parent_paper_title
stringclasses
63 values
parent_paper_arxiv_id
stringclasses
63 values
citation_shorthand
stringlengths
2
56
raw_citation_text
stringlengths
9
63
cited_paper_title
stringlengths
5
161
cited_paper_arxiv_link
stringlengths
32
37
cited_paper_abstract
stringlengths
406
1.92k
has_metadata
bool
1 class
is_arxiv_paper
bool
2 classes
bib_paper_authors
stringlengths
2
2.44k
bib_paper_year
float64
1.97k
2.03k
bib_paper_month
stringclasses
16 values
bib_paper_url
stringlengths
20
116
bib_paper_doi
stringclasses
269 values
bib_paper_journal
stringlengths
3
148
original_title
stringlengths
5
161
search_res_title
stringlengths
4
122
search_res_url
stringlengths
22
267
search_res_content
stringlengths
19
1.92k
Unlocking the Power of Rehearsal in Continual Learning: A Theoretical Perspective
2506.00205v1
rypescdivide
\cite{rypescdivide}
Divide and not forget: Ensemble of selectively trained experts in Continual Learning
http://arxiv.org/abs/2401.10191v3
Class-incremental learning is becoming more popular as it helps models widen their applicability while not forgetting what they already know. A trend in this area is to use a mixture-of-expert technique, where different models work together to solve the task. However, the experts are usually trained all at once using w...
true
true
Rype{\'s}{\'c}, Grzegorz and Cygert, Sebastian and Khan, Valeriya and Trzcinski, Tomasz and Zieli{\'n}ski, Bartosz Micha{\l} and Twardowski, Bart{\l}omiej
2,024
null
null
null
null
Divide and not forget: Ensemble of selectively trained experts in Continual Learning
Ensemble of selectively trained experts in Continual Learning - arXiv
https://arxiv.org/abs/2401.10191
Divide and not forget: Ensemble of selectively trained experts in Continual Learning. Authors:Grzegorz Rypeść, Sebastian Cygert, Valeriya Khan,
Unlocking the Power of Rehearsal in Continual Learning: A Theoretical Perspective
2506.00205v1
kirkpatrick2017overcoming
\cite{kirkpatrick2017overcoming}
Overcoming catastrophic forgetting in neural networks
http://arxiv.org/abs/1612.00796v2
The ability to learn tasks in a sequential fashion is crucial to the development of artificial intelligence. Neural networks are not, in general, capable of this and it has been widely thought that catastrophic forgetting is an inevitable feature of connectionist models. We show that it is possible to overcome this lim...
true
true
Kirkpatrick, James and Pascanu, Razvan and Rabinowitz, Neil and Veness, Joel and Desjardins, Guillaume and Rusu, Andrei A and Milan, Kieran and Quan, John and Ramalho, Tiago and Grabska-Barwinska, Agnieszka and others
2,017
null
null
null
Proceedings of the national academy of sciences
Overcoming catastrophic forgetting in neural networks
Overcoming catastrophic forgetting in neural networks
http://arxiv.org/pdf/1612.00796v2
The ability to learn tasks in a sequential fashion is crucial to the development of artificial intelligence. Neural networks are not, in general, capable of this and it has been widely thought that catastrophic forgetting is an inevitable feature of connectionist models. We show that it is possible to overcome this lim...
Unlocking the Power of Rehearsal in Continual Learning: A Theoretical Perspective
2506.00205v1
magistrielastic
\cite{magistrielastic}
Elastic Feature Consolidation for Cold Start Exemplar-Free Incremental Learning
http://arxiv.org/abs/2402.03917v3
Exemplar-Free Class Incremental Learning (EFCIL) aims to learn from a sequence of tasks without having access to previous task data. In this paper, we consider the challenging Cold Start scenario in which insufficient data is available in the first task to learn a high-quality backbone. This is especially challenging f...
true
true
Magistri, Simone and Trinci, Tomaso and Soutif, Albin and van de Weijer, Joost and Bagdanov, Andrew D
2,024
null
null
null
null
Elastic Feature Consolidation for Cold Start Exemplar-Free Incremental Learning
[2402.03917] Elastic Feature Consolidation for Cold Start Exemplar ...
https://arxiv.org/abs/2402.03917
Exemplar-Free Class Incremental Learning (EFCIL) aims to learn from a sequence of tasks without having access to previous task data. In this
Unlocking the Power of Rehearsal in Continual Learning: A Theoretical Perspective
2506.00205v1
saha2021gradient
\cite{saha2021gradient}
Gradient Projection Memory for Continual Learning
http://arxiv.org/abs/2103.09762v1
The ability to learn continually without forgetting the past tasks is a desired attribute for artificial learning systems. Existing approaches to enable such learning in artificial neural networks usually rely on network growth, importance based weight update or replay of old data from the memory. In contrast, we propo...
true
true
Saha, Gobinda and Garg, Isha and Roy, Kaushik
2,021
null
null
null
null
Gradient Projection Memory for Continual Learning
Gradient Projection Memory for Continual Learning
http://arxiv.org/pdf/2103.09762v1
The ability to learn continually without forgetting the past tasks is a desired attribute for artificial learning systems. Existing approaches to enable such learning in artificial neural networks usually rely on network growth, importance based weight update or replay of old data from the memory. In contrast, we propo...
Unlocking the Power of Rehearsal in Continual Learning: A Theoretical Perspective
2506.00205v1
lin2022trgp
\cite{lin2022trgp}
TRGP: Trust Region Gradient Projection for Continual Learning
http://arxiv.org/abs/2202.02931v1
Catastrophic forgetting is one of the major challenges in continual learning. To address this issue, some existing methods put restrictive constraints on the optimization space of the new task for minimizing the interference to old tasks. However, this may lead to unsatisfactory performance for the new task, especially...
true
true
Lin, Sen and Yang, Li and Fan, Deliang and Zhang, Junshan
2,022
null
null
null
International Conference on Learning Representations(ICLR)
TRGP: Trust Region Gradient Projection for Continual Learning
TRGP: Trust Region Gradient Projection for Continual Learning
http://arxiv.org/pdf/2202.02931v1
Catastrophic forgetting is one of the major challenges in continual learning. To address this issue, some existing methods put restrictive constraints on the optimization space of the new task for minimizing the interference to old tasks. However, this may lead to unsatisfactory performance for the new task, especially...
Unlocking the Power of Rehearsal in Continual Learning: A Theoretical Perspective
2506.00205v1
cbrs2020
\cite{cbrs2020}
Online continual learning from imbalanced data
null
null
true
false
Chrysakis, Aristotelis and Moens, Marie-Francine
2,020
null
null
null
null
Online continual learning from imbalanced data
Online Continual Learning from Imbalanced Data
https://proceedings.mlr.press/v119/chrysakis20a.html
We aim to evaluate memory population methods that are used in online continual learning, when dealing with highly imbalanced and temporally correlated streams
Unlocking the Power of Rehearsal in Continual Learning: A Theoretical Perspective
2506.00205v1
infors2022
\cite{infors2022}
Information-theoretic Online Memory Selection for Continual Learning
http://arxiv.org/abs/2204.04763v1
A challenging problem in task-free continual learning is the online selection of a representative replay memory from data streams. In this work, we investigate the online memory selection problem from an information-theoretic perspective. To gather the most information, we propose the \textit{surprise} and the \textit{...
true
true
Sun, Shengyang and Calandriello, Daniele and Hu, Huiyi and Li, Ang and Titsias, Michalis
2,022
null
null
null
null
Information-theoretic Online Memory Selection for Continual Learning
Information-theoretic Online Memory Selection for Continual Learning
https://openreview.net/forum?id=IpctgL7khPp
We present information-theoretic algorithms to tackle the online memory selection problem in task-free and data imbalanced continual learning.
Unlocking the Power of Rehearsal in Continual Learning: A Theoretical Perspective
2506.00205v1
shin2017continual
\cite{shin2017continual}
Continual Learning with Deep Generative Replay
http://arxiv.org/abs/1705.08690v3
Attempts to train a comprehensive artificial intelligence capable of solving multiple tasks have been impeded by a chronic problem called catastrophic forgetting. Although simply replaying all previous data alleviates the problem, it requires large memory and even worse, often infeasible in real world applications wher...
true
true
Shin, Hanul and Lee, Jung Kwon and Kim, Jaehong and Kim, Jiwon
2,017
null
null
null
Advances in Neural Information Processing Systems(Neurips)
Continual Learning with Deep Generative Replay
Continual Learning with Deep Generative Replay
http://arxiv.org/pdf/1705.08690v3
Attempts to train a comprehensive artificial intelligence capable of solving multiple tasks have been impeded by a chronic problem called catastrophic forgetting. Although simply replaying all previous data alleviates the problem, it requires large memory and even worse, often infeasible in real world applications wher...
Unlocking the Power of Rehearsal in Continual Learning: A Theoretical Perspective
2506.00205v1
chaudhry2018efficient
\cite{chaudhry2018efficient}
Efficient Lifelong Learning with A-GEM
null
null
true
false
Chaudhry, Arslan and Ranzato, Marc’Aurelio and Rohrbach, Marcus and Elhoseiny, Mohamed
2,018
null
null
null
null
Efficient Lifelong Learning with A-GEM
Efficient Lifelong Learning with A-GEM
http://arxiv.org/pdf/1812.00420v2
In lifelong learning, the learner is presented with a sequence of tasks, incrementally building a data-driven prior which may be leveraged to speed up learning of a new task. In this work, we investigate the efficiency of current lifelong approaches, in terms of sample complexity, computational and memory cost. Towards...
Unlocking the Power of Rehearsal in Continual Learning: A Theoretical Perspective
2506.00205v1
chaudhry2019continual
\cite{chaudhry2019continual}
Continual learning with tiny episodic memories
null
null
true
false
Dokania, P and Torr, P and Ranzato, M
2,019
null
null
null
null
Continual learning with tiny episodic memories
[PDF] Continual Learning with Tiny Episodic Memories
https://tajanthan.github.io/il/docs/cler.pdf
In this work, we empirically analyze the effective- ness of a very small episodic memory in a CL setup where each training example is only seen once.
Unlocking the Power of Rehearsal in Continual Learning: A Theoretical Perspective
2506.00205v1
rebuffi2017icarl
\cite{rebuffi2017icarl}
iCaRL: Incremental Classifier and Representation Learning
http://arxiv.org/abs/1611.07725v2
A major open problem on the road to artificial intelligence is the development of incrementally learning systems that learn about more and more concepts over time from a stream of data. In this work, we introduce a new training strategy, iCaRL, that allows learning in such a class-incremental way: only the training dat...
true
true
Rebuffi, Sylvestre-Alvise and Kolesnikov, Alexander and Sperl, Georg and Lampert, Christoph H
2,017
null
null
null
null
iCaRL: Incremental Classifier and Representation Learning
iCaRL: Incremental Classifier and Representation Learning
http://arxiv.org/pdf/1611.07725v2
A major open problem on the road to artificial intelligence is the development of incrementally learning systems that learn about more and more concepts over time from a stream of data. In this work, we introduce a new training strategy, iCaRL, that allows learning in such a class-incremental way: only the training dat...
