{"slug": "which-open-source-model-would-you-recommend-for-generating-video-recommendations", "title": "Which open-source model would you recommend for generating video recommendations in an OTT platform?", "summary": "TensorFlow Recommenders (TFRS) with a two-tower retrieval model is the recommended open-source starting point for video recommendation in an OTT platform, according to a technical response. The response suggests using TFRS's basic retrieval tutorial, and for sequential watch histories, BERT4Rec or SASRec via RecBole, with sentence-transformers/all-MiniLM-L6-v2 for cold-start content embeddings. The recommended progression is to start with TFRS two-tower retrieval plus a ranker, and only adopt BERT4Rec if sequential data improves offline metrics.", "body_md": "I would not start with a general-purpose LLM for this. Video recommendation is mainly a **retrieval-and-ranking** problem, and the best model depends on what interaction data you have.\n\nA practical open-source starting point is **TensorFlow Recommenders (TFRS)** with a two-tower retrieval model:\n\nThe official movie-retrieval tutorial is close to this use case and includes training, evaluation, and export: [https://www.tensorflow.org/recommenders/examples/basic_retrieval](https://www.tensorflow.org/recommenders/examples/basic_retrieval)\n\nIf you have ordered watch histories, compare sequential models such as **BERT4Rec** or SASRec through RecBole. BERT4Rec is specifically designed to model sequences of user interactions: [BERT4Rec — RecBole 1.2.1 documentation](https://recbole.io/docs/user_guide/model/sequential/bert4rec.html)\n\nFor a cold-start baseline, you can embed titles, descriptions, genres, and tags with a text encoder such as `sentence-transformers/all-MiniLM-L6-v2`\n\nand recommend semantically similar videos. Its model card lists semantic search as an intended use: [sentence-transformers/all-MiniLM-L6-v2 · Hugging Face](https://huggingface.co/sentence-transformers/all-MiniLM-L6-v2) . However, that is an item-content encoder, not a complete personalized recommender.\n\nMy suggested progression would be:\n\nSo, if I had to choose one free starting stack: **TFRS two-tower retrieval plus a ranker**, with content embeddings for cold start. I would choose BERT4Rec only after confirming that sequential watch history improves the offline metrics.", "url": "https://wpnews.pro/news/which-open-source-model-would-you-recommend-for-generating-video-recommendations", "canonical_source": "https://discuss.huggingface.co/t/which-open-source-model-would-you-recommend-for-generating-video-recommendations-in-an-ott-platform/178458#post_2", "published_at": "2026-08-04 23:29:20+00:00", "updated_at": "2026-08-04 23:34:25.893007+00:00", "lang": "en", "topics": ["machine-learning", "artificial-intelligence"], "entities": ["TensorFlow Recommenders", "BERT4Rec", "SASRec", "RecBole", "sentence-transformers/all-MiniLM-L6-v2", "Hugging Face"], "alternates": {"html": "https://wpnews.pro/news/which-open-source-model-would-you-recommend-for-generating-video-recommendations", "markdown": "https://wpnews.pro/news/which-open-source-model-would-you-recommend-for-generating-video-recommendations.md", "text": "https://wpnews.pro/news/which-open-source-model-would-you-recommend-for-generating-video-recommendations.txt", "jsonld": "https://wpnews.pro/news/which-open-source-model-would-you-recommend-for-generating-video-recommendations.jsonld"}}