Monolith: Real Time Recommendation System With Collisionless Embedding Table · HackerPapers discussions

Monolith: Real Time Recommendation System With Collisionless Embedding Table — HN submission

Fri, 20 Feb 2026 16:47:44 GMT

Monolith – The research paper behind TikToks algorithm (2022)

2 points; 0 discussion replies on HN. Paper and source history. A submission is not an endorsement.

Monolith: Real Time Recommendation System With Collisionless Embedding Table — mentioned by jamesblonde

Fri, 06 Feb 2026 13:27:26 GMT

I gave a talk at PyData Berlin on how to build your own TikTok recommendation algorithm. The TikTok personalized recommendation engine is the world's most valuable AI. It's TikTok's differentiation. It updates recommendations within 1 second of you clicking - at human perceivable latency. If your AI recommender has poor feature freshness, it will be perceived as slow, not intelligent - no matter how good the recommendations are. TikTok's recommender is partly built on European Technology (Apache Flink for real-time feature computation), along with Kafka, and distributed model training…

Paper and discussion history. A mention is not an endorsement.

Monolith: Real Time Recommendation System With Collisionless Embedding Table — mentioned by supo

Sat, 24 May 2025 01:19:57 GMT

If this was a solved problem, every e-com website would already run some variant of https://arxiv.org/pdf/2209.07663 for all their shopping surfaces and some version of "deep research" on every search query. They don't, so the problem isn't solved. And a big part of the problem is still the integration of behavioral, structured and unstructured signals into the underlying retrieval of products and content - which is what the article is about. Also, the company is doing fine, not sure where that last sentence came from ^_^

Paper and discussion history. A mention is not an endorsement.

Monolith: Real Time Recommendation System With Collisionless Embedding Table — mentioned by Grimblewald

Fri, 17 Jan 2025 00:43:03 GMT

the tiktok algo is genuinely impressive. What's cool is that the engineers published some works explaining how it functions. https://arxiv.org/abs/2209.07663 It's an interesting read if you're into recommender systems or AI in general. What amazes me is that despite this published work google and meta still can't produce a decent social media algorithm, so it's either incompetence or malice.

Paper and discussion history. A mention is not an endorsement.

Monolith: Real Time Recommendation System With Collisionless Embedding Table — HN submission

Thu, 26 Dec 2024 14:44:34 GMT

Monolith: Real Time Recommendation System with Collisionless Embedding Table

1 point; 0 discussion replies on HN. Paper and source history. A submission is not an endorsement.

Monolith: Real Time Recommendation System With Collisionless Embedding Table — HN submission

Fri, 20 Dec 2024 09:36:32 GMT

Monolith: Real Time Recommendation System with Collisionless Embedding Table

2 points; 0 discussion replies on HN. Paper and source history. A submission is not an endorsement.

Monolith: Real Time Recommendation System With Collisionless Embedding Table — HN submission

Mon, 16 Dec 2024 14:10:39 GMT

2022 Recommendation System paper from ByteDance – the parent company of TikTo

1 point; 1 discussion reply on HN. Paper and source history. A submission is not an endorsement.

Monolith: Real Time Recommendation System With Collisionless Embedding Table — mentioned by wyldberry

Fri, 06 Dec 2024 18:09:20 GMT

At best this comment implies there's actual an actual smoking gun related to what functionally are nation-state capability and operations to influence populations external to their own, specifically related to TikTok/ByteDance. If you have an AP government level understanding of how China and the USA work, then you can probably understand that this is an intersection of: government laws, great power competition, and wealth seeking behaviors. When you control the algorithm, everyone downstream of it is at the mercy of the power structures and incentives of your top level leadership [0].…

Paper and discussion history. A mention is not an endorsement.

Monolith: Real Time Recommendation System With Collisionless Embedding Table — HN submission

Tue, 08 Oct 2024 07:51:48 GMT

Real Time Recommendation System with Collisionless Embedding Table (2022)

1 point; 0 discussion replies on HN. Paper and source history. A submission is not an endorsement.

Monolith: Real Time Recommendation System With Collisionless Embedding Table — mentioned by jerojero

Tue, 14 Nov 2023 11:38:34 GMT

If you think about it, the secret sauce tends to be in the data they're able to capture and input into the recommendation algorithms rather than the recommendation algorithm itself. I mean, even Tiktok has papers detailing their recommendation algorithm [1]. So does Youtube and plenty of others. It's all in the data now a days. [1] https://arxiv.org/pdf/2209.07663.pdf

Paper and discussion history. A mention is not an endorsement.

Monolith: Real Time Recommendation System With Collisionless Embedding Table — HN submission

Sun, 08 Oct 2023 14:05:40 GMT

Real-Time Recommendation System with Collisionless Embedding Table (2022) [pdf]

2 points; 0 discussion replies on HN. Paper and source history. A submission is not an endorsement.

Monolith: Real Time Recommendation System With Collisionless Embedding Table — mentioned by cwillu

Fri, 14 Apr 2023 20:19:00 GMT

Referenced paper at https://arxiv.org/pdf/2209.07663.pdf

Paper and discussion history. A mention is not an endorsement.

Monolith: Real Time Recommendation System With Collisionless Embedding Table — mentioned by O__________O

Fri, 17 Feb 2023 17:18:30 GMT

Relate HN comments: - https://news.ycombinator.com/item?id=33494796 Related paper: - https://arxiv.org/abs/2209.07663 Related GitHub: - https://github.com/bytedance/monolith

Paper and discussion history. A mention is not an endorsement.

Monolith: Real Time Recommendation System With Collisionless Embedding Table — mentioned by dajoh

Fri, 03 Feb 2023 15:16:13 GMT

Also, here's the paper on it: https://arxiv.org/pdf/2209.07663.pdf

Paper and discussion history. A mention is not an endorsement.

Monolith: Real Time Recommendation System With Collisionless Embedding Table — mentioned by sgwizdak

Wed, 18 Jan 2023 02:56:03 GMT

This paper by various folks at Bytedance hints at their recommender architecture: https://arxiv.org/pdf/2209.07663.pdf

Paper and discussion history. A mention is not an endorsement.

Monolith: Real Time Recommendation System With Collisionless Embedding Table — mentioned by CharlieSweeting

Sat, 17 Dec 2022 12:17:21 GMT

Arvind's making the case that the algorithm isn't particularly special and it's all about the system design which hooks users. It's definitely true that a huge part of their user retention isn't some magical algorithm, but they have made some genuine algorithmic/training improvements which are interesting. They published their methodology a few months ago (https://arxiv.org/abs/2209.07663), which wasn't referenced in the article (only an official blog post and a few reverse engineering attempts that I don't think are enough to conclude that their recommender is just run of the mill).

Paper and discussion history. A mention is not an endorsement.

Monolith: Real Time Recommendation System With Collisionless Embedding Table — HN submission

Mon, 07 Nov 2022 23:36:55 GMT

Paper on 'Monolith: Real Time Recommendation System' by Bytedance Inc (TikTok)

3 points; 0 discussion replies on HN. Paper and source history. A submission is not an endorsement.

Monolith: Real Time Recommendation System With Collisionless Embedding Table — HN submission

Mon, 07 Nov 2022 10:52:32 GMT

TikTok published an article on their recommendation system architecture

3 points; 0 discussion replies on HN. Paper and source history. A submission is not an endorsement.

Monolith: Real Time Recommendation System With Collisionless Embedding Table — HN submission

Sun, 06 Nov 2022 17:13:55 GMT

Monolith: Real time recommendation system with collisionless embedding table

70 points; 6 discussion replies on HN. Paper and source history. A submission is not an endorsement.