R to @NeuralNeron: great post tho!

R to @shapedai: "RAG to work well you need to do some careful training along with your model" Imo the completion model do not need a complete retraining at all, if your retrieval is accurate + your prompting game is on point. Finetuning embedding model, reranking then completion/instruct

R to @shapedai: #RetrievalAugmentedGeneration #RAG #DataScience #LLM #MachineLearning #Data #MultimodalData #RelevanceSearch #Recsys #RecommendationSystems #ai #SemanticSearch

R to @shapedai: Curious to hear if you’ve played around with RAG or LLMs for recsys and what your learnings have been.

We’ve been testing Retrieval-Augmented Generation (RAG) to see if it improves LLM performance for recommendation systems! https://www.shaped.ai/blog/rag-for-recsys-a-magic-formula

R to @nickpiscotty: Formidable Founders (FF029) Tullie Murrell (@tulliemurrell), CEO @shapedai Real-time AI for personalization. Backed by @tribecap @arjunsethi @SusaVentures @GFC_SEA @roguecapital @golda @Liquid2V @uncommon_cap @tikhon @jamiequint

R to @shapedai: #AI #Personalization #recsys #llm #machinelearning

R to @shapedai: shaped.ai/blog/personalizati…

Which marketplace has the best personalization? Top platforms such as Etsy, eBay, Amazon, and Alibaba use AI to drive many personalization features that increase conversion and enhance users experience. Read more about it in our latest blog post linked in the next tweet 👇!

R to @shapedai: Take a look at the write up here: https://www.shaped.ai/blog/exploring-benefits-of-llms-in-recsys And here's the research paper it's based off: https://arxiv.org/pdf/2306.05817v2.pdf

R to @shapedai: It also highlights the challenges of integrating these models in a production setting (e.g. inference latency, bias, long-text input, etc...).

How can large language models (LLMs) be used to help your users find what they're looking for? We wrote a post that looks at the benefits of using Large Language Models for content/product recommendations.

R to @loriberenberg: This was a great time (as usual) with the @NYAICommunity!

R to @loriberenberg: .@AliceAlbrecht of @recollect_ai and @tulliemurrell of @shapedai talked about the hype cycle and generative ai. there’s untapped potential in LLMs for enterprise applications… and they won’t be chat! think search, fraud detection, security.

R to @shapedai: #SyntheticData #AI #MachineLearning #DataScience #Privacy #FutureOfAI #Gartner #ArtificialIntelligence #UnbiasedData

R to @shapedai: Ready to learn more about synthetic data and its potential to revolutionize the AI landscape? Check out our full blog post: https://www.shaped.ai/blog/how-synthetic-data-is-used-to-train-machine-learning-models

R to @shapedai: Gartner predicts that by 2024, 60% of data needed for AI and analytics projects will be artificially generated. Synthetic data is here to stay and will continue to shape the future of AI.

R to @shapedai: Real-world applications are already benefiting from synthetic data! From NVIDIA Omniverse Replicator for autonomous vehicles 🚗 to Curai in healthcare 🏥, the impact is tangible and growing rapidly.

R to @shapedai: Our blog post covers various generation methods, such as SMOTE, Borderline SMOTE, and KMeans SMOTE. Learn how each technique contributes to creating high-quality, unbiased, and scalable data for your AI projects

R to @shapedai: 🔑 Advantages of synthetic data include generating large training datasets without manual labeling, reducing restrictions associated with regulated or sensitive data, and alleviating privacy concerns