Unlocking the Power of Rehearsal in Continual Learning: A Theoretical Perspective
2506.00205v1
gargtic2024
\cite{gargtic2024}
TiC-CLIP: Continual Training of CLIP Models
null
null
true
false
Garg, Saurabh and Farajtabar, Mehrdad and Pouransari, Hadi and Vemulapalli, Raviteja and Mehta, Sachin and Tuzel, Oncel and Shankar, Vaishaal and Faghri, Fartash
2,024
null
null
null
null
TiC-CLIP: Continual Training of CLIP Models
TiC-CLIP: Continual Training of CLIP Models
http://arxiv.org/pdf/2310.16226v3
Keeping large foundation models up to date on latest data is inherently expensive. To avoid the prohibitive costs of constantly retraining, it is imperative to continually train these models. This problem is exacerbated by the lack of any large scale continual learning benchmarks or baselines. We introduce the first se...
Unlocking the Power of Rehearsal in Continual Learning: A Theoretical Perspective
2506.00205v1
lopez2017gradient
\cite{lopez2017gradient}
Gradient Episodic Memory for Continual Learning
http://arxiv.org/abs/1706.08840v6
One major obstacle towards AI is the poor ability of models to solve new problems quicker, and without forgetting previously acquired knowledge. To better understand this issue, we study the problem of continual learning, where the model observes, once and one by one, examples concerning a sequence of tasks. First, we ...
true
true
Lopez-Paz, David and Ranzato, Marc'Aurelio
2,017
null
null
null
Advances in Neural Information Processing Systems(Neurips)
Gradient Episodic Memory for Continual Learning
Gradient Episodic Memory for Continual Learning
http://arxiv.org/pdf/1706.08840v6
One major obstacle towards AI is the poor ability of models to solve new problems quicker, and without forgetting previously acquired knowledge. To better understand this issue, we study the problem of continual learning, where the model observes, once and one by one, examples concerning a sequence of tasks. First, we ...
Unlocking the Power of Rehearsal in Continual Learning: A Theoretical Perspective
2506.00205v1
bennani2020generalisation
\cite{bennani2020generalisation}
Generalisation Guarantees for Continual Learning with Orthogonal Gradient Descent
http://arxiv.org/abs/2006.11942v4
In Continual Learning settings, deep neural networks are prone to Catastrophic Forgetting. Orthogonal Gradient Descent was proposed to tackle the challenge. However, no theoretical guarantees have been proven yet. We present a theoretical framework to study Continual Learning algorithms in the Neural Tangent Kernel reg...
true
true
Bennani, Mehdi Abbana and Sugiyama, Masashi
2,020
null
null
null
null
Generalisation Guarantees for Continual Learning with Orthogonal Gradient Descent
Generalisation Guarantees for Continual Learning with Orthogonal ...
https://ui.adsabs.harvard.edu/abs/2020arXiv200611942A/abstract
In Continual Learning settings, deep neural networks are prone to Catastrophic Forgetting. Orthogonal Gradient Descent was proposed to tackle the challenge.
Unlocking the Power of Rehearsal in Continual Learning: A Theoretical Perspective
2506.00205v1
doan2021theoretical
\cite{doan2021theoretical}
A Theoretical Analysis of Catastrophic Forgetting through the NTK Overlap Matrix
http://arxiv.org/abs/2010.04003v2
Continual learning (CL) is a setting in which an agent has to learn from an incoming stream of data during its entire lifetime. Although major advances have been made in the field, one recurring problem which remains unsolved is that of Catastrophic Forgetting (CF). While the issue has been extensively studied empirica...
true
true
Doan, Thang and Bennani, Mehdi Abbana and Mazoure, Bogdan and Rabusseau, Guillaume and Alquier, Pierre
2,021
null
null
null
null
A Theoretical Analysis of Catastrophic Forgetting through the NTK Overlap Matrix
A Theoretical Analysis of Catastrophic Forgetting through the NTK ...
https://arxiv.org/abs/2010.04003
In this paper, we show that the impact of CF increases as two tasks increasingly align. We introduce a measure of task similarity called the NTK overlap matrix.
Unlocking the Power of Rehearsal in Continual Learning: A Theoretical Perspective
2506.00205v1
yin2020optimization
\cite{yin2020optimization}
Optimization and Generalization of Regularization-Based Continual Learning: a Loss Approximation Viewpoint
http://arxiv.org/abs/2006.10974v3
Neural networks have achieved remarkable success in many cognitive tasks. However, when they are trained sequentially on multiple tasks without access to old data, their performance on early tasks tend to drop significantly. This problem is often referred to as catastrophic forgetting, a key challenge in continual lear...
true
true
Yin, Dong and Farajtabar, Mehrdad and Li, Ang and Levine, Nir and Mott, Alex
2,020
null
null
null
arXiv preprint arXiv:2006.10974
Optimization and Generalization of Regularization-Based Continual Learning: a Loss Approximation Viewpoint
‪Dong Yin‬ - ‪Google Scholar‬
https://scholar.google.com/citations?user=YtM8P88AAAAJ&hl=en
Optimization and Generalization of Regularization-Based Continual Learning: a Loss Approximation Viewpoint. D Yin, M Farajtabar, A Li, N Levine, A Mott. arXiv
Unlocking the Power of Rehearsal in Continual Learning: A Theoretical Perspective
2506.00205v1
cao2022provable
\cite{cao2022provable}
Provable Lifelong Learning of Representations
http://arxiv.org/abs/2110.14098v2
In lifelong learning, tasks (or classes) to be learned arrive sequentially over time in arbitrary order. During training, knowledge from previous tasks can be captured and transferred to subsequent ones to improve sample efficiency. We consider the setting where all target tasks can be represented in the span of a smal...
true
true
Cao, Xinyuan and Liu, Weiyang and Vempala, Santosh S
2,022
null
null
null
null
Provable Lifelong Learning of Representations
Provable Lifelong Learning of Representations
http://arxiv.org/pdf/2110.14098v2
In lifelong learning, tasks (or classes) to be learned arrive sequentially over time in arbitrary order. During training, knowledge from previous tasks can be captured and transferred to subsequent ones to improve sample efficiency. We consider the setting where all target tasks can be represented in the span of a smal...
Unlocking the Power of Rehearsal in Continual Learning: A Theoretical Perspective
2506.00205v1
li2022provable
\cite{li2022provable}
Provable and Efficient Continual Representation Learning
http://arxiv.org/abs/2203.02026v2
In continual learning (CL), the goal is to design models that can learn a sequence of tasks without catastrophic forgetting. While there is a rich set of techniques for CL, relatively little understanding exists on how representations built by previous tasks benefit new tasks that are added to the network. To address t...
true
true
Li, Yingcong and Li, Mingchen and Asif, M Salman and Oymak, Samet
2,022
null
null
null
arXiv preprint arXiv:2203.02026
Provable and Efficient Continual Representation Learning
Provable and Efficient Continual Representation Learning
http://arxiv.org/pdf/2203.02026v2
In continual learning (CL), the goal is to design models that can learn a sequence of tasks without catastrophic forgetting. While there is a rich set of techniques for CL, relatively little understanding exists on how representations built by previous tasks benefit new tasks that are added to the network. To address t...
Unlocking the Power of Rehearsal in Continual Learning: A Theoretical Perspective
2506.00205v1
evron2023continual
\cite{evron2023continual}
Continual Learning in Linear Classification on Separable Data
http://arxiv.org/abs/2306.03534v1
We analyze continual learning on a sequence of separable linear classification tasks with binary labels. We show theoretically that learning with weak regularization reduces to solving a sequential max-margin problem, corresponding to a special case of the Projection Onto Convex Sets (POCS) framework. We then develop u...
true
true
Evron, Itay and Moroshko, Edward and Buzaglo, Gon and Khriesh, Maroun and Marjieh, Badea and Srebro, Nathan and Soudry, Daniel
2,023
null
null
null
null
Continual Learning in Linear Classification on Separable Data
Continual Learning in Linear Classification on Separable Data
http://arxiv.org/pdf/2306.03534v1
We analyze continual learning on a sequence of separable linear classification tasks with binary labels. We show theoretically that learning with weak regularization reduces to solving a sequential max-margin problem, corresponding to a special case of the Projection Onto Convex Sets (POCS) framework. We then develop u...
Unlocking the Power of Rehearsal in Continual Learning: A Theoretical Perspective
2506.00205v1
evron2022catastrophic
\cite{evron2022catastrophic}
How catastrophic can catastrophic forgetting be in linear regression?
http://arxiv.org/abs/2205.09588v2
To better understand catastrophic forgetting, we study fitting an overparameterized linear model to a sequence of tasks with different input distributions. We analyze how much the model forgets the true labels of earlier tasks after training on subsequent tasks, obtaining exact expressions and bounds. We establish conn...
true
true
Evron, Itay and Moroshko, Edward and Ward, Rachel and Srebro, Nathan and Soudry, Daniel
2,022
null
null
null
null
How catastrophic can catastrophic forgetting be in linear regression?
How catastrophic can catastrophic forgetting be in linear regression?
http://arxiv.org/pdf/2205.09588v2
To better understand catastrophic forgetting, we study fitting an overparameterized linear model to a sequence of tasks with different input distributions. We analyze how much the model forgets the true labels of earlier tasks after training on subsequent tasks, obtaining exact expressions and bounds. We establish conn...
Unlocking the Power of Rehearsal in Continual Learning: A Theoretical Perspective
2506.00205v1
lin2023theory
\cite{lin2023theory}
Theory on Forgetting and Generalization of Continual Learning
http://arxiv.org/abs/2302.05836v1
Continual learning (CL), which aims to learn a sequence of tasks, has attracted significant recent attention. However, most work has focused on the experimental performance of CL, and theoretical studies of CL are still limited. In particular, there is a lack of understanding on what factors are important and how they ...
true
true
Lin, Sen and Ju, Peizhong and Liang, Yingbin and Shroff, Ness
2,023
null
null
null
null
Theory on Forgetting and Generalization of Continual Learning
Theory on Forgetting and Generalization of Continual Learning
http://arxiv.org/pdf/2302.05836v1
Continual learning (CL), which aims to learn a sequence of tasks, has attracted significant recent attention. However, most work has focused on the experimental performance of CL, and theoretical studies of CL are still limited. In particular, there is a lack of understanding on what factors are important and how they ...
Unlocking the Power of Rehearsal in Continual Learning: A Theoretical Perspective
2506.00205v1
ding2024understanding
\cite{ding2024understanding}
Understanding Forgetting in Continual Learning with Linear Regression
http://arxiv.org/abs/2405.17583v1
Continual learning, focused on sequentially learning multiple tasks, has gained significant attention recently. Despite the tremendous progress made in the past, the theoretical understanding, especially factors contributing to catastrophic forgetting, remains relatively unexplored. In this paper, we provide a general ...
true
true
Ding, Meng and Ji, Kaiyi and Wang, Di and Xu, Jinhui
2,024
null
null
null
null
Understanding Forgetting in Continual Learning with Linear Regression
Understanding Forgetting in Continual Learning with Linear Regression
http://arxiv.org/pdf/2405.17583v1
Continual learning, focused on sequentially learning multiple tasks, has gained significant attention recently. Despite the tremendous progress made in the past, the theoretical understanding, especially factors contributing to catastrophic forgetting, remains relatively unexplored. In this paper, we provide a general ...
