How to fit Qwen3.8-27B into 16GB of VRAM and run it on a single RTX 3080 card: the best quantizations and Llama.cpp flags I've found TL;DR: UD-IQ3_XXS with KV cache quantization … »
How to fit Qwen 3.6 35B A3B into 16GB of VRAM, & run it with Llama.cpp on an RTX 3080 The belly hangs over the belt, but it fits … »
Underrated reasons to dislike AI The big arguments for and against AI have been endlessly discussed, and I don’t feel I have much to add. AGI and existential risk; human obsolescence; power use; cybersecurity; safety + censorship; slop; … »
PSA: local AI handwriting recognition is usable now To my delight and surprise, local Qwen3-VL:8b WER is acceptable, even on my cryptic handwriting … »
How to increase Ollama context length OLLAMA_CONTEXT_LENGTH environment variable didn’t have an effect, but there’s another way … »
How to pause and resume a resource-intensive process on Linux with SIGSTOP: specifically, Ollama LLM inference Keep your system resources to yourself when you need them, and put them to good use when you don’t … »
OCR Typewritten Documents with a Local Vision Model (Qwen3-VL:8B + Ollama) It’s now possible to get results better than Tesseract, without relying on cloud services … »
