{
  "$type": "site.standard.document",
  "bskyPostRef": {
    "cid": "bafyreiej7fryyntzgrlmxukrvfit2u43p2cwsfpnlyiwvzt6wihls3qnwa",
    "uri": "at://did:plc:pgryn3ephfd2xgft23qokfzt/app.bsky.feed.post/3mkmtvwhepzq2"
  },
  "path": "/t/custom-batches-in-sentence-transformers-for-multiplenegativesrankingloss/175587#post_3",
  "publishedAt": "2026-04-29T08:08:22.000Z",
  "site": "https://discuss.huggingface.co",
  "textContent": "Thank you a lot for your answer, I’ll read it carefully and try to implement it.\nSince it appears to be (at least partially) LLM-generated, I hope you don’t mind a couple of clarifying questions (I’m new to hf, and I am not sure what’s the ai-generated content policy here):\n\n  1. Did you review the solution yourself? If so, are you familiar with the `sentence-transformers` library?\n\n  2. Which LLM did you use to generate the answer? I’ve asked ChatGPT 5.4 several times about this issue, but it consistently suggests writing a custom training loop that bypasses `SentenceTransformerTrainer` and its additional features.\n\n\n\n\nThanks again!",
  "title": "Custom batches in sentence-transformers for MultipleNegativesRankingLoss"
}