ever stared at an LLM output like “bro what are you thinking?” yeah. same. so I built a thing — not to tame the model, but to eavesdrop on its internal monologue.
born from late nights, too much tea, and a borderline unhealthy obsession with why machines say what they say, this debugger lets you peek into the mind of the model — its whispers, its logic spirals, its “I-think-I-know-the-answer” energy. live-streamed. markdown’d. vibes included.
what I learned? transparency isn’t just ethical — it’s ✨existential✨. reasoning is recursion in disguise. and Streamlit? the easiest way to turn madness into UI.
this started as curiosity. then turned into a project. now it’s the prelude to a whole research paper.
read it. fork it. judge it. break it. just don’t say I didn’t warn you — the machine thinks. and now you can watch it do so.
Log in or sign up for Devpost to join the conversation.