LoRA, PEFT, instruction tuning and domain adaptation — adapting a pretrained model without paying to train one.
arXiv:2609.20886v1 Announce Type: cross Abstract: Business intelligence (BI) is a cornerstone of enterprise decision-making and is widely used by enterprise users in software such as Power BI and Tab...
The paper presents a Deep Operator Network that reconstructs full‑field 4D mitral regurgitation hemodynamics from sparse planar velocity data and boundary pressure traces. The network is pretrained on a URANS database of eleven orifice phantoms and then fine‑tuned on new cases, achieving rapid predictions in minutes. While the adaptation improves flow topology in the observed plane, reconstruction error increases sharply with distance from that plane, limiting physical consistency in the surrounding volume.
The paper presents a three‑stage, parameter‑efficient approach to improve automatic target recognition (ATR) with synthetic aperture sonar (SAS) data by adapting DINOv3 Vision Transformers. Stage 1 applies Low‑Rank Adaptation (LoRA) while freezing the backbone, which significantly boosts the area under the precision‑recall curve from 0.300 to 0.679. Subsequent hard‑negative mining and supervised contrastive learning stages show negligible impact, indicating that a single LoRA adaptation is sufficient for effective underwater ATR.
The paper introduces a three‑stage pipeline to improve accented conversational ASR for speakers from India, Indonesia, and Latin America. It uses heuristic SQL filters to curate entity‑rich training data, regional LoRA adapters fine‑tuned on Qwen2.5‑Omni‑3B to generate both verbatim and corrected transcripts, and a six‑category error taxonomy validated by an LLM judge. The approach raises entity recall to 80‑85% and filler recall to 76‑86%, while keeping WER low (6‑10%) and outperforming Whisper and a commercial ASR on entity recall.
LiteMedCoT-VL is a parameter‑efficient pipeline that transfers chain‑of‑thought reasoning from a 235B teacher model to a 2B student model using LoRA fine‑tuning on explanation‑enriched data. The approach enables a compact vision‑language model to perform medical visual question answering without relying on image captions, achieving 64.9% accuracy on the PMC‑VQA benchmark—an 11‑point improvement over the zero‑shot Qwen3‑VL‑4B baseline. Visual grounding analysis confirms that the model bases its predictions on image content rather than textual priors.
Rhamba is a region‑aware pretraining framework for resting‑state fMRI that combines anatomically guided masking with hybrid Attention‑Mamba architectures. The study pretrained models on the ABIDE dataset using three masking strategies (Any, Majority, Pure) and evaluated four architectural variants, finding that the Mamba‑Attention (MA) hybrid achieved the best average AUROC on downstream schizophrenia and ADHD classification tasks. Explainable AI via Integrated Gradients highlighted that performance depends on the interaction between masking strategy and architecture rather than a single dominant configuration.
arXiv:2609.21227v1 Announce Type: cross Abstract: Factual hallucination is commonly defined by incorrect factual outputs. We study a paraphrase-induced hallucination setting, where a model answers a...
Draft-OPD introduces an on‑policy distillation method for speculative draft models, addressing the mismatch between supervised fine‑tuning and inference by letting the target model supervise the drafter on draft‑induced states. The approach uses target‑assisted rollouts for stable continuations and replays drafting from error positions exposed during verification, enabling the drafter to learn from both accepted and rejected proposals. Experiments demonstrate that Draft‑OPD achieves more than five‑fold lossless acceleration across diverse tasks, outperforming prior draft models such as EAGLE‑3 and DFlash by 23 % and 13 % respectively.
KnowDemo is a framework that generates diverse robot demonstrations from human videos by leveraging structured manipulation knowledge. It uses a vision‑language model to extract task requirements and permissible execution variations, then resolves these against target‑scene entities to guide candidate generation and screening before motion planning. The resulting demonstrations feature multimodal behavior, alternative contact strategies, and valid subtask orders, and have been shown to improve planning success and enable sim‑to‑real policy transfer across three tasks.
The paper investigates the effect of using semantic positive pairs—different instances of the same class—in self‑supervised visual representation learning. By creating matched ImageNet‑1K subsets of augmented pairs and manually curated semantic pairs, the authors compare contrastive and non‑contrastive SSL methods under identical training conditions. Across transfer learning and object detection tasks, semantic‑pair pretraining consistently outperforms augmented‑pair pretraining, with contrastive methods like SimCLR showing the largest gains, indicating that semantic pairs foster additional invariances beyond standard augmentations.
