mlsanityData to 7 Oct 2026

Federated Reasoning Distillation Framework with Model Learnability-Aware Data Allocation

arXiv 2602.18749 ↗First posted 21 Feb 2026cs.AIcited by 0

On the map: LLM reasoning → Prompting and chain-of-thought reasoning

Abstract (arXiv)

Data allocation plays a critical role in federated large language model (LLM) and small language models (SLMs) reasoning collaboration. Nevertheless, existing data allocation methods fail to address an under-explored challenge in collaboration: bidirectional model learnability gap, where client-side SLMs cannot identify high-reward samples matching their learnability constraints for effective knowledge transfer from LLMs, while LLMs struggle to select samples contributing novel knowledge beyond their existing data. Furthermore, these collaboration frameworks face another key challenge: domain-agnostic reasoning transfer, where existing reasoning transfer methods fail to flexibly adapt to the local domain data, preventing SLMs from effectively acquiring step-by-step reasoning abilities within from general LLM. To address these challenges, we propose LaDa, a federated reasoning distillation framework with model learnability-aware data allocation. It introduces a model learnability-aware data filter that adaptively allocates high-reward samples based on the learnability gap between each SLM and LLM pair, effectively facilitating bidirectional knowledge transfer. We further design a domain adaptive reasoning distillation method that aligns joint probabilities of reasoning paths on filtered high-reward samples through contrastive distillation learning between SLM and LLM, enabling SLM to capture underlying reasoning patterns under local data distribution. LaDa operates as a plug-in module for existing collaboration frameworks, adapting knowledge transfer based on model learnability gaps.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Placed in Prompting and chain-of-thought reasoning, one of 17 directions in LLM reasoning. Builds on 17 of its 137 references that are in the map. First posted on arXiv 21 Feb 2026.

Builds on

17 of 137 references.

Closest ancestors · 1

Same direction, newest first

  1. 2025Small Models Struggle to Learn from Strong Reasoners17 Feb

Other references · 16

  1. 2015Distilling the Knowledge in a Neural Networkcited by 10,076
  2. 2023MAmmoTH: Building Math Generalist Models through Hybrid Instruction Tuningcited by 483
  3. 2018Approximation Methods for Bilevel Programmingcited by 231
  4. 2024Mutual Reasoning Makes Smaller LLMs Stronger Problem-Solverscited by 137
  5. 2020Neural Thompson Samplingcited by 127
  6. 2020Federated Knowledge Distillationcited by 51
  7. 2025Efficient Multivariate Time Series Forecasting via Calibrated Language Models with Privileged Knowledge Distillationcited by 20
  8. 2025Adaptive Batch-Wise Sample Scheduling for Direct Preference Optimizationcited by 16
  9. 2023Mutual Enhancement of Large and Small Language Models with Cross-Silo Knowledge Transfercited by 13
  10. 2025Contrastive Private Data Synthesis via Weighted Multi-PLM Fusioncited by 11
  11. 2024Neural Active Learning Beyond Banditscited by 10
  12. 2024Federated Co-tuning Framework for Large and Small Language Modelscited by 5
4 more
  1. 2025PPC-GPT: Federated Task-Specific Compression of Large Language Models via Pruning and Chain-of-Thought Distillationcited by 4
  2. 2024OFL-W3: A One-shot Federated Learning System on Web 3.0cited by 3
  3. 2025CE-LSLM: Efficient Large-Small Language Model Inference and Communication via Cloud-Edge Collaborationcited by 2
  4. 2024FedCoT: Federated Chain-of-Thought Distillation for Large Language Modelscited by 1

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2024Improving Mathematical Reasoning Capabilities of Small Language Models via Feedback-Driven Distillation
  2. 2026FERA: Uncertainty-Aware Federated Reasoning for Large Language Models
  3. 2025FedCoT: Communication-Efficient Federated Reasoning Enhancement for Large Language Models
  4. 2023Mixed Distillation Helps Smaller Language Model Better Reasoning
  5. 2026Offline Guidance, Online Reasoning: Reusing LLM Feedback for Small Language Models
  6. 2024Beyond Answers: Transferring Reasoning Capabilities to Smaller LLMs Using Multi-Teacher Knowledge Distillation
  7. 2024Unconstrained Model Merging for Enhanced LLM Reasoning
  8. 2023Effective Distillation of Table-based Reasoning Ability from LLMs