This is nice and useful because the new GPT-OSS model uses this technique. Kudos... | Hacker News

Hacker Newsnew | past | comments | ask | show | jobs | submit

		am17an 3 months ago \| parent \| context \| favorite \| on: How attention sinks keep language models stable This is nice and useful because the new GPT-OSS model uses this technique. Kudos to the original authors!

diggan 3 months ago [–]

And, as always, the FOSS ecosystem moves quickly, llama.cpp already fully support them! https://github.com/ggml-org/llama.cpp/pull/15157

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact