outputs. I'm excited to be presenting at the Databricks Data + AI Summit on how we at Dynamo AI accelerate this last mile for enterprise customers!

16

I wonder what the click rate is for those LinkedIn AI suggestions on the bottom of posts. Corollary: I wonder what percentage of those clicks are accidental and whether it is three digits.

3

5 Comments

What can your playlist tell you about the future of #GenAI? At a recent GenAI Meetup, Kenneth Longshaw dove into how Graph Neural Networks (GNNs) build connections and the power of context in data, for everything from playlists to traffic forecasting. 🎵 GNNs and Playlists: GNNs are powerful tools that analyze and interpret the complex relationships within data, similar to how song recommendations are personalized to you when creating a Spotify playlist. Songs aren’t isolated entities: they are part of a vast and dynamic network of artists, albums, genres, tempo, and more. By mapping out and understanding these connections, GNNs simplify the comprehension of complex data networks, making it easier to grasp how individual elements are interconnected within the broader system. This means more relevant recommendations based on your listening history and songs already in your playlists. 🔑 The Key Ingredient: Context is the cornerstone of meaningful data analysis. GNNs weave rich layers of context into data to revolutionize our problem-solving approaches. In healthcare, this means creating treatment plans with a nuanced understanding of each patient's unique medical history, lifestyle, and even genetic predispositions, leading to truly personalized care. In finance, GNNs transform market trend predictions by integrating a multitude of factors — economic indicators, geopolitical events, and consumer behavior patterns — to offer forecasts with unprecedented precision. 💡 Smarter Decisions: When contexts have insights, decision-makers across sectors are empowered to make smarter, more proactive decisions. By integrating context, we can anticipate supply chain disruptions before they unfold, tailor customer experiences by accurately predicting individual needs and preferences, and navigate the complexities of various industries with greater agility and foresight. This fosters innovation and efficiency and creates the foundation for strategic advantage. As we continue to explore and apply these insights to our own products at simplySOLV, we are so excited to unlock innovative solutions and drive meaningful progress in our increasingly data-driven world! Are you using GNNs in your work? How have you witnessed them impacting your industry?

7

Just found an excellent Towards Data Science article by Mariya Mansurova on Medium about causal analysis with controlled experiments and observational data: https://lnkd.in/e2fq76hZ A few highlights: 1️⃣ The treatment effect size and confidence determined with classical (frequentist) A/B/n tests are equal to the respective coefficient value and significance in a linear regression 2️⃣ Adding more independent variables to the linear regression increases confidence in the treatment (assuming more of the observed residual variance can be explained by those variables) 3️⃣ We can reduce bias from confounding variables (that impact the treatment and our target metric) by also adding them as features to the linear regression 4️⃣ When applying the above to observational data, we can come to the same causal conclusions if we include all confounders (of course this won't be entirely feasible and as much art as science in a lot of cases) As a bonus, Mariya shares her Python notebook code to reproduce the described analysis incl. many beautiful graphs. #experimentation #causalinference #linearregression #statistics #machinelearning

18

1 Comment

As I keep on saying : Prompt Engineering has never been a thing. It's wrong for 2 reasons : it makes you think it's deterministic (one great prompt= one great output, whereas LLMs are inherently stochastic machines, exactly like your Most Significant Other), and; it tricks you into thinking you need to be an engineer to get it right. Guess what: IT engineers have the most difficult time of their life with LLMs! And now we have a third, mike-dropping reason: smart bots can now engineer your most perfect prompt for you: just ask away, and add on!

