Saved in:
| Main Authors: | , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2605.23702 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866918518659743744 |
|---|---|
| author | Salle, Alexandre Niu, Chenglei Mahapatra, Suchismit Chen, Xiaoxiao Sedhain, Suvash Wang, Yaqi Shahryari, Shervin Agrawal, Saurabh Chen, Qiang Tamir, Michael |
| author_facet | Salle, Alexandre Niu, Chenglei Mahapatra, Suchismit Chen, Xiaoxiao Sedhain, Suvash Wang, Yaqi Shahryari, Shervin Agrawal, Saurabh Chen, Qiang Tamir, Michael |
| contents | Personalized discovery systems often train separate models for item ranking, carousel ranking, and search, even though these tasks expose complementary signals from the same viewer journey: watches shape carousel and item ranking, search queries reveal intent even when they do not lead to a catalog match, and watch history helps interpret search as rewatching, continuation, or new discovery. We introduce the user story, a serialized representation that turns a user's cross-surface history - attributes, sessions, watch events with surface and carousel context, and search events - into a single token sequence. By interleaving pretrained language tokens with domain-specific event tokens, user stories let heterogeneous recommendation and search tasks be expressed as prompted next-token prediction over a shared grammar. TubiFM is one instantiation of this approach: a Llama 3.2 1B-based model trained on user stories and prompted to rank items, carousels, or search results without task-specific architectures. In offline evaluation, this single model outperforms specialist baselines across item, carousel, and search ranking. In online A/B tests, TubiFM significantly improves search total viewing time (TVT) by $+3.9\%$ and carousel TVT by $+0.30\%$. Item ranking is statistically neutral on TVT ($+0.14\%$), but matches a mature production stack; across all three tasks, TubiFM serves on L40S GPUs and reduces p99 ranking latency from 500ms to 200ms. These results show that shared user stories can improve discovery while simplifying ranking systems. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_23702 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | TubiFM: Unified Item, Carousel, and Search Ranking for Streaming Discovery Salle, Alexandre Niu, Chenglei Mahapatra, Suchismit Chen, Xiaoxiao Sedhain, Suvash Wang, Yaqi Shahryari, Shervin Agrawal, Saurabh Chen, Qiang Tamir, Michael Information Retrieval Personalized discovery systems often train separate models for item ranking, carousel ranking, and search, even though these tasks expose complementary signals from the same viewer journey: watches shape carousel and item ranking, search queries reveal intent even when they do not lead to a catalog match, and watch history helps interpret search as rewatching, continuation, or new discovery. We introduce the user story, a serialized representation that turns a user's cross-surface history - attributes, sessions, watch events with surface and carousel context, and search events - into a single token sequence. By interleaving pretrained language tokens with domain-specific event tokens, user stories let heterogeneous recommendation and search tasks be expressed as prompted next-token prediction over a shared grammar. TubiFM is one instantiation of this approach: a Llama 3.2 1B-based model trained on user stories and prompted to rank items, carousels, or search results without task-specific architectures. In offline evaluation, this single model outperforms specialist baselines across item, carousel, and search ranking. In online A/B tests, TubiFM significantly improves search total viewing time (TVT) by $+3.9\%$ and carousel TVT by $+0.30\%$. Item ranking is statistically neutral on TVT ($+0.14\%$), but matches a mature production stack; across all three tasks, TubiFM serves on L40S GPUs and reduces p99 ranking latency from 500ms to 200ms. These results show that shared user stories can improve discovery while simplifying ranking systems. |
| title | TubiFM: Unified Item, Carousel, and Search Ranking for Streaming Discovery |
| topic | Information Retrieval |
| url | https://arxiv.org/abs/2605.23702 |