Kobie Crawford

Interested in integrating SageMaker Processing jobs with PrestoDB for efficient batch transforms and frequent model retraining 🚀 ? Releasing a new AWS blog with Amit Arora, Johnny Chivers and Antara Raisa containing an end to end an open source solution that uses Amazon SageMaker Pipelines and PrestoDB. This blog dives into an end-to-end workflow that seamlessly connects SageMaker Processing jobs to your PrestoDB data sources, enabling optimized batch inference and scheduled model retraining. From data fetching and model training to batch transforms and real-time deployment, you can now use this approach for a variety of use cases all through minimal/no-code changes! Do check it out and let me know if you have any questions/feedback. Blog link: https://lnkd.in/evsgxt3P Open-Source GitHub Repository: https://lnkd.in/e8nsWvGN #AWS #Machinelearning #AmazonSageMaker #MLOps

13

1 Comment

In case you missed it, we have now released the second issue of our Confluent Chronicles – a fun and entertaining take on the fundamentals of stream processing, Apache Flink, and Confluent! We had such great feedback on the first issue that we couldn’t just stop there – you can check out and download the latest edition here: https://lnkd.in/gEFxA7af Would love to hear what you think of it!   #ApacheFlink #ApacheKafka #Confluent

39

1 Comment

𝐀𝐠𝐞𝐧𝐭𝐢𝐜 𝐖𝐨𝐫𝐤𝐟𝐥𝐨𝐰: 𝐏𝐢𝐨𝐧𝐞𝐞𝐫𝐢𝐧𝐠 𝐭𝐡𝐞 𝐅𝐮𝐭𝐮𝐫𝐞 Recently I happened to see this nice video ( 𝘩𝘵𝘵𝘱𝘴://𝘸𝘸𝘸.𝘺𝘰𝘶𝘵𝘶𝘣𝘦.𝘤𝘰𝘮/𝘸𝘢𝘵𝘤𝘩?𝘷=𝘴𝘢𝘭78𝘈𝘊𝘵𝘎𝘛𝘤 ) by Andrew Ng in which he talks about Agentic workflows. In Language Model (LLM) development, we provide a zero-shot prompt and receive an answer. We may review it and add more prompts to get it to a desired shape. This is exactly what the agentic workflow does. Basically, 𝘢 𝒃𝒖𝒏𝒄𝒉 𝒐𝒇 𝑨𝑰 𝒂𝒈𝒆𝒏𝒕𝒔 𝒘𝒐𝒓𝒌 𝒕𝒐𝒈𝒆𝒕𝒉𝒆𝒓 𝒊𝒏 𝒂𝒏 𝒊𝒕𝒆𝒓𝒂𝒕𝒊𝒗𝒆 𝒘𝒂𝒚 𝒕𝒐 𝒑𝒓𝒐𝒅𝒖𝒄𝒆 𝒂 𝒅𝒆𝒔𝒊𝒓𝒆𝒅 𝒐𝒖𝒕𝒑𝒖𝒕. Moreover, he even claims the output produced by agentic workflow is much better than the output produced single LLM which might sound obvious but what surprised me was, as per the “𝑯𝒖𝒎𝒂𝒏𝑬𝒗𝒂𝒍” coding benchmark, agents-based approach beating plain LLM by quite a margin (see the picture below). The plain 𝐆𝐏𝐓 3.5 was at 48% but 𝑳𝑫𝑩+𝑹𝒆𝒇𝒍𝒆𝒄𝒕𝒊𝒐𝒏 (𝘈 𝘓𝘢𝘳𝘨𝘦 𝘓𝘢𝘯𝘨𝘶𝘢𝘨𝘦 𝘔𝘰𝘥𝘦𝘭 𝘋𝘦𝘣𝘶𝘨𝘨𝘦𝘳 𝘷𝘪𝘢 𝘝𝘦𝘳𝘪𝘧𝘺𝘪𝘯𝘨 𝘙𝘶𝘯𝘵𝘪𝘮𝘦 𝘌𝘹𝘦𝘤𝘶𝘵𝘪𝘰𝘯 𝘚𝘵𝘦𝘱-𝘣𝘺-𝘴𝘵𝘦𝘱) fared at impressive 95.1% with the same GPT 3.5. The plain 𝐆𝐏𝐓 4 was at 67% but 𝑨𝒈𝒆𝒏𝒕𝑪𝒐𝒅𝒆𝒓 (𝘔𝘶𝘭𝘵𝘪-𝘈𝘨𝘦𝘯𝘵-𝘣𝘢𝘴𝘦𝘥 𝘊𝘰𝘥𝘦 𝘎𝘦𝘯𝘦𝘳𝘢𝘵𝘪𝘰𝘯 𝘸𝘪𝘵𝘩 𝘐𝘵𝘦𝘳𝘢𝘵𝘪𝘷𝘦 𝘛𝘦𝘴𝘵𝘪𝘯𝘨 𝘢𝘯𝘥 𝘖𝘱𝘵𝘪𝘮𝘪𝘴𝘢𝘵𝘪𝘰𝘯) fared at whooping 96.3% with the same GPT 4.  You can get more information about the evaluation and the models at 𝘩𝘵𝘵𝘱𝘴://𝘱𝘢𝘱𝘦𝘳𝘴𝘸𝘪𝘵𝘩𝘤𝘰𝘥𝘦.𝘤𝘰𝘮/𝘴𝘰𝘵𝘢/𝘤𝘰𝘥𝘦-𝘨𝘦𝘯𝘦𝘳𝘢𝘵𝘪𝘰𝘯-𝘰𝘯-𝘩𝘶𝘮𝘢𝘯𝘦𝘷𝘢𝘭. He also mentioned the following are the 4 design patterns that are emerging in the agentic workflow. 1.    Reflection 2.    Tool use 3.    Planning 4.    Multi-agent collaboration If you are interested in learning more, instead of me explaining each of these patterns, I would encourage watching the video :) #ai #agent #agenticworkflow #gpt

22

*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

57

3 Comments

Meet Imply's Team at #AWSSummit Berlin on 15 + 16 May 2024 at STATION Berlin! 💻 🇩🇪 Learn more about the managed version of Apache Druid —a time-series data storage and analysis engine with subsecond queries across data at scale. 🐨 🧙🏼‍♂️ #ApacheKafka #ApacheDruid #eventdrivenarchitecture #streamingdata #realtimeanalytics #AWSIoTCore #cloudengineering Raj Prabagaran, MBAHellmar BeckerTom Mack

12

My latest blog post details how to set up your local Windows computer to fine-tune Llama3 and similar models. #generativeai #llm #windows #artificialintelligence

5

I'm speaking at SREday San Francisco 2024! Join me on June 22nd for a day full of Site Reliability, DevOps and Cloud! Keynote: Achieving DORA outcomes by building reliability capabilities into your platform Infrastructure breaks, but systems can persist! We really want systems that withstand unavoidable failures. Abstractly, we understand that. In this talk, Steve presents concepts like spanning failure domains and using generic mitigations, as well as introducing a lab environment where teams can experiment with these directly and a community where we can discuss it all, memes and all. Grab a ticket with 20% off with my code STEVE20. https://lnkd.in/gZAdxes7 #SREday #SiteReliability #DevOps #Cloud

157

5 Comments

Need accuracy & speed in your GenAI stack? Superlinked + Redis = smarter and faster apps. Tyler Hutcherson Ben Gutkovich Daniel Svonava Ash Vijayakanthan Jim Allen Wallace Albin Bergh Yael Fainsilber Adrienne Cohen #RAG #genai #vectordb

20

3 Comments

Truth. This blog articulates the shortcomings of most tiered storage solutions in delivering predictable performance at scale. Without some rebuild that harnesses cloud primitives like AWS S3 and GCP GCS, mere incremental adjustments to existing products won't adequately serve users. From a product manager's perspective, sometimes incremental adjustments isn't a path to sustainable product adoption and success.

9

Learn how to power real-time data streaming applications with Couchbase Capella and Amazon MSK in our latest blog post

3

Interested to know how Coroot can improve your observability ? Here is the list of Coroot's most amazing features! https://lnkd.in/emDrcgX9 #coroot #opensource #observability

4

With the announcement of S3-native-streams (Freight clusters), here is a commentary on Confluent strategy regarding object storage, streaming and an open data architecture. https://lnkd.in/dTg6DKVy

48

6 Comments

This is a great first step in understanding how scaling dynamics extend beyond language to other types of sequential data. Time is a universal axis, understanding how models learn optimally over data collected over time is a crucial step in the broader AI research agenda.

26

Creating an EKS cluster can be done with (relatively) simple AWS CLI commands. There are a lot of commands to run, and while they can be wrapped in a script and parameterized - it's still not a very good solution. https://hubs.la/Q02s3Gg_0 #eks #kubernetes #k8s #aws #sre

4

"The Heart of the #datamesh Beats Real-Time with #apachekafka" -> Video recording from my tech talk at the Wix Kafka meetup in Tel Aviv. If there were a buzzword of the hour, it would undoubtedly be “data mesh” and related debates around #dataproduct! This new architectural paradigm unlocks analytic and transactional data at scale and enables rapid access to an ever-growing number of distributed domain datasets for various usage scenarios. The data mesh addresses the most common weaknesses of the traditional centralized data lake or data platform architecture. And the heart of a decentralized data mesh infrastructure must be real-time, reliable, and scalable to provide access to data products in good data quality, no matter if the application is real-time, batch or a request-response API. Learn how the de facto standard for #datastreaming, #Kafka, plays a crucial role in building a data mesh and data products. https://lnkd.in/dMT5uqBw

74

🤯 This is a major news for the data space to be excited about! Branded as "continuous queries", #GCP is bringing stream processing directly into #BigQuery ! This makes it incredibly easy to build pipelines that drive both continuous insights, reactive operations, and autonomous workflows - and will no doubt push other vendors to align with the #StreamingDataWarehouse benchmark. 💡 But keep in mind that your outputs are only as continuous as your input. That means that in order to leverage the power of continuous queries, you first need the power of Continuous Ingestion. 🚀 Fortunately, off-the-shelf solutions like Popsink let you stream data directly into Continuous Queries in BigQuery - whether from #Salesforce, #Postgres, an #AS400, or anything in-between. Give a shot to discover the power of end-to-end continuous processing. #ContinuousQueries #StreamProcessing #DataAnalytics #DataPipelines

3

Check out this full-code tutorial on how to fine-tune and evaluate open source #LLMs from Hugging Face using Amazon SageMaker! 🤝 Did we mention Comet is integrated with SageMaker and Hugging Face?

10

More details about the mighty phi-3 family of SLMs. Such a key development as GenAI Dev becomes more about always optimizing and orchestrating with LLMs routing to LLMs and SLMs continuously to balance complexity, latency, cost and more! https://lnkd.in/gGQYqNKt

9