Arize AI’s Post

Tomorrow: Pick up some practical ways to improve your genAI development process--with embeddings tracing and experiments visualization. 🧪 Sally-Ann DeLucia will be demonstrating how these tools are designed to help you: 📊 Track experiments more efficiently to achieve better results 🧬 Diagnose issues using detailed embedding traces ⚡Accelerate iteration with a clear, visual understanding of your data Our aim here is to simplify the complexities of high-dimensional embeddings, providing you with deeper insights into model performance and optimization. You’ll also learn how to access a comprehensive breakdown of labels for your experiments. Sign Up: https://lnkd.in/gvAvTNAk

🎯 LeetCode Daily Challenge - Day 24 Dec 2024 🚀 Today's challenge: "3203. Find Minimum Diameter After Merging Two Trees" Problem: There exist two undirected trees with n and m nodes, numbered from 0 to n - 1 and from 0 to m - 1, respectively. You are given two 2D integer arrays edges1 and edges2 of lengths n - 1 and m - 1, respectively, where edges1[i] = [ai, bi] indicates that there is an edge between nodes ai and bi in the first tree and edges2[i] = [ui, vi] indicates that there is an edge between nodes ui and vi in the second tree. You must connect one node from the first tree with another node from the second tree with an edge. Return the minimum possible diameter of the resulting tree. The diameter of a tree is the length of the longest path between any two nodes in the tree.   Example 1: Input: edges1 = [[0,1],[0,2],[0,3]], edges2 = [[0,1]] Output: 3 Explanation: We can obtain a tree of diameter 3 by connecting node 0 from the first tree with any node from the second tree. Approach Find the farthest node starting from node 0(can be any other node). Now from that node find the other farthest node which will give us the diameter of the tree(d1 and d2). Now the answer could be max(d1, d2, (d1+1)/2 + (d2+1)/2 + 1) Complexity Time complexity: O(n) Space complexity: O(n)

🚀 Exciting news! I’m thrilled to be teaming up with Frontend Queens this October 15th to dive into the world of Retrieval-Augmented Generation (RAG) models! 🎉 I'll be sharing insights on how to implement RAG models and exploring how they can empower your projects. Don't miss out on this opportunity to expand your knowledge and connect with an amazing community. See you there! 🤖👩🏻💻💡 #FrontendQueens #RAGmodels #AI #MachineLearning #TechTalk

🦙🎉 We've expanded our Prompt Engineering with Llama 2 course to include the new Llama 3 model by AI at Meta! Interact with these models through API calls and: 🌟 Learn prompt engineering best practices for Llama 2 & 3 🧪 Experiment with few-shot and chain-of-thought prompting 💻 Code alongside Code Llama and learn to use Llama Guard to check prompts and responses Get started and let us know what you think of this update! Sign up here: https://hubs.la/Q02vH6kr0

Building Agentic RAG - Llama3 edition 🦙💫 We’re excited to collaborate with some folks at AI at Meta to release an entire set of cookbooks on agentic RAG: from routing and tool use to layering on an agent reasoning loop to building a multi-document agent. This is directly inspired by our DeepLearning.AI course on Agentic RAG. Here we’re able to show that you’re also able to use purely local models (llama 3) for complex agent workflows over your data. There are a variety of integrations to help run Llama 3, from Groq to Fireworks AI This includes the following: L1: Router Engine L2: Tool Calling L3: Building an Agent Reasoning Loop L4: Building a Multi-Document Agent Github resources here: https://lnkd.in/e7_2z7ka

In our coming episode on the 29th of November, meet Marco Basile to get an insight on how to build an AI company fast & get 480x acceleration! #𝐀𝐈 #𝐈𝐧𝐧𝐨𝐯𝐚𝐭𝐢𝐨𝐧 #𝐁𝐮𝐬𝐢𝐧𝐞𝐬𝐬𝐓𝐫𝐚𝐧𝐬𝐟𝐨𝐫𝐦𝐚𝐭𝐢𝐨𝐧 #𝐒𝐭𝐚𝐫𝐭𝐮𝐩𝐬 #𝐄𝐧𝐭𝐞𝐫𝐩𝐫𝐢𝐬𝐞𝐬 #𝐖𝐞𝐛𝐢𝐧𝐚𝐫

Want to elevate your RAG system's performance? BeyondLLM is your go-to solution!  Say goodbye to uncertainty about whether your LangChain RAG system is providing top-notch responses. BeyondLLM is here to empower your evaluation process! Our tool offers immediate insights on three essential metrics: ✅ Context Relevancy: Ensure your retrieval aligns perfectly with user needs. ✅ Answer Relevancy: Confirm that responses genuinely address user queries. ✅ Groundedness: Keep your AI anchored to factual information. Ready to change your RAG evaluation process? Explore our comprehensive cookbook on GitHub and dive deeper with our insightful blog post! GitHub Cookbook : https://lnkd.in/g4f2_Jmg Release Blog: https://lnkd.in/gWQ5wRtW

Want to see how I compressed 6-week development cycles into 48-hour sprints? Join me with Brain Station 23 and FARZANA AFRIN TISHA to discover: - How I turned $32k development work into $3 daily investments - The exact patterns that led to 480x acceleration - Practical steps you can implement immediately No theory. No enterprise-scale complexity. Just real patterns that work for businesses ready to move fast. I'll share my journey from AI skeptic to practical implementation in 16 weeks, alongside AI Strategist Hande Ocak Basev. November 29th 4PM-5PM CET / 11 AM ET Link in the comments 👇🏼

Models are getting more transparent AND smaller. This is a good thing. Transparency in AI research is crucial for fostering innovation and collaboration. However, many open-source language models only release the model weights and architecture, leaving the training data and process as a black box. ChuXin is a new 1.6B parameter language model that lays everything out in the open—from the training data to the evaluation code. Through lightweight continual pretraining, the model's context length has been extended to a1M tokens, enabling strong performance on challenging long-form retrieval tasks. I'm glad to see more and more models in the 1.5-4B parameter range and the more we know about them, the better. ↓ Liked this post? Follow the link under my name and never miss a paper highlight again 💡

🚨 We still have a few spots left! Join us tomorrow (Thursday, July 18th) for a live webinar + Q&A on how to build production-ready compound AI systems with Chains! ⛓️ 🚀 ✅ Save your spot: https://lnkd.in/eRHYsJXP Don't miss our Co-Founder and CTO Amir Haghighat along with Marius Killinger, one of the engineers behind Chains, while they break down how you can build performant, reliable, and efficient multi-model ML inference workflows. Chains enables users to: ⛓️ Assemble distinct computational steps into cohesive workflows ⛓️ Combine business logic with ML models ⛓️ Customize hardware and scaling for each component in an inference pipeline RSVP now and secure your spot!