AI Decision System
Replayable, offline-verifiable records for AI-assisted go/no-go decisions: experiment promotions, typed model answers and human-approved actions.
Explore a real language model's architecture, tensors, and live generation in 3D — all from real data.
Developer toolsTypeScript · PythonGitHub @SudharsanselvarajOnlineOwner unconfirmed
Have you tried it? One tap is enough. No sign-in needed.
Interactive 3D visualization platform for exploring transformer architectures, tensors, and real-time LLM inference.
Taken from the repository README
Loading the full README. The saved excerpt is shown below.
Stop reading architecture diagrams. Watch your model think.
A real 3D inspector wired into a real forward pass — not an illustration of one.
Every other LLM visualizer gives you one of two things: a static architecture diagram, or a
"visualization" quietly running on fake numbers. TokenPrint does neither. Load
Qwen2.5-0.5B-Instruct live or drop in any local .gguf, and watch attention heads, RoPE
rotations, and the residual stream move in real 3D — every value traced back to
named_parameters(), a real forward pass, or the GGUF binary header itself.Read the full README ↗The full README is unavailable right now. Read it at the original link. The saved excerpt is shown below.
Stop reading architecture diagrams. Watch your model think.
A real 3D inspector wired into a real forward pass — not an illustration of one.
Every other LLM visualizer gives you one of two things: a static architecture diagram, or a
"visualization" quietly running on fake numbers. TokenPrint does neither. Load
Qwen2.5-0.5B-Instruct live or drop in any local .gguf, and watch attention heads, RoPE
rotations, and the residual stream move in real 3D — every value traced back to
named_parameters(), a real forward pass, or the GGUF binary header itself.Read the full README ↗