{
"$type": "site.standard.document",
"bskyPostRef": {
"cid": "bafyreigne6r2xxpkpnaugsis57zrqvdz556rliaefkoruewomfyrebw4f4",
"uri": "at://did:plc:pgryn3ephfd2xgft23qokfzt/app.bsky.feed.post/3mejhi3b27kr2"
},
"path": "/t/request-arxiv-endorsement-for-new-mech-interp-paper-on-llm-self-referential-circuits/173241#post_4",
"publishedAt": "2026-02-10T15:44:20.000Z",
"site": "https://discuss.huggingface.co",
"tags": [
"Zenodo",
"When Models Examine Themselves: Vocabulary-Activation Correspondence in..."
],
"textContent": "Zenodo\n\n### When Models Examine Themselves: Vocabulary-Activation Correspondence in...\n\nLarge language models produce rich introspective language when prompted for self-examination, but whether this language reflects internal computation or sophisticated confabulation has remained unclear. We show that self-referential vocabulary tracks...\n\nUpdate: a more concise version with formatting adjustments - welcoming any feedback and discussions.",
"title": "[Request] arXiv endorsement for new mech interp paper on LLM self-referential circuits"
}