mlsanityData to 14 Oct 2026

Large Language Model's Multi-Capability Alignment in Biomedical Domain

arXiv 2508.04278 ↗First posted 6 Aug 2025cs.AIcited by 0

Not on the map. Why?

Abstract (arXiv)

BalancedBio is a theoretically grounded framework for parameter-efficient biomedical reasoning, addressing multi-capability integration in domain-specific AI alignment. It establishes the Biomedical Multi-Capability Convergence Theorem, proving orthogonal gradient spaces are essential to prevent capability interference for safe deployment. Key innovations include: (1) Medical Knowledge Grounded Synthetic Generation (MKGSG), extending Source2Synth with clinical workflow constraints and medical ontology validation for factual accuracy and safety; and (2) Capability Aware Group Relative Policy Optimization, deriving optimal hybrid reward weighting to maintain orthogonality in RL, using a reward model with rule-based and model-based scores adapted to biomedical tasks. Mathematical analysis proves Pareto-optimal convergence, preserving performance across capabilities. It achieves state-of-the-art results in its parameter class: domain expertise (80.95% BIOMED-MMLU, +15.32% over baseline), reasoning (61.94%, +7.75%), instruction following (67.95%, +6.44%), and integration (86.7%, +18.5%). Theoretical safety guarantees include bounds on capability preservation and clinical accuracy. Real-world deployment yields 78% cost reduction, 23% improved diagnostic accuracy, and 89% clinician acceptance. This work provides a principled methodology for biomedical AI alignment, enabling efficient reasoning with essential safety and reliability, with the 0.5B model version to be released.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 22 of its 32 references that are in the map. First posted on arXiv 6 Aug 2025.

Builds on

22 of 32 references.

References · 22

  1. 2020Language Models are Few-Shot Learnerscited by 32,419
  2. 2022Training language models to follow instructions with human feedbackcited by 15,971
  3. 2019Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformercited by 14,599
  4. 2021LoRA: Low-Rank Adaptation of Large Language Modelscited by 14,380
  5. 2023Llama 2: Open Foundation and Fine-Tuned Chat Modelscited by 13,028
  6. 2024DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Modelscited by 9,379
  7. 2020Retrieval-Augmented Generation for Knowledge-Intensive NLP Taskscited by 9,297
  8. 2022ReAct: Synergizing Reasoning and Acting in Language Modelscited by 8,695
  9. 2020Measuring Massive Multitask Language Understandingcited by 7,963
  10. 2020Scaling Laws for Neural Language Modelscited by 6,586
  11. 2022Self-Consistency Improves Chain of Thought Reasoning in Language Modelscited by 5,708
  12. 2019BioBERT: a pre-trained biomedical language representation model for biomedical text miningcited by 1,947
10 more
  1. 2022Large Language Models Encode Clinical Knowledgecited by 1,830
  2. 2021WebGPT: Browser-assisted question-answering with human feedbackcited by 1,745
  3. 2019PubMedQA: A Dataset for Biomedical Research Question Answeringcited by 1,288
  4. 2020Domain-Specific Language Model Pretraining for Biomedical Natural Language Processingcited by 1,009
  5. 2019Publicly Available Clinical BERT Embeddingscited by 1,002
  6. 2022BioGPT: Generative Pre-trained Transformer for Biomedical Text Generation and Miningcited by 538
  7. 2022One Embedder, Any Task: Instruction-Finetuned Text Embeddingscited by 333
  8. 2023ChatDoctor: A Medical Chat Model Fine-Tuned on a Large Language Model Meta-AI (LLaMA) Using Medical Domain Knowledgecited by 325
  9. 2023Benchmarking Large Language Models on CMExam -- A Comprehensive Chinese Medical Exam Datasetcited by 91
  10. 2024Source2Synth: Synthetic Data Generation and Curation Grounded in Real Data Sourcescited by 28

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2025Aligning LLMs with Biomedical Knowledge using Balanced Fine-Tuning
  2. 2024MedCare: Advancing Medical LLMs through Decoupling Clinical Alignment and Knowledge Aggregation
  3. 2026Improving the Safety and Trustworthiness of Medical AI via Multi-Agent Evaluation Loops
  4. 2025MedBioLM: Optimizing Medical and Biological QA with Fine-Tuned Large Language Models and Retrieval-Augmented Generation
  5. 2025The Forgotten Shield: Safety Grafting in Parameter-Space for Medical MLLMs
  6. 2026Neurosymbolic Alignment for Physiologically-Safe Clinical Language Models
  7. 2025Black-Box Behavioral Distillation Breaks Safety Alignment in Medical LLMs
  8. 2023MedAgents: Large Language Models as Collaborators for Zero-shot Medical Reasoning