Unlocking the Power of Rehearsal in Continual Learning: A Theoretical Perspective
2506.00205v1
goldfarb2023analysis
\cite{goldfarb2023analysis}
Analysis of Catastrophic Forgetting for Random Orthogonal Transformation Tasks in the Overparameterized Regime
http://arxiv.org/abs/2207.06475v1
Overparameterization is known to permit strong generalization performance in neural networks. In this work, we provide an initial theoretical analysis of its effect on catastrophic forgetting in a continual learning setup. We show experimentally that in permuted MNIST image classification tasks, the generalization perf...
true
true
Goldfarb, Daniel and Hand, Paul
2,023
null
null
null
null
Analysis of Catastrophic Forgetting for Random Orthogonal Transformation Tasks in the Overparameterized Regime
[PDF] Analysis of Catastrophic Forgetting for Random Orthogonal ...
https://proceedings.mlr.press/v206/goldfarb23a/goldfarb23a.pdf
Missing: 04/08/2025
Unlocking the Power of Rehearsal in Continual Learning: A Theoretical Perspective
2506.00205v1
zhao2024statistical
\cite{zhao2024statistical}
A Statistical Theory of Regularization-Based Continual Learning
http://arxiv.org/abs/2406.06213v1
We provide a statistical analysis of regularization-based continual learning on a sequence of linear regression tasks, with emphasis on how different regularization terms affect the model performance. We first derive the convergence rate for the oracle estimator obtained as if all data were available simultaneously. Ne...
true
true
Zhao, Xuyang and Wang, Huiyuan and Huang, Weiran and Lin, Wei
2,024
null
null
null
null
A Statistical Theory of Regularization-Based Continual Learning
[PDF] A Statistical Theory of Regularization-Based Continual Learning
https://openreview.net/pdf?id=A54CXWn9VB
We provide a statistical analysis of regularization- based continual learning on a sequence of linear regression tasks, with emphasis on how differ-.
Unlocking the Power of Rehearsal in Continual Learning: A Theoretical Perspective
2506.00205v1
li2024theory
\cite{li2024theory}
Theory on Mixture-of-Experts in Continual Learning
http://arxiv.org/abs/2406.16437v3
Continual learning (CL) has garnered significant attention because of its ability to adapt to new tasks that arrive over time. Catastrophic forgetting (of old tasks) has been identified as a major issue in CL, as the model adapts to new tasks. The Mixture-of-Experts (MoE) model has recently been shown to effectively mi...
true
true
Li, Hongbo and Lin, Sen and Duan, Lingjie and Liang, Yingbin and Shroff, Ness B
2,024
null
null
null
arXiv preprint arXiv:2406.16437
Theory on Mixture-of-Experts in Continual Learning
Theory on Mixture-of-Experts in Continual Learning
https://openreview.net/forum?id=7XgKAabsPp
by H Li · Cited by 24 — This paper provides a theoretical study of Mixture-of-Experts (MoE) models for Continual Learning (CL). Specifically, it examines the CL of linear regression
Unlocking the Power of Rehearsal in Continual Learning: A Theoretical Perspective
2506.00205v1
banayeeanzade2024theoretical
\cite{banayeeanzade2024theoretical}
Theoretical Insights into Overparameterized Models in Multi-Task and Replay-Based Continual Learning
http://arxiv.org/abs/2408.16939v2
Multi-task learning (MTL) is a machine learning paradigm that aims to improve the generalization performance of a model on multiple related tasks by training it simultaneously on those tasks. Unlike MTL, where the model has instant access to the training data of all tasks, continual learning (CL) involves adapting to n...
true
true
Banayeeanzade, Mohammadamin and Soltanolkotabi, Mahdi and Rostami, Mohammad
2,024
null
null
null
arXiv preprint arXiv:2408.16939
Theoretical Insights into Overparameterized Models in Multi-Task and Replay-Based Continual Learning
Theoretical Insights into Overparameterized Models in Multi-Task...
https://openreview.net/forum?id=4zGPT0ZwnU
The paper provides theoretical insights into multi-task learning (MTL) and replay-based continual learning (CL) in overparameterized settings, using linear
Timing is important: Risk-aware Fund Allocation based on Time-Series Forecasting
2505.24835v1
Informer
\cite{Informer}
Informer: Beyond Efficient Transformer for Long Sequence Time-Series Forecasting
http://arxiv.org/abs/2012.07436v3
Many real-world applications require the prediction of long sequence time-series, such as electricity consumption planning. Long sequence time-series forecasting (LSTF) demands a high prediction capacity of the model, which is the ability to capture precise long-range dependency coupling between output and input effici...
true
true
Haoyi Zhou and Shanghang Zhang and Jieqi Peng and Shuai Zhang and Jianxin Li and Hui Xiong and Wancai Zhang
2,021
null
null
10.1609/AAAI.V35I12.17325
null
Informer: Beyond Efficient Transformer for Long Sequence Time-Series Forecasting
zhouhaoyi/Informer2020: The GitHub repository for the paper ...
https://github.com/zhouhaoyi/Informer2020
This is the origin Pytorch implementation of Informer in the following paper: Informer: Beyond Efficient Transformer for Long Sequence Time-Series Forecasting.
Timing is important: Risk-aware Fund Allocation based on Time-Series Forecasting
2505.24835v1
Pyraformer
\cite{Pyraformer}
Pyraformer: Low-Complexity Pyramidal Attention for Long-Range Time Series Modeling and Forecasting
null
null
true
false
Shizhan Liu and Hang Yu and Cong Liao and Jianguo Li and Weiyao Lin and Alex X. Liu and Schahram Dustdar
2,022
null
https://openreview.net/forum?id=0EXmFzUn5I
null
null
Pyraformer: Low-Complexity Pyramidal Attention for Long-Range Time Series Modeling and Forecasting
Pyraformer: Low-Complexity Pyramidal Attention for Long-Range ...
https://openreview.net/forum?id=0EXmFzUn5I
We propose a multiresolution pyramidal attention mechanism for long-range dependence modeling and time series forecasting.
Timing is important: Risk-aware Fund Allocation based on Time-Series Forecasting
2505.24835v1
LogSparse
\cite{LogSparse}
Enhancing the Locality and Breaking the Memory Bottleneck of Transformer on Time Series Forecasting
http://arxiv.org/abs/1907.00235v3
Time series forecasting is an important problem across many domains, including predictions of solar plant energy output, electricity consumption, and traffic jam situation. In this paper, we propose to tackle such forecasting problem with Transformer [1]. Although impressed by its performance in our preliminary study, ...
true
true
Shiyang Li and Xiaoyong Jin and Yao Xuan and Xiyou Zhou and Wenhu Chen and Yu{-}Xiang Wang and Xifeng Yan
2,019
null
https://proceedings.neurips.cc/paper/2019/hash/6775a0635c302542da2c32aa19d86be0-Abstract.html
null
null
Enhancing the Locality and Breaking the Memory Bottleneck of Transformer on Time Series Forecasting
Enhancing the Locality and Breaking the Memory ...
http://papers.neurips.cc/paper/8766-enhancing-the-locality-and-breaking-the-memory-bottleneck-of-transformer-on-time-series-forecasting.pdf
by S Li · Cited by 2226 — We successfully apply Transformer architecture to time series forecasting and perform extensive experiments on both synthetic and real datasets to validate
Timing is important: Risk-aware Fund Allocation based on Time-Series Forecasting
2505.24835v1
PatchTST
\cite{PatchTST}
A Time Series is Worth 64 Words: Long-term Forecasting with Transformers
http://arxiv.org/abs/2211.14730v2
We propose an efficient design of Transformer-based models for multivariate time series forecasting and self-supervised representation learning. It is based on two key components: (i) segmentation of time series into subseries-level patches which are served as input tokens to Transformer; (ii) channel-independence wher...
true
true
Yuqi Nie and Nam H. Nguyen and Phanwadee Sinthong and Jayant Kalagnanam
2,023
null
https://openreview.net/forum?id=Jbdc0vTOcol
null
null
A Time Series is Worth 64 Words: Long-term Forecasting with Transformers
PatchTST (ICLR 2023) - GitHub
https://github.com/yuqinie98/PatchTST
This is an offical implementation of PatchTST: A Time Series is Worth 64 Words: Long-term Forecasting with Transformers. Our model has been included in
Timing is important: Risk-aware Fund Allocation based on Time-Series Forecasting
2505.24835v1
Crossformer
\cite{Crossformer}
Crossformer: Transformer Utilizing Cross-Dimension Dependency for Multivariate Time Series Forecasting
null
null
true
false
Yunhao Zhang and Junchi Yan
2,023
null
https://openreview.net/forum?id=vSVLM2j9eie
null
null
Crossformer: Transformer Utilizing Cross-Dimension Dependency for Multivariate Time Series Forecasting
Crossformer: Transformer Utilizing Cross-Dimension ...
https://openreview.net/forum?id=vSVLM2j9eie
by Y Zhang · Cited by 1238 — We propose Crossformer, a Transformer-based model that explicitly utilizes cross-dimension dependency for multivariate time series forecasting.
Timing is important: Risk-aware Fund Allocation based on Time-Series Forecasting
2505.24835v1
Autoformer
\cite{Autoformer}
Autoformer: Decomposition Transformers with Auto-Correlation for Long-Term Series Forecasting
http://arxiv.org/abs/2106.13008v5
Extending the forecasting time is a critical demand for real applications, such as extreme weather early warning and long-term energy consumption planning. This paper studies the long-term forecasting problem of time series. Prior Transformer-based models adopt various self-attention mechanisms to discover the long-ran...
true
true
Haixu Wu and Jiehui Xu and Jianmin Wang and Mingsheng Long
2,021
null
https://proceedings.neurips.cc/paper/2021/hash/bcc0d400288793e8bdcd7c19a8ac0c2b-Abstract.html
null
null
Autoformer: Decomposition Transformers with Auto-Correlation for Long-Term Series Forecasting
[PDF] Decomposition Transformers with Auto-Correlation for Long-Term ...
https://ise.thss.tsinghua.edu.cn/~mlong/doc/Autoformer-nips21.pdf
Autoformer achieves a 38% relative improvement under the long-term setting on six bench- marks, covering five real-world applications: energy, traffic,
Timing is important: Risk-aware Fund Allocation based on Time-Series Forecasting
2505.24835v1
FEDformer
\cite{FEDformer}
FEDformer: Frequency Enhanced Decomposed Transformer for Long-term Series Forecasting
http://arxiv.org/abs/2201.12740v3
Although Transformer-based methods have significantly improved state-of-the-art results for long-term series forecasting, they are not only computationally expensive but more importantly, are unable to capture the global view of time series (e.g. overall trend). To address these problems, we propose to combine Transfor...
