DingoOz/llm-visuals

Kept llm-visuals open beside llama-server while tuning a long-context run. Watching TTFT and context fill live beat guessing from logs. It's a Rust TUI that finds llama.cpp, Ollama, vLLM or SGLang by…

Proof of the value in local AI. 18 hours of autonomous coding with 2 models running over 2 GPUs has generated 925,000 tokens of working code. No cloud, 0% subscription usage, no 5 hour limits and it…

Steal my coding workflow: I dictate my thoughts to Grok and ask for a goal / loop prompt for Grok Build. Start Grok Build using local LLM Models (Qwen3.8-27B and Qwen3.5-4B). Then hit enter and walk…

I am getting a lot of value out of this open source tool llm-visuals when testing my local models like Qwen3.8-27B.

I am in love with this new tool that shows me how the MoE experts are being used with Qwen3.6-35B-A3B. Realtime look inside a local AI's 'brain'. I will open source this tomorrow.