AIBearisharXiv – CS AI · 2d ago7/10
🧠Researchers benchmarked five physics foundation models across 8 physical dynamics and 25 test regimes, revealing that current models function as conditional rather than universal generalists. The study demonstrates that model performance heavily depends on physical regime, temporal scale, and distribution shifts, with pretraining and scaling unable to reliably overcome these limitations.
AIBullisharXiv – CS AI · 3d ago7/10
🧠PromptEmbedder introduces a dual-LLM framework that decouples text embedding from specific model architectures, achieving comparable performance to LoRA while reducing GPU memory by 40% and accelerating training 3.7x. The innovation enables efficient transfer across different LLM backbones by retraining only a lightweight alignment matrix rather than entire models.
AIBullisharXiv – CS AI · May 127/10
🧠Researchers introduce OPT-BENCH, a framework for training LLMs on NP-hard optimization problems using quality-aware reinforcement learning. Testing on Qwen2.5-7B achieves 93.1% success rate and 46.6% quality ratio, substantially outperforming GPT-4o, with demonstrated transfer benefits across mathematics, logic, and reasoning tasks.
🧠 GPT-4
AIBullisharXiv – CS AI · May 117/10
🧠Researchers introduce rubric-grounded reinforcement learning, a framework that trains AI models using structured, multi-criterion rewards from an LLM judge rather than binary outcomes. Training Llama-3.1-8B on scientific documents achieved 71.7% normalized reward and demonstrated improved performance on multiple reasoning benchmarks, suggesting that document-grounded training signals can produce generalizable reasoning capabilities.
🧠 Llama
AIBullisharXiv – CS AI · May 97/10
🧠Researchers introduce ScaleLogic, a synthetic reasoning framework that systematically studies how reinforcement learning improves LLM reasoning across varying task difficulty and logical complexity. The study reveals that RL training compute follows a power law with reasoning depth, with scaling efficiency improving when models train on more expressively complex logic, suggesting that training content quality matters as much as training volume.
AIBullisharXiv – CS AI · May 97/10
🧠Researchers introduce LANTERN, a framework that uses large language models to automatically generate task descriptions and intelligently aggregate knowledge from multiple source tasks for reinforcement learning. The system achieves 40-60% improvements in sample efficiency by adaptively weighting source policies based on task similarity and managing teacher-student knowledge transfer through uncertainty-aware gating.
AIBullisharXiv – CS AI · Apr 207/10
🧠Researchers have developed an exascale workflow using graph foundation models trained on 544+ million atomistic structures to accelerate materials discovery. The system can screen 1.1 billion structures in 50 seconds—a task requiring years of traditional computation—and demonstrates strong transfer learning capabilities across diverse chemical applications.
AIBullisharXiv – CS AI · Apr 147/10
🧠Researchers propose a method to adapt 2D multimodal large language models for 3D medical imaging analysis, introducing a Text-Guided Hierarchical Mixture of Experts framework that enables task-specific feature extraction. The approach demonstrates improved performance on medical report generation and visual question answering tasks while reusing pre-trained parameters from 2D models.
AIBullisharXiv – CS AI · Mar 46/104
🧠Researchers have developed EvoSkill, an automated framework that enables AI agents to discover and refine domain-specific skills through iterative failure analysis. The system demonstrated significant performance improvements on specialized tasks, with accuracy gains of 7.3% on financial data analysis and 12.1% on search-augmented QA, while showing transferable capabilities across different domains.
AIBullisharXiv – CS AI · Mar 47/103
🧠Researchers propose a new IMPRINT framework for transfer learning that improves foundation model adaptation to new tasks without parameter optimization. The framework identifies three key components and introduces a clustering-based variant that outperforms existing methods by 4%.
AIBullisharXiv – CS AI · Mar 47/103
🧠Researchers developed D2E (Desktop to Embodied AI), a framework that uses desktop gaming data to pretrain AI models for robotics tasks. Their 1B-parameter model achieved 96.6% success on manipulation tasks and 83.3% on navigation, matching performance of models up to 7 times larger while using scalable desktop data instead of expensive physical robot training data.
AIBullisharXiv – CS AI · Mar 47/103
🧠Researchers developed a new neural solver model using GCON modules and energy-based loss functions that achieves state-of-the-art performance across multiple graph combinatorial optimization tasks. The study demonstrates effective transfer learning between related optimization problems through computational reducibility-informed pretraining strategies, representing progress toward foundational AI models for combinatorial optimization.
AINeutralarXiv – CS AI · Mar 37/105
🧠Researchers identified that fine-tuning non-robust pretrained AI models with robust objectives can lead to poor performance, termed 'suboptimal transfer.' They propose Epsilon-Scheduling, a novel training technique that adjusts perturbation strength during training to improve both task adaptation and adversarial robustness.