OneBid is a unified auto‑bidding foundation model that consolidates diverse cost‑per‑X (oCPX) advertising scenarios into a single framework. It builds on Decision Transformer by conditioning on two atomic signals—Return‑to‑Go for conversion value and Cost‑to‑Go for cost ratio—and incorporates value‑aware regularization. A sequence‑level Mixture‑of‑Experts architecture captures cross‑scenario knowledge while preserving low latency, and a Critic‑guided Relative Offline Policy optimization (CROP) aligns the backbone with scenario‑specific preferences without unsafe online exploration. In production at Kuaishou, OneBid achieved a 2.2% overall ADVV increase and up to 13.1% in the ROAS scenario.
The paper introduces PA‑RL, a reinforcement‑learning framework that uses artificial potential fields as the action representation for contact‑rich robotic manipulation. Instead of directly commanding motion, the policy adjusts potential‑field parameters, which a Cartesian impedance controller then executes, decoupling task strategy from low‑level control. In peg‑in‑hole experiments, PA‑RL achieved a 100% success rate in simulation, outperformed baselines in torque and acceleration variation, and transferred to a real robot without fine‑tuning.
arXiv:2606.03963v4 Announce Type: replace-cross Abstract: Deep reinforcement learning enables autonomous robots to learn complex navigation tasks, but still relies heavily on time consuming manual re...
MemeLens is a unified multilingual, multitask Vision‑Language Model designed to improve meme understanding across a wide range of tasks such as hate, misogyny, propaganda, sentiment, and humour. The authors consolidated 38 public meme datasets, mapping their labels into a shared taxonomy of 20 tasks covering harm, targets, figurative intent, and affect, and conducted extensive experiments to show that multimodal training and a unified approach outperform fine‑tuning on individual datasets. All experimental resources, the model, and the datasets are released publicly for community use.
The paper introduces the COmmunity-centered Peer Engaged Support (COPES) dataset and a three‑axis evaluation framework to gauge how well Large Language Models (LLMs) align with community perspectives on mental‑health support queries. Experiments show that fine‑tuning LLMs on COPES improves strategy alignment and emotion‑tone alignment by over 50% for general‑purpose models, yet these gains are uneven across subreddits and coping strategies. The study also finds that post‑training shifts the model’s recommendations toward problem‑focused advice while reducing emotion‑focused responses, indicating persistent disparities in performance across different communities and needs.
The paper examines whether fine‑tuning large language models (LLMs) with personality‑labelled data improves their ability to act as socially interactive agents. Two small open‑weight LLMs were fine‑tuned on a corpus of personality‑labelled social media posts and dialogues, and the resulting models were evaluated in various social interaction scenarios by independent LLM judges. The findings show that the fine‑tuned models do not outperform their baseline counterparts in role‑playing personalities, though they offer comparable text quality and increased linguistic diversity for the Qwen models; low inter‑rater agreement limits confidence in the results, suggesting future work should focus on training data quality and domain alignment.
SynthDemo‑RL introduces a teacher‑student framework that uses an automated teacher to generate successful manipulation trajectories from simulator‑privileged state, which are then distilled into a Vision‑Language‑Action (VLA) student via supervised fine‑tuning. The student is further refined with PPO using binary task‑success rewards. On the LIBERO‑PRO benchmark, SynthDemo‑RL rescues all 27 previously unsolvable tasks and achieves near‑perfect success rates, matching performance that would otherwise require human demonstrations.
Attention-Aware Routing (AAR) augments the router in Mixture-of-Experts language models with temporal and spectral features derived from a sliding window of attention weights, thereby separating contextual information from the token’s hidden state. By keeping the base transformer frozen and training only routing parameters, AAR achieves a +3.37‑point improvement on GSM8K over a routing‑only baseline and demonstrates that routing changes propagate through the residual stream to reshape attention without directly updating the attention mechanism. The method also reduces long diverging generations, shows depth‑sensitivity affecting retrieval versus reasoning, and offers a controlled probe of routing‑relevant information across layers.
BoostAPR is a three-stage framework that improves automated program repair by using execution-grounded reinforcement learning with dual reward models. The approach first fine‑tunes a model on execution‑verified demonstrations, then trains a sequence‑level assessor and a line‑level credit allocator from execution outcomes, and finally applies PPO optimization where the line‑level model redistributes rewards to critical edit regions. Evaluated on SWE‑Gym and four benchmarks, BoostAPR achieves significant gains, including 40.7% on SWE‑bench Verified and 95.0% on QuixBugs, demonstrating strong cross‑language generalization.
Fine‑tuning reshapes internal representations of large language models, affecting attention patterns and layer‑wise activations. The study shows that components identified by EAP as important for task performance cluster in specific layers, yet these layers do not align with those undergoing the largest representational changes. Additionally, overlapping EAP components across different tasks do not guarantee cross‑task transfer and can even degrade performance when tasks differ in nature.