true
true
Tian Zhou and Ziqing Ma and Qingsong Wen and Xue Wang and Liang Sun and Rong Jin
2,022
null
https://proceedings.mlr.press/v162/zhou22g.html
null
null
FEDformer: Frequency Enhanced Decomposed Transformer for Long-term Series Forecasting
MAZiqing/FEDformer
https://github.com/MAZiqing/FEDformer
Frequency Enhanced Decomposed Transformer (FEDformer) is more efficient than standard Transformer with a linear complexity to the sequence length.See more
Timing is important: Risk-aware Fund Allocation based on Time-Series Forecasting
2505.24835v1
MICN
\cite{MICN}
{MICN:} Multi-scale Local and Global Context Modeling for Long-term Series Forecasting
null
null
true
false
Huiqiang Wang and Jian Peng and Feihu Huang and Jince Wang and Junhui Chen and Yifei Xiao
2,023
null
https://openreview.net/forum?id=zt53IDUR1U
null
null
{MICN:} Multi-scale Local and Global Context Modeling for Long-term Series Forecasting
Modeling Temporal Symmetry: Dual-Component Framework for ...
https://www.mdpi.com/2073-8994/17/4/577
Micn: Multi-scale local and global context modeling for long-term series forecasting. In Proceedings of the Eleventh International Conference on Learning
Timing is important: Risk-aware Fund Allocation based on Time-Series Forecasting
2505.24835v1
TimesNet
\cite{TimesNet}
TimesNet: Temporal 2D-Variation Modeling for General Time Series Analysis
http://arxiv.org/abs/2210.02186v3
Time series analysis is of immense importance in extensive applications, such as weather forecasting, anomaly detection, and action recognition. This paper focuses on temporal variation modeling, which is the common key problem of extensive analysis tasks. Previous methods attempt to accomplish this directly from the 1...
true
true
Haixu Wu and Tengge Hu and Yong Liu and Hang Zhou and Jianmin Wang and Mingsheng Long
2,023
null
https://openreview.net/forum?id=ju\_Uqw384Oq
null
null
TimesNet: Temporal 2D-Variation Modeling for General Time Series Analysis
The complete code and scripts of TimesNet ...
https://github.com/thuml/TimesNet
GitHub - thuml/TimesNet: About Code release for "TimesNet: Temporal 2D-Variation Modeling for General Time Series Analysis" (ICLR 2023), https://openreview.net/pdf?id=ju_Uqw384Oq * GitHub Advanced Security Enterprise-grade security features About Code release for "TimesNet: Temporal 2D-Variation Modeling for General ...
Timing is important: Risk-aware Fund Allocation based on Time-Series Forecasting
2505.24835v1
DLinear
\cite{DLinear}
Are Transformers Effective for Time Series Forecasting?
http://arxiv.org/abs/2205.13504v3
Recently, there has been a surge of Transformer-based solutions for the long-term time series forecasting (LTSF) task. Despite the growing performance over the past few years, we question the validity of this line of research in this work. Specifically, Transformers is arguably the most successful solution to extract t...
true
true
Ailing Zeng and Muxi Chen and Lei Zhang and Qiang Xu
2,023
null
null
10.1609/AAAI.V37I9.26317
null
Are Transformers Effective for Time Series Forecasting?
Are Transformers Effective for Time Series Forecasting?
http://arxiv.org/pdf/2205.13504v3
Recently, there has been a surge of Transformer-based solutions for the long-term time series forecasting (LTSF) task. Despite the growing performance over the past few years, we question the validity of this line of research in this work. Specifically, Transformers is arguably the most successful solution to extract t...
Timing is important: Risk-aware Fund Allocation based on Time-Series Forecasting
2505.24835v1
TSMixer
\cite{TSMixer}
TSMixer: Lightweight MLP-Mixer Model for Multivariate Time Series Forecasting
http://arxiv.org/abs/2306.09364v4
Transformers have gained popularity in time series forecasting for their ability to capture long-sequence interactions. However, their high memory and computing requirements pose a critical bottleneck for long-term forecasting. To address this, we propose TSMixer, a lightweight neural architecture exclusively composed ...
true
true
Vijay Ekambaram and Arindam Jati and Nam Nguyen and Phanwadee Sinthong and Jayant Kalagnanam
2,023
null
null
10.1145/3580305.3599533
null
TSMixer: Lightweight MLP-Mixer Model for Multivariate Time Series Forecasting
Lightweight MLP-Mixer Model for Multivariate Time Series Forecasting
https://dl.acm.org/doi/10.1145/3580305.3599533
TSMixer is designed for multivariate forecasting and representation learning on patched time series, providing an efficient alternative to Transformers.
Timing is important: Risk-aware Fund Allocation based on Time-Series Forecasting
2505.24835v1
DeepAR
\cite{DeepAR}
DeepAR: Probabilistic Forecasting with Autoregressive Recurrent Networks
http://arxiv.org/abs/1704.04110v3
Probabilistic forecasting, i.e. estimating the probability distribution of a time series' future given its past, is a key enabler for optimizing business processes. In retail businesses, for example, forecasting demand is crucial for having the right inventory available at the right time at the right place. In this pap...
true
true
David Salinas and Valentin Flunkert and Jan Gasthaus and Tim Januschowski
2,020
null
null
https://doi.org/10.1016/j.ijforecast.2019.07.001
International Journal of Forecasting
DeepAR: Probabilistic Forecasting with Autoregressive Recurrent Networks
DeepAR: Probabilistic Forecasting with Autoregressive Recurrent Networks
http://arxiv.org/pdf/1704.04110v3
Probabilistic forecasting, i.e. estimating the probability distribution of a time series' future given its past, is a key enabler for optimizing business processes. In retail businesses, for example, forecasting demand is crucial for having the right inventory available at the right time at the right place. In this pap...
Timing is important: Risk-aware Fund Allocation based on Time-Series Forecasting
2505.24835v1
D3VAE
\cite{D3VAE}
Generative Time Series Forecasting with Diffusion, Denoise, and Disentanglement
null
null
true
false
Yan Li and Xinjiang Lu and Yaqing Wang and Dejing Dou
2,022
null
http://papers.nips.cc/paper\_files/paper/2022/hash/91a85f3fb8f570e6be52b333b5ab017a-Abstract-Conference.html
null
null
Generative Time Series Forecasting with Diffusion, Denoise, and Disentanglement
Generative Time Series Forecasting with Diffusion, Denoise, and Disentanglement
http://arxiv.org/pdf/2301.03028v1
Time series forecasting has been a widely explored task of great importance in many applications. However, it is common that real-world time series data are recorded in a short time period, which results in a big gap between the deep model and the limited and noisy time series. In this work, we propose to address the t...
Timing is important: Risk-aware Fund Allocation based on Time-Series Forecasting
2505.24835v1
CF-RNN
\cite{CF-RNN}
Conformal Time-series Forecasting
null
null
true
false
Kamile Stankeviciute and Ahmed M. Alaa and Mihaela van der Schaar
2,021
null
https://proceedings.neurips.cc/paper/2021/hash/312f1ba2a72318edaaa995a67835fad5-Abstract.html
null
null
Conformal Time-series Forecasting
Conformal Time-Series Forecasting
https://proceedings.neurips.cc/paper/2021/file/312f1ba2a72318edaaa995a67835fad5-Paper.pdf
by K Stankeviciute · 2021 · Cited by 189 — In this work, we extend the inductive conformal prediction framework to the time-series forecasting setup, and propose a lightweight uncertainty estimation
Timing is important: Risk-aware Fund Allocation based on Time-Series Forecasting
2505.24835v1
EnbPI
\cite{EnbPI}
Conformal prediction interval for dynamic time-series
null
null
true
false
Chen Xu and Yao Xie
2,021
null
http://proceedings.mlr.press/v139/xu21h.html
null
null
Conformal prediction interval for dynamic time-series
[PDF] Conformal Prediction Interval for Dynamic Time-Series
https://proceedings.mlr.press/v139/xu21h/xu21h.pdf
Abstract. We develop a method to construct distribution- free prediction intervals for dynamic time-series, called EnbPI that wraps around any bootstrap.
Timing is important: Risk-aware Fund Allocation based on Time-Series Forecasting
2505.24835v1
EnbPI2
\cite{EnbPI2}
Conformal prediction for time series
http://arxiv.org/abs/2010.09107v15
We develop a general framework for constructing distribution-free prediction intervals for time series. Theoretically, we establish explicit bounds on conditional and marginal coverage gaps of estimated prediction intervals, which asymptotically converge to zero under additional assumptions. We obtain similar bounds on...
true
true
Chen Xu and Yao Xie
2,023
null
null
10.1109/TPAMI.2023.3272339
{IEEE} Trans. Pattern Anal. Mach. Intell.
Conformal prediction for time series
Conformal prediction for time series
http://arxiv.org/pdf/2010.09107v15
We develop a general framework for constructing distribution-free prediction intervals for time series. Theoretically, we establish explicit bounds on conditional and marginal coverage gaps of estimated prediction intervals, which asymptotically converge to zero under additional assumptions. We obtain similar bounds on...
Timing is important: Risk-aware Fund Allocation based on Time-Series Forecasting
2505.24835v1
Prescriptive
\cite{Prescriptive}
From Predictive to Prescriptive Analytics
http://arxiv.org/abs/1402.5481v4
In this paper, we combine ideas from machine learning (ML) and operations research and management science (OR/MS) in developing a framework, along with specific methods, for using data to prescribe optimal decisions in OR/MS problems. In a departure from other work on data-driven optimization and reflecting our practic...
true
true
Dimitris Bertsimas and Nathan Kallus
2,020
null
null
10.1287/MNSC.2018.3253
Manag. Sci.
From Predictive to Prescriptive Analytics
From Predictive to Prescriptive Analytics
http://arxiv.org/pdf/1402.5481v4
In this paper, we combine ideas from machine learning (ML) and operations research and management science (OR/MS) in developing a framework, along with specific methods, for using data to prescribe optimal decisions in OR/MS problems. In a departure from other work on data-driven optimization and reflecting our practic...
Timing is important: Risk-aware Fund Allocation based on Time-Series Forecasting
2505.24835v1
PtO-bound
\cite{PtO-bound}
Generalization Bounds in the Predict-then-Optimize Framework
http://arxiv.org/abs/1905.11488v3
The predict-then-optimize framework is fundamental in many practical settings: predict the unknown parameters of an optimization problem, and then solve the problem using the predicted values of the parameters. A natural loss function in this environment is to consider the cost of the decisions induced by the predicted...
true
true
Othman El Balghiti and Adam N. Elmachtoub and Paul Grigas and Ambuj Tewari
2,019
null
https://proceedings.neurips.cc/paper/2019/hash/a70145bf8b173e4496b554ce57969e24-Abstract.html
null
null
Generalization Bounds in the Predict-then-Optimize Framework
[PDF] Generalization Bounds in the Predict-then-Optimize Framework
https://www.ambujtewari.com/research/elbalghiti22generalization.pdf
The predict-then-optimize framework is fundamental in many practical settings: predict the unknown param- eters of an optimization problem, and then solve
Timing is important: Risk-aware Fund Allocation based on Time-Series Forecasting
2505.24835v1
PTOCA
\cite{PTOCA}
A Predict-Then-Optimize Couriers Allocation Framework for Emergency Last-mile Logistics
null
null
true
false
Kaiwen Xia and Li Lin and Shuai Wang and Haotian Wang and Desheng Zhang and Tian He
2,023
null
null
10.1145/3580305.3599766
null
A Predict-Then-Optimize Couriers Allocation Framework for Emergency Last-mile Logistics
A Predict-Then-Optimize Couriers Allocation ... - BibSonomy
https://www.bibsonomy.org/bibtex/1340e4e6e70a4e7eaefc9a9d0241a9e82
A Predict-Then-Optimize Couriers Allocation Framework for Emergency Last-mile Logistics. K. Xia, L. Lin, S. Wang, H. Wang, D. Zhang, and T. He.
