FP8 weight edit. Not the official Flash recipe. No quoted tok/s.
Params
unknown
Fastest decode
—
Refusal
Abliterated
Setups
0
Runs on
Quoted one-stream decode on real boxes.
No quoted pair yet.
Get it running
- 1. DownloadPuts the weights on this computer.
huggingface-cli download dealignai/GLM-5.3-Flash-ABLITERATED-FP8 - 2. Start itUse this only if the download is GGUF.
llama-server -hf dealignai/GLM-5.3-Flash-ABLITERATED-FP8 - Open the weightsHugging Face ↗