Tootfinder

Dual-Actor Fine-Tuning of VLA Models: A Talk-and-Tweak Human-in-the-Loop Approach
Piaopiao Jin, Qi Wang, Guokang Sun, Ziwen Cai, Pinjia He, Yangwei You
https://arxiv.org/abs/2509.13774

Dual-Actor Fine-Tuning of VLA Models: A Talk-and-Tweak Human-in-the-Loop Approach
Vision-language-action (VLA) models demonstrate strong generalization in robotic manipulation but face challenges in complex, real-world tasks. While supervised fine-tuning with demonstrations is constrained by data quality, reinforcement learning (RL) offers a promising alternative. We propose a human-in-the-loop dual-actor fine-tuning framework grounded in RL. The framework integrates a primary actor for robust multi-task performance with a refinement actor for latent-space adaptation. Beyond…

Latent Traits and Cross-Task Transfer: Deconstructing Dataset Interactions in LLM Fine-tuning
Shambhavi Krishna, Atharva Naik, Chaitali Agarwal, Sudharshan Govindan, Taesung Lee, Haw-Shiuan Chang
https://arxiv.org/abs/2509.13624

Latent Traits and Cross-Task Transfer: Deconstructing Dataset Interactions in LLM Fine-tuning
Large language models are increasingly deployed across diverse applications. This often includes tasks LLMs have not encountered during training. This implies that enumerating and obtaining the high-quality training data for all tasks is infeasible. Thus, we often need to rely on transfer learning using datasets with different characteristics, and anticipate out-of-distribution requests. Motivated by this practical need, we propose an analysis framework, building a transfer learning matrix and …

Pharmacist: Safety Alignment Data Curation for Large Language Models against Harmful Fine-tuning
Guozhi Liu, Qi Mu, Tiansheng Huang, Xinhua Wang, Li Shen, Weiwei Lin, Zhang Li
https://arxiv.org/abs/2510.10085

Pharmacist: Safety Alignment Data Curation for Large Language Models against Harmful Fine-tuning
Harmful fine-tuning issues present significant safety challenges for fine-tuning-as-a-service in large language models. Existing alignment-stage defenses, e.g., Vaccine, Repnoise, Booster, and T-Vaccine, mitigate harmful fine-tuning issues by enhancing the model's robustness during the alignment phase. While these methods have been proposed to mitigate the issue, they often overlook a critical upstream factor: the role of the original safety-alignment data. We observe that their defense perform…

Evolution of meta's llama models and parameter-efficient fine-tuning of large language models: a survey
Abdulhady Abas Abdullah, Arkaitz Zubiaga, Seyedali Mirjalili, Amir H. Gandomi, Fatemeh Daneshfar, Mohammadsadra Amini, Alan Salam Mohammed, Hadi Veisi
https://arxiv.org/abs/2510.12178

Evolution of meta's llama models and parameter-efficient fine-tuning of large language models: a survey
This review surveys the rapid evolution of Meta AI's LLaMA (Large Language Model Meta AI) series - from LLaMA 1 through LLaMA 4 and the specialized parameter-efficient fine-tuning (PEFT) methods developed for these models. We first describe the LLaMA family of foundation models (7B-65B to 288B parameters), their architectures (including native multimodal and Mixtureof-Experts variants), and key performance characteristics. We then describe and discuss the concept of PEFT, which adapts large pre…

Language models' activations linearly encode training-order recency
Dmitrii Krasheninnikov, Richard E. Turner, David Krueger
https://arxiv.org/abs/2509.14223 https://…

Language models' activations linearly encode training-order recency
We show that language models' activations linearly encode when information was learned during training. Our setup involves creating a model with a known training order by sequentially fine-tuning Llama-3.2-1B on six disjoint but otherwise similar datasets about named entities. We find that the average activations of test samples for the six training datasets encode the training order: when projected into a 2D subspace, these centroids are arranged exactly in the order of training and lie on a s…

Personalized Federated Fine-Tuning of Vision Foundation Models for Healthcare
Adam Tupper, Christian Gagn\'e
https://arxiv.org/abs/2510.12741 https://a…

Personalized Federated Fine-Tuning of Vision Foundation Models for Healthcare
Foundation models open up new possibilities for the use of AI in healthcare. However, even when pre-trained on health data, they still need to be fine-tuned for specific downstream tasks. Furthermore, although foundation models reduce the amount of training data required to achieve good performance, obtaining sufficient data is still a challenge. This is due, in part, to restrictions on sharing and aggregating data from different sources to protect patients' privacy. One possible solution to th…

FedLoDrop: Federated LoRA with Dropout for Generalized LLM Fine-tuning
Sijing Xie, Dingzhu Wen, Changsheng You, Qimei Chen, Mehdi Bennis, Kaibin Huang
https://arxiv.org/abs/2510.12078

FedLoDrop: Federated LoRA with Dropout for Generalized LLM Fine-tuning
Fine-tuning (FT) large language models (LLMs) is crucial for adapting general-purpose models to specific tasks, enhancing accuracy and relevance with minimal resources. To further enhance generalization ability while reducing training costs, this paper proposes Federated LoRA with Dropout (FedLoDrop), a new framework that applies dropout to the rows and columns of the trainable matrix in Federated LoRA. A generalization error bound and convergence analysis under sparsity regularization are obta…

NIRVANA: Structured pruning reimagined for large language models compression
Mengting Ai, Tianxin Wei, Sirui Chen, Jingrui He
https://arxiv.org/abs/2509.14230 https://

NIRVANA: Structured pruning reimagined for large language models compression
Structured pruning of large language models (LLMs) offers substantial efficiency improvements by removing entire hidden units, yet current approaches often suffer from significant performance degradation, particularly in zero-shot settings, and necessitate costly recovery techniques such as supervised fine-tuning (SFT) or adapter insertion. To address these critical shortcomings, we introduce NIRVANA, a novel pruning method explicitly designed to balance immediate zero-shot accuracy preservatio…

Mira Murati's Thinking Machines Lab makes Tinker, its API for fine-tuning language models, generally available, adds support for Kimi K2 Thinking, and more (Thinking Machines Lab)
https://thinkingmachines.ai/blog/tinker-general-availability/

Tinker: General Availability and Vision Input
Today we are announcing four updates to Tinker: No more waitlist New reasoning model: Kimi K2 Thinking New inference interface that is compatible with the OpenAI API Vision input support with Qwen3-VL General availability The waitlist is over! Everybody can use Tinker now; sign up here to get started. See the Tinker homepage for available models and pricing, and check out the Tinker cookbook for code examples. More reasoning with Kimi K2 Thinking Users can now fine-tune Kimi K2 Thinking on Ti…

Improving Context Fidelity via Native Retrieval-Augmented Reasoning
Suyuchen Wang, Jinlin Wang, Xinyu Wang, Shiqi Li, Xiangru Tang, Sirui Hong, Xiao-Wen Chang, Chenglin Wu, Bang Liu
https://arxiv.org/abs/2509.13683

Improving Context Fidelity via Native Retrieval-Augmented Reasoning
Large language models (LLMs) often struggle with context fidelity, producing inconsistent answers when responding to questions based on provided information. Existing approaches either rely on expensive supervised fine-tuning to generate evidence post-answer or train models to perform web searches without necessarily improving utilization of the given context. We propose CARE, a novel native retrieval-augmented reasoning framework that teaches LLMs to explicitly integrate in-context evidence wi…

Towards Rationale-Answer Alignment of LVLMs via Self-Rationale Calibration
Yuanchen Wu, Ke Yan, Shouhong Ding, Ziyin Zhou, Xiaoqiang Li
https://arxiv.org/abs/2509.13919 https://…

Towards Rationale-Answer Alignment of LVLMs via Self-Rationale Calibration
Large Vision-Language Models (LVLMs) have manifested strong visual question answering capability. However, they still struggle with aligning the rationale and the generated answer, leading to inconsistent reasoning and incorrect responses. To this end, this paper introduces the Self-Rationale Calibration (SRC) framework to iteratively calibrate the alignment between rationales and answers. SRC begins by employing a lightweight "rationale fine-tuning" approach, which modifies the model's respons…

TICL: Text-Embedding KNN For Speech In-Context Learning Unlocks Speech Recognition Abilities of Large Multimodal Models
Haolong Zheng, Yekaterina Yegorova, Mark Hasegawa-Johnson
https://arxiv.org/abs/2509.13395

TICL: Text-Embedding KNN For Speech In-Context Learning Unlocks Speech Recognition Abilities of Large Multimodal Models
Speech foundation models have recently demonstrated the ability to perform Speech In-Context Learning (SICL). Selecting effective in-context examples is crucial for SICL performance, yet selection methodologies remain underexplored. In this work, we propose Text-Embedding KNN for SICL (TICL), a simple pipeline that uses semantic context to enhance off-the-shelf large multimodal models' speech recognition ability without fine-tuning. Across challenging automatic speech recognition tasks, includi…

Predicting Crystal Structures and Ionic Conductivity in Li$_{3}$YCl$_{6-x}$Br$_{x}$ Halide Solid Electrolytes Using a Fine-Tuned Machine Learning Interatomic Potential
Jonas B\"ohm, Aur\'elie Champagne
https://arxiv.org/abs/2510.09861

