Challenge: Existing metrics fail to distinguish mandatory behaviors required for task success from non-mandatory patterns that reflect a model’s autonomous preferences.
Approach: They propose to use response pattern similarity and action graph similarity to isolate non-mandatory behaviors from mandatory behaviors.
Outcome: Evaluating 18 models from 8 providers on -Bench and 2-Bench against Claude Sonnet 4.5, the authors find that within-family model pairs score 5.9 pp higher in response pattern similarity and action graph similarity .

Similar Papers

Quantification of Large Language Model Distillation (2025.acl-long)

Copied to clipboard

Challenge: Existing studies have revealed the robustness degra-dation caused by data distillation.
Approach: They propose a framework to evaluate and quantify model distillation . they aim to identify identity cognition contradictions and analyse multi-granularity response similarities across models to measure the extent of homogenization.
Outcome: The proposed framework addresses two key aspects: (1) Identifying identity cognition contradictions to assess discrepancies in how models perceive and represent identity-related information; (2) Analyzing multi-granularity response similarities across models to measure the extent of homogenization.
Do LLM Agents Really Mimic Humans? Diagnosing and Aligning Microeconomic Behaviors in Macro-ABMs (2026.findings-acl)

Copied to clipboard

Challenge: Existing studies focus on replicating macro-level stylized facts while neglecting verification of micro-level decision-making.
Approach: They propose a framework that replicates macro-level stylized facts while ignoring micro-level decision-making.
Outcome: The proposed framework improves alignment with human trends and captures behavioral heterogeneity.
Self-Distillation Bridges Distribution Gap in Language Model Fine-Tuning (2024.acl-long)

Copied to clipboard

Challenge: Experimental results show that fine-tuning of large language models for specific tasks can be challenging . distribution shift during fine-timing can lead to performance degradation in general task capabilities .
Approach: They propose a new approach that bridges the distribution gap between task datasets and LLMs by guiding fine-tuning with a distilled dataset generated by the model itself.
Outcome: The proposed approach achieves comparable or superior performance on downstream tasks compared to the vanilla approach.
I2D2: Inductive Knowledge Distillation with NeuroLogic and Self-Imitation (2023.acl-long)

Copied to clipboard

Challenge: Empirical results suggest that scale is not the only way to build commonsense capabilities.
Approach: They propose a commonsense distillation framework that can achieve a competitive level of commonsensing without relying on the benefits of scale.
Outcome: The proposed framework breaks the dependence on the extreme-scale teacher model with two innovations: (1) the novel adaptation of NeuroLogic Decoding to enhance the generation quality of the weak, off-the-shelf language models, and (2) self-imitation learning to iteratively learn from the model’s own enhanced commonsense acquisition capabilities.
PrefIx: Understand and Adapt to User Preference in Human-Agent Interaction (2026.findings-acl)

Copied to clipboard

Challenge: Current benchmarks evaluate task accuracy but overlook how agents interact . Preference-aware agents show 7.6% average UX improvement and 18.5% gain in preference alignment.
Approach: They propose a configurable environment that evaluates both what agents accomplish and how they interact.
Outcome: The proposed model improves performance and improves user experience by 7.6% and 18.5% respectively.
Beyond Logits: Aligning Feature Dynamics for Effective Knowledge Distillation (2025.acl-long)

Copied to clipboard

Challenge: Knowledge distillation (KD) compresses large language models into lightweight versions called student models.
Approach: They propose to align the entire feature dynamics between teacher and student models by using two additional loss terms to achieve this.
Outcome: The proposed method matches the entire feature dynamics between teacher and student models rather than just the final states.
Aligning LLMs with Individual Preferences via Interaction (2025.coling-main)

Copied to clipboard

Challenge: Existing studies on LLMs alignment focus on generalizing their behavior to generalized values such as helpfulness, harmlessness, and honesty.
Approach: They train large language models to "interact to align" to implicitly infer user preferences . they use a multi-turn preference dataset to generate a personalized alignment .
Outcome: The proposed method enables dynamic, personalized alignment via interaction with a multi-turn preference dataset.
Can Compact Language Models Search Like Agents? Distillation-Guided Policy Optimization for Preserving Agentic RAG Capabilities (2026.acl-long)

Copied to clipboard

Challenge: Existing agentic RAG systems rely on large language models with billions of parameters.
Approach: They propose a method to elicit agentic RAG behaviors from compact models . they propose ARC, which uses cold-start initialization and teacher guidance .
Outcome: The proposed method outperforms the larger teacher model in some cases.
SmartAD: Capacity-Aligned Agent Distillation for Small Language Models (2026.findings-acl)

Copied to clipboard

Challenge: Large language models (LLMs) show strong reasoning and decision-making ability, but their high inference cost motivates transferring agentic skills to small language models.
Approach: They propose a capacity-aligned agent distillation framework that trains SLMs on full reason–act–observe trajectories from a tool-using teacher.
Outcome: The proposed framework outperforms all baselines on multi-hop QA and math benchmarks with 1.5B and 3B models.
Causal Distillation for Language Models (2022.naacl-main)

Copied to clipboard

Challenge: Distillation efforts have led to language models that are more compact and efficient without serious drops in performance.
Approach: They propose to augment distillation with a third objective that encourages the student model to imitate the causal dynamics of the teacher through a distillation interchange intervention training objective (DIITO).
Outcome: The proposed method lowers perplexity on the WikiText-103M corpus and improves on the GLUE benchmark, SQuAD, and CoNLL-2003.

What is GenGO?

GenGO is an NLP powered publication search system. It currenctly indexes 30k+ papers from ACL Anthology, and implements multi-aspect summarization, semantic search, and more!

Information

About
Limitations