TinyStories-33M WebGPU

Client-side text generation. Inference runs on your GPU via transformers.js and the ONNX Runtime Web WebGPU backend — no server-side compute.