Anand Singh
513 posts
@FederalAsh14
Finance guy interested in new tech and VC.
Dan says he's got Qwen 3.5 397B-A17B - a 209GB on disk MoE model - running on an M3 Mac at ~5.7 tokens per second using only 5.5 GB of active memory (!) by quantizing and then streaming weights from SSD (at ~17GB/s), since MoE models only use a small subset of their weights for each token
@MeganTStevenson Will be curious to see your and others' experiences. I just tried (using `--hybrid docling-fast --hybrid-mode full` and `--enrich-formula` and though super fast it was not good for the stuff other than text Attached is how gemini rated the conversions I got from the same paper
someone just open-sourced a tool that converts pdfs to markdown at 100 pages per second. 100% free. runs entirely on cpu. no expensive gpus needed.
Today, we're launching the world's largest open-source dataset of computer-use recordings. 10,000+ hours across Salesforce, Blender, Photoshop and more, to automate the next level of white-collar work. Link in the comments :) @markov__ai
🚨 Want to parse complex PDFs with SOTA accuracy, 100% locally? 📄🔍 At just 0.9B parameters, you can drop GLM-OCR straight into LM Studio and run it on almost any machine! 🥔 🧠 0.9B total parameters 💾 Runs on < 1.5GB VRAM (or ~1GB quantized!) 💸 Zero API costs 🔒 Total data privacy Desktop document AI is officially here. 💻⚡
Coronavirus is not a health emergency: officials
