{
"$type": "site.standard.document",
"bskyPostRef": {
"cid": "bafyreihkgpihc6bsxjfpr2fx6kkzdh5miw6ow3pb3lr5chcjxyn46gwycq",
"uri": "at://did:plc:pgryn3ephfd2xgft23qokfzt/app.bsky.feed.post/3mjr4xkr3ebk2"
},
"path": "/t/szukam-feedbacku-wlasna-reprezentacja-semantyczna-bryla-dla-malych-modeli/175350#post_3",
"publishedAt": "2026-04-18T07:31:59.000Z",
"site": "https://discuss.huggingface.co",
"textContent": "Thank you for this incredible analysis. This is exactly\nthe kind of feedback I was hoping for.\n\nYour priorities match my intuition — I suspected that\naffect and is_core do most of the work, and I’m planning\nablation experiments to confirm this.\n\nThe hybrid approach (raw + bryla) is something I haven’t\ntried yet — that’s now high on my list.\n\nI’ll look into BabyLM and the papers you referenced.\nThis gives me a clear roadmap.\n\nReally appreciate you taking the time.",
"title": "Szukam feedbacku — własna reprezentacja semantyczna \"bryła\" dla małych modeli"
}