Timing is important: Risk-aware Fund Allocation based on Time-Series Forecasting
2505.24835v1
PTOFA
\cite{PTOFA}
A Predict-Then-Optimize Customer Allocation Framework for Online Fund Recommendation
http://arxiv.org/abs/2503.03165v1
With the rapid growth of online investment platforms, funds can be distributed to individual customers online. The central issue is to match funds with potential customers under constraints. Most mainstream platforms adopt the recommendation formulation to tackle the problem. However, the traditional recommendation reg...
true
true
Tang, Xing and Weng, Yunpeng and Lyu, Fuyuan and Liu, Dugang and He, Xiuqiang
2,025
null
null
null
arXiv preprint arXiv:2503.03165
A Predict-Then-Optimize Customer Allocation Framework for Online Fund Recommendation
[Literature Review] A Predict-Then-Optimize Customer ...
https://www.themoonlight.io/en/review/a-predict-then-optimize-customer-allocation-framework-for-online-fund-recommendation
The paper presents a novel Predict-Then-Optimize Fund Allocation (PTOFA) framework designed to address the challenges of matching funds with
Timing is important: Risk-aware Fund Allocation based on Time-Series Forecasting
2505.24835v1
PTO-PNO-Benchmark
\cite{PTO-PNO-Benchmark}
Benchmarking PtO and PnO Methods in the Predictive Combinatorial Optimization Regime
http://arxiv.org/abs/2311.07633v5
Predictive combinatorial optimization, where the parameters of combinatorial optimization (CO) are unknown at the decision-making time, is the precise modeling of many real-world applications, including energy cost-aware scheduling and budget allocation on advertising. Tackling such a problem usually involves a predict...
true
true
Geng, Haoyu and Ruan, Hang and Wang, Runzhong and Li, Yang and Wang, Yang and Chen, Lei and Yan, Junchi
2,024
null
https://openreview.net/forum?id=cX57Pbw8vS
null
null
Benchmarking PtO and PnO Methods in the Predictive Combinatorial Optimization Regime
Benchmarking PtO and PnO Methods in the Predictive ...
https://arxiv.org/html/2311.07633v5
Predictive combinatorial optimization is a family of Combinatorial Optimization (CO) problems where the problem parameters are unknown during decision-making.
Timing is important: Risk-aware Fund Allocation based on Time-Series Forecasting
2505.24835v1
PtOorPnO
\cite{PtOorPnO}
Predict-then-optimize or predict-and-optimize? An empirical evaluation of cost-sensitive learning strategies
null
null
true
false
Toon Vanderschueren and Tim Verdonck and Bart Baesens and Wouter Verbeke
2,022
null
null
10.1016/J.INS.2022.02.021
Inf. Sci.
Predict-then-optimize or predict-and-optimize? An empirical evaluation of cost-sensitive learning strategies
Predict-then-optimize or predict-and-optimize? An ...
https://repository.uantwerpen.be/link/irua/189575
by T Vanderschueren · 2022 · Cited by 81 — Predict-then-optimize or predict-and-optimize? An empirical evaluation of cost-sensitive learning strategies. Author. Vanderschueren, Toon. Verdonck, Tim.
Timing is important: Risk-aware Fund Allocation based on Time-Series Forecasting
2505.24835v1
PnO-bound
\cite{PnO-bound}
Risk Bounds and Calibration for a Smart Predict-then-Optimize Method
http://arxiv.org/abs/2108.08887v2
The predict-then-optimize framework is fundamental in practical stochastic decision-making problems: first predict unknown parameters of an optimization model, then solve the problem using the predicted values. A natural loss function in this setting is defined by measuring the decision error induced by the predicted p...
true
true
Heyuan Liu and Paul Grigas
2,021
null
https://proceedings.neurips.cc/paper/2021/hash/b943325cc7b7422d2871b345bf9b067f-Abstract.html
null
null
Risk Bounds and Calibration for a Smart Predict-then-Optimize Method
[PDF] Risk Bounds and Calibration for a Smart Predict-then-Optimize Method
https://papers.neurips.cc/paper/2021/file/b943325cc7b7422d2871b345bf9b067f-Paper.pdf
The predict-then-optimize framework is fundamental in practical stochastic decision-making problems: first predict unknown parameters of an optimization.
Timing is important: Risk-aware Fund Allocation based on Time-Series Forecasting
2505.24835v1
DFL-Survey
\cite{DFL-Survey}
Decision-Focused Learning: Foundations, State of the Art, Benchmark and Future Opportunities
http://arxiv.org/abs/2307.13565v4
Decision-focused learning (DFL) is an emerging paradigm that integrates machine learning (ML) and constrained optimization to enhance decision quality by training ML models in an end-to-end system. This approach shows significant potential to revolutionize combinatorial decision-making in real-world applications that o...
true
true
Jayanta Mandi and James Kotary and Senne Berden and Maxime Mulamba and Victor Bucarey and Tias Guns and Ferdinando Fioretto
2,024
null
null
10.1613/JAIR.1.15320
J. Artif. Intell. Res.
Decision-Focused Learning: Foundations, State of the Art, Benchmark and Future Opportunities
View of Decision-Focused Learning: Foundations, State ...
https://jair.org/index.php/jair/article/view/15320/27076
by J Mandi · 2024 · Cited by 107 — Return to Article Details Decision-Focused Learning: Foundations, State of the Art, Benchmark and Future Opportunities Download Download PDF. Thumbnails
Timing is important: Risk-aware Fund Allocation based on Time-Series Forecasting
2505.24835v1
OptNet
\cite{OptNet}
OptNet: Differentiable Optimization as a Layer in Neural Networks
http://arxiv.org/abs/1703.00443v5
This paper presents OptNet, a network architecture that integrates optimization problems (here, specifically in the form of quadratic programs) as individual layers in larger end-to-end trainable deep networks. These layers encode constraints and complex dependencies between the hidden states that traditional convoluti...
true
true
Brandon Amos and J. Zico Kolter
2,017
null
http://proceedings.mlr.press/v70/amos17a.html
null
null
OptNet: Differentiable Optimization as a Layer in Neural Networks
OptNet: Differentiable Optimization as a Layer in Neural Networks
http://arxiv.org/pdf/1703.00443v5
This paper presents OptNet, a network architecture that integrates optimization problems (here, specifically in the form of quadratic programs) as individual layers in larger end-to-end trainable deep networks. These layers encode constraints and complex dependencies between the hidden states that traditional convoluti...
Timing is important: Risk-aware Fund Allocation based on Time-Series Forecasting
2505.24835v1
Cvxpylayers
\cite{Cvxpylayers}
Differentiable Convex Optimization Layers
http://arxiv.org/abs/1910.12430v1
Recent work has shown how to embed differentiable optimization problems (that is, problems whose solutions can be backpropagated through) as layers within deep learning architectures. This method provides a useful inductive bias for certain problems, but existing software for differentiable optimization layers is rigid...
true
true
Akshay Agrawal and Brandon Amos and Shane T. Barratt and Stephen P. Boyd and Steven Diamond and J. Zico Kolter
2,019
null
https://proceedings.neurips.cc/paper/2019/hash/9ce3c52fc54362e22053399d3181c638-Abstract.html
null
null
Differentiable Convex Optimization Layers
Differentiable Convex Optimization Layers
http://arxiv.org/pdf/1910.12430v1
Recent work has shown how to embed differentiable optimization problems (that is, problems whose solutions can be backpropagated through) as layers within deep learning architectures. This method provides a useful inductive bias for certain problems, but existing software for differentiable optimization layers is rigid...
Timing is important: Risk-aware Fund Allocation based on Time-Series Forecasting
2505.24835v1
NCE
\cite{NCE}
Contrastive Losses and Solution Caching for Predict-and-Optimize
http://arxiv.org/abs/2011.05354v2
Many decision-making processes involve solving a combinatorial optimization problem with uncertain input that can be estimated from historic data. Recently, problems in this class have been successfully addressed via end-to-end learning approaches, which rely on solving one optimization problem for each training instan...
true
true
Maxime Mulamba and Jayanta Mandi and Michelangelo Diligenti and Michele Lombardi and Victor Bucarey and Tias Guns
2,021
null
null
10.24963/IJCAI.2021/390
null
Contrastive Losses and Solution Caching for Predict-and-Optimize
[PDF] Contrastive Losses and Solution Caching for Predict-and-Optimize
https://people.cs.kuleuven.be/~tias.guns/files/ijcai21_nce_solpool.pdf
In contrast, our con- trastive losses, coupled with a solution caching mechanism, do away with repeatedly solving the optimization problem.
Timing is important: Risk-aware Fund Allocation based on Time-Series Forecasting
2505.24835v1
SPO+
\cite{SPO+}
Smart "Predict, then Optimize"
http://arxiv.org/abs/1710.08005v5
Many real-world analytics problems involve two significant challenges: prediction and optimization. Due to the typically complex nature of each challenge, the standard paradigm is predict-then-optimize. By and large, machine learning tools are intended to minimize prediction error and do not account for how the predict...
true
true
Adam N. Elmachtoub and Paul Grigas
2,022
null
null
10.1287/MNSC.2020.3922
Manag. Sci.
Smart "Predict, then Optimize"
Smart "Predict, then Optimize"
http://arxiv.org/pdf/1710.08005v5
Many real-world analytics problems involve two significant challenges: prediction and optimization. Due to the typically complex nature of each challenge, the standard paradigm is predict-then-optimize. By and large, machine learning tools are intended to minimize prediction error and do not account for how the predict...
Aligning Protein Conformation Ensemble Generation with Physical Feedback
2505.24203v1
jumper2021highly
\cite{jumper2021highly}
Highly accurate protein structure prediction with AlphaFold
null
null
true
false
Jumper, John and Evans, Richard and Pritzel, Alexander and Green, Tim and Figurnov, Michael and Ronneberger, Olaf and Tunyasuvunakool, Kathryn and Bates, Russ and {\v{Z}}{\'\i}dek, Augustin and Potapenko, Anna and others
2,021
null
null
null
Nature
Highly accurate protein structure prediction with AlphaFold
Highly accurate protein structure prediction with AlphaFold - Nature
https://www.nature.com/articles/s41586-021-03819-2
Highly accurate protein structure prediction with AlphaFold | Nature We validated an entirely redesigned version of our neural network-based model, AlphaFold, in the challenging 14th Critical Assessment of protein Structure Prediction (CASP14)15."), demonstrating accuracy competitive with experimental structures in a m...