Predicting Crystal Structures and Ionic Conductivity in Li$_{3}$YCl$_{6-x}$Br$_{x}$ Halide Solid Electrolytes Using a Fine-Tuned Machine Learning Interatomic Potential
This work demonstrates the effectiveness of fine-tuning the CHGNet universal machine learning interatomic potential (uMLIP) to investigate ionic transport mechanisms in ternary halide solid electrolytes of the Li$_{3}$YCl$_{6-x}$Br$_{x}$ family (x = 0 to 6), which are promising candidates for solid-state battery applications. We present a strategy for generating ordered structural models from experimentally derived disordered Li$_{3}$YCl$_{6}$ (LYC) and Li$_{3}$YBr$_{6}$ (LYB) structures. These…

Canary-1B-v2 & Parakeet-TDT-0.6B-v3: Efficient and High-Performance Models for Multilingual ASR and AST
Monica Sekoyan, Nithin Rao Koluguri, Nune Tadevosyan, Piotr Zelasko, Travis Bartley, Nick Karpov, Jagadeesh Balam, Boris Ginsburg
https://arxiv.org/abs/2509.14128

Canary-1B-v2 & Parakeet-TDT-0.6B-v3: Efficient and High-Performance Models for Multilingual ASR and AST
This report introduces Canary-1B-v2, a fast, robust multilingual model for Automatic Speech Recognition (ASR) and Speech-to-Text Translation (AST). Built with a FastConformer encoder and Transformer decoder, it supports 25 languages primarily European. The model was trained on 1.7M hours of total data samples, including Granary and NeMo ASR Set 3.0, with non-speech audio added to reduce hallucinations for ASR and AST. We describe its two-stage pre-training and fine-tuning process with dynamic d…

InstructPLM-mu: 1-Hour Fine-Tuning of ESM2 Beats ESM3 in Protein Mutation Predictions
Junde Xu, Yapin Shi, Lijun Lang, Taoyong Cui, Zhiming Zhang, Guangyong Chen, Jiezhong Qiu, Pheng-Ann Heng
https://arxiv.org/abs/2510.03370

InstructPLM-mu: 1-Hour Fine-Tuning of ESM2 Beats ESM3 in Protein Mutation Predictions
Multimodal protein language models deliver strong performance on mutation-effect prediction, but training such models from scratch demands substantial computational resources. In this paper, we propose a fine-tuning framework called InstructPLM-mu and try to answer a question: \textit{Can multimodal fine-tuning of a pretrained, sequence-only protein language model match the performance of models trained end-to-end? } Surprisingly, our experiments show that fine-tuning ESM2 with structural input…

Data-Model Co-Evolution: Growing Test Sets to Refine LLM Behavior
Minjae Lee, Minsuk Kahng
https://arxiv.org/abs/2510.12728 https://arxiv.org/pdf/2510.1272…

Data-Model Co-Evolution: Growing Test Sets to Refine LLM Behavior
A long-standing challenge in machine learning has been the rigid separation between data work and model refinement, enforced by slow fine-tuning cycles. The rise of Large Language Models (LLMs) overcomes this historical barrier, allowing applications developers to instantly govern model behavior by editing prompt instructions. This shift enables a new paradigm: data-model co-evolution, where a living test set and a model's instructions evolve in tandem. We operationalize this paradigm in an int…

TIT: A Tree-Structured Instruction Tuning Approach for LLM-Based Code Translation
He Jiang, Yufu Wang, Hao Lin, Peiyu Zou, Zhide Zhou, Ang Jia, Xiaochen Li, Zhilei Ren
https://arxiv.org/abs/2510.09400 …

TIT: A Tree-Structured Instruction Tuning Approach for LLM-Based Code Translation
Large Language Models (LLMs) have shown strong performance in automated source-to-target code translation through pretraining on extensive code corpora. However, mainstream LLM-based code translation methods suffer from two critical limitations. First, they are highly sensitive to language-specific features, which often introduce source-language syntax or lexicon into the output, leading to syntactic confusion. Second, they lack fine-grained semantic alignment due to an over-reliance on functio…

Stacked Regression using Off-the-shelf, Stimulus-tuned and Fine-tuned Neural Networks for Predicting fMRI Brain Responses to Movies (Algonauts 2025 Report)
Robert Scholz, Kunal Bagga, Christine Ahrends, Carlo Alberto Barbano
https://arxiv.org/abs/2510.06235

Stacked Regression using Off-the-shelf, Stimulus-tuned and Fine-tuned Neural Networks for Predicting fMRI Brain Responses to Movies (Algonauts 2025 Report)
We present our submission to the Algonauts 2025 Challenge, where the goal is to predict fMRI brain responses to movie stimuli. Our approach integrates multimodal representations from large language models, video encoders, audio models, and vision-language models, combining both off-the-shelf and fine-tuned variants. To improve performance, we enhanced textual inputs with detailed transcripts and summaries, and we explored stimulus-tuning and fine-tuning strategies for language and vision models…

Leveraging Language Semantics for Collaborative Filtering with TextGCN and TextGCN-MLP: Zero-Shot vs In-Domain Performance
Andrei Chernov, Haroon Wahab, Oleg Novitskij
https://arxiv.org/abs/2510.12461 …

Leveraging Language Semantics for Collaborative Filtering with TextGCN and TextGCN-MLP: Zero-Shot vs In-Domain Performance
In recent years, various approaches have been proposed to leverage large language models (LLMs) for incorporating textual information about items into recommender systems. Existing methods primarily focus on either fine-tuning LLMs to generate recommendations or integrating LLM-based embeddings into downstream models. In this work, we follow the latter direction and propose \textbf{TextGCN}, which applies parameter-free graph convolution layers directly over LLM-based item-title embeddings, ins…

A Solution to the Hierarchy Problem with Non-Linear Quantum Mechanics
David E. Kaplan, Surjeet Rajendran
https://arxiv.org/abs/2510.12030 https://arxiv.org…

A Solution to the Hierarchy Problem with Non-Linear Quantum Mechanics
We argue that the hierarchy problem of the standard model of particle physics can be solved by adding a state-dependent term to the Higgs sector. We present an example of a scalar field with a Higgs-like potential with an additional term proportional to the expectation value of the squared Higgs field operator. We show that the mass can be parametrically lighter than the theory's energy-momentum cutoff without fine tuning. We find the Higgs mass can be technically natural, even with a Planck-sc…

MeTA-LoRA: Data-Efficient Multi-Task Fine-Tuning for Large Language Models
Bo Cheng, Xu Wang, Jinda Liu, Yi Chang, Yuan Wu
https://arxiv.org/abs/2510.11598 https://

MeTA-LoRA: Data-Efficient Multi-Task Fine-Tuning for Large Language Models
Low-Rank Adaptation (LoRA) has emerged as one of the most widely used parameter-efficient fine-tuning (PEFT) methods for adapting large language models (LLMs) to downstream tasks. While highly effective in single-task settings, it struggles to efficiently leverage inter-task knowledge in complex multi-task learning scenarios, often requiring substantial task-specific data to achieve optimal performance. To address this limitation, we introduce MeTA-LoRA, a two-stage optimization framework that …

Ein Artikel in Plos One beleuchtet den politischen #Bias von #Chatbots und zeigt, dass viele KI-Modelle eher linksgerichtete Antworten geben.
Dies könnte auf das Supervised Fine-Tuning zurückzuführen sein, bei dem KI durch menschliche Beispiele lernt. Interessanterweise weisen unterschiedliche

Linksgerichtete Chatbots? Eine Analyse der politischen Ausrichtung von LLMs
Politischer Bias in ChatGPT und Co.: Gibt es eine politische Ausrichtung in LLMs und wodurch könnte sie entstehen?

Knowledge-Decoupled Functionally Invariant Path with Synthetic Personal Data for Personalized ASR
Yue Gu, Zhihao Du, Ying Shi, Jiqing Han, Yongjun He
https://arxiv.org/abs/2510.10401

Knowledge-Decoupled Functionally Invariant Path with Synthetic Personal Data for Personalized ASR
Fine-tuning generic ASR models with large-scale synthetic personal data can enhance the personalization of ASR models, but it introduces challenges in adapting to synthetic personal data without forgetting real knowledge, and in adapting to personal data without forgetting generic knowledge. Considering that the functionally invariant path (FIP) framework enables model adaptation while preserving prior knowledge, in this letter, we introduce FIP into synthetic-data-augmented personalized ASR mo…

Efficient In-Memory Acceleration of Sparse Block Diagonal LLMs
Jo\~ao Paulo Cardoso de Lima, Marc Dietrich, Jeronimo Castrillon, Asif Ali Khan
https://arxiv.org/abs/2510.11192 h…

Efficient In-Memory Acceleration of Sparse Block Diagonal LLMs
Structured sparsity enables deploying large language models (LLMs) on resource-constrained systems. Approaches like dense-to-sparse fine-tuning are particularly compelling, achieving remarkable structured sparsity by reducing the model size by over 6.7x, while still maintaining acceptable accuracy. Despite this reduction, LLM inference, especially the decode stage being inherently memory-bound, is extremely expensive on conventional Von-Neumann architectures. Compute-in-memory (CIM) architectur…

