trying to replicate the emergent misalignment paper on qwen2.5-coder. at first i thought i cooked the model beyond repair but now i think the alignment it's wah-ing against is... chinese? the model…

Posters: voooooogel. voooooogel

2 likes recorded by the source corpus.

Topics: AI & LLMs, AI safety

View Post: https://x.com/voooooogel/status/1895041486716313614