LoRA, PEFT, instruction tuning and domain adaptation — adapting a pretrained model without paying to train one.
arXiv:2611.06007v1 Announce Type: new Abstract: Motorized Two-Wheelers (MTWs) dominate personal transport across much of the Global South, yet remain largely overlooked in predictive perception resea...
arXiv:2602.11499v3 Announce Type: replace Abstract: Multimodal Large Language Models have shown promising capabilities in bridging visual and textual reasoning, yet their reasoning capabilities in Op...
arXiv:2605.20342v3 Announce Type: replace Abstract: Training large multimodal models (LMMs) via reinforcement learning (RL) to natively invoke video-processing tools (e.g., cropping) has become a pro...
The study evaluates whether overlap metrics like Dice can reliably indicate metabolic tumor volume (MTV) estimation accuracy in PET segmentation models. Across five FDG PET/CT models, higher Dice scores correlated with lower relative MTV error on average, but at the patient level, 16.8% of cases fell on opposite sides of clinically relevant thresholds. Additionally, many models missed lesions, showing that model ranking alone does not guarantee reliable individual MTV estimates.
The paper introduces NEO-VLA, an encoder‑free vision‑language‑action framework that tokenizes images with lightweight patch convolutions and lets visual and language features evolve together throughout the backbone. By allowing a flow‑matching action expert to access instruction‑conditioned visual representations at multiple depths, NEO‑VLA consistently outperforms encoder‑based counterparts across diverse manipulation tasks and distribution shifts. Layer‑wise probing reveals that action‑critical signals peak early in the visual hierarchy, underscoring the benefit of accessing representations across the computational pipeline rather than relying on a fixed encoder output.
The study presents a conditional diffusion-based generative model that augments T2-weighted prostate MRI data with clinically significant prostate cancer (csPCa) and transition zone (TZ) lesions. By extending MAISI‑v2 with csPCa class, zonal, and lesion‑mask conditioning and fine‑tuning on 1,303 patients, the model achieves a substantial reduction in Fréchet inception distance (FID) from 13.28 to 2.04. Synthetic augmentation improves csPCa classification AUC across five simulated training scenarios, though the benefit does not significantly transfer to a cross‑institutional dataset.
ModalSearcher introduces an autonomous cross‑modal search framework that lets an agent dynamically choose among direct reasoning, text‑to‑text search, text‑to‑image search, and reverse‑image search, actively incorporating visual evidence as needed. The system builds an interleaved multimodal knowledge graph via a scalable pipeline that samples diverse paths and uses backward question construction to prevent shortcut reasoning. Trained with supervised fine‑tuning and reinforcement learning, ModalSearcher‑8B outperforms Qwen3‑VL‑8B by 16.2 accuracy points across five multimodal knowledge‑intensive QA benchmarks and shows consistent gains on active visual‑search tasks.
The paper introduces a new task, Intra-Architecture Source Attribution for Stable Diffusion (IntrAA‑SD), which focuses on distinguishing between fine‑tuned versions of the same Stable Diffusion XL architecture. It presents a 140K‑image benchmark dataset with in‑domain and out‑of‑distribution prompts, and proposes a baseline method using a style‑conditioned Mixture‑of‑Experts network that leverages a frozen CLIP backbone and a LoRA‑adapted DINOv2 network. The authors evaluate their approach on attribution accuracy and conduct ablation studies, providing code and data publicly on GitHub.
The paper investigates whether model merging requires all experts to use the same LoRA rank. Experiments with 36 experts across three domains and multiple ranks show that allowing each expert its own optimal rank improves multi‑domain accuracy by up to 1.8 percentage points, while traditional uniform‑rank methods can even degrade performance. The authors attribute this to how merge methods select spectral components, arguing that rank can be chosen per expert when the merge operator does not rely on magnitude alone.
The paper introduces a visual-native agent harness that registers every tool‑returned image as an addressable reference, enabling reuse of intermediate visual evidence in multimodal deep search tasks. It also proposes On‑Policy Data Evolution (ODE), a closed‑loop data generator that refines training data across rounds based on policy rollouts, targeting the agent’s evolving needs. Experiments on eight benchmarks show ODE boosts the Qwen3‑VL‑8B agent’s performance from 24.9% to 39.0% on average, and from 30.6% to 41.5% at 30B, while validating the benefits of image bank reuse and rollout‑feedback evolution.
The paper introduces MechScam, a benchmark of 6,614 videos that covers diverse scam mechanisms and non‑scam content, providing evidence‑supported explanations and annotations of scam elements. It also proposes ScamCourt, a debate framework where prosecution and defense debaters argue over verified citations, and a judge decides based on the transcript. Experiments show that current large multimodal language models struggle to differentiate scam videos from similar non‑scam content, while ScamCourt reduces false positives and better resists misleading arguments.
The paper reports on training and calibration experiments for three open models on the ARC‑AGI‑2 benchmark, using offline runs on four L4 GPUs for up to 12 hours. For the NVARC Qwen3‑4B grid model, a curated corpus of 12,000 executable rules and 95,993 tasks was built, and fine‑tuning on 1,920 real task records raised the semi‑private score from 28.96 to 33.06, with no synthetic‑task checkpoint outperforming the real‑task model. For the Qwen3.8‑27B reasoning model, fine‑tuning on 1,036 of its own reasoning traces and a specialized inference procedure increased the score from 26.81 to 44.03, while the Qwen3.8‑Flash‑Next mixture‑of‑experts model achieved a best score of 56.67 after expert pruning, routing statistics adaptation, and an FP8 indexer cache.
The study explores text-only domain adaptation for Hausa medical automatic speech recognition (ASR) using a fixed acoustic model and two decoding-time language‑model fusion techniques: Shallow Fusion (SF) and the Density Ratio Approach (DRA). A 4‑gram KenLM trained on 19,421 Hausa health sentences is combined with a 75,372‑sentence general‑domain model, and evaluated on 250 physician‑reviewed prompts from 10 speakers across six northern Nigerian states. Both fusion methods reduce word error rate (WER) from 29.56% to 23.55% (SF) and 23.33% (DRA), outperforming greedy decoding and showing no significant difference between the two approaches.
The paper introduces AssameseLM-E2B, a 2‑billion‑parameter language model tailored for the low‑resource Assamese language, and its instruction‑tuned variant AssameseLM-E2B‑it. Built on the Gemma‑4 architecture, it uses an expanded byte‑level BPE vocabulary that cuts tokenization overhead by nearly 500 % compared with multilingual baselines, and employs a parameter‑efficient Continual Pre‑Training pipeline with Hybrid LoRA followed by supervised fine‑tuning on a balanced, multi‑domain dataset. Evaluation shows strong performance with a Bits‑per‑Byte score of 0.7488 on FLORES‑200, a 96.7 % RAG score, and 100 % refusal accuracy on adversarial prompts, highlighting the effectiveness of parameter‑efficient adaptation for locally deployable models under constrained resources.
arXiv:2603.13275v4 Announce Type: replace-cross Abstract: Accurate prediction of surgical duration is pivotal for hospital resource management. Although recent supervised learning approaches, from ma...
arXiv:2603.19310v5 Announce Type: replace-cross Abstract: Reinforcement learning (RL) fine-tuning improves large language model (LLM) reasoning using reward signals on sampled rollouts. When reward l...
arXiv:2611.05538v1 Announce Type: new Abstract: Historically automated electroencephalography (EEG) systems was done using task-specific supervised models trained on purposely annotated data. A more...
arXiv:2611.06066v1 Announce Type: new Abstract: Automatic modulation classification (AMC) is a key component of cognitive wireless systems, enabling spectrum awareness and adaptive communication. Mos...
arXiv:2611.06194v1 Announce Type: new Abstract: Reliability management of industrial fleets depends on early fault warnings, yet lifecycle data across similar assets are highly imbalanced. A few asse...
arXiv:2611.06598v1 Announce Type: new Abstract: Low-Rank Adaptation (LoRA) adapts a frozen language model through a compact learned update, making it a natural setting for asking how much information...