Zac Zuo

DeepSeek made waves with their R1 language model, but their multimodal capabilities (especially image understanding) are not good enough: But, they are rapidly evolving. DeepSeek-VL2, their new open-source family of Mixture-of-Experts (MoE) vision-language models, is a big step forward, achieving strong performance with a much smaller activated parameter count, thanks to its MoE design. And...

Hi everyone! Pika just dropped another amazing new feature: Pikadditions! This looks like the easiest way yet to make really unique and eye-catching videos. Basically, you can now seamlessly add any image (person, object, whatever!) to any video – your own footage or a favorite clip. Just upload a video, an image, add a short prompt, and Pika's AI handles the rest (blending, sizing, movement...

Hi everyone, Sharing s1, a new approach to improving LLM performance at test time. This work comes from researchers at Stanford University and the University of Washington, and offers some exciting results: · Strong Reasoning: Matches the performance of larger models (like o1-preview) on reasoning tasks. · Minimal Data: Achieves this with only 1,000 examples (insane, right?). · Budget...

Hi everyone! 👋 As someone who reads a lot of arXiv papers (especially with the rapid pace of LLM research!), I've always wished for a better way to see direct discussions within the academic community about new research. Social media and blogs are scattered – alphaXiv is different. It's a new platform from arXiv Labs, built specifically for in-context discussion of preprints. So, here's what...

Hey everyone! 👋 Check out Mistral Small 3 – it's setting a new benchmark for "small" LLMs (under 70B)! 🚀 This 24B parameter model from Mistral AI offers performance comparable to much larger models, but with a focus on efficiency. So here's the key features: · Powerful & Efficient: State-of-the-art results with low latency (150 tokens/s). · Locally Deployable: Runs on a single RTX 4090...

Hi everyone, Le Chat by Mistral AI is now available as a mobile app for iOS and Android! The Le Chat app brings powerful AI capabilities to your fingertips, including: 💬 Natural Conversations: Chat with advanced AI models. 🔍 Real-time Web Search: Access up-to-date information. 📄 Document Analysis: Summarize and extract insights from documents. Check it out on your phone!

Hi everyone, Open WebUI is a self-hosted, open-source AI platform designed for ease of use and extensibility. It offers a range of features for interacting with a variety of LLMs: 🔗 Broad LLM Compatibility: Works with Ollama and OpenAI-compatible APIs (e.g., LMStudio, Mistral), plus built-in RAG, web browsing/search, and image generation. ✨ Highly Extensible: Includes a model builder, Python...

Here's the official DeepSeek-recommended list of integrations: "Awesome DeepSeek Integrations" (super direct name, that's their typical style, obviously!). It connects DeepSeek's API to a variety of apps, and the list is constantly growing. What integrations are you most interested in?

This forum update on PH is a good step! Reddit-like structure makes all the topics easier to follow. For 'Product Forums' section, it seems like not every product is able to have its own sub-page there, are the sub-pages within that section still by invitation now?

If you're already an Adobe Acrobat subscriber, there's a great new reason to check out the AI Assistant add-on. They've added powerful contract analysis features: Summarize: Get key points from any PDF. Ask: Chat with your docs for quick answers. NEW - Contracts: Easily understand terms, compare versions (up to 10!). It's all integrated seamlessly with the Acrobat you already use. And try ...

Ok, so this one is my favorite AI emoji generator so far. 😉 AI Emojis crafts personalized emojis using the brilliance of AI with a simple prompt. 🎨 Prompt-Driven AI Design: Simply tell it your mood, and watch it become an emoji. 📱 iMessage Integration: Enhance conversations with bespoke emoji stickers. 💾 Easy Download: Archive your special creations for repeated use. 📤 Instant Sharing: Spread...