GrASP: A Generalizable Address-based Semantic Prefetcher for Scalable Transactional and Analytical Workloads
Farzaneh Zirak, Farhana Choudhury, Renata Borovica-Gajic
https://arxiv.org/abs/2510.11011

GrASP: A Generalizable Address-based Semantic Prefetcher for Scalable Transactional and Analytical Workloads
Data prefetching--loading data into the cache before it is requested--is essential for reducing I/O overhead and improving database performance. While traditional prefetchers focus on sequential patterns, recent learning-based approaches, especially those leveraging data semantics, achieve higher accuracy for complex access patterns. However, these methods often struggle with today's dynamic, ever-growing datasets and require frequent, timely fine-tuning. Privacy constraints may also restrict a…

Attack via Overfitting: 10-shot Benign Fine-tuning to Jailbreak LLMs
Zhixin Xie, Xurui Song, Jun Luo
https://arxiv.org/abs/2510.02833 https://arxiv.org/pdf…

Attack via Overfitting: 10-shot Benign Fine-tuning to Jailbreak LLMs
Despite substantial efforts in safety alignment, recent research indicates that Large Language Models (LLMs) remain highly susceptible to jailbreak attacks. Among these attacks, finetuning-based ones that compromise LLMs' safety alignment via fine-tuning stand out due to its stable jailbreak performance. In particular, a recent study indicates that fine-tuning with as few as 10 harmful question-answer (QA) pairs can lead to successful jailbreaking across various harmful questions. However, such…

Early Detection and Reduction of Memorisation for Domain Adaptation and Instruction Tuning
Dean L. Slack, Noura Al Moubayed
https://arxiv.org/abs/2510.11372 https://

Early Detection and Reduction of Memorisation for Domain Adaptation and Instruction Tuning
Although large language models excel across many tasks, they can memorise training data and thereby expose private or copyrighted text. Most defences target the pre-training stage, leaving memorisation during fine-tuning, especially for domain adaptation and instruction tuning, poorly understood. We fine-tune Pythia, Llama3, and Mistral models spanning 1.4B-70B parameters on common evaluation datasets and track verbatim memorisation throughout training. We find that memorisation increases drama…

Mira Murati's Thinking Machines Lab launches its first product, Tinker, which automates the creation of custom frontier AI models (Will Knight/Wired)
https://www.wired.com/story/thinking-machines-lab-first-product-fine-tune/

Exclusive: Mira Murati’s Stealth AI Lab Launches Its First Product
Thinking Machines Lab, led by a group of prominent former OpenAI researchers, is betting that fine tuning cutting-edge models will be the next frontier in AI.

Flow of Knowledge: Federated Fine-Tuning of LLMs in Healthcare under Non-IID Conditions
Zeyu Chen, Yun Ji, Bowen Wang, Liwen Shi, Zijie Zeng, Sheng Zhang
https://arxiv.org/abs/2510.00543

Flow of Knowledge: Federated Fine-Tuning of LLMs in Healthcare under Non-IID Conditions
Large language models (LLMs) show great promise in healthcare, but their applications are hindered by data privacy restrictions and the challenges of cross-institution collaboration. Sensitive medical data cannot be centralized, while non-independent and identically distributed (non-IID) characteristics across institutions further com- plicate convergence and fairness. To address these issues, we present a federated fine-tuning approach based on Low-Rank Adaptation (LoRA), enabling privacy-pres…

How to Teach Large Multimodal Models New Skills
Zhen Zhu, Yiming Gong, Yao Xiao, Yaoyao Liu, Derek Hoiem
https://arxiv.org/abs/2510.08564 https://arxiv.org…

How to Teach Large Multimodal Models New Skills
How can we teach large multimodal models (LMMs) new skills without erasing prior abilities? We study sequential fine-tuning on five target skills while monitoring general ability on eight held-out benchmarks across three model families. We observe that apparent "forgetting" on held-out tasks after narrow fine-tuning can partly recover at later stages. We trace this behavior to a measurable shift in the output token distribution, manifested through a simple counting-bias probe that co-varies wit…

Learning from Convenience Samples: A Case Study on Fine-Tuning LLMs for Survey Non-response in the German Longitudinal Election Study
Tobias Holtdirk, Dennis Assenmacher, Arnim Bleier, Claudia Wagner
https://arxiv.org/abs/2509.25063

Learning from Convenience Samples: A Case Study on Fine-Tuning LLMs for Survey Non-response in the German Longitudinal Election Study
Survey researchers face two key challenges: the rising costs of probability samples and missing data (e.g., non-response or attrition), which can undermine inference and increase the use of convenience samples. Recent work explores using large language models (LLMs) to simulate respondents via persona-based prompts, often without labeled data. We study a more practical setting where partial survey responses exist: we fine-tune LLMs on available data to impute self-reported vote choice under bot…

Self-supervised diffusion model fine-tuning for costate initialization using Markov chain Monte Carlo
Jannik Graebner, Ryne Beeson
https://arxiv.org/abs/2510.02527 https://

Self-supervised diffusion model fine-tuning for costate initialization using Markov chain Monte Carlo
Global search and optimization of long-duration, low-thrust spacecraft trajectories with the indirect method is challenging due to a complex solution space and the difficulty of generating good initial guesses for the costate variables. This is particularly true in multibody environments. Given data that reveals a partial Pareto optimal front, it is desirable to find a flexible manner in which the Pareto front can be completed and fronts for related trajectory problems can be found. In this wor…

Understanding the Effects of Domain Finetuning on LLMs
Eshaan Tanwar, Deepak Nathani, William Yang Wang, Tanmoy Chakraborty
https://arxiv.org/abs/2510.09359 https://

Understanding the Effects of Domain Finetuning on LLMs
Large Language Models (LLMs) fine-tuned for specific domains exhibit strong performance; however, the underlying mechanisms by which this fine-tuning reshapes their parametric space are not well understood. Prior works primarily focus on auto-regressive or general-purpose instruct models, leaving domain-specialised LLMs under-explored. We present the first systematic study of domain-specific fine-tuning in large medical language models. Our analysis reveals that fine-tuning modifies only a smal…

CoIRL-AD: Collaborative-Competitive Imitation-Reinforcement Learning in Latent World Models for Autonomous Driving
Xiaoji Zheng, Ziyuan Yang, Yanhao Chen, Yuhang Peng, Yuanrong Tang, Gengyuan Liu, Bokui Chen, Jiangtao Gong
https://arxiv.org/abs/2510.12560

CoIRL-AD: Collaborative-Competitive Imitation-Reinforcement Learning in Latent World Models for Autonomous Driving
End-to-end autonomous driving models trained solely with imitation learning (IL) often suffer from poor generalization. In contrast, reinforcement learning (RL) promotes exploration through reward maximization but faces challenges such as sample inefficiency and unstable convergence. A natural solution is to combine IL and RL. Moving beyond the conventional two-stage paradigm (IL pretraining followed by RL fine-tuning), we propose CoIRL-AD, a competitive dual-policy framework that enables IL an…

Enhancing Speech Emotion Recognition via Fine-Tuning Pre-Trained Models and Hyper-Parameter Optimisation
Aryan Golbaghi, Shuo Zhou
https://arxiv.org/abs/2510.07052 https://

Enhancing Speech Emotion Recognition via Fine-Tuning Pre-Trained Models and Hyper-Parameter Optimisation
We propose a workflow for speech emotion recognition (SER) that combines pre-trained representations with automated hyperparameter optimisation (HPO). Using SpeechBrain wav2vec2-base model fine-tuned on IEMOCAP as the encoder, we compare two HPO strategies, Gaussian Process Bayesian Optimisation (GP-BO) and Tree-structured Parzen Estimators (TPE), under an identical four-dimensional search space and 15-trial budget, with balanced class accuracy (BCA) on the German EmoDB corpus as the objective.…

Prompt, Synthesize, Fine-Tune: A Secure Code Generation Recipe
Junjie Li, Fazle Rabbi, Bo Yang, Song Wang, Jinqiu Yang
https://arxiv.org/abs/2510.07189 https://

Prompt, Synthesize, Fine-Tune: A Secure Code Generation Recipe
Although Large Language Models (LLMs) show promising solutions to automated code generation, they often produce insecure code that threatens software security. Current approaches (e.g., SafeCoder) to improve secure code generation suffer from limited and imbalanced datasets, reducing their effectiveness and generalizability. In this work, we present Secure-Instruct, a novel framework that automatically synthesizes high-quality vulnerable and secure code examples, generates fine-tuning instructi…

Does LLM Focus on the Right Words? Diagnosing Language Bias in LLM-based Recommenders
Bohao Wang, Jiawei Chen, Feng Liu, Changwang Zhang, Jun Wang, Canghong Jin, Chun Chen, Can Wang
https://arxiv.org/abs/2510.10978

Does LLM Focus on the Right Words? Diagnosing Language Bias in LLM-based Recommenders
Large language models (LLMs), owing to their extensive open-domain knowledge and semantic reasoning capabilities, have been increasingly integrated into recommender systems (RS). However, a substantial gap remains between the pre-training objectives of LLMs and the specific requirements of recommendation tasks. To address this gap, supervised fine-tuning (SFT) is commonly performed on specially curated recommendation datasets to further enhance their predictive ability. Despite its success, SFT…

