Results 51 to 60 of about 226,203 (262)

Parameter-Efficient Sparsity for Large Language Models Fine-Tuning

Proceedings of the Thirty-First International Joint Conference on Artificial Intelligence, 2022
With the dramatically increased number of parameters in language models, sparsity methods have received ever-increasing research focus to compress and accelerate the models. While most research focuses on how to accurately retain appropriate weights while maintaining the performance of the compressed model, there are challenges in the computational ...
Li, Yuchao, Luo, Fuli, Tan, Chuanqi, Wang, Mengdi, Huang, Songfang, Li, Shen, Bai, Junjie   +6 more
openaire   +2 more sources

Developing evidence‐based, cost‐effective P4 cancer medicine for driving innovation in prevention, therapeutics, patient care and reducing healthcare inequalities

Molecular Oncology, EarlyView.
The cancer problem is increasing globally with projections up to the year 2050 showing unfavourable outcomes in terms of incidence and cancer‐related deaths. The main challenges are prevention, improved therapeutics resulting in increased cure rates and enhanced health‐related quality of life.
Ulrik Ringborg, Joachim von Braun, Julio Celis, Anton Berns, Michael Baumann, Tit Albreht, Nancy Abou‐Zeid, Vanderlei Bagnato, Christian Brandts, Chien‐Jen Chen, Massimiliano di Pietro, Manjit Dosanjh, Thomas Dubois, Alexander Eggermont, Angelika Eggert, Ingemar Ernberg, Sara Faithfull, Johannes Förner, Stefan Fröhling, Manuel Heitor, Leroy Hood, Wei Jiang, Bengt Jönsson, Ravi Kannan, Maria Leptin, Su Li, Peter Lindgren, Douglas Lowy, Jun Ma, Alex Markham, Péter Nagy, Simon Oberst, M. Iqbal Parker, Danielle Rodin, Kevin Ryan, Joachim Schüz, Richard Sullivan, Josep Tabernero, Peter Turkson, Oliver Várhelyi, Harold Varmus, Chijie Wang, Elisabete Weiderpass, Nils Wilking   +43 more
wiley   +1 more source

Large language models for PHM: a review of optimization techniques and applications

Autonomous Intelligent Systems
The rapid advancement of Large Language Models (LLMs) has created unprecedented opportunities for industrial automation, process optimization, and decision support systems.
Tingyi Yu, Junya Tang, Qingyun Yu, Li Li, Ying Liu, Raul Poler   +5 more
doaj   +1 more source

Parameter-Efficient Fine-Tuning With Adapters

In the arena of language model fine-tuning, the traditional approaches, such as Domain-Adaptive Pretraining (DAPT) and Task-Adaptive Pretraining (TAPT), although effective, but computational intensive. This research introduces a novel adaptation method utilizing the UniPELT framework as a base and added a PromptTuning Layer, which significantly reduces
Chen, Keyu, Pang, Yuan, Yang, Zi
openaire   +2 more sources

Targeted modulation of IGFL2‐AS1 reveals its translational potential in cervical adenocarcinoma

Molecular Oncology, EarlyView.
Cervical adenocarcinoma patients face worse outcomes than squamous cell carcinoma counterparts despite similar treatment. The identification of IGFL2‐AS1's differential expression provides a molecular basis for distinguishing these histotypes, paving the way for personalized therapies and improved survival in vulnerable populations globally.
Ricardo Cesar Cintra, Lucca Paolo Hsu Helmich, Daniel Rodrigues de Bastos, Laura Sichero, Patrícia Savio de Araujo‐Souza, Fabio Passetti, Luisa Lina Villa   +6 more
wiley   +1 more source

Strong Baselines for Parameter-Efficient Few-Shot Fine-Tuning

Proceedings of the AAAI Conference on Artificial Intelligence
Few-shot classification (FSC) entails learning novel classes given only a few examples per class after a pre-training (or meta-training) phase on a set of base classes. Recent works have shown that simply fine-tuning a pre-trained Vision Transformer (ViT) on new test classes is a strong approach for FSC. Fine-tuning ViTs, however, is expensive in time,
Basu, Samyadeep, Massiceti, Daniela, Hu, Shell Xu, Feizi, Soheil   +3 more
openaire   +2 more sources

Engineering tandem VHHs to target different epitopes to enhance antibody‐dependent cell‐mediated cytotoxicity

FEBS Open Bio, EarlyView.
Tandem VHH targeting distinct EGFR epitopes were engineered into a monovalent bispecific antibody (7D12‐EGA1‐Fc) with more potent ADCC without increasing affinity to EGFR. Structural modeling of 7D12‐EGA1‐Fc showed cross‐linking of separate EGFR domains to enhance CD16a engagement on NK cells.
Yuqiang Xu, Hao Jiang, Limin Chen, Fulai Zhou, Ying Jin, Mark L. Chiu   +5 more
wiley   +1 more source

Naturalness of Neutralino Dark Matter

, 2013
We investigate the level of fine-tuning of neutralino Dark Matter below 200 GeV in the low-energy phenomenological minimal supersymmetric Standard Model taking into account the newest results from XENON100 and the Large Hadron Collider as well as all ...
Grothaus, Philipp, Lindner, Manfred, Takanishi, Yasutaka   +2 more
core   +1 more source

Mycobacterial cell division arrest and smooth‐to‐rough envelope transition using CRISPRi‐mediated genetic repression systems

FEBS Open Bio, EarlyView.
CRISPRI‐mediated gene silencing and phenotypic exploration in nontuberculous mycobacteria. In this Research Protocol, we describe approaches to control, monitor, and quantitatively assess CRISPRI‐mediated gene silencing in M. smegmatis and M. abscessus model organisms.
Vanessa Point, Wafaa Achache, Janïs Laudouze, Eliana Sepulveda Ramos, Mickaël Maziero, Céline Crauste, Stéphane Canaan, Pierre Santucci   +7 more
wiley   +1 more source

Comparison between parameter-efficient techniques and full fine-tuning: A case study on multilingual news article classification.

PLoS ONE
Adapters and Low-Rank Adaptation (LoRA) are parameter-efficient fine-tuning techniques designed to make the training of language models more efficient. Previous results demonstrated that these methods can even improve performance on some classification ...
Olesya Razuvayevskaya, Ben Wu, João A Leite, Freddy Heppell, Ivan Srba, Carolina Scarton, Kalina Bontcheva, Xingyi Song   +7 more
doaj   +1 more source