Aligning Protein Conformation Ensemble Generation with Physical Feedback
2505.24203v1
noe2019boltzmann
\cite{noe2019boltzmann}
Boltzmann Generators -- Sampling Equilibrium States of Many-Body Systems with Deep Learning
http://arxiv.org/abs/1812.01729v2
Computing equilibrium states in condensed-matter many-body systems, such as solvated proteins, is a long-standing challenge. Lacking methods for generating statistically independent equilibrium samples in "one shot", vast computational effort is invested for simulating these system in small steps, e.g., using Molecular...
true
true
No{\'e}, Frank and Olsson, Simon and K{\"o}hler, Jonas and Wu, Hao
2,019
null
null
null
Science
Boltzmann Generators -- Sampling Equilibrium States of Many-Body Systems with Deep Learning
(PDF) Boltzmann generators: Sampling equilibrium states ...
https://www.researchgate.net/publication/335645955_Boltzmann_generators_Sampling_equilibrium_states_of_many-body_systems_with_deep_learning
Boltzmann generators: Sampling equilibrium states of many-body systems with deep learning. September 2019; Science 365(6457):eaaw1147. DOI
Aligning Protein Conformation Ensemble Generation with Physical Feedback
2505.24203v1
arts2023two
\cite{arts2023two}
Two for One: Diffusion Models and Force Fields for Coarse-Grained Molecular Dynamics
null
null
true
false
Arts, Marloes and Satorras, Victor Garcia and Huang, Chin-Wei and Zuegner, Daniel and Federici, Marco and Clementi, Cecilia and No{\'e}, Frank and Pinsler, Robert and Berg, Rianne van den
2,023
null
null
null
arXiv preprint arXiv:2302.00600
Two for One: Diffusion Models and Force Fields for Coarse-Grained Molecular Dynamics
Two for One: Diffusion Models and Force Fields for Coarse-Grained ...
https://arxiv.org/abs/2302.00600
In this work, we leverage connections between score-based generative models, force fields and molecular dynamics to learn a CG force field
Aligning Protein Conformation Ensemble Generation with Physical Feedback
2505.24203v1
jing2023eigenfold
\cite{jing2023eigenfold}
EigenFold: Generative Protein Structure Prediction with Diffusion Models
http://arxiv.org/abs/2304.02198v1
Protein structure prediction has reached revolutionary levels of accuracy on single structures, yet distributional modeling paradigms are needed to capture the conformational ensembles and flexibility that underlie biological function. Towards this goal, we develop EigenFold, a diffusion generative modeling framework f...
true
true
Jing, Bowen and Erives, Ezra and Pao-Huang, Peter and Corso, Gabriele and Berger, Bonnie and Jaakkola, Tommi
2,023
null
null
null
arXiv preprint arXiv:2304.02198
EigenFold: Generative Protein Structure Prediction with Diffusion Models
EigenFold: Generative Protein Structure Prediction with Diffusion Models
http://arxiv.org/pdf/2304.02198v1
Protein structure prediction has reached revolutionary levels of accuracy on single structures, yet distributional modeling paradigms are needed to capture the conformational ensembles and flexibility that underlie biological function. Towards this goal, we develop EigenFold, a diffusion generative modeling framework f...
Aligning Protein Conformation Ensemble Generation with Physical Feedback
2505.24203v1
lu2024str2str
\cite{lu2024str2str}
Str2Str: A Score-based Framework for Zero-shot Protein Conformation Sampling
http://arxiv.org/abs/2306.03117v3
The dynamic nature of proteins is crucial for determining their biological functions and properties, for which Monte Carlo (MC) and molecular dynamics (MD) simulations stand as predominant tools to study such phenomena. By utilizing empirically derived force fields, MC or MD simulations explore the conformational space...
true
true
Lu, Jiarui and Zhong, Bozitao and Zhang, Zuobai and Tang, Jian
2,024
null
null
null
null
Str2Str: A Score-based Framework for Zero-shot Protein Conformation Sampling
Codebase of the paper "Str2Str: A Score-based Framework for Zero ...
https://github.com/lujiarui/Str2Str
Str2Str is a score-based framework (which means it can accommodate any diffusion/flow matching architecture) for protein conformation sampling in a zero-shot
Aligning Protein Conformation Ensemble Generation with Physical Feedback
2505.24203v1
zheng2024predicting
\cite{zheng2024predicting}
Towards Predicting Equilibrium Distributions for Molecular Systems with Deep Learning
http://arxiv.org/abs/2306.05445v1
Advances in deep learning have greatly improved structure prediction of molecules. However, many macroscopic observations that are important for real-world applications are not functions of a single molecular structure, but rather determined from the equilibrium distribution of structures. Traditional methods for obtai...
true
true
Zheng, Shuxin and He, Jiyan and Liu, Chang and Shi, Yu and Lu, Ziheng and Feng, Weitao and Ju, Fusong and Wang, Jiaxi and Zhu, Jianwei and Min, Yaosen and others
2,024
null
null
null
Nature Machine Intelligence
Towards Predicting Equilibrium Distributions for Molecular Systems with Deep Learning
Towards Predicting Equilibrium Distributions for Molecular Systems ...
https://arxiv.org/abs/2306.05445
In this paper, we introduce a novel deep learning framework, called Distributional Graphormer (DiG), in an attempt to predict the equilibrium distribution of
Aligning Protein Conformation Ensemble Generation with Physical Feedback
2505.24203v1
wang2024proteinconformationgenerationforceguided
\cite{wang2024proteinconformationgenerationforceguided}
Protein Conformation Generation via Force-Guided SE(3) Diffusion Models
http://arxiv.org/abs/2403.14088v2
The conformational landscape of proteins is crucial to understanding their functionality in complex biological processes. Traditional physics-based computational methods, such as molecular dynamics (MD) simulations, suffer from rare event sampling and long equilibration time problems, hindering their applications in ge...
true
true
Yan Wang and Lihao Wang and Yuning Shen and Yiqun Wang and Huizhuo Yuan and Yue Wu and Quanquan Gu
2,024
null
https://arxiv.org/abs/2403.14088
null
null
Protein Conformation Generation via Force-Guided SE(3) Diffusion Models
Official Implemetation of ConfDiff (ICML'24) - Protein Conformation ...
https://github.com/bytedance/ConfDiff
A force-guided SE(3) diffusion model for protein conformation generation. ConfDiff can generate protein conformations with rich diversity while preserving high
Aligning Protein Conformation Ensemble Generation with Physical Feedback
2505.24203v1
jing2024alphafoldmeetsflowmatching
\cite{jing2024alphafoldmeetsflowmatching}
AlphaFold Meets Flow Matching for Generating Protein Ensembles
http://arxiv.org/abs/2402.04845v2
The biological functions of proteins often depend on dynamic structural ensembles. In this work, we develop a flow-based generative modeling approach for learning and sampling the conformational landscapes of proteins. We repurpose highly accurate single-state predictors such as AlphaFold and ESMFold and fine-tune them...
true
true
Bowen Jing and Bonnie Berger and Tommi Jaakkola
2,024
null
https://arxiv.org/abs/2402.04845
null
null
AlphaFold Meets Flow Matching for Generating Protein Ensembles
AlphaFold Meets Flow Matching for Generating Protein Ensembles
http://arxiv.org/pdf/2402.04845v2
The biological functions of proteins often depend on dynamic structural ensembles. In this work, we develop a flow-based generative modeling approach for learning and sampling the conformational landscapes of proteins. We repurpose highly accurate single-state predictors such as AlphaFold and ESMFold and fine-tune them...
Aligning Protein Conformation Ensemble Generation with Physical Feedback
2505.24203v1
lu2024structure
\cite{lu2024structure}
Structure Language Models for Protein Conformation Generation
http://arxiv.org/abs/2410.18403v2
Proteins adopt multiple structural conformations to perform their diverse biological functions, and understanding these conformations is crucial for advancing drug discovery. Traditional physics-based simulation methods often struggle with sampling equilibrium conformations and are computationally expensive. Recently, ...
true
true
Lu, Jiarui and Chen, Xiaoyin and Lu, Stephen Zhewen and Shi, Chence and Guo, Hongyu and Bengio, Yoshua and Tang, Jian
2,024
null
null
null
arXiv preprint arXiv:2410.18403
Structure Language Models for Protein Conformation Generation
Structure Language Models for Protein Conformation Generation
http://arxiv.org/pdf/2410.18403v2
Proteins adopt multiple structural conformations to perform their diverse biological functions, and understanding these conformations is crucial for advancing drug discovery. Traditional physics-based simulation methods often struggle with sampling equilibrium conformations and are computationally expensive. Recently, ...
Aligning Protein Conformation Ensemble Generation with Physical Feedback
2505.24203v1
jing2024generative
\cite{jing2024generative}
Generative Modeling of Molecular Dynamics Trajectories
http://arxiv.org/abs/2409.17808v1
Molecular dynamics (MD) is a powerful technique for studying microscopic phenomena, but its computational cost has driven significant interest in the development of deep learning-based surrogate models. We introduce generative modeling of molecular trajectories as a paradigm for learning flexible multi-task surrogate m...
true
true
Jing, Bowen and St{\"a}rk, Hannes and Jaakkola, Tommi and Berger, Bonnie
2,024
null
null
null
arXiv preprint arXiv:2409.17808
Generative Modeling of Molecular Dynamics Trajectories
Generative Modeling of Molecular Dynamics Trajectories
http://arxiv.org/pdf/2409.17808v1
Molecular dynamics (MD) is a powerful technique for studying microscopic phenomena, but its computational cost has driven significant interest in the development of deep learning-based surrogate models. We introduce generative modeling of molecular trajectories as a paradigm for learning flexible multi-task surrogate m...
Aligning Protein Conformation Ensemble Generation with Physical Feedback
2505.24203v1
kreutzer2018reliability
\cite{kreutzer2018reliability}
Reliability and Learnability of Human Bandit Feedback for Sequence-to-Sequence Reinforcement Learning
http://arxiv.org/abs/1805.10627v3
We present a study on reinforcement learning (RL) from human bandit feedback for sequence-to-sequence learning, exemplified by the task of bandit neural machine translation (NMT). We investigate the reliability of human bandit feedback, and analyze the influence of reliability on the learnability of a reward estimator,...
true
true
Julia Kreutzer and Joshua Uyheng and S. Riezler
2,018
null
null
10.18653/v1/P18-1165
Annual Meeting of the Association for Computational Linguistics
Reliability and Learnability of Human Bandit Feedback for Sequence-to-Sequence Reinforcement Learning
Reliability and Learnability of Human Bandit Feedback for ...
https://aclanthology.org/P18-1165/
Missing: 04/08/2025
Aligning Protein Conformation Ensemble Generation with Physical Feedback
2505.24203v1
stiennon2020learning
\cite{stiennon2020learning}
Learning to summarize from human feedback
http://arxiv.org/abs/2009.01325v3
As language models become more powerful, training and evaluation are increasingly bottlenecked by the data and metrics used for a particular task. For example, summarization models are often trained to predict human reference summaries and evaluated using ROUGE, but both of these metrics are rough proxies for what we r...
true
true
Nisan Stiennon and Long Ouyang and Jeff Wu and Daniel M. Ziegler and Ryan J. Lowe and Chelsea Voss and Alec Radford and Dario Amodei and Paul Christiano
2,020
null
null
null
Neural Information Processing Systems
Learning to summarize from human feedback
Learning to summarize from human feedback
http://arxiv.org/pdf/2009.01325v3
As language models become more powerful, training and evaluation are increasingly bottlenecked by the data and metrics used for a particular task. For example, summarization models are often trained to predict human reference summaries and evaluated using ROUGE, but both of these metrics are rough proxies for what we r...