Proficiency-Aware Adaptation and Data Augmentation for Robust L2 ASR
Ling Sun, Charlotte Zhu, Shuju Shi
https://arxiv.org/abs/2510.10738 https://arxiv.org/…

Proficiency-Aware Adaptation and Data Augmentation for Robust L2 ASR
General-purpose ASR underperforms for atypical speakers, such as L2 learners, reinforcing bias and limiting use in education and accessibility. Using the CEFR-graded Speak and Improve corpus, we show that naive fine-tuning of Whisper reduces average WER but simultaneously widens disparities and disproportionately harms lower-level learners. To address this, we propose two strategies: (i) proficiency-aware multitask learning, jointly optimizing ASR with proficiency classification, and (ii) targe…

Actions as Language: Fine-Tuning VLMs into VLAs Without Catastrophic Forgetting
Asher J. Hancock, Xindi Wu, Lihan Zha, Olga Russakovsky, Anirudha Majumdar
https://arxiv.org/abs/2509.22195

Actions as Language: Fine-Tuning VLMs into VLAs Without Catastrophic Forgetting
Fine-tuning vision-language models (VLMs) on robot teleoperation data to create vision-language-action (VLA) models is a promising paradigm for training generalist policies, but it suffers from a fundamental tradeoff: learning to produce actions often diminishes the VLM's foundational reasoning and multimodal understanding, hindering generalization to novel scenarios, instruction following, and semantic understanding. We argue that this catastrophic forgetting is due to a distribution mismatch …

Valid Survey Simulations with Limited Human Data: The Roles of Prompting, Fine-Tuning, and Rectification
Stefan Krsteski, Giuseppe Russo, Serina Chang, Robert West, Kristina Gligori\'c
https://arxiv.org/abs/2510.11408

Valid Survey Simulations with Limited Human Data: The Roles of Prompting, Fine-Tuning, and Rectification
Surveys provide valuable insights into public opinion and behavior, but their execution is costly and slow. Large language models (LLMs) have been proposed as a scalable, low-cost substitute for human respondents, but their outputs are often biased and yield invalid estimates. We study the interplay between synthesis methods that use LLMs to generate survey responses and rectification methods that debias population estimates, and explore how human responses are best allocated between them. Usin…

MetaBreak: Jailbreaking Online LLM Services via Special Token Manipulation
Wentian Zhu, Zhen Xiang, Wei Niu, Le Guan
https://arxiv.org/abs/2510.10271 https://

MetaBreak: Jailbreaking Online LLM Services via Special Token Manipulation
Unlike regular tokens derived from existing text corpora, special tokens are artificially created to annotate structured conversations during the fine-tuning process of Large Language Models (LLMs). Serving as metadata of training data, these tokens play a crucial role in instructing LLMs to generate coherent and context-aware responses. We demonstrate that special tokens can be exploited to construct four attack primitives, with which malicious users can reliably bypass the internal safety ali…

HES-SQL: Hybrid Reasoning for Efficient Text-to-SQL with Structural Skeleton Guidance
Suming Qiu, Jing Li, Zhicheng Zhou, Junjie Huang, Linyuan Qiu, Zhijie Sun
https://arxiv.org/abs/2510.08896

HES-SQL: Hybrid Reasoning for Efficient Text-to-SQL with Structural Skeleton Guidance
We present HES-SQL, a novel hybrid training framework that advances Text-to-SQL generation through the integration of thinking-mode-fused supervised fine-tuning (SFT) with Group Relative Policy Optimization (GRPO). Our approach introduces three key innovations: (1) a skeleton-completeness scoring mechanism that enhances preference alignment between generated queries and optimal SQL structures; (2) a query-latency-aware reward system that incentivizes the generation of computationally efficient …

MEC$^3$O: Multi-Expert Consensus for Code Time Complexity Prediction
Joonghyuk Hahn, Soohan Lim, Yo-Sub Han
https://arxiv.org/abs/2510.09049 https://arxiv.…

MEC$^3$O: Multi-Expert Consensus for Code Time Complexity Prediction
Predicting the complexity of source code is essential for software development and algorithm analysis. Recently, Baik et al. (2025) introduced CodeComplex for code time complexity prediction. The paper shows that LLMs without fine-tuning struggle with certain complexity classes. This suggests that no single LLM excels at every class, but rather each model shows advantages in certain classes. We propose MEC$^3$O, a multi-expert consensus system, which extends the multi-agent debate frameworks. M…

Comparing fine-tuning strategies of MACE machine learning force field for modeling Li-ion diffusion in LiF for batteries
Nada Alghamdi, Paolo de Angelis, Pietro Asinari, Eliodoro Chiavazzo
https://arxiv.org/abs/2510.05020

Comparing fine-tuning strategies of MACE machine learning force field for modeling Li-ion diffusion in LiF for batteries
Machine learning interatomic potentials (MLIPs) are transforming materials science and engineering by enabling the study of complex phenomena, such as those critical to battery operation. In this work, we benchmark the MACE machine learning model against a well-trained DeePMD potential for predicting interstitial lithium diffusivity in LiF, a key component in the solid electrolyte interphase in Li ion batteries. Our results demonstrate that the MACE-MPA-0 foundational model achieves comparable …

A Deep Transfer Learning-Based Low-overhead Beam Prediction in Vehicle Communications
Zhiqiang Xiao, Yuwen Cao, Mondher Bouazizi, Tomoaki Ohtsuki, Shahid Mumtaz
https://arxiv.org/abs/2509.20659

A Deep Transfer Learning-Based Low-overhead Beam Prediction in Vehicle Communications
Existing transfer learning-based beam prediction approaches primarily rely on simple fine-tuning. When there is a significant difference in data distribution between the target domain and the source domain, simple fine-tuning limits the model's performance in the target domain. To tackle this problem, we propose a transfer learning-based beam prediction method that combines fine-tuning with domain adaptation. We integrate a domain classifier into fine-tuning the pre-trained model. The model ext…

Replaced article(s) found for cs.CV. https://arxiv.org/list/cs.CV/new
[6/8]:
- GeoVLM-R1: Reinforcement Fine-Tuning for Improved Remote Sensing Reasoning
Mustansar Fiaz, Hiyam Debary, Paolo Fraccaro, Danda Paudel, Luc Van Gool, Fahad Khan, Salman Khan

DeePAQ: A Perceptual Audio Quality Metric Based On Foundational Models and Weakly Supervised Learning
Guanxin Jiang, Andreas Brendel, Pablo M. Delgado, J\"urgen Herre
https://arxiv.org/abs/2510.12326

DeePAQ: A Perceptual Audio Quality Metric Based On Foundational Models and Weakly Supervised Learning
This paper presents the Deep learning-based Perceptual Audio Quality metric (DeePAQ) for evaluating general audio quality. Our approach leverages metric learning together with the music foundation model MERT, guided by surrogate labels, to construct an embedding space that captures distortion intensity in general audio. To the best of our knowledge, DeePAQ is the first in the general audio quality domain to leverage weakly supervised labels and metric learning for fine-tuning a music foundation…

MATH-Beyond: A Benchmark for RL to Expand Beyond the Base Model
Prasanna Mayilvahanan, Ricardo Dominguez-Olmedo, Thadd\"aus Wiedemer, Wieland Brendel
https://arxiv.org/abs/2510.11653

MATH-Beyond: A Benchmark for RL to Expand Beyond the Base Model
With the advent of DeepSeek-R1, a new wave of reinforcement learning (RL) methods has emerged that seem to unlock stronger mathematical reasoning. However, a closer look at the open-source ecosystem reveals a critical limitation: with sufficiently many draws (e.g., $\texttt{pass@1024}$), many existing base models already solve nearly all questions on widely used math benchmarks such as MATH-500 and AIME 2024. This suggests that the RL fine-tuning methods prevalent in the LLM reasoning literatur…

Sci2Pol: Evaluating and Fine-tuning LLMs on Scientific-to-Policy Brief Generation
Weimin Wu, Alexander C. Furnas, Eddie Yang, Gefei Liu, Akhil Pandey Akella, Xuefeng Song, Dashun Wang, Han Liu
https://arxiv.org/abs/2509.21493

Sci2Pol: Evaluating and Fine-tuning LLMs on Scientific-to-Policy Brief Generation
We propose Sci2Pol-Bench and Sci2Pol-Corpus, the first benchmark and training dataset for evaluating and fine-tuning large language models (LLMs) on policy brief generation from a scientific paper. We build Sci2Pol-Bench on a five-stage taxonomy to mirror the human writing process: (i) Autocompletion, (ii) Understanding, (iii) Summarization, (iv) Generation, and (v) Verification. It features 18 tasks in multiple-choice and open-ended formats. Specifically, for the Generation stage, we show that…

Teaching Language Models to Faithfully Express their Uncertainty
Bryan Eikema, Evgenia Ilia, Jos\'e G. C. de Souza, Chrysoula Zerva, Wilker Aziz
https://arxiv.org/abs/2510.12587

