Sachin Kotwani’s Post

Just finished the course “Google Gemini for Developers” by Lynn Langit! Check it out: https://lnkd.in/eZFZ5uYV #generativeai. Nice to see how Google is making building custom LLM accessible with Vertex AI Agent Builder, more info here: https://lnkd.in/eK4Pdzyx

Gemini’s is here! Google has released a new multi-modal model named Gemini. It shows impressive capabilities and is said to beat GPT-4 on almost benchmarks, which is great if you can live with all the ‘BUT’s … >> Gemini comes in three flavours Ultra, Pro and Nano, with Ultra being the most performant, GPT4-beating version, BUT … that version is not available until next year. >> The model has a truly multi-modal architecture BUT … it can only reply with text currently. >> Gemini Pro can now be used everywhere through Bard BUT … not if you’re in Europe. And so on … What is truly impressive though is the way the model can pick up on minor nuances in image and video input and draw correct conclusions from it (=> duck inference => shell game). It can also reason its way into building a minor website for your inspiration and to guide you through choices (=> birthday party planning) This announcement is not all there in terms of actual release, but the game is definitely on for the LLM/LLM heavyweight crown. An interesting future lies ahead… #generativeai #google #openai #llm #lmm https://lnkd.in/dBCJXAPn

🚀 Excited about the new Gemini Flash model introduced by Google at I/O 2024! 🌟 This lightweight, efficient model supports multimodal reasoning, making it a versatile tool for various AI applications. Whether you're generating content or analyzing images, Gemini Flash offers impressive performance. Check out my latest blog post to get started with the Gemini Flash model, complete with a step-by-step guide and sample code. Dive in and explore the potential of this powerful AI model! 💡 👉 https://lnkd.in/d-9cK423

It's that time of the week again. A new update is out! 🧞 GPT-4 for teams 💨 Faster loading of AI Chats 🪨 Improved reliability for Snippets 🐞 Bug fixes for Theme Studio, Calculator, and more Full changelog 👉 https://lnkd.in/emswZvs4

Notebook collection is really helpful for quick execution and learning

OpenAI recently held its first developer event, and they announced some advancements in AI technology! Here are the key highlights: 🔹 OpenAI has reduced pricing across many parts of platform. 🔹 GPT-4 Turbo with a powerful 128K context for enhanced memory and performance. Now it's more capable and affordable. 🔹 DALL-E 3 API: OpenAI's text-to-image model, DALL-E 3, is now available through an API with built-in moderation tools. Users can now create stunning images with ease! 🔹 New Text-to-Speech APIs: OpenAI introduced the Audio API, offering access to generative AI models with six preset voices. 🔹 Assistants API: OpenAI launched new API that allows developers to build their own agent-like experiences, from coding assistants to AI-powered vacation planners. Check more in the article ⬇️

Exciting news from Hugging Face! 🚀 Check out the newly added notebook in the Open Source AI Cookbook, demonstrating a seamless transition from OpenAI models to Open LLMs using the Messages API from Text Generation Inference. What recipes are you eager to see added next? Share your thoughts! 👀 #AI #OpenSource #TechInnovation #AICommunity

"𝐓𝐡𝐞 𝐅𝐨𝐮𝐫 𝐖𝐚𝐫𝐬 ⚔ 𝐨𝐟 𝐭𝐡𝐞 𝐀𝐈 𝐒𝐭𝐚𝐜𝐤 (𝐃𝐞𝐜 𝟐𝟎𝟐𝟑 𝐑𝐞𝐜𝐚𝐩)" - great recent episode from the "Latent Space" podcast 🎧. It ... . ... provides a comprehensive review of the following four major “wars” in the AI landscape in 2023 and offers insights into the outlook for AI in 2024: (𝟏) 𝐃𝐚𝐭𝐚 💿: OpenAI's partnership with Axel Springer and the lawsuit from The New York Times demanding the shutdown of GPTs, along with the rise of interest in synthetic data. (𝟐) 𝐆𝐏𝐔/𝐈𝐧𝐟𝐞𝐫𝐞𝐧𝐜𝐞 ⚡️: The price fluctuations of Mixtral output tokens, benchmark drama between Anyscale and other inference providers, and advancements in new model architectures (Mamba, RWKV) and GPU resources (Modular, tinycorp, Apple MLX). (𝟑) 𝐌𝐮𝐥𝐭𝐢𝐦𝐨𝐝𝐚𝐥𝐢𝐭𝐲 🖼️: Updates from Midjourney, Assembly AI, Replicate, and Suno AI, as well as the ongoing development of “God Models” by OpenAI and Google. (𝟒) 𝐑𝐀𝐆/𝐎𝐩𝐬 ⚙️: The debate between Vector DBs and new alternatives like turbopuffer, the competition between LangChain and LlamaIndex, and the advancements in LLMOps and framework-driven tooling. This episode provides a great overview of these four wars, discussing key developments, partnerships, lawsuits, and advancements in each area. It offers valuable insights into the major trends and challenges in the AI field during 2023 and sets the stage for what to expect this year. Don't miss it! 𝐿𝑖𝑛𝑘 𝑡𝑜 𝑆𝑝𝑜𝑡𝑖𝑓𝑦 🔗 https://lnkd.in/ePhmmPjf

Worried about being locked-in with an OAI model? There's an easy way to swap it with an Open Source model without breaking your LangChain and LlamaIndex integrations. Check out the latest Open Source AI cookbook addition by Andrew Reed 👇

This collection of code-driven recipes empowers you to tackle diverse tasks using the power of #opensource tools and models. From #text_analysis to #image_classification, each recipe provides a step-by-step guide, ensuring you can whip up impressive AI solutions in no time. So, grab your favourite coding utensils and dive into the Cookbook to get the AI-driven dish of your liking.