Aligning Protein Conformation Ensemble Generation with Physical Feedback
2505.24203v1
ouyang2022training
\cite{ouyang2022training}
Training language models to follow instructions with human feedback
null
null
true
false
Long Ouyang and Jeff Wu and Xu Jiang and Diogo Almeida and Carroll L. Wainwright and Pamela Mishkin and Chong Zhang and Sandhini Agarwal and Katarina Slama and Alex Ray and John Schulman and Jacob Hilton and Fraser Kelton and Luke E. Miller and Maddie Simens and Amanda Askell and P. Welinder and P. Christiano and J. Le...
2,022
null
null
null
Neural Information Processing Systems
Training language models to follow instructions with human feedback
Training language models to follow instructions with human feedback
http://arxiv.org/pdf/2203.02155v1
Making language models bigger does not inherently make them better at following a user's intent. For example, large language models can generate outputs that are untruthful, toxic, or simply not helpful to the user. In other words, these models are not aligned with their users. In this paper, we show an avenue for alig...
Aligning Protein Conformation Ensemble Generation with Physical Feedback
2505.24203v1
black2023training
\cite{black2023training}
Training Diffusion Models with Reinforcement Learning
http://arxiv.org/abs/2305.13301v4
Diffusion models are a class of flexible generative models trained with an approximation to the log-likelihood objective. However, most use cases of diffusion models are not concerned with likelihoods, but instead with downstream objectives such as human-perceived image quality or drug effectiveness. In this paper, we ...
true
true
Black, Kevin and Janner, Michael and Du, Yilun and Kostrikov, Ilya and Levine, Sergey
2,023
null
null
null
arXiv preprint arXiv:2305.13301
Training Diffusion Models with Reinforcement Learning
Training Diffusion Models with Reinforcement Learning
http://arxiv.org/pdf/2305.13301v4
Diffusion models are a class of flexible generative models trained with an approximation to the log-likelihood objective. However, most use cases of diffusion models are not concerned with likelihoods, but instead with downstream objectives such as human-perceived image quality or drug effectiveness. In this paper, we ...
Aligning Protein Conformation Ensemble Generation with Physical Feedback
2505.24203v1
fan2024reinforcement
\cite{fan2024reinforcement}
Reinforcement learning for fine-tuning text-to-image diffusion models
null
null
true
false
Fan, Ying and Watkins, Olivia and Du, Yuqing and Liu, Hao and Ryu, Moonkyung and Boutilier, Craig and Abbeel, Pieter and Ghavamzadeh, Mohammad and Lee, Kangwook and Lee, Kimin
2,024
null
null
null
Advances in Neural Information Processing Systems
Reinforcement learning for fine-tuning text-to-image diffusion models
DPOK: Reinforcement Learning for Fine-tuning Text-to-Image ... - arXiv
https://arxiv.org/abs/2305.16381
We propose using online reinforcement learning (RL) to fine-tune text-to-image models. We focus on diffusion models, defining the fine-tuning task as an RL
Aligning Protein Conformation Ensemble Generation with Physical Feedback
2505.24203v1
rafailov2024direct
\cite{rafailov2024direct}
Direct Preference Optimization: Your Language Model is Secretly a Reward Model
http://arxiv.org/abs/2305.18290v3
While large-scale unsupervised language models (LMs) learn broad world knowledge and some reasoning skills, achieving precise control of their behavior is difficult due to the completely unsupervised nature of their training. Existing methods for gaining such steerability collect human labels of the relative quality of...
true
true
Rafailov, Rafael and Sharma, Archit and Mitchell, Eric and Manning, Christopher D and Ermon, Stefano and Finn, Chelsea
2,024
null
null
null
Advances in Neural Information Processing Systems
Direct Preference Optimization: Your Language Model is Secretly a Reward Model
Direct Preference Optimization: Your Language Model is Secretly a ...
https://arxiv.org/abs/2305.18290
**arXiv:2305.18290** (cs) View a PDF of the paper titled Direct Preference Optimization: Your Language Model is Secretly a Reward Model, by Rafael Rafailov and 5 other authors View a PDF of the paper titled Direct Preference Optimization: Your Language Model is Secretly a Reward Model, by Rafael Rafailov and 5 other a...
Aligning Protein Conformation Ensemble Generation with Physical Feedback
2505.24203v1
Wallace_2024_CVPR
\cite{Wallace_2024_CVPR}
Diffusion Model Alignment Using Direct Preference Optimization
http://arxiv.org/abs/2311.12908v1
Large language models (LLMs) are fine-tuned using human comparison data with Reinforcement Learning from Human Feedback (RLHF) methods to make them better aligned with users' preferences. In contrast to LLMs, human preference learning has not been widely explored in text-to-image diffusion models; the best existing app...
true
true
Wallace, Bram and Dang, Meihua and Rafailov, Rafael and Zhou, Linqi and Lou, Aaron and Purushwalkam, Senthil and Ermon, Stefano and Xiong, Caiming and Joty, Shafiq and Naik, Nikhil
2,024
June
null
null
null
Diffusion Model Alignment Using Direct Preference Optimization
Diffusion Model Alignment Using Direct Preference Optimization
http://arxiv.org/pdf/2311.12908v1
Large language models (LLMs) are fine-tuned using human comparison data with Reinforcement Learning from Human Feedback (RLHF) methods to make them better aligned with users' preferences. In contrast to LLMs, human preference learning has not been widely explored in text-to-image diffusion models; the best existing app...
Aligning Protein Conformation Ensemble Generation with Physical Feedback
2505.24203v1
zhou2024antigen
\cite{zhou2024antigen}
Antigen-Specific Antibody Design via Direct Energy-based Preference Optimization
http://arxiv.org/abs/2403.16576v3
Antibody design, a crucial task with significant implications across various disciplines such as therapeutics and biology, presents considerable challenges due to its intricate nature. In this paper, we tackle antigen-specific antibody sequence-structure co-design as an optimization problem towards specific preferences...
true
true
Zhou, Xiangxin and Xue, Dongyu and Chen, Ruizhe and Zheng, Zaixiang and Wang, Liang and Gu, Quanquan
2,024
null
null
null
arXiv preprint arXiv:2403.16576
Antigen-Specific Antibody Design via Direct Energy-based Preference Optimization
Antigen-Specific Antibody Design via Direct Energy-based ...
https://openreview.net/forum?id=GN2GXjPyN8&referrer=%5Bthe%20profile%20of%20Xiangxin%20Zhou%5D(%2Fprofile%3Fid%3D~Xiangxin_Zhou1)
Summary: This paper applies direct preference optimization to antibody design. Specifically, it uses Rosetta binding energy to guide a pre-trained diffusion
Aligning Protein Conformation Ensemble Generation with Physical Feedback
2505.24203v1
alford2017rosetta
\cite{alford2017rosetta}
The Rosetta all-atom energy function for macromolecular modeling and design
null
null
true
false
Alford, Rebecca F and Leaver-Fay, Andrew and Jeliazkov, Jeliazko R and O’Meara, Matthew J and DiMaio, Frank P and Park, Hahnbeom and Shapovalov, Maxim V and Renfrew, P Douglas and Mulligan, Vikram K and Kappel, Kalli and others
2,017
null
null
null
Journal of chemical theory and computation
The Rosetta all-atom energy function for macromolecular modeling and design
The Rosetta all-atom energy function for macromolecular ...
https://pmc.ncbi.nlm.nih.gov/articles/PMC5717763/
by RF Alford · 2017 · Cited by 1630 — The goal of this paper is to describe the energy calculations used by the Rosetta macromolecular modeling program: we explain the underlying physical concepts,
Aligning Protein Conformation Ensemble Generation with Physical Feedback
2505.24203v1
gu2024aligning
\cite{gu2024aligning}
Aligning Target-Aware Molecule Diffusion Models with Exact Energy Optimization
http://arxiv.org/abs/2407.01648v2
Generating ligand molecules for specific protein targets, known as structure-based drug design, is a fundamental problem in therapeutics development and biological discovery. Recently, target-aware generative models, especially diffusion models, have shown great promise in modeling protein-ligand interactions and gener...
true
true
Gu, Siyi and Xu, Minkai and Powers, Alexander and Nie, Weili and Geffner, Tomas and Kreis, Karsten and Leskovec, Jure and Vahdat, Arash and Ermon, Stefano
2,024
null
null
null
arXiv preprint arXiv:2407.01648
Aligning Target-Aware Molecule Diffusion Models with Exact Energy Optimization
[PDF] Aligning Target-Aware Molecule Diffusion Models with Exact Energy ...
https://proceedings.neurips.cc/paper_files/paper/2024/file/4ddfe69f164eae70abc86f0f9cbed7e8-Paper-Conference.pdf
ALIDIFF aligns target-aware diffusion models with preferred properties, shifting chemical distribution towards higher binding affinity and structural
Aligning Protein Conformation Ensemble Generation with Physical Feedback
2505.24203v1
cheng2024decomposed
\cite{cheng2024decomposed}
Decomposed Direct Preference Optimization for Structure-Based Drug Design
http://arxiv.org/abs/2407.13981v2
Diffusion models have achieved promising results for Structure-Based Drug Design (SBDD). Nevertheless, high-quality protein subpocket and ligand data are relatively scarce, which hinders the models' generation capabilities. Recently, Direct Preference Optimization (DPO) has emerged as a pivotal tool for aligning genera...
true
true
Cheng, Xiwei and Zhou, Xiangxin and Yang, Yuwei and Bao, Yu and Gu, Quanquan
2,024
null
null
null
arXiv preprint arXiv:2407.13981
Decomposed Direct Preference Optimization for Structure-Based Drug Design
Decomposed Direct Preference Optimization for Structure ...
https://arxiv.org/abs/2407.13981
Image 2: arxiv logo>q-bio> arXiv:2407.13981 **arXiv:2407.13981** (q-bio) View a PDF of the paper titled Decomposed Direct Preference Optimization for Structure-Based Drug Design, by Xiwei Cheng and 4 other authors (or arXiv:2407.13981v2 [q-bio.BM] for this version) View a PDF of the paper titled Decomposed Direct Pre...
CLONE: Customizing LLMs for Efficient Latency-Aware Inference at the Edge
2506.02847v1
GPT4
\cite{GPT4}
{GPT-4} Technical Report
null
null
true
false
OpenAI
2,023
null
https://doi.org/10.48550/arXiv.2303.08774
10.48550/ARXIV.2303.08774
CoRR
{GPT-4} Technical Report
(PDF) GPT-4 Technical Report - ResearchGate
https://www.researchgate.net/publication/383739523_GPT-4_Technical_Report
We report the development of GPT-4, a large-scale, multimodal model which can accept image and text inputs and produce text outputs.
CLONE: Customizing LLMs for Efficient Latency-Aware Inference at the Edge
2506.02847v1
PaLM
\cite{PaLM}
PaLM: Scaling Language Modeling with Pathways
http://arxiv.org/abs/2204.02311v5
Large language models have been shown to achieve remarkable performance across a variety of natural language tasks using few-shot learning, which drastically reduces the number of task-specific training examples needed to adapt the model to a particular application. To further our understanding of the impact of scale o...
true
true
Aakanksha Chowdhery and Sharan Narang and Jacob Devlin and Maarten Bosma and Gaurav Mishra and Adam Roberts and Paul Barham and Hyung Won Chung and Charles Sutton and ...