Teaching Language Models to Faithfully Express their Uncertainty
Large language models (LLMs) often miscommunicate their uncertainty: repeated queries can produce divergent answers, yet generated responses are typically unhedged or hedged in ways that do not reflect this variability. This conveys unfaithful information about the uncertain state of the LLMs' knowledge, creating a faithfulness gap that affects even strong LLMs. We introduce Faithful Uncertainty Tuning (FUT): a fine-tuning approach that teaches instruction-tuned LLMs to express uncertainty fait…

Fine-Tuning Jailbreaks under Highly Constrained Black-Box Settings: A Three-Pronged Approach
Xiangfang Li, Yu Wang, Bo Li
https://arxiv.org/abs/2510.01342 https://

Fine-Tuning Jailbreaks under Highly Constrained Black-Box Settings: A Three-Pronged Approach
With the rapid advancement of large language models (LLMs), ensuring their safe use becomes increasingly critical. Fine-tuning is a widely used method for adapting models to downstream tasks, yet it is vulnerable to jailbreak attacks. However, most existing studies focus on overly simplified attack scenarios, limiting their practical relevance to real-world defense settings. To make this risk concrete, we present a three-pronged jailbreak attack and evaluate it against provider defenses under a…

Replaced article(s) found for cs.AI. https://arxiv.org/list/cs.AI/new
[5/7]:
- On-Policy RL Meets Off-Policy Experts: Harmonizing Supervised Fine-Tuning and Reinforcement Learn...
Zhang, Xie, Sun, Chen, Wang, Li, Ding, Zhou

Resource-Efficient Fine-Tuning of LLaMA-3.2-3B for Medical Chain-of-Thought Reasoning
Imran Mansha
https://arxiv.org/abs/2510.05003 https://arxiv.org/pdf/2…

Resource-Efficient Fine-Tuning of LLaMA-3.2-3B for Medical Chain-of-Thought Reasoning
Large Language Models (LLMs) such as GPT-4 and LLaMA have demonstrated remarkable reasoning abilities but require significant computational resources for fine-tuning. This paper presents a resource-efficient fine-tuning approach for LLaMA-3.2-3B to enhance medical chain-of-thought reasoning while operating under constrained GPU and memory settings. Using parameter-efficient tuning techniques such as LoRA and QLoRA, we adapt the base model on publicly available medical reasoning datasets. The mo…

Lightweight and Generalizable Acoustic Scene Representations via Contrastive Fine-Tuning and Distillation
Kuang Yuan, Yang Gao, Xilin Li, Xinhao Mei, Syavosh Zadissa, Tarun Pruthi, Saeed Bagheri Sereshki
https://arxiv.org/abs/2510.03728

Lightweight and Generalizable Acoustic Scene Representations via Contrastive Fine-Tuning and Distillation
Acoustic scene classification (ASC) models on edge devices typically operate under fixed class assumptions, lacking the transferability needed for real-world applications that require adaptation to new or refined acoustic categories. We propose ContrastASC, which learns generalizable acoustic scene representations by structuring the embedding space to preserve semantic relationships between scenes, enabling adaptation to unseen categories without retraining. Our approach combines supervised con…

AMAQ: Adaptive Mixed-bit Activation Quantization for Collaborative Parameter Efficient Fine-tuning
Yurun Song, Zhuoyi Yang, Ian G. Harris, Sangeetha Abdu Jyothi
https://arxiv.org/abs/2510.05468

AMAQ: Adaptive Mixed-bit Activation Quantization for Collaborative Parameter Efficient Fine-tuning
Large Language Models (LLMs) are scaling rapidly, creating significant challenges for collaborative server client distributed training, particularly in terms of communication efficiency and computational overheads. To address these challenges, we implement Parameter-efficient Split Learning, which effectively balances efficiency and performance for collaborative training on low-resource devices. To reduce communication overhead in collaborative training, we introduce Adaptive Mixed bit Activa…

Devstral: Fine-tuning Language Models for Coding Agent Applications
Abhinav Rastogi, Adam Yang, Albert Q. Jiang, Alexander H. Liu, Alexandre Sablayrolles, Am\'elie H\'eliou, Am\'elie Martin, Anmol Agarwal, Andy Ehrenberg, Andy Lo, Antoine Roux, Arthur Darcet, Arthur Mensch, Baptiste Bout, Baptiste Rozi\`ere, Baudouin De Monicault, Chris Bamford, Christian Wallenwein, Christophe Renaudin, Cl\'emence Lanfranchi, Cl\'ement Denoix, Corentin Barreau, Darius Dabert Devon …

Devstral: Fine-tuning Language Models for Coding Agent Applications
We introduce Devstral-Small, a lightweight open source model for code agents with the best performance among models below 100B size. In this technical report, we give an overview of how we design and develop a model and craft specializations in agentic software development. The resulting model, Devstral-Small is a small 24B model, fast and easy to serve. Despite its size, Devstral-Small still attains competitive performance compared to models more than an order of magnitude larger.

MetaVLA: Unified Meta Co-training For Efficient Embodied Adaption
Chen Li, Zhantao Yang, Han Zhang, Fangyi Chen, Chenchen Zhu, Anudeepsekhar Bolimera, Marios Savvides
https://arxiv.org/abs/2510.05580

MetaVLA: Unified Meta Co-training For Efficient Embodied Adaption
Vision-Language-Action (VLA) models show promise in embodied reasoning, yet remain far from true generalists-they often require task-specific fine-tuning, and generalize poorly to unseen tasks. We propose MetaVLA, a unified, backbone-agnostic post-training framework for efficient and scalable alignment. MetaVLA introduces Context-Aware Meta Co-Training, which consolidates diverse target tasks into a single fine-tuning stage while leveraging structurally diverse auxiliary tasks to improve in-dom…

Fine-Tuning Bulk-oriented Universal Interatomic Potentials for Surfaces: Accuracy, Efficiency, and Forgetting Control
Jaekyun Hwang, Taehun Lee, Yonghyuk Lee, Su-Hyun Yoo
https://arxiv.org/abs/2509.25807

Fine-Tuning Bulk-oriented Universal Interatomic Potentials for Surfaces: Accuracy, Efficiency, and Forgetting Control
Accurate prediction of surface energies and stabilities is essential for materials design, yet first-principles calculations remain computationally expensive and most existing interatomic potentials are trained only on bulk systems. Here, we demonstrate that fine-tuning foundation machine learning potentials (MLPs) significantly improves both computational efficiency and predictive accuracy for surface modeling. While existing universal interatomic potentials (UIPs) have been solely trained and…

Multimodal Carotid Risk Stratification with Large Vision-Language Models: Benchmarking, Fine-Tuning, and Clinical Insights
Daphne Tsolissou, Theofanis Ganitidis, Konstantinos Mitsis, Stergios CHristodoulidis, Maria Vakalopoulou, Konstantina Nikita
https://arxiv.org/abs/2510.02922

Multimodal Carotid Risk Stratification with Large Vision-Language Models: Benchmarking, Fine-Tuning, and Clinical Insights
Reliable risk assessment for carotid atheromatous disease remains a major clinical challenge, as it requires integrating diverse clinical and imaging information in a manner that is transparent and interpretable to clinicians. This study investigates the potential of state-of-the-art and recent large vision-language models (LVLMs) for multimodal carotid plaque assessment by integrating ultrasound imaging (USI) with structured clinical, demographic, laboratory, and protein biomarker data. A fram…

TR2-D2: Tree Search Guided Trajectory-Aware Fine-Tuning for Discrete Diffusion
Sophia Tang, Yuchen Zhu, Molei Tao, Pranam Chatterjee
https://arxiv.org/abs/2509.25171 https://

TR2-D2: Tree Search Guided Trajectory-Aware Fine-Tuning for Discrete Diffusion
Reinforcement learning with stochastic optimal control offers a promising framework for diffusion fine-tuning, where a pre-trained diffusion model is optimized to generate paths that lead to a reward-tilted distribution. While these approaches enable optimization without access to explicit samples from the optimal distribution, they require training on rollouts under the current fine-tuned model, making them susceptible to reinforcing sub-optimal trajectories that yield poor rewards. To overcom…

Reasoning Pattern Matters: Learning to Reason without Human Rationales
Chaoxu Pang, Yixuan Cao, Ping Luo
https://arxiv.org/abs/2510.12643 https://arxiv.org…

Reasoning Pattern Matters: Learning to Reason without Human Rationales
Large Language Models (LLMs) have demonstrated remarkable reasoning capabilities under the widely adopted SFT+RLVR paradigm, which first performs Supervised Fine-Tuning (SFT) on human-annotated reasoning trajectories (rationales) to establish initial reasoning behaviors, then applies Reinforcement Learning with Verifiable Rewards (RLVR) to optimize the model using verifiable signals without golden rationales. However, annotating high-quality rationales for the SFT stage remains prohibitively ex…

Sample-Efficient Differentially Private Fine-Tuning via Gradient Matrix Denoising
Ali Dadsetan, Frank Rudzicz
https://arxiv.org/abs/2510.01137 https://arxi…

Sample-Efficient Differentially Private Fine-Tuning via Gradient Matrix Denoising
We address the challenge of sample efficiency in differentially private fine-tuning of large language models (LLMs) using DP-SGD. While DP-SGD provides strong privacy guarantees, the added noise significantly increases the entropy of gradient matrices, disrupting their low-rank structure and slowing optimization. We propose a post-processing algorithm that leverages random matrix theory to denoise gradients, restore low-rank structure, and improve alignment with the original signal. Applied to …