9

One of the biggest impediments to AI adoption are hallucinations. Hallucinations are essentially false answers generated by AI models. For example, last year, a New York lawyer faced reprimand for using generative ai to help produce a legal brief. However, the model output cited court cases that never even occurred! The below link is a fantastic post on how Palantir's dynamic ontology (the digital twin of your enterprise) helps reduce hallucinations. Reducing the risk of hallucinations is critical for using Generative AI effectively and responsibly. Palantir AIP enables you to ground LLMs in data from the ontology, letting you seamlessly provide trusted data to the model and help it produce more accurate results. This is a very approachable article for non-technical readers interested in AI adoption too. Palantir Technologies

18

Prompting for scale will soon be obsolete... i think...i hope. Prompting is an error-prone, discovery-based process, unsuitable for scalable AI. Many folks rely on black-box LLMs behind API walls (e.g., Claude, GPT-4), lacking direct control over LLM model weights. Programmatically custom-pruned LLMs will be the future of controllable, production AI, not prompting. hint--> #steeringvectors

16

6 Comments

I'm thrilled to share an exciting update from our team! We've expanded the Amazon Titan family with the addition of Titan Text Premier, a large language model optimized for performance and cost-efficiency in retrieval-augmented generation (RAG) and agent-based applications. Titan Text Premier represents a significant milestone, but it's just the beginning! Stay tuned for more updates from AWS and AGI #InnovateWithAWS! #GenAI #AI

136

I just published Vol. 57 of "Top Information Retrieval Papers of the Week" on Substack. My Substack newsletter features the 7-10 most notable research papers on information retrieval (including recommender systems, search & ranking, etc.) from each week, with a brief summary, and links to the paper/codebase. This week’s newsletter highlights the following research work: 📚 Training IR Models with Minimal Labels via Prompt-Optimized Query Synthesis, from Xian et al. 📚 Probing the Interplay of Parametric and Non-Parametric Memory in RAG Language Models, from UMass Amherst 📚 Prompt Engineering's Impact on Zero-Shot LLM Ranking Effectiveness, from The University of Queensland 📚 A Systematic Analysis of Retrieval-Augmented Code Generation, from Wang et al. 📚 Benchmarking Long-Context Language Models for Real-World Tasks, from Google DeepMind 📚 A Tournament-Inspired Approach for Zero-Shot Document Ranking with Large Language Models, from Renmin University 📚 Multifaceted Criteria for Determining Retrieval Need in Augmented Language Generation, from Fudan University 📚 An LLM-based reranking Approach for Balancing Accuracy, Diversity and Fairness, from CityU 📚 LLM-Guided Reinforcement Learning for Novel Item Recommendation in Large-Scale Systems, from Microsoft 📚 Automating Prompt Engineering for Zero-Shot Passage Relevance Ranking, from Jin et al. #InformationRetrieval #ResearchPapers #CuratedContent #Newsletter  #substack

42

1 Comment

I just published Vol. 54 of "Top Information Retrieval Papers of the Week" on Substack. My Substack newsletter features the 7-10 most notable research papers on information retrieval (including recommender systems, search & ranking, etc.) from each week, with a brief summary, and links to the paper/codebase. This week’s newsletter highlights the following research work: 📚 Graph Retrieval-Augmented Generation for Context-Aware Reasoning, from Emory University 📚 Enabling Fine-Grained Ranking at Varying Granularities via Multi-Vector Embeddings, from Reddy et al. 📚 Learning to Ask Clarification Questions for Ambiguous Retrieval Queries, from UC Berkeley 📚 Principled and Practical Multi-Vector Retrieval with Approximation Guarantees, from Google Research 📚 Towards User-Centric Recommendations via LLM-Based Simulation and Tool Learning, from USTC 📚 Building Pseudo-Graph Databases from Raw Texts for Retrieval-Augmented Language Models, from Renmin University 📚 Retrieval-Augmented Early Exiting for Efficient Large Language Model Inference, from Huang et al. 📚 A State-of-the-Art Generalist Embedding Model, from NVIDIA 📚 Enhancing Multilingual Language Model Training via Repeat Ranking, from Devine et al. 📚 A Real-World Evaluation of HNSW for Vector Retrieval, from Marqo #InformationRetrieval #ResearchPapers #CuratedContent #Newsletter  #substack

46

2 Comments

AI is not slowing down, and the scaling laws continue to hold. Claude 3.5 Sonnet is now probably the most powerful model available, but Sonnet isn't even Anthropic's biggest model—Opus is. A particularly impressive number here is 59.4% on GPQA diamond questions. GPQA is a benchmark of PhD-level questions for which you can't Google the answers, and they don't make the data available online for people to easily train on, so the contamination risk is lower. In November 2023, GPT-4 scored 38.8% on GPQA with few-shot prompting. Claude 3.5 Sonnet gets 59.4% with 0-shot prompting (while GPT-4 has improved, with GPT-4o, to 53.6%).