2,023
null
http://jmlr.org/papers/v24/22-1144.html
null
J. Mach. Learn. Res.
PaLM: Scaling Language Modeling with Pathways
PaLM: Scaling Language Modeling with Pathways
http://arxiv.org/pdf/2204.02311v5
Large language models have been shown to achieve remarkable performance across a variety of natural language tasks using few-shot learning, which drastically reduces the number of task-specific training examples needed to adapt the model to a particular application. To further our understanding of the impact of scale o...
CLONE: Customizing LLMs for Efficient Latency-Aware Inference at the Edge
2506.02847v1
llama
\cite{llama}
LLaMA: Open and Efficient Foundation Language Models
http://arxiv.org/abs/2302.13971v1
We introduce LLaMA, a collection of foundation language models ranging from 7B to 65B parameters. We train our models on trillions of tokens, and show that it is possible to train state-of-the-art models using publicly available datasets exclusively, without resorting to proprietary and inaccessible datasets. In partic...
true
true
Hugo Touvron and Thibaut Lavril and Gautier Izacard and Xavier Martinet and Marie{-}Anne Lachaux and Timoth{\'{e}}e Lacroix and Baptiste Rozi{\`{e}}re and Naman Goyal and Eric ...
2,023
null
https://doi.org/10.48550/arXiv.2302.13971
10.48550/ARXIV.2302.13971
CoRR
LLaMA: Open and Efficient Foundation Language Models
LLaMA: Open and Efficient Foundation Language Models
http://arxiv.org/pdf/2302.13971v1
We introduce LLaMA, a collection of foundation language models ranging from 7B to 65B parameters. We train our models on trillions of tokens, and show that it is possible to train state-of-the-art models using publicly available datasets exclusively, without resorting to proprietary and inaccessible datasets. In partic...
CLONE: Customizing LLMs for Efficient Latency-Aware Inference at the Edge
2506.02847v1
Llama_2
\cite{Llama_2}
Llama 2: Open Foundation and Fine-Tuned Chat Models
http://arxiv.org/abs/2307.09288v2
In this work, we develop and release Llama 2, a collection of pretrained and fine-tuned large language models (LLMs) ranging in scale from 7 billion to 70 billion parameters. Our fine-tuned LLMs, called Llama 2-Chat, are optimized for dialogue use cases. Our models outperform open-source chat models on most benchmarks ...
true
true
Hugo Touvron and Louis Martin and Kevin Stone and Peter Albert and Amjad Almahairi and Yasmine Babaei and Nikolay Bashlykov and Soumya Batra and Prajjwal Bhargava and ...
2,023
null
https://doi.org/10.48550/arXiv.2307.09288
10.48550/ARXIV.2307.09288
CoRR
Llama 2: Open Foundation and Fine-Tuned Chat Models
Llama 2: Open Foundation and Fine-Tuned Chat Models - Meta AI
https://ai.meta.com/research/publications/llama-2-open-foundation-and-fine-tuned-chat-models/
We develop and release Llama 2, a collection of pretrained and fine-tuned large language models (LLMs) ranging in scale from 7 billion to 70 billion parameters.
CLONE: Customizing LLMs for Efficient Latency-Aware Inference at the Edge
2506.02847v1
anthropic_claude
\cite{anthropic_claude}
Claude: A Family of AI Models
null
null
true
false
Anthropic
2,024
null
https://www.anthropic.com/product
null
null
Claude: A Family of AI Models
Introducing the next generation of Claude - Anthropic
https://www.anthropic.com/news/claude-3-family
The family includes three state-of-the-art models in ascending order of capability: Claude 3 Haiku, Claude 3 Sonnet, and Claude 3 Opus. ### Claude 3 model family The Claude 3 models can power live customer chats, auto-completions, and data extraction tasks where responses must be immediate and in real-time. We’ve devel...
CLONE: Customizing LLMs for Efficient Latency-Aware Inference at the Edge
2506.02847v1
gemma
\cite{gemma}
Gemma: Open Models Based on Gemini Research and Technology
http://arxiv.org/abs/2403.08295v4
This work introduces Gemma, a family of lightweight, state-of-the art open models built from the research and technology used to create Gemini models. Gemma models demonstrate strong performance across academic benchmarks for language understanding, reasoning, and safety. We release two sizes of models (2 billion and 7...
true
true
Thomas Mesnard and Cassidy Hardin and Robert Dadashi and Surya Bhupatiraju and Shreya Pathak and Laurent Sifre and Morgane Rivi{\`{e}}re and Mihir Sanjay Kale and Juliette Love...
2,024
null
https://doi.org/10.48550/arXiv.2403.08295
10.48550/ARXIV.2403.08295
CoRR
Gemma: Open Models Based on Gemini Research and Technology
Gemma: Open Models Based on Gemini Research and Technology
http://arxiv.org/pdf/2403.08295v4
This work introduces Gemma, a family of lightweight, state-of-the art open models built from the research and technology used to create Gemini models. Gemma models demonstrate strong performance across academic benchmarks for language understanding, reasoning, and safety. We release two sizes of models (2 billion and 7...
CLONE: Customizing LLMs for Efficient Latency-Aware Inference at the Edge
2506.02847v1
AIIndex2024
\cite{AIIndex2024}
AI Index Report
null
null
true
false
stanford
null
null
null
null
null
AI Index Report
AI Index 2025: State of AI in 10 Charts | Stanford HAI
https://hai.stanford.edu/news/ai-index-2025-state-of-ai-in-10-charts
* AI Index * AI Index Report The 2025 AI Index Report, published on April 7, 2025, is an independent initiative at the Stanford Institute for Human-Centered Artificial Intelligence (HAI), led by the AI Index Steering Committee, an interdisciplinary group of experts from across academia and industry. Each year, the ...
CLONE: Customizing LLMs for Efficient Latency-Aware Inference at the Edge
2506.02847v1
apple_ai
\cite{apple_ai}
Apple Intelligence
null
null
true
false
Apple
2,024
null
null
null
null
Apple Intelligence
How to get Apple Intelligence
https://support.apple.com/en-us/121115
* Apple * Apple Watch * Explore All Apple Watch * Why Apple Watch * Shop Apple Watch * Apple Watch Support * Shop Apple Vision Pro * Apple Vision Pro Support * Apple TV Support * Apple One * Apple TV+ Support * Apple Watch * Made by Apple Apple Intelligence is available in beta starting with i...
CLONE: Customizing LLMs for Efficient Latency-Aware Inference at the Edge
2506.02847v1
song2024powerinfer
\cite{song2024powerinfer}
PowerInfer: Fast Large Language Model Serving with a Consumer-grade GPU
http://arxiv.org/abs/2312.12456v2
This paper introduces PowerInfer, a high-speed Large Language Model (LLM) inference engine on a personal computer (PC) equipped with a single consumer-grade GPU. The key principle underlying the design of PowerInfer is exploiting the high locality inherent in LLM inference, characterized by a power-law distribution in ...
true
true
Song, Yixin and Mi, Zeyu and Xie, Haotong and Chen, Haibo
2,024
null
null
null
null
PowerInfer: Fast Large Language Model Serving with a Consumer-grade GPU
Fast Large Language Model Serving with a Consumer-grade GPU
https://arxiv.org/abs/2312.12456
This paper introduces PowerInfer, a high-speed Large Language Model (LLM) inference engine on a personal computer (PC) equipped with a single consumer-grade
CLONE: Customizing LLMs for Efficient Latency-Aware Inference at the Edge
2506.02847v1
yuan2023mobile
\cite{yuan2023mobile}
Mobile Foundation Model as Firmware
http://arxiv.org/abs/2308.14363v3
In today's landscape, smartphones have evolved into hubs for hosting a multitude of deep learning models aimed at local execution. A key realization driving this work is the notable fragmentation among these models, characterized by varied architectures, operators, and implementations. This fragmentation imposes a sign...
true
true
Yuan, Jinliang and Yang, Chen and Cai, Dongqi and Wang, Shihe and Yuan, Xin and Zhang, Zeling and Li, Xiang and Zhang, Dingge and Mei, Hanzi and Jia, Xianqing and others
2,023
null
null
null
arXiv preprint arXiv:2308.14363
Mobile Foundation Model as Firmware
Mobile Foundation Model as Firmware
http://arxiv.org/pdf/2308.14363v3
In today's landscape, smartphones have evolved into hubs for hosting a multitude of deep learning models aimed at local execution. A key realization driving this work is the notable fragmentation among these models, characterized by varied architectures, operators, and implementations. This fragmentation imposes a sign...
CLONE: Customizing LLMs for Efficient Latency-Aware Inference at the Edge
2506.02847v1
app_3
\cite{app_3}
Drive Like a Human: Rethinking Autonomous Driving with Large Language Models
http://arxiv.org/abs/2307.07162v1
In this paper, we explore the potential of using a large language model (LLM) to understand the driving environment in a human-like manner and analyze its ability to reason, interpret, and memorize when facing complex scenarios. We argue that traditional optimization-based and modular autonomous driving (AD) systems fa...
true
true
Fu, Daocheng and Li, Xin and Wen, Licheng and Dou, Min and Cai, Pinlong and Shi, Botian and Qiao, Yu
2,024
null
null
null
null
Drive Like a Human: Rethinking Autonomous Driving with Large Language Models
Drive Like a Human: Rethinking Autonomous Driving with ...
https://www.computer.org/csdl/proceedings-article/wacvw/2024/702800a910/1WbOYt2kbVC
by D Fu · 2024 · Cited by 238 — In this paper, we explore the potential of using a large language model (LLM) to understand the driving environment in a human-like manner.
CLONE: Customizing LLMs for Efficient Latency-Aware Inference at the Edge
2506.02847v1
app5_robot
\cite{app5_robot}
BAT: Behavior-Aware Human-Like Trajectory Prediction for Autonomous Driving
http://arxiv.org/abs/2312.06371v2
The ability to accurately predict the trajectory of surrounding vehicles is a critical hurdle to overcome on the journey to fully autonomous vehicles. To address this challenge, we pioneer a novel behavior-aware trajectory prediction model (BAT) that incorporates insights and findings from traffic psychology, human beh...
true
true
Haicheng Liao and Zhenning Li and Huanming Shen and Wenxuan Zeng and Dongping Liao and Guofa Li and Shengbo Eben Li and Chengzhong Xu
2,023
null
https://doi.org/10.48550/arXiv.2312.06371
10.48550/ARXIV.2312.06371
CoRR
BAT: Behavior-Aware Human-Like Trajectory Prediction for Autonomous Driving
Behavior-Aware Human-Like Trajectory Prediction for Autonomous ...
https://github.com/Petrichor625/BATraj-Behavior-aware-Model
Introducing a real-time dynamic geometric graph method for the continuous representation of driving behavior in trajectory prediction for autonomous driving.
CLONE: Customizing LLMs for Efficient Latency-Aware Inference at the Edge
2506.02847v1
app6_laptops
\cite{app6_laptops}
Creating Large Language Models on Your Laptop
null
null
true
false
Xinyu Ye and Zhe Wang and Haihao Shen and Yu Luo and Hanwen Chang
2,