media MarkTechPost · 1d ago · releases

Yandex replaces 15+ stage recommendation cascade with single generative Sona model

from English

Yandex has introduced Sona, a generative AI model that consolidates candidate generation and ranking into a single system, replacing the traditional multi-stage recommendation cascades. The company tested this architecture in a seven-day live production experiment on its smart speakers.

Replaced more than 15 candidate generators, pre-ranking, and ranking stages with one served transformer.
Eliminates hand-engineered features by using only logged event fields and learned Semantic IDs.
Uses a frozen 0.6B-parameter Teacher Ranker for distillation during training, which is removed at serving time.
Achieved +4.53% Active Users, +6.30% Total Listening Time, and +11.42% Likes in online A/B tests.
Outperformed the previous Argus system with 2.35x the uplift on Active Users.

The model demonstrates that a single generative recommender can effectively replace complex cascades while improving user engagement metrics.

Importance 2/3 MarkTechPost Research paper Retrieval & RAG

Read original

arxiv arXiv cs.AI · 7d ago · 29 views

RareDx uses graph-grounded policy optimization to improve rare-disease diagnosis

The authors introduce RareDx, a system that couples controlled evidence use with knowledge-graph-grounded policy optimization to address the long-tail reasoning problem of rare-disease diagnosis. The training pipeline combines Top-10 post-training with RareDx-KGPO, which projects predictions into a canonical disease graph and integrates curated graded relevance, ontology proximity, biomedical similarity, and phenotype consistency.

arxiv arXiv cs.AI · 8d ago · 48 views

Highlight-Then-Summarize compresses evidence to improve long-context understanding

Researchers propose Highlight-Then-Summarize (H2S), a compress-then-reason paradigm that identifies question-relevant evidence and integrates it into a compact summary before generating an answer. The team constructs H2S-Dataset with 6,647 examples and introduces H2S-RL to provide process-level rewards for evidence selection.

arxiv arXiv cs.CL · 8d ago · 36 views

Highlight-Then-Summarize compresses evidence for better long-context understanding

Researchers propose Highlight-Then-Summarize (H2S), a compress-then-reason paradigm that identifies question-relevant evidence and integrates it into a compact summary before generating an answer. To support this approach, the team constructed H2S-Dataset with 6,647 examples from 11 benchmark families and introduced H2S-RL for process-level rewards.

media Hugging Face Forums · 9d ago · 37 views

Biological JEPA models disease progression using biological constraints

A new open-source project called Biological JEJA combines JEPA-style representation learning with biological constraints to model how diseases progress over time. The approach was tested on real ADNI data from 2,347 patients and compared against several other models.

arxiv arXiv cs.CL · 12d ago · 34 views

R3Con applies relevance realization principles to improve large-scale reasoning

Researchers propose R3Con, a harness that operationalizes cognitive theory principles for constructing effective representations of very large contexts. The system addresses the challenge of assembling interdependent information scattered across vast sources beyond standard model context limits.