LlamaParse 🤝 Quivr 🧠

Quivr (YC W24) 🧠 (@_StanGirard) is a popular open-source platform where you can create personalized AI assistants over your data.

We’re excited to partner with Quivr to introduce LlamaParse - parse any complex document (.pdf, .pptx, .md) through our advanced

GPT-4o support now available in create-llama! Get 90% of the way through building a chatbot over your data just by answering a few questions.

Announcing LlamaIndex on Vertex AI 💫

We are excited to partner with the Vertex AI team (Google Cloud) to feature a brand-new RAG API on Vertex, powered by LlamaIndex 🦙 advanced modules that enable e2e indexing, embedding, retrieval, and generation.

It is simultaneously 💡easy

Google vs OpenAI is the beef we’re all here for

Dumping updates from I/O in this 🧵

Use GPT-4o for Data Extraction from Images! 🏞️🗂️

OpenAI just released GPT-4o, where the 'o' stands for omni, alluding to the fact that this latest and greatest model is capable of working in text, vision, and audio modalities.

I recently shared a post on how one could use

Congrats to our own Ravi Theja, co-creator of Navarasa, for having his work featured at Google I/O today! Navarasa is a fine-tuned version of Google's Gemma model, designed to work well with 15 different Indic languages.

Check out the video: youtube.com/watch?v=b4Gs-t…

And learn

Build a local, private research assistant running on your laptop in a snap with llamafile from Mozilla!

llamafiles are fun: no need to install anything, just download the file and run it, and you get a local LLM and embedding model that you can use directly from LlamaIndex. In

OpenAI just announced a new AI model it's calling GPT-4o

Comprehensive Tour of Llama3 Use Cases with ollama, Groq Inc, Replicate 🦙

To celebrate the Llama 3 hackathon hosted by AI at Meta this past weekend, we’ve created a set of cookbooks that show you how to use Llama 3 for 7️⃣ different use cases 🔥, from easy-to-hard.

1. Basic

This is a comprehensive article by kingzzm on how to evaluate RAG systems - a core piece of any AI engineer’s toolkit.

It covers 4 evaluation libraries (TruLens, Ragas, UpTrain, DeepEval) and the comprehensive set of evaluation metrics that these libraries support.

1. Metrics

Using RAG for Content Moderation 👮

This article by CloudRaft shows you how to setup a RAG pipeline to ensure that user-generated images adhere within a predefined set of content moderation rules.

1. Caption user-uploaded image to convert it to text.
2. Match against indexed

Building a Financial Agent that can Perform Reflection 📈🔎

This is a great example by Hanane Dupouy on showing you how to build an agent with reflection to help analyze stock prices.

There’s different ways to implement reflection:
1. Implementing CRITIC (tool use) to provide

Generating PowerPoints with Llama 3 🦙📊

This is an awesome article by Arslan Shahid (e/acc) on how to use LlamaIndex 🦙 to build a Llama3 RAG pipeline that doesn’t just answer questions, but can generate a Powerpoint slide deck.

Python-pptx is a neat library to programmatically create

RAG with Trustworthiness scores!🎗️

Even with sources/citations, it can be challenging to determine the accuracy of a generated RAG response right away.

Cleanlab offers an impressive “Trustworthy Language Model” that assigns a trustworthiness score to each answer generated by

Automatically preserve chat history beyond the size of your token window with our Chat Summary Memory Buffer!

Long-running chats can easily exceed this size of your token window. Truncating the oldest entries in the chat log works, but can lose valuable context. This technique

You asked for it, and it's here: Google Firestore serverless document DB and vector store integration!

✅ Serverless, JSON-compatible database that easily scales, with no partitioning or maintenance
✅ Fully customizable security and data validation rules to ensure the data is

Friends don't let friends ship naive top-k RAG in 2024 x.com/granawkins/sta…