Simon Smith’s Post

Meet the new HPT 1.5 Air, the best Multimodal LLama3. And it’s open-sourced. In March this year, HyperGAI launched HPT Air along with our proprietary model HPT Pro, and we’ve been busy ever since! Today, we are glad to announce the release of HPT 1.5 Air, the best open-sourced 8B Multimodal LLaMA 3. HPT 1.5 Air follows the same recipe as its predecessor, HPT 1.0 Air, with a visual encoder, the H-Former, and the LLM. Although we kept the H-Former the same, we upgraded the visual encoder and changed the LLM to the LLaMA 3 8B Chat version. Thus, our HPT 1.5 Air is open sourced and fully available, empowering the users in developing many applications. HPT 1.5 Air demonstrates improved capabilities in: - Multimodal understanding & complex reasoning - Real world, contextual understanding HPT 1.5 Air outperforms GPT- 4V and Gemini 1.0 Pro on the SEED-I, ScienceQA, and MMStar benchmarks, despite having significantly fewer parameters. HPT 1.5 Air is now available on Hugging Face and GitHub, with a public demo coming soon. Take a look at our blog here: <https://lnkd.in/gXW5xZU9> Huggingface: https://lnkd.in/dCDwhdKQ

Yep, ChatGPT is great (if you know how to use it). What if I told you that you can easily train it with your own data? Learn how to build a custom AI Chatbot 😎 (it's easy I promise)

You can now integrate Decipher AI (YC W24) by just copy and pasting a single script on your website or application. No engineer required :) Within minutes, our AI will analyze session replays to uncover hidden bugs that are costing you user trust and churn. You even get rich technical context to understand and fix the bugs, fast.

Only you can use the power of Chat GPT to make your website stand out. The truth is only you can use power of Chat GPT to make your website stand out against others too.

tech blog Research says that chat GPT4 can be used for voice based scams: We know that voice cloning is being made easier due to technology. TThere are always good reasons to have a voice cloned, and so having the tech to do it is good. But with the good uses come the bad. I was not aware that voice cloning was in chat gpt at all, but then […] check it out!

For every Luzmo users (and future Luzmo users 😉 ) Our lab has just released new cool features 🚀 1️⃣ GenBI GPT : our first custom chat GPT available in the GPT store. Thanks to it, you can use the power of chat GPT with the information store in your dataset. This will allow you for exemple to directly ask questions about your data or create new dataset in Luzmo from an image analysis with GPT 4o. 2️⃣ Be able to log in with your Luzmo account in instachart to save dataset and dashboard created in your Luzmo account.

Create Custom Chatbots alongside GPT4o, Claude Opus, and Gemini 1.5 with ChatLLM 🔥🔥 In this video we are going to see how you can use GPT4o, Claude Opus and Gemini 1.5 side by side with Abacus.AI. You can also upload PDF documents and your own data to chat with and get correct responses based on the data. Another cool thing is that you can upload any data to the platform and create a custom chatbot with deployment in a few clicks. There are also enterprise connectors like teams, drive, slack and a ton of others connectors which can be hooked up to one of the chatbots. I'm showing all of it in this video and even create a custom chatbot with PDF documents and deployment in just a few minutes. You can then have a deployed endpoint you can send requests to and get responses back. Just like OpenAI's APIs but with your own chatbot!

ChatGPT's custom GPTs just got even more powerful. Now they can run spreadsheet calculations! 🚀 From drafting emails to generating on-brand copy, custom GPTs are already proving incredibly versatile. But up until now, there’s been a big limitation: they couldn’t run complex calculations. That changes today with GRID’s Spreadsheet Engine API. Now, custom GPTs can integrate seamlessly with your spreadsheet models, allowing them to handle everything from pricing calculations to financial forecasts with accuracy and speed. Check out how this works in the demo below ⬇️ and take a look at the first comment for a sneak peek at building a GPT that interacts with your models in just a few minutes! 👀

Want to build a Retrieval-Augmented Generation (RAG) system that is completely free, runs locally on your machine, and keeps your data private? And best is: it's completely No Code, so you don't have to be a developer to run it! In this step-by-step tutorial, I’ll show you how to set up a RAG pipeline using GPT4ALL—a powerful, open-source system that works offline. It comes with a beautiful UI, and it makes creating and managing local file collections a breeze. To understand it you can watch the video from https://lnkd.in/e5iXNQiy

Discover how to train a smaller model to run faster on your phone, or an embedded device, for object detection. (Easy to understand!) What’s fascinating is using a large model like ChatGPT - which is slow and power-hungry with a 3-second response time - to train a much smaller model that runs significantly faster on a low-end device (0.1 seconds, 0.3 MB). https://lnkd.in/ezcejDa6