AIBullisharXiv – CS AI · 2d ago6/10
🧠Researchers developed an uncertainty-aware transfer learning framework using Temporal Fusion Transformers to enable energy forecasting models trained on one building to work effectively on different buildings with minimal retraining. The approach achieved 93.2% prediction interval coverage and demonstrated that freezing most model parameters while fine-tuning only output layers produces superior cross-building generalization compared to full model retraining.
AINeutralarXiv – CS AI · 2d ago6/10
🧠Researchers benchmark supervised fine-tuned vision-language models against frontier zero-shot AI baselines on screen-conditioned action prediction using the PiSAR dataset. A fine-tuned Qwen3-VL-8B model substantially outperforms GPT and Claude zero-shot approaches (0.783 vs 0.459-0.482 semantic similarity), but the same training recipe fails on Gemma-4-26B, revealing critical architecture-to-method misalignment in model optimization.
🧠 GPT-5🧠 Claude🧠 Opus
AINeutralarXiv – CS AI · 2d ago6/10
🧠Researchers propose a learning-based visual peg-in-hole system that trains on multiple shapes in simulation and adapts to unseen shapes in real-world environments with minimal sim-to-real transfer costs. The approach decouples perception from control through modular networks, achieving 100% success rates on EV charging systems with only hundreds of auto-labeled training samples.
AINeutralarXiv – CS AI · 2d ago6/10
🧠Researchers introduce a neuron-centric model fusion algorithm that combines independently trained neural networks without retraining by matching intermediate representations and using neuron attribution scores. The method outperforms existing approaches in zero-shot and non-IID scenarios across multiple architectures including VGGs, ResNets, and Vision Transformers.
AINeutralarXiv – CS AI · 2d ago6/10
🧠Researchers introduce StreamSynth, a new framework enabling large language models to learn and improve synthetic data generation across sequential tasks by accumulating experience and transferring knowledge between related synthesis problems. The SynLearner framework demonstrates that LLMs can leverage historical task insights to enhance future data generation quality, establishing synthetic data creation as an experience-driven process rather than isolated operations.
AINeutralarXiv – CS AI · 2d ago6/10
🧠Researchers propose an accuracy-aware pruning mechanism for CNNs that improves upon existing Layer-wise Relevance Propagation (LRP) methods to reduce model size without degrading performance in transfer learning scenarios with limited data. The approach dynamically adjusts pruning rates using harmonic mean of class accuracy, achieving 15% improvement in compression efficiency while maintaining task-specific accuracy.
AINeutralarXiv – CS AI · 3d ago6/10
🧠FLORO is a multimodal geospatial foundation model that learns from diverse remote sensing data across multiple sensor types and resolutions with minimal pretraining data. Despite using significantly smaller datasets than competing models, FLORO demonstrates strong transfer learning performance on ecological and environmental applications, achieving competitive results on scene classification, segmentation, and regression tasks.
AINeutralarXiv – CS AI · 3d ago6/10
🧠Researchers propose a multi-dimensional evaluation framework for EEG foundation models that tests performance under realistic biomedical constraints like limited labeled data and reduced sensor coverage. Analysis of models including LaBraM, CSBrain, and CBraMod reveals foundation models excel at long-context tasks but struggle with short-window Brain-Computer Interface applications and channel constraints compared to supervised alternatives.
AINeutralarXiv – CS AI · 3d ago6/10
🧠Researchers demonstrate that Transformers develop analogical reasoning—the ability to transfer relational patterns across different domains—through two key mechanisms: geometric alignment of structures in embedding space and functor application. This mechanistic understanding bridges cognitive science and neural network architecture, with findings validated across both synthetic tasks and pretrained large language models.
AINeutralarXiv – CS AI · 4d ago6/10
🧠Researchers propose G-Substrate, a novel graph framework that treats graph structures as persistent substrates across multiple data modalities and tasks rather than isolated, task-specific constructs. The approach uses unified structural schemas and role-based training to enable graph representations to accumulate knowledge across heterogeneous domains, demonstrating superior performance compared to traditional isolated and multi-task learning methods.
AINeutralarXiv – CS AI · 4d ago6/10
🧠Researchers introduce CogAdapt, a framework that adapts clinical ECG foundation models to wearable cognitive load assessment by bridging the gap between hospital-grade 12-lead sensors and 3-lead wearable devices. The approach achieves strong cross-subject generalization on benchmark datasets, demonstrating the feasibility of transferring pre-trained medical models to consumer health applications.
AIBullisharXiv – CS AI · May 126/10
🧠Researchers introduce CORTEG, a framework that adapts pretrained scalp-EEG foundation models to intracranial ECoG recordings, enabling brain-computer interfaces to learn across patients with minimal calibration time. The approach demonstrates competitive or superior performance on finger trajectory and audio envelope decoding tasks while reducing per-patient training requirements to 10-30 minutes.
