This is a heavily interactive web application, and JavaScript is required. Simple HTML interfaces are possible, but that is not what this is.
Post
Daniël de Kok
danieldk.eu
did:plc:xbq7pvqlraojw25b27syamzp
And on the heels of Flash Attention 3, we also have the vLLM fork of Flash Attention 3:
https://huggingface.co/kernels/kernels-community/vllm-flash-attn3/tree/v1/build
This is used for e.g. gpt-oss, since that model uses attention sinks.
2026-07-02T06:24:14.125Z