2

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

I loved reading this expertly told article; kudos to Yasemin Saplakoglu! It is the story of what might, in hindsight, be remembered as the most impactful innovation driven by machine learning of our time -- and yes, I have heard of that ChatGPT thing ;) (It is an interesting and funny aside that the Alphafold 2 paper talks about #machinelearning, but has zero mentions of #AI.. 😅)

15

Check out our latest post on Clappy.ai, "Leverage Layout-Aware RAG for More Accurate Retrieval of Knowledge from Documents." Discover how to harness the power of Retrieval-Augmented Generation (RAG) for documents in PDF, Doc, and Markdown formats. Dive into the specifics of implementation and explore our reference repository to get started with advanced RAG for your knowledge base. Don't miss out on this essential read! Continue reading here for all the details: The Power of Hierarchical Indexing with RAG. (https://lnkd.in/gbiVHfcH)

2

Just wrapped a webinar with PEI about data unification alongside Pete Keenan. It's a good thing he knows what he's talking about - I mostly just smiled and nodded and said "totally" every once in a while after he talked. It's clear from a survey we fielded during the webinar that even the biggest venture and private equity managers haven't solved the golden record problem. As more data piles into the private market alongside more demanding institutional LPs, the problem is only going to get bigger, and you can't just slap an LLM on it. Comment or DM if you're in VC or PE finance and want to learn more - I know some smart people for you to talk to.

21

3 Comments

Tips a hat to SourseAI - “A narrow ICP definition and deep customer empathy” Tanya Hyams-Young & Paul Jefferies are seeing the fly wheel effect of vertical SaaS. Note to Aussie startups that take this route - think hard about the TAM, here and abroad. You’ll probably conclude you’re going to need to move to “the rest of the world” pretty quickly. See you in a week EMEA! #verticalsaas #ai #decisionintelligence #telecoms #mvno

11

5 Comments

I just published Vol. 52 of "Top Information Retrieval Papers of the Week" on Substack. My Substack newsletter features the 7-10 most notable research papers on information retrieval (including recommender systems, search & ranking, etc.) from each week, with a brief summary, and links to the paper/codebase. This week’s newsletter highlights the following research work: 📚 A Comprehensive Survey of Retrieval-Augmented Large Language Models: Architectures, Training, and Applications, from PolyU 📚 An Analysis Framework for Benchmarking Retrieval-Augmented Generation Systems, from Tu et al. 📚 Distilling Cross-Encoders from Large Language Models for Efficient Passage Re-ranking, from Schlatt et al. 📚 A Comprehensive Benchmark for Assessing Retrieval-Augmented Language Models in Biomedicine, from UMTC 📚 Improving Factual Accuracy of Large Language Models for Healthcare using Targeted Knowledge Graph Retrieval, from Huawei 📚 Identifying Shopping Product Questions in Voice Assistants via Graph Attention Networks, from Amazon 📚 A Learnable Tokenizer for Generative Recommendations, from NUS 📚 An Enhanced Retrieval-Augmented Generation Framework for Accurate, Efficient, and Personalized Language Models, from UTS 📚 Constructing Synthetic Test Collections Using Large Language Models for IR Evaluation, from Rahmani et al. 📚 Enhancing Sequential Recommendation with Soft Contrastive Learning, from Renmin University #InformationRetrieval #ResearchPapers #CuratedContent #Newsletter  #substack

70

2 Comments