Pith. sign in

REVIEW 10 cited by

HYDRA: Model Factorization Framework for Black-Box LLM Personalization

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2406.02888 v3 pith:KNZWYCEW submitted 2024-06-05 cs.CL cs.AIcs.LG

classification cs.CLcs.AIcs.LG
keywords modeluser-specifichydrausersblack-boxcaptureknowledgepersonalization
verification ladder T0 review T1 audit T2 compute T3 formal

Signed reviews

No signed human review yet.

0 comments
read the original abstract

Personalization has emerged as a critical research area in modern intelligent systems, focusing on mining users' behavioral history and adapting to their preferences for delivering tailored experiences. Despite the remarkable few-shot capabilities exhibited by black-box large language models (LLMs), the inherent opacity of their model parameters presents significant challenges in aligning the generated output with individual expectations. Existing solutions have primarily focused on prompt design to incorporate user-specific profiles and behaviors; however, such approaches often struggle to generalize effectively due to their inability to capture shared knowledge among all users. To address these challenges, we propose HYDRA, a model factorization framework that captures both user-specific behavior patterns from historical data and shared general knowledge among all users to deliver personalized generation. In order to capture user-specific behavior patterns, we first train a reranker to prioritize the most useful information from top-retrieved relevant historical records. By combining the prioritized history with the corresponding query, we train an adapter to align the output with individual user-specific preferences, eliminating the reliance on access to inherent model parameters of black-box LLMs. Both the reranker and the adapter can be decomposed into a base model with multiple user-specific heads, resembling a hydra. The base model maintains shared knowledge across users, while the multiple personal heads capture user-specific preferences. Experimental results demonstrate that HYDRA outperforms existing state-of-the-art prompt-based methods by an average relative improvement of 9.01% across five diverse personalization tasks in the LaMP benchmark. Our implementation is available at https://github.com/night-chen/HYDRA.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 10 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Synthetic Interaction Data for Scalable Personalization in Large Language Models

    cs.LG 2026-02 conditional novelty 6.0 of 10

    PersonaGym simulates noisy multi-turn user–assistant interactions to build PersonaAtlas, and PPOpt learns to rewrite user prompts from interaction history, improving judged personalization on synthetic benchmarks.

  2. Aligning LLMs by Predicting Preferences from User Writing Samples

    cs.CL 2025-05 conditional novelty 6.0 of 10

    PROSE uses iterative refinement and cross-sample verification to infer personalized writing preferences from user demonstrations, outperforming CIPHER by 33% on the new PLUME benchmark.

  3. What Does Success Look Like? Catalyzing Meeting Intentionality with AI-Assisted Prospective Reflection

    cs.HC 2025-05 conditional novelty 6.0 of 10

    A study with 18 employees found that a generative AI Meeting Purpose Assistant can help people clarify meeting goals, anticipate challenges, and change how they prepare, with social and technical barriers to adoption.

  4. ComMer: a Framework for Compressing and Merging User Data for Personalization

    cs.CL 2025-01 unverdicted novelty 6.0 of 10

    ComMer compresses and merges user documents into compact inputs that improve personalized skill learning under tight inference budgets, at the cost of detail on knowledge-intensive tasks.

  5. AI PERSONA: Towards Life-long Personalization of LLMs

    cs.CL 2024-12 conditional novelty 6.0 of 10

    AI PERSONA introduces a training-free, prompt-based framework that maintains and updates per-user persona profiles, plus a synthetic benchmark on which it outperforms no-persona and RAG baselines, approaching the orac...

  6. On the Way to LLM Personalization: Learning to Remember User Conversations

    cs.CL 2024-11 conditional novelty 6.0 of 10

    Finetuning a LoRA adapter on self-generated question-answer pairs lets Llama 3 8B recall conversation topics with 81.5% accuracy, close to RAG at 83.5% but without retrieval.

  7. Adaptive Graph of Thoughts: Test-Time Adaptive Reasoning Unifying Chain, Tree, and Graph Structures

    cs.AI 2025-02 conditional novelty 5.0 of 10

    AGoT is a recursive graph-based prompting framework that decomposes LLM queries into nested subgraphs and reports large relative gains on some benchmarks, though headline GPQA gains rely on a shuffled subset.

  8. A Survey on Training-free Alignment of Large Language Models

    cs.CL 2025-08 conditional novelty 4.0 of 10

    A survey that catalogs and categorizes training-free LLM alignment methods into pre-decoding, in-decoding, and post-decoding, with a limited experimental comparison on one model.

  9. Embedding-to-Prefix: Parameter-Efficient Personalization for Pre-Trained Large Language Models

    cs.CL 2025-05 reject novelty 4.0 of 10

    E2P projects pre-computed user embeddings into a single soft prefix token for frozen LLMs, reporting gains on four personalization tasks, though its reproduction scripts write zero embeddings.

  10. Multi-Dimensional Insights: Benchmarking Real-World Personalization in Large Multimodal Models

    cs.AI 2024-12 conditional novelty 4.0 of 10

    The MDI benchmark evaluates large multimodal models on age-stratified, real-world multiple-choice questions and finds GPT-4o leading at about 79 percent average accuracy.

Pith tools