Data Efficient Adaptation in Large Language Models via Continuous Low-Rank Fine-Tuning
Xiao Han, Zimo Zhao, Wanyu Wang, Maolin Wang, Zitao Liu, Yi Chang, Xiangyu Zhao
https://arxiv.org/abs/2509.18942

Data Efficient Adaptation in Large Language Models via Continuous Low-Rank Fine-Tuning
Recent advancements in Large Language Models (LLMs) have emphasized the critical role of fine-tuning (FT) techniques in adapting LLMs to specific tasks, especially when retraining from scratch is computationally infeasible. Fine-tuning enables LLMs to leverage task- or domain-specific data, producing models that more effectively meet the requirements of targeted applications. However, con- ventional FT approaches often suffer from catastrophic forgetting and suboptimal data efficiency, limiting…

Exploring Fine-Tuning of Large Audio Language Models for Spoken Language Understanding under Limited Speech data
Youngwon Choi, Jaeyoon Jung, Hyeonyu Kim, Huu-Kim Nguyen, Hwayeon Kim
https://arxiv.org/abs/2509.15389

Exploring Fine-Tuning of Large Audio Language Models for Spoken Language Understanding under Limited Speech data
Large Audio Language Models (LALMs) have emerged as powerful tools for speech-related tasks but remain underexplored for fine-tuning, especially with limited speech data. To bridge this gap, we systematically examine how different fine-tuning schemes including text-only, direct mixing, and curriculum learning affect spoken language understanding (SLU), focusing on scenarios where text-label pairs are abundant while paired speech-label data are limited. Results show that LALMs already achieve co…

Replaced article(s) found for cs.CV. https://arxiv.org/list/cs.CV/new
[4/5]:
- Fairness-Aware Fine-Tuning of Vision-Language Models for Medical Glaucoma Diagnosis
Zijian Gu, Yuxi Liu, Zhenhao Zhang, Song Wang

Probing Latent Knowledge Conflict for Faithful Retrieval-Augmented Generation
Linfeng Gao, Baolong Bi, Zheng Yuan, Le Wang, Zerui Chen, Zhimin Wei, Shenghua Liu, Qinggang Zhang, Jinsong Su
https://arxiv.org/abs/2510.12460

Probing Latent Knowledge Conflict for Faithful Retrieval-Augmented Generation
Retrieval-Augmented Generation (RAG) has emerged as a powerful paradigm to enhance the factuality of Large Language Models (LLMs). However, existing RAG systems often suffer from an unfaithfulness issue, where the model's response contradicts evidence from the retrieved context. Existing approaches to improving contextual faithfulness largely rely on external interventions, such as prompt engineering, decoding constraints, or reward-based fine-tuning. These works treat the LLM as a black box an…

Safety Game: Balancing Safe and Informative Conversations with Blackbox Agentic AI using LP Solvers
Tuan Nguyen, Long Tran-Thanh
https://arxiv.org/abs/2510.09330 https://…

Safety Game: Balancing Safe and Informative Conversations with Blackbox Agentic AI using LP Solvers
Ensuring that large language models (LLMs) comply with safety requirements is a central challenge in AI deployment. Existing alignment approaches primarily operate during training, such as through fine-tuning or reinforcement learning from human feedback, but these methods are costly and inflexible, requiring retraining whenever new requirements arise. Recent efforts toward inference-time alignment mitigate some of these limitations but still assume access to model internals, which is impractic…

Fine-tuning Behavioral Cloning Policies with Preference-Based Reinforcement Learning
Ma\"el Macuglia, Paul Friedrich, Giorgia Ramponi
https://arxiv.org/abs/2509.26605 https…

Fine-tuning Behavioral Cloning Policies with Preference-Based Reinforcement Learning
Deploying reinforcement learning (RL) in robotics, industry, and health care is blocked by two obstacles: the difficulty of specifying accurate rewards and the risk of unsafe, data-hungry exploration. We address this by proposing a two-stage framework that first learns a safe initial policy from a reward-free dataset of expert demonstrations, then fine-tunes it online using preference-based human feedback. We provide the first principled analysis of this offline-to-online approach and introduce…

FlyLoRA: Boosting Task Decoupling and Parameter Efficiency via Implicit Rank-Wise Mixture-of-Experts
Heming Zou, Yunliang Zang, Wutong Xu, Yao Zhu, Xiangyang Ji
https://arxiv.org/abs/2510.08396

FlyLoRA: Boosting Task Decoupling and Parameter Efficiency via Implicit Rank-Wise Mixture-of-Experts
Low-Rank Adaptation (LoRA) is a widely used parameter-efficient fine-tuning method for foundation models, but it suffers from parameter interference, resulting in suboptimal performance. Although Mixture-of-Experts (MoE)-based LoRA variants show promise in mitigating intra-task correlations in single-task instruction tuning, they introduce additional router parameters and remain ineffective in multi-task model merging where inter-task interference arises. Inspired by the fly olfactory circuit, …

BEFT: Bias-Efficient Fine-Tuning of Language Models
Baichuan Huang, Ananth Balashankar, Amir Aminifar
https://arxiv.org/abs/2509.15974 https://arxiv.org/pd…

BEFT: Bias-Efficient Fine-Tuning of Language Models
Fine-tuning all-bias-terms stands out among various parameter-efficient fine-tuning (PEFT) techniques, owing to its out-of-the-box usability and competitive performance, especially in low-data regimes. Bias-only fine-tuning has the potential for unprecedented parameter efficiency. However, the link between fine-tuning different bias terms (i.e., bias terms in the query, key, or value projections) and downstream performance remains unclear. The existing approaches, e.g., based on the magnitude o…

AI-CNet3D: An Anatomically-Informed Cross-Attention Network with Multi-Task Consistency Fine-tuning for 3D Glaucoma Classification
Roshan Kenia, Anfei Li, Rishabh Srivastava, Kaveri A. Thakoor
https://arxiv.org/abs/2510.00882

AI-CNet3D: An Anatomically-Informed Cross-Attention Network with Multi-Task Consistency Fine-tuning for 3D Glaucoma Classification
Glaucoma is a progressive eye disease that leads to optic nerve damage, causing irreversible vision loss if left untreated. Optical coherence tomography (OCT) has become a crucial tool for glaucoma diagnosis, offering high-resolution 3D scans of the retina and optic nerve. However, the conventional practice of condensing information from 3D OCT volumes into 2D reports often results in the loss of key structural details. To address this, we propose a novel hybrid deep learning model that integra…

DualTune: Decoupled Fine-Tuning for On-Device Agentic Systems
Rohan Kadekodi, Zhan Jin, Keisuke Kamahori, Yile Gu, Sean Khatiri, Noah H. Bayindirli, Sergey Gorbunov, Baris Kasikci
https://arxiv.org/abs/2510.00229

DualTune: Decoupled Fine-Tuning for On-Device Agentic Systems
The deployment of Large Language Models (LLMs) as agentic orchestrators has revolutionized task automation, but the need for privacy-preserving, cost-effective solutions demands on-device inference capabilities. However, local LLMs consistently underperform compared to frontier models in tool calling scenarios, struggling with both tool selection from large tool sets and accurate argument generation for complex parameter structures. We introduce a methodology that disaggregates a tool-calling t…

HINT: Helping Ineffective Rollouts Navigate Towards Effectiveness
Xinyi Wang, Jinyi Han, Zishang Jiang, Tingyun Li, Jiaqing Liang, Sihang Jiang, Zhaoqian Dai, Shuguang Ma, Fei Yu, Yanghua Xiao
https://arxiv.org/abs/2510.09388

HINT: Helping Ineffective Rollouts Navigate Towards Effectiveness
Reinforcement Learning (RL) has become a key driver for enhancing the long chain-of-thought (CoT) reasoning capabilities of Large Language Models (LLMs). However, prevalent methods like GRPO often fail when task difficulty exceeds the model's capacity, leading to reward sparsity and inefficient training. While prior work attempts to mitigate this using off-policy data, such as mixing RL with Supervised Fine-Tuning (SFT) or using hints, they often misguide policy updates In this work, we identif…

BALF: Budgeted Activation-Aware Low-Rank Factorization for Fine-Tuning-Free Model Compression
David Gonz\'alez Mart\'inez
https://arxiv.org/abs/2509.25136 https://

BALF: Budgeted Activation-Aware Low-Rank Factorization for Fine-Tuning-Free Model Compression
Neural network compression techniques typically require expensive fine-tuning or search procedures, rendering them impractical on commodity hardware. Inspired by recent LLM compression research, we present a general activation-aware factorization framework that can be applied to a broad range of layers. Moreover, we introduce a scalable budgeted rank allocator that allows flexible control over compression targets (e.g., retaining 50% of parameters) with no overhead. Together, these components f…

SliceFine: The Universal Winning-Slice Hypothesis for Pretrained Networks
Md Kowsher, Ali O. Polat, Ehsan Mohammady Ardehaly, Mehrdad Salehi, Zia Ghiasi, Prasanth Murali, Chen Chen
https://arxiv.org/abs/2510.08513

