Zobrazeno 1 - 10
of 88
pro vyhledávání: '"Jain Paras P"'
Autor:
Guilberteau, Jules, Jain, Paras, Jolly, Mohit Kumar, Duteil, Nastassia Pouradier, Pouchol, Camille
Phenotypic heterogeneity along the epithelial-mesenchymal (E-M) axis contributes to cancer metastasis and drug resistance. Recent experimental efforts have collated detailed time-course data on the emergence and dynamics of E-M heterogeneity in a cel
Externí odkaz:
http://arxiv.org/abs/2309.09569
Cloud applications are increasingly distributing data across multiple regions and cloud providers. Unfortunately, wide-area bulk data transfers are often slow, bottlenecking applications. We demonstrate that it is possible to significantly improve in
Externí odkaz:
http://arxiv.org/abs/2210.07259
Fine-tuning models on edge devices like mobile phones would enable privacy-preserving personalization over sensitive data. However, edge training has historically been limited to relatively small models with simple architectures because training is b
Externí odkaz:
http://arxiv.org/abs/2207.07697
Autor:
Wu, Zhanghao, Jain, Paras, Wright, Matthew A., Mirhoseini, Azalia, Gonzalez, Joseph E., Stoica, Ion
Graph neural networks are powerful architectures for structured datasets. However, current methods struggle to represent long-range dependencies. Scaling the depth or width of GNNs is insufficient to broaden receptive fields as larger GNNs encounter
Externí odkaz:
http://arxiv.org/abs/2201.08821
Question answering models struggle to generalize to novel compositions of training patterns, such to longer sequences or more complex test structures. Current end-to-end models learn a flat input embedding which can lose input syntax context. Prior a
Externí odkaz:
http://arxiv.org/abs/2111.03642
Autor:
Ichnowski, Jeffrey, Jain, Paras, Stellato, Bartolomeo, Banjac, Goran, Luo, Michael, Borrelli, Francesco, Gonzalez, Joseph E., Stoica, Ion, Goldberg, Ken
First-order methods for quadratic optimization such as OSQP are widely used for large-scale machine learning and embedded optimal control, where many related problems must be rapidly solved. These methods face two persistent challenges: manual hyperp
Externí odkaz:
http://arxiv.org/abs/2107.10847
Recent work learns contextual representations of source code by reconstructing tokens from their context. For downstream semantic understanding tasks like summarizing code in English, these representations should ideally capture program functionality
Externí odkaz:
http://arxiv.org/abs/2007.04973
Akademický článek
Tento výsledek nelze pro nepřihlášené uživatele zobrazit.
K zobrazení výsledku je třeba se přihlásit.
K zobrazení výsledku je třeba se přihlásit.
Autor:
Jain, Paras, Jain, Ajay, Nrusimha, Aniruddha, Gholami, Amir, Abbeel, Pieter, Keutzer, Kurt, Stoica, Ion, Gonzalez, Joseph E.
We formalize the problem of trading-off DNN training time and memory requirements as the tensor rematerialization optimization problem, a generalization of prior checkpointing strategies. We introduce Checkmate, a system that solves for optimal remat
Externí odkaz:
http://arxiv.org/abs/1910.02653
Publikováno v:
Compiler Construction (CC) 2019
Modern microprocessors are equipped with Single Instruction Multiple Data (SIMD) or vector instructions which expose data level parallelism at a fine granularity. Programmers exploit this parallelism by using low-level vector intrinsics in their code
Externí odkaz:
http://arxiv.org/abs/1902.02816