Zobrazeno 1 - 10
of 320
pro vyhledávání: '"Kim Changhoon"'
Transformer-based large-scale pre-trained models achieve great success, and fine-tuning, which tunes a pre-trained model on a task-specific dataset, is the standard practice to utilize these models for downstream tasks. Recent work has developed adap
Externí odkaz:
http://arxiv.org/abs/2412.03587
Autor:
Patel, Maitreya, Kusumba, Abhiram, Cheng, Sheng, Kim, Changhoon, Gokhale, Tejas, Baral, Chitta, Yang, Yezhou
Contrastive Language-Image Pretraining (CLIP) models maximize the mutual information between text and visual modalities to learn representations. This makes the nature of the training data a significant factor in the efficacy of CLIP for downstream t
Externí odkaz:
http://arxiv.org/abs/2411.02545
Autor:
Nath, Utkarsh, Goel, Rajeev, Jeon, Eun Som, Kim, Changhoon, Min, Kyle, Yang, Yezhou, Yang, Yingzhen, Turaga, Pavan
To address the data scarcity associated with 3D assets, 2D-lifting techniques such as Score Distillation Sampling (SDS) have become a widely adopted practice in text-to-3D generation pipelines. However, the diffusion models used in these techniques a
Externí odkaz:
http://arxiv.org/abs/2408.05938
In the evolving landscape of text-to-image (T2I) diffusion models, the remarkable capability to generate high-quality images from textual descriptions faces challenges with the potential misuse of reproducing sensitive content. To address this critic
Externí odkaz:
http://arxiv.org/abs/2405.16341
Autor:
Hong, Jinyung, Jeon, Eun Som, Kim, Changhoon, Park, Keun Hee, Nath, Utkarsh, Yang, Yezhou, Turaga, Pavan, Pavlic, Theodore P.
Biased attributes, spuriously correlated with target labels in a dataset, can problematically lead to neural networks that learn improper shortcuts for classifications and limit their capabilities for out-of-distribution (OOD) generalization. Althoug
Externí odkaz:
http://arxiv.org/abs/2403.14140
Text-to-image (T2I) diffusion models, notably the unCLIP models (e.g., DALL-E-2), achieve state-of-the-art (SOTA) performance on various compositional T2I benchmarks, at the cost of significant computational resources. The unCLIP stack comprises T2I
Externí odkaz:
http://arxiv.org/abs/2312.04655
The rapid advancement of generative models, facilitating the creation of hyper-realistic images from textual descriptions, has concurrently escalated critical societal concerns such as misinformation. Although providing some mitigation, traditional f
Externí odkaz:
http://arxiv.org/abs/2306.04744
Autor:
Le, Yanfang, Lee, Jeongkeun, Blendin, Jeremias, Chen, Jiayi, Nikolaidis, Georgios, Pan, Rong, Soule, Robert, Akella, Aditya, Segura, Pedro Yebenes, singhvi, Arjun, Li, Yuliang, Meng, Qingkai, Kim, Changhoon, Arslan, Serhat
State-of-the-art congestion control algorithms for data centers alone do not cope well with transient congestion and high traffic bursts. To help with these, we revisit the concept of direct \emph{backward} feedback from switches and propose Back-to-
Externí odkaz:
http://arxiv.org/abs/2305.00538
Generative models have enabled the creation of contents that are indistinguishable from those taken from nature. Open-source development of such models raised concerns about the risks of their misuse for malicious purposes. One potential risk mitigat
Externí odkaz:
http://arxiv.org/abs/2304.09752
Autor:
Ibanez, Stephen, Mallery, Alex, Arslan, Serhat, Jepsen, Theo, Shahbaz, Muhammad, Kim, Changhoon, McKeown, Nick
Many recent papers have demonstrated fast in-network computation using programmable switches, running many orders of magnitude faster than CPUs. The main limitation of writing software for switches is the constrained programming model and limited sta
Externí odkaz:
http://arxiv.org/abs/2212.06658