← All concepts

post training alignment

1 articles · 4 co-occurring · 0 contradictions · 0 briefs

The author's core claim is that post-training creates irreversible patterns in how models handle specific interface formats. This is a concrete example of post-training effects on behavior.

The author's core claim is that post-training creates irreversible patterns in how models handle specific interface formats. This is a concrete example of post-training effects on behavior.

query this concept
$ db.articles("post-training-alignment")
$ db.cooccurrence("post-training-alignment")
$ db.contradictions("post-training-alignment")