Today’s top AI Highlights:

1. Microsoft’s new opensource Phi 3.5 models with MoE and vision

2. Extract structured data in just a few clicks with LlamaIndex

3. New text-to-image model rated better than FLUX and DALL.E 3

4. Autonomous AI agent for comprehensive online research

5. Build and evaluate RAG systems in 5-7 lines of code

& so much more!

Read time: 3 mins

Latest Developments

Microsoft’s Small LMs Compete with Gemini and GPT-4o 💪

Microsoft has released its next series of opensource small language models. The new Phi 3.5 series features Phi-3.5-mini, Phi-3.5-MoE, and Phi-3.5-vision, designed to be lightweight and powerful, focused on high-quality reasoning, and supporting 128K token context length. They are trained on a mix of synthetic and filtered public web data, further tuned for instruction following and safety.

Key Highlights:

1. Phi-3.5-mini - This lightweight model with 3.8B parameters excels in multilingual tasks and long context scenarios, outperforming larger models like Mistral-7B and Llama 3.1 8B in Multilingual MMLU and long context question answering.

2. Phi-3.5-MoE - Utilizing the Mixture-of-Experts architecture, Phi-3.5-MoE achieves impressive reasoning capabilities with only 6.6B active parameters, surpassing larger models like Llama 3.1 8B and Gemma 2 9, and even competes with GPT-4o-mini.

3. Phi-3.5-vision - This model introduces powerful visual understanding to the series, performing competitively against larger models like Gemini 1.5 Flash in tasks like image captioning, visual question answering, and document intelligence benchmarks.

4. Opensource - All Phi 3.5 models are opensourced under the MIT license and are available on Hugging Face and Azure AI Studio.

Stop Wrestling with Data and Get What You Need 📊

Your unstructured data is a goldmine but it doesn’t have to be an archaeological dig. LlamaIndex has released LlamaExtract, a new managed service that lets you easily extract structured data from your unstructured documents. It offers both schema inference from your documents and the ability to extract values based on a provided schema. You can access LlamaExtract through a user-friendly UI or via an API.

Key Highlights:

1. Automated Schema Inference - Upload your documents, and LlamaExtract will intelligently infer the underlying schema, saving you the time and effort of manual definition. You can further customize this inferred schema to perfectly match your needs.

2. Streamlined Extraction Workflow - Whether using the intuitive UI for rapid prototyping or the powerful API for programmatic access, LlamaExtract offers a smooth and efficient way to extract structured data from your documents.

3. Use Cases - From extracting key information from resumes and invoices to structuring product catalogs, LlamaExtract is adaptable to various document types and extraction tasks, making it a powerful tool in your LLM toolkit.

Quick Bites

1. OpenAI is partnering with Condé Nast to bring content from brands like Vogue and Wired into ChatGPT and the new SearchGPT prototype. This is in a series of partnerships of OpenAI with publication houses like Time and Vox Media to bring reliable information and also avoid copyright infringement.
   

2. AI chip and inference startup Recogni has announced a patented logarithmic number system for AI, Pareto, that radically simplifies AI compute by turning multiplications into additions, making Recogni’s chips smaller, faster, and less energy-hungry. Tested on various AI models, Pareto delivered more efficient AI operations without compromising accuracy.
   

3. Ideogram has launched Ideogram 2.0, a free text-to-image model with five distinct styles, along with an iOS app, a beta API, and Ideogram Search. The model is rated better than Flux Pro and DALL·E 3 by human evaluators.

Tools of the Trade

1. GPT Researcher: An autonomous AI agent designed for fast, detailed, and unbiased online research. It tackles issues like misinformation and speed by parallelizing tasks, producing long, comprehensive reports using a variety of reliable sources.

2. Pipeshift AI: A cloud platform that helps teams fine-tune and run opensource LLMs for production, offering better performance and ownership compared to GPT/Claude.

3. Beyond LLM: Opensource all-in-one toolkit that lets you build and evaluate RAG and LLM applications in just 5-7 lines of code. It automates key integrations and supports custom evaluation metrics, helping reduce LLM hallucinations.

4. Awesome LLM Apps: Build awesome LLM apps using RAG to interact with data sources like GitHub, Gmail, PDFs, and YouTube videos through simple text. These apps will let you retrieve information, engage in chat, and extract insights directly from content on these platforms.

Hot Takes

1. Honestly, I'm thinking about this a lot. Maybe it makes sense to build products, that would work with better models and just wait. ~
   Niklas Sikorra
   

2. BTW the only people who haven't visited Stackoverflow in months are the ones who don't write that much code.
   Creators != Programmers ~
   Jaydeep Karale

Meme of the Day

Tech bros trying every weekly model variant pretending they can feel a difference

That’s all for today! See you tomorrow with more such AI-filled content.

Real-time AI Updates 🚨

⚡️ Follow me on Twitter @Saboo_Shubham for lightning-fast AI updates and never miss what’s trending!

Unwind AI - Twitter | LinkedIn | Instagram | Facebook

Awesome LLM Apps | Sponsor Us

PS: We curate this AI newsletter every day for FREE, your support is what keeps us going. If you find value in what you read, share it with at least one (or 20) of your friends!