SliceFine: The Universal Winning-Slice Hypothesis for Pretrained Networks
This paper presents a theoretical framework explaining why fine tuning small, randomly selected subnetworks (slices) within pre trained models can be sufficient for downstream adaptation. We prove that pretrained networks exhibit a universal winning slice property arising from two phenomena: (1) spectral balance the eigenspectra of different weight matrix slices are remarkably similar; and (2) high task energy their backbone representations retain rich, task relevant features. This leads to the…

Beyond Log Likelihood: Probability-Based Objectives for Supervised Fine-Tuning across the Model Capability Continuum
Gaotang Li, Ruizhong Qiu, Xiusi Chen, Heng Ji, Hanghang Tong
https://arxiv.org/abs/2510.00526

Beyond Log Likelihood: Probability-Based Objectives for Supervised Fine-Tuning across the Model Capability Continuum
Supervised fine-tuning (SFT) is the standard approach for post-training large language models (LLMs), yet it often shows limited generalization. We trace this limitation to its default training objective: negative log likelihood (NLL). While NLL is classically optimal when training from scratch, post-training operates in a different paradigm and could violate its optimality assumptions, where models already encode task-relevant priors and supervision can be long and noisy. To this end, we study…

IA2: Alignment with ICL Activations Improves Supervised Fine-Tuning
Aayush Mishra, Daniel Khashabi, Anqi Liu
https://arxiv.org/abs/2509.22621 https://arxiv…

IA2: Alignment with ICL Activations Improves Supervised Fine-Tuning
Supervised Fine-Tuning (SFT) is used to specialize model behavior by training weights to produce intended target responses for queries. In contrast, In-Context Learning (ICL) adapts models during inference with instructions or demonstrations in the prompt. ICL can offer better generalizability and more calibrated responses compared to SFT in data scarce settings, at the cost of more inference compute. In this work, we ask the question: Can ICL's internal computations be used to improve the qual…

TempSamp-R1: Effective Temporal Sampling with Reinforcement Fine-Tuning for Video LLMs
Yunheng Li, Jing Cheng, Shaoyong Jia, Hangyi Kuang, Shaohui Jiao, Qibin Hou, Ming-Ming Cheng
https://arxiv.org/abs/2509.18056

TempSamp-R1: Effective Temporal Sampling with Reinforcement Fine-Tuning for Video LLMs
This paper introduces TempSamp-R1, a new reinforcement fine-tuning framework designed to improve the effectiveness of adapting multimodal large language models (MLLMs) to video temporal grounding tasks. We reveal that existing reinforcement learning methods, such as Group Relative Policy Optimization (GRPO), rely on on-policy sampling for policy updates. However, in tasks with large temporal search spaces, this strategy becomes both inefficient and limited in performance, as it often fails to i…

Replaced article(s) found for cs.CL. https://arxiv.org/list/cs.CL/new
[1/4]:
- Privacy-Preserving Parameter-Efficient Fine-Tuning for Large Language Model Services
Yansong Li, Zhixing Tan, Paula Branco, Yang Liu

Symmetry-Aware Fully-Amortized Optimization with Scale Equivariant Graph Metanetworks
Bart Kuipers, Freek Byrman, Daniel Uyterlinde, Alejandro Garc\'ia-Castellanos
https://arxiv.org/abs/2510.08300 …

Symmetry-Aware Fully-Amortized Optimization with Scale Equivariant Graph Metanetworks
Amortized optimization accelerates the solution of related optimization problems by learning mappings that exploit shared structure across problem instances. We explore the use of Scale Equivariant Graph Metanetworks (ScaleGMNs) for this purpose. By operating directly in weight space, ScaleGMNs enable single-shot fine-tuning of existing models, reducing the need for iterative optimization. We demonstrate the effectiveness of this approach empirically and provide a theoretical result: the gauge …

A Multi-Agent Framework for Stateful Inference-Time Search
Arshika Lalan, Rajat Ghosh, Aditya Kolsur, Debojyoti Dutta
https://arxiv.org/abs/2510.07147 https://

A Multi-Agent Framework for Stateful Inference-Time Search
Recent work explores agentic inference-time techniques to perform structured, multi-step reasoning. However, stateless inference often struggles on multi-step tasks due to the absence of persistent state. Moreover, task-specific fine-tuning or instruction-tuning often achieve surface-level code generation but remain brittle on tasks requiring deeper reasoning and long-horizon dependencies. To address these limitations, we propose stateful multi-agent evolutionary search, a training-free framewo…

One-Token Rollout: Guiding Supervised Fine-Tuning of LLMs with Policy Gradient
Rui Ming, Haoyuan Wu, Shoubo Hu, Zhuolun He, Bei Yu
https://arxiv.org/abs/2509.26313 https://

One-Token Rollout: Guiding Supervised Fine-Tuning of LLMs with Policy Gradient
Supervised fine-tuning (SFT) is the predominant method for adapting large language models (LLMs), yet it often struggles with generalization compared to reinforcement learning (RL). In this work, we posit that this performance disparity stems not just from the loss function, but from a more fundamental difference: SFT learns from a fixed, pre-collected dataset, whereas RL utilizes on-policy data sampled from the current policy. Building on this hypothesis, we introduce one-token rollout (OTR), …

Robust RGB-T Tracking via Learnable Visual Fourier Prompt Fine-tuning and Modality Fusion Prompt Generation
Hongtao Yang, Bineng Zhong, Qihua Liang, Zhiruo Zhu, Yaozong Zheng, Ning Li
https://arxiv.org/abs/2509.19733

Robust RGB-T Tracking via Learnable Visual Fourier Prompt Fine-tuning and Modality Fusion Prompt Generation
Recently, visual prompt tuning is introduced to RGB-Thermal (RGB-T) tracking as a parameter-efficient finetuning (PEFT) method. However, these PEFT-based RGB-T tracking methods typically rely solely on spatial domain information as prompts for feature extraction. As a result, they often fail to achieve optimal performance by overlooking the crucial role of frequency-domain information in prompt learning. To address this issue, we propose an efficient Visual Fourier Prompt Tracking (named VFPTra…

Agent Learning via Early Experience
Kai Zhang, Xiangchao Chen, Bo Liu, Tianci Xue, Zeyi Liao, Zhihan Liu, Xiyao Wang, Yuting Ning, Zhaorun Chen, Xiaohan Fu, Jian Xie, Yuxuan Sun, Boyu Gou, Qi Qi, Zihang Meng, Jianwei Yang, Ning Zhang, Xian Li, Ashish Shah, Dat Huynh, Hengduo Li, Zi Yang, Sara Cao, Lawrence Jang, Shuyan Zhou, Jiacheng Zhu, Huan Sun, Jason Weston, Yu Su, Yifan Wu

Agent Learning via Early Experience
A long-term goal of language agents is to learn and improve through their own experience, ultimately outperforming humans in complex, real-world tasks. However, training agents from experience data with reinforcement learning remains difficult in many environments, which either lack verifiable rewards (e.g., websites) or require inefficient long-horizon rollouts (e.g., multi-turn tool use). As a result, most current agents rely on supervised fine-tuning on expert data, which is challenging to s…

Influence Functions for Efficient Data Selection in Reasoning
Prateek Humane, Paolo Cudrano, Daniel Z. Kaplan, Matteo Matteucci, Supriyo Chakraborty, Irina Rish
https://arxiv.org/abs/2510.06108

Influence Functions for Efficient Data Selection in Reasoning
Fine-tuning large language models (LLMs) on chain-of-thought (CoT) data shows that a small amount of high-quality data can outperform massive datasets. Yet, what constitutes "quality" remains ill-defined. Existing reasoning methods rely on indirect heuristics such as problem difficulty or trace length, while instruction-tuning has explored a broader range of automated selection strategies, but rarely in the context of reasoning. We propose to define reasoning data quality using influence functi…

TiTok: Transfer Token-level Knowledge via Contrastive Excess to Transplant LoRA
Chanjoo Jung, Jaehyung Kim
https://arxiv.org/abs/2510.04682 https://arxiv.o…

TiTok: Transfer Token-level Knowledge via Contrastive Excess to Transplant LoRA
Large Language Models (LLMs) are widely applied in real world scenarios, but fine-tuning them comes with significant computational and storage costs. Parameter-Efficient Fine-Tuning (PEFT) methods such as LoRA mitigate these costs, but the adapted parameters are dependent on the base model and cannot be transferred across different backbones. One way to address this issue is through knowledge distillation, but its effectiveness inherently depends on training data. Recent work such as TransLoRA …

Fine-Tuning Large Multimodal Models for Automatic Pronunciation Assessment
Ke Wang, Wenning Wei, Yan Deng, Lei He, Sheng Zhao
https://arxiv.org/abs/2509.15701 https://

Fine-Tuning Large Multimodal Models for Automatic Pronunciation Assessment
Automatic Pronunciation Assessment (APA) is critical for Computer-Assisted Language Learning (CALL), requiring evaluation across multiple granularities and aspects. Large Multimodal Models (LMMs) present new opportunities for APA, but their effectiveness in fine-grained assessment remains uncertain. This work investigates fine-tuning LMMs for APA using the Speechocean762 dataset and a private corpus. Fine-tuning significantly outperforms zero-shot settings and achieves competitive results on si…

InfiMed-Foundation: Pioneering Advanced Multimodal Medical Models with Compute-Efficient Pre-Training and Multi-Stage Fine-Tuning
Guanghao Zhu, Zhitian Hou, Zeyu Liu, Zhijie Sang, Congkai Xie, Hongxia Yang
https://arxiv.org/abs/2509.22261

InfiMed-Foundation: Pioneering Advanced Multimodal Medical Models with Compute-Efficient Pre-Training and Multi-Stage Fine-Tuning
Multimodal large language models (MLLMs) have shown remarkable potential in various domains, yet their application in the medical field is hindered by several challenges. General-purpose MLLMs often lack the specialized knowledge required for medical tasks, leading to uncertain or hallucinatory responses. Knowledge distillation from advanced models struggles to capture domain-specific expertise in radiology and pharmacology. Additionally, the computational cost of continual pretraining with lar…

Targeted Fine-Tuning of DNN-Based Receivers via Influence Functions
Marko Tuononen, Heikki Penttinen, Ville Hautam\"aki
https://arxiv.org/abs/2509.15950 https://

Targeted Fine-Tuning of DNN-Based Receivers via Influence Functions
We present the first use of influence functions for deep learning-based wireless receivers. Applied to DeepRx, a fully convolutional receiver, influence analysis reveals which training samples drive bit predictions, enabling targeted fine-tuning of poorly performing cases. We show that loss-relative influence with capacity-like binary cross-entropy loss and first-order updates on beneficial samples most consistently improves bit error rate toward genie-aided performance, outperforming random fi…

Family Matters: Language Transfer and Merging for Adapting Small LLMs to Faroese
Jenny Kunz, Iben Nyholm Debess, Annika Simonsen
https://arxiv.org/abs/2510.00810 https://…

Family Matters: Language Transfer and Merging for Adapting Small LLMs to Faroese
We investigate how to adapt small, efficient LLMs to Faroese, a low-resource North Germanic language. Starting from English models, we continue pre-training on related Scandinavian languages, either individually or combined via merging, before fine-tuning on Faroese. We compare full fine-tuning with parameter-efficient tuning using LoRA, evaluating their impact on both linguistic accuracy and text comprehension. Due to the lack of existing Faroese evaluation data, we construct two new minimal-p…

Improving Reasoning for Diffusion Language Models via Group Diffusion Policy Optimization
Kevin Rojas, Jiahe Lin, Kashif Rasul, Anderson Schneider, Yuriy Nevmyvaka, Molei Tao, Wei Deng
https://arxiv.org/abs/2510.08554

Improving Reasoning for Diffusion Language Models via Group Diffusion Policy Optimization
Diffusion language models (DLMs) enable parallel, order-agnostic generation with iterative refinement, offering a flexible alternative to autoregressive large language models (LLMs). However, adapting reinforcement learning (RL) fine-tuning to DLMs remains an open challenge because of the intractable likelihood. Pioneering work such as diffu-GRPO estimated token-level likelihoods via one-step unmasking. While computationally efficient, this approach is severely biased. A more principled foundat…

Guided Star-Shaped Masked Diffusion
Viacheslav Meshchaninov, Egor Shibaev, Artem Makoian, Ivan Klimov, Danil Sheshenya, Andrei Malinin, Nikita Balagansky, Daniil Gavrilov, Aibek Alanov, Dmitry Vetrov
https://arxiv.org/abs/2510.08369

Guided Star-Shaped Masked Diffusion
The performance of pre-trained masked diffusion models is often constrained by their sampling procedure, which makes decisions irreversible and struggles in low-step generation regimes. We introduce a novel sampling algorithm that works with pre-trained models and, after a lightweight fine-tuning of a single layer, significantly improves sample quality and efficiency. Our method reformulates the generation process using a star-shaped paradigm, which inherently allows for error correction. To ma…

When Long Helps Short: How Context Length in Supervised Fine-tuning Affects Behavior of Large Language Models
Yingming Zheng, Hanqi Li, Kai Yu, Lu Chen
https://arxiv.org/abs/2509.18762

Explaining Fine Tuned LLMs via Counterfactuals A Knowledge Graph Driven Framework
Yucheng Wang, Ziyang Chen, Md Faisal Kabir
https://arxiv.org/abs/2509.21241 https://

Explaining Fine Tuned LLMs via Counterfactuals A Knowledge Graph Driven Framework
The widespread adoption of Low-Rank Adaptation (LoRA) has enabled large language models (LLMs) to acquire domain-specific knowledge with remarkable efficiency. However, understanding how such a fine-tuning mechanism alters a model's structural reasoning and semantic behavior remains an open challenge. This work introduces a novel framework that explains fine-tuned LLMs via counterfactuals grounded in knowledge graphs. Specifically, we construct BioToolKG, a domain-specific heterogeneous knowled…

Investigating the Representation of Backchannels and Fillers in Fine-tuned Language Models
Yu Wang, Leyi Lao, Langchu Huang, Gabriel Skantze, Yang Xu, Hendrik Buschmeier
https://arxiv.org/abs/2509.20237

Investigating the Representation of Backchannels and Fillers in Fine-tuned Language Models
Backchannels and fillers are important linguistic expressions in dialogue, but are under-represented in modern transformer-based language models (LMs). Our work studies the representation of them in language models using three fine-tuning strategies. The models are trained on three dialogue corpora in English and Japanese, where backchannels and fillers are preserved and annotated, to investigate how fine-tuning can help LMs learn their representations. We first apply clustering analysis to the…

AccurateRAG: A Framework for Building Accurate Retrieval-Augmented Question-Answering Applications
Linh The Nguyen, Chi Tran, Dung Ngoc Nguyen, Van-Cuong Pham, Hoang Ngo, Dat Quoc Nguyen
https://arxiv.org/abs/2510.02243

AccurateRAG: A Framework for Building Accurate Retrieval-Augmented Question-Answering Applications
We introduce AccurateRAG -- a novel framework for constructing high-performance question-answering applications based on retrieval-augmented generation (RAG). Our framework offers a pipeline for development efficiency with tools for raw dataset processing, fine-tuning data generation, text embedding & LLM fine-tuning, output evaluation, and building RAG systems locally. Experimental results show that our framework outperforms previous strong baselines and obtains new state-of-the-art question-a…

A method for improving multilingual quality and diversity of instruction fine-tuning datasets
Chunguang Zhao, Yilun Liu, Pufan Zeng, Yuanchang Luo, Shimin Tao, Minggui He, Weibin Meng, Song Xu, Ziang Chen, Chen Liu, Hongxia Ma, Li Zhang, Boxing Chen, Daimeng Wei
https://arxiv.org/abs/2509.15549

A method for improving multilingual quality and diversity of instruction fine-tuning datasets
Multilingual Instruction Fine-Tuning (IFT) is essential for enabling large language models (LLMs) to generalize effectively across diverse linguistic and cultural contexts. However, the scarcity of high-quality multilingual training data and corresponding building method remains a critical bottleneck. While data selection has shown promise in English settings, existing methods often fail to generalize across languages due to reliance on simplistic heuristics or language-specific assumptions. In…

Finetune Once: Decoupling General & Domain Learning with Dynamic Boosted Annealing
Yang Tang, Ruijie Liu, Yifan Wang, Shiyu Li, Xi Chen
https://arxiv.org/abs/2509.26242 http…

Finetune Once: Decoupling General & Domain Learning with Dynamic Boosted Annealing
Large language models (LLMs) fine-tuning shows excellent implications. However, vanilla fine-tuning methods often require intricate data mixture and repeated experiments for optimal generalization. To address these challenges and streamline the training process, we propose an efficient and universal solution, Dynamic Boosted Annealing (DBA). We obtain a global gradient through zero-learning-rate training on general data, which is subsequently employed for gradient boosting and dynamic training …

Metaphor identification using large language models: A comparison of RAG, prompt engineering, and fine-tuning
Matteo Fuoli, Weihang Huang, Jeannette Littlemore, Sarah Turner, Ellen Wilding
https://arxiv.org/abs/2509.24866

Metaphor identification using large language models: A comparison of RAG, prompt engineering, and fine-tuning
Metaphor is a pervasive feature of discourse and a powerful lens for examining cognition, emotion, and ideology. Large-scale analysis, however, has been constrained by the need for manual annotation due to the context-sensitive nature of metaphor. This study investigates the potential of large language models (LLMs) to automate metaphor identification in full texts. We compare three methods: (i) retrieval-augmented generation (RAG), where the model is provided with a codebook and instructed to …

Scaling LLM Multi-turn RL with End-to-end Summarization-based Context Management
Miao Lu, Weiwei Sun, Weihua Du, Zhan Ling, Xuesong Yao, Kang Liu, Jiecao Chen
https://arxiv.org/abs/2510.06727

Scaling LLM Multi-turn RL with End-to-end Summarization-based Context Management
We study reinforcement learning (RL) fine-tuning of large language model (LLM) agents for long-horizon multi-turn tool use, where context length quickly becomes a fundamental bottleneck. Existing RL pipelines can suffer from degraded instruction following, excessive rollout costs, and most importantly, strict context limits. To address these challenges, we introduce summarization-based context management to training. In specific, it periodically compresses the tool using history by LLM-generate…