Patrice Bourgougnon’s Post

Cohere Compressed embeddings can save you 32x on your vector DB cost, while keeping 100% search quality. Now also available on AWS Bedrock.

🌟 Cohere Embed v3, supporting int8 and binary embeddings - outperforms competitors in 18 languages (nDCG@10) while slashing memory costs by 100x—from $130k to $1,300 annually for 250 million embeddings 💫 Amazon Web Services (AWS) #AI #SemanticSearch #EmbeddingTechnology #MachineLearning

Amazon Bedrock now supports compressed embeddings from Cohere Embed. ☁️ https://go.aws/4eBniXm Cohere Embed is a leading text embedding model most frequently used to power RAG & semantic search systems. Compressed embeddings (int8 and binary) enable developers and businesses to build more efficient #generativeAI applications without compromising on performance. #AWS

Great Amazon Web Services (AWS) #reinvent2023 #recap today ! A quick overview of what I anticipate using most frequently in the future. 𝐆𝐞𝐧𝐞𝐀𝐈: - Amazon Q: Generative AI assistant designed for work that can be tailored to customer's business, data code and operations. - Amazon Bedrock: build and scale generative AI applications with foundation models. ° Bedrock’s Model Evaluation feature: helps you compare different models and versions with your own custom dataset: https://lnkd.in/euiU4jTv ° Amazon Bedrock Playground: provide you a console environment to experiment with running inference on different models and with different configurations. ° Knowledge base (coming soon): you can give FMs and agents contextual information from your company’s private data sources for Retrieval Augmented Generation (RAG) to deliver more relevant, accurate, and customized responses. - Amazon Titan (High-performing foundation models from Amazon) : Image Generator, Text Lite, Text Express. 𝐎𝐛𝐬𝐞𝐫𝐯𝐚𝐛𝐢𝐥𝐢𝐭𝐲:  - Amazon CloudWatch - Application Signals:  automatically instrument your applications on AWS so that you can monitor current application health and track long-term application performance against your business objectives. - My Application:  is an extension of Console Home that helps you manage and monitor the cost, health, security posture, and performance of your applications on AWS - https://lnkd.in/eV4mA-Fu. 𝐃𝐚𝐭𝐚 𝐐𝐮𝐚𝐥𝐢𝐭𝐲: - AWS Glue Data Quality Anomaly Detection & Dynamic Rules. - Zero-ETL Integrations: ° Integration with Amazon Redshift. ° Integration with Amazon OpenSearch. 𝐀𝐖𝐒 𝐆𝐥𝐮𝐞 𝐃𝐚𝐭𝐚 𝐂𝐚𝐭𝐚𝐥𝐨𝐠 Supports Multi-Engine Views: create views across databases without concern about engine and with simpler security. 𝐒𝟑 𝐄𝐱𝐩𝐫𝐞𝐬𝐬 𝐎𝐧𝐞 𝐙𝐨𝐧𝐞: a high-performance, single-zone Amazon S3 storage class that is purpose-built to deliver consistent, single-digit millisecond data access for your most latency-sensitive applications. 𝐑𝐞𝐝𝐬𝐡𝐢𝐟𝐭: - Support for Apache Iceberg - Serverless AI-Scaling You can find all news on recap-all.pdf. Thank you to all speakers: Olivier Leplus, Christian Kamwangala, Adrien Lebret, Guillaume Delacour, Florian Blond ☁️, khalil MOUAKHER, Martin Coulon & Nasma BRAHMI

I'm excited to share that I recently completed the 'Serverless LLM Apps with Amazon Bedrock' course, expanding my knowledge in the cutting-edge field of Generative AI. If you haven't done it yet you can enroll below. This course provided great hands-on experience in building serverless applications powered by #Amazon_Bedrock and Amazon Web Services (AWS) Serverless Solutions. #genai #aws Great course Mike Chambers DeepLearning.AI! https://lnkd.in/d4jKbpeF

I was initially thrilled to hear about the new features announced at AWS re:Invent, particularly regarding Amazon Q and expanded LLM support in Bedrock. However, my experience over the past two days has been disappointing. Here are some specifics: Amazon Q with Native Retriever and S3: The synchronisation feature that crawls and indexes data is disappointingly slow. For instance, crawling a mere 5 KB text file took an unexpectedly long time, and indexing continued for over two hours before I stopped it. Kendra Retriever for Indexing a Confluence Page: Here, the crawling of a single Confluence page using a regex filter was somewhat faster, completing in about 20 minutes. However, the indexing is still ongoing, and I plan to let it finish to see the total duration. Interestingly, despite the indexing process being incomplete, I was able to query and retrieve results from the indexed content via the Kendra console. Issues with Kendra Retriever on Amazon Q: I also attempted to use the Kendra retriever with Amazon Q, but it didn’t work. This might be due to incomplete indexing, or perhaps there's something I'm missing in the setup. For context, six months ago, I developed a similar interface and backend system. It was designed for uploading various file types (including PDFs requiring extensive OCR, text files, docs, etc.), scanning Confluence spaces and pages, and crawling web pages. System that I built using open source libraries like Langchain, Unstrcutured.io in a serverless fashion, indexes content in Qdrant vector DB. This setup powers our internal chatbot using #RAG, assisting our sales and support teams in quickly finding answers. The efficiency and speed of this in house built system stand in stark contrast to my current experience with Amazon Q and its associated tools. I think AWS can make it more simpler for anyone to use, build and release their own custom chatbots with their own specific business domain knowledge. Will keep myself tuned. #AWS #reinvent2023 #qdrant #RAG #langchain #genai

Derive new insights from your data with machine learning (ML) on AWS. Learn how LMA can get you started.

🚀 Exciting news! Check out this insightful blog post on federated learning on AWS using FedML, Amazon EKS, and Amazon SageMaker. This post explains the challenges and solutions for implementing machine learning in regulated industries and the use of federated learning to maintain data privacy. It also dives into a detailed use case of heart disease data and features the FedML framework, FedML Octopus, and FedML MLOps. Discover more about the solution architecture and how Amazon EKS and SageMaker are leveraged in this comprehensive approach. Read the full blog post here: [Federated Learning on AWS using FedML Amazon EKS and Amazon SageMaker](https://ift.tt/v5RoTlU). Let's continue advancing knowledge and innovation in the world of machine learning! #AWS #MachineLearning #FedML #AmazonEKS #AmazonSageMaker

Hey everyone, Today, I want to dive into something that's been on my mind: How can we truly scale Machine Learning (ML) with the muscle of Google Cloud Platform (GCP)? Scaling Architecture: If you're like me and think that deploying ML models should be as smooth as butter, check out GCP’s Kubernetes Engine. It takes the headache out of scaling, letting you focus on what you love—solving problems with data. High-Frequency Trading: Ever thought about the math that goes behind high-speed trading? Algorithms like t-SNE and PCA are game-changers, but they require serious computational power. GCP’s Compute Engine makes it manageable by putting GPUs to work, enabling real-time decisions. AutoML and Reinforcement Learning: Implementing advanced techniques like Reinforcement Learning can be daunting. Google’s AutoML offers pre-optimized algorithms, meaning we can deploy models like PPO or DDPG faster than ever. Security First: Let’s be real—none of this matters if it's not secure. With GCP’s IAM and VPC, you can sleep better knowing that your models are locked down tight. Big Data, Big Impact: Google Cloud's BigQuery ML allows you to run models directly where your data lives. Less data movement, more action. #MachineLearning #GoogleCloud #DataScience #HighFrequencyTrading #ReinforcementLearning #Security #BigQuery

Derive new insights from your data with machine learning (ML) on AWS. Learn how LMA can help you get started.