OneRank: Unified Transformer-Native Ranking Architecture for Multi-Task Recommendation
Summary
OneRank proposes a Transformer-native multi-task ranking framework that integrates feature encoding and prediction to reduce inter-task interference and improve ranking performance in recommender systems.
View Cached Full Text
Cached at: 06/16/26, 11:33 AM
Paper page - OneRank: Unified Transformer-Native Ranking Architecture for Multi-Task Recommendation
Source: https://huggingface.co/papers/2606.16838 Authors:
,
,
,
,
,
,
,
,
,
Abstract
OneRank presents a Transformer-native multi-task learning framework that integrates feature encoding and prediction to reduce inter-task interference and improve ranking performance in recommender systems.
Multi-task learning(MTL) is essential inrecommender systemsto enable complementary learning among diverse user feedback. While modern industrial practices have shifted from DNNs toTransformer-centric architectures to strengthen sequence modeling and scaling capacity, they still decouple feature encoding from multi-task prediction, treating theTransformeras a task-agnostic encoder. This design fundamentally limits the performance and scalability by (1) creating an information bottleneck under heterogeneous task objectives, (2) inducing gradient interference that leads to the seesaw phenomenon, and (3) forcing a dataflow transition in which attention-based, context-adaptive representation learning is converted to static feed-forward task prediction with incompatible information read-write dynamics. We propose OneRank, aTransformer-native multi-task ranking framework that eliminates encoder-predictor separation and introducestask-private channelsfor forward representation learning and backward optimization, enabling task-specialized learning while reducing inter-task interference. In the forward pass, OneRank learns task-specific representations bottom-up throughtask-conditioned information selection,candidate-aware contextualization, and controlledcross-task interaction. In the backward pass, cross-taskgradient detachmentisolates task-private parameter updates from shared knowledge extraction modules, preventingnegative transfer. We further replace static task-specific MLP scorers withdynamic matching-based scoringfor context-aware personalized ranking. By internalizing multi-task reasoning within theTransformerstack, OneRank establishes a unified and scalable architectural paradigm. Offline and online experiments on large-scale industrial datasets show that OneRank significantly outperforms state-of-the-art baselines while maintaining computational efficiency.
View arXiv pageView PDFAdd to collection
Get this paper in your agent:
hf papers read 2606\.16838
Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash
Models citing this paper0
No model linking this paper
Cite arxiv.org/abs/2606.16838 in a model README.md to link it from this page.
Datasets citing this paper0
No dataset linking this paper
Cite arxiv.org/abs/2606.16838 in a dataset README.md to link it from this page.
Spaces citing this paper0
No Space linking this paper
Cite arxiv.org/abs/2606.16838 in a Space README.md to link it from this page.
Collections including this paper0
No Collection including this paper
Add this paper to acollectionto link it from this page.
Similar Articles
Expand More, Shrink Less: Shaping Effective-Rank Dynamics for Dense Scaling in Recommendation
This paper proposes RankElastor, a novel architecture that mitigates embedding collapse in dense scaling of recommendation models by introducing parameterized full mixing and GLU-improved P-FFNs, achieving robust scaling and improved performance on large-scale datasets.
The Entropic Bound for Transformers: Why Static Rank Fails and Attention-Native Rank Recovers
This paper introduces the Entropic Bound, a spectral measure of task-intrinsic capacity for transformers, proving that the intrinsic rank of the token-mixing operator provides a tight lower bound on required model capacity. It shows that while a naive transfer from linear attention fails for real attention, an attention-native intrinsic rank restores the full theoretical structure.
Field-Aware RankMixer with Dual-Stream Bilinear Fusion for the Tencent UNI-REC Challenge
This paper presents FA-RankMixer, a model that combines field-aware semantic tokenization, target-aware multi-domain DIN, RankMixer blocks, and group-wise bilinear fusion for unified pCVR prediction in the Tencent UNI-REC Challenge. The solution ranked 9th on the official leaderboard.
Training transformers where every layer W = V·Uᵀ from initialization reveals a corpus-determined optimal rank - looking for arXiv endorser (cs.LG) [D]
This paper proposes Native Factorized Weights for transformers, where every linear layer is trained as a product of two low-rank matrices from initialization. Experiments show a corpus-determined optimal rank that minimizes validation loss and a generalization band, outperforming dense baselines with fewer parameters.
Task Decomposition-Guided Reranking for Adaptive Agent Skill Retrieval
Proposes SkillReranker, an inference-time reranking framework that decomposes tasks and skills to build a directed acyclic execution graph and uses a cross-encoder for scoring, improving task performance and efficiency in agent skill selection.