{
  "$type": "site.standard.document",
  "bskyPostRef": {
    "cid": "bafyreib34sv3mvalfl2adfqwmierjlxrijqieymsbqdcfkwllmxzwyuzja",
    "uri": "at://did:plc:pgryn3ephfd2xgft23qokfzt/app.bsky.feed.post/3mnzzpamtdxj2"
  },
  "path": "/t/can-an-llm-lose-conceptual-continuity-while-remaining-coherent/176469#post_9",
  "publishedAt": "2026-06-11T18:58:58.000Z",
  "site": "https://discuss.huggingface.co",
  "textContent": "I am taking your comment as an interesting side-quest for a later moment! Now I want to show you (and anyone reading) a comment about the current state of the system:\n\n  * Uses proven working TIS components\n  * NIAH: 100% at all budgets (structural learning works)\n  * LITM: 46.1% @ 50% (oracle-tier, semantic learning blocked)\n\n\n\nThis means that my system solves Needle-In-A-Haystack problem at KV budgets from 25% to 100% while the main problem, the Lost-In-The-Middle problem, which is a generalization of NIAH, is what’s becoming my main focus right now!\n\nBy the way .. can you sponsor me at ArXiv later on, when I’m ready to publish? Thanks!",
  "title": "Can an LLM lose conceptual continuity while remaining coherent?"
}