{"slug": "weight-training-samples", "title": "Weight training samples", "summary": "Goldener 7.1.0 introduces the GoldDensityWeigher, a tool that weights training samples by embedding density, according to a Hugging Face blog post by Yann-CV. The post describes reusing the resulting weights for random or curriculum sampling and for loss computation, and its author is recruiting researchers to test the approach on real use cases and potentially co-publish results.", "body_md": "[Yann-CV](https://discuss.huggingface.co/u/Yann-CV)\n1\n \nEver wonder if you could weight your training samples from foundational embedding?\n\nIt is now possible with Goldener 7.1.0!\n\nIn [GOLDENER <br> Weighting training samples by embedding density](https://huggingface.co/blog/Yann-CV/goldener-weighting-samples-by-embeddings) , we introduce how to use our `GoldDensityWeigher` and then reuse the obtained weights for sampling (random or curriculum) and loss computation.\n\nLet us know if you wanna give it a try!\n\nWe are actively looking for researchers willing to test it on real use cases and potentially publish something together ?", "url": "https://wpnews.pro/news/weight-training-samples", "canonical_source": "https://discuss.huggingface.co/t/weight-training-samples/191993#post_1", "published_at": "2026-11-17 19:59:16+00:00", "updated_at": "2026-11-17 21:12:17.793513+00:00", "lang": "en", "topics": ["machine-learning", "ai-research", "developer-tools", "large-language-models"], "entities": ["Goldener", "GoldDensityWeigher", "Hugging Face", "Yann-CV"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/weight-training-samples", "markdown": "https://wpnews.pro/news/weight-training-samples.md", "text": "https://wpnews.pro/news/weight-training-samples.txt", "jsonld": "https://wpnews.pro/news/weight-training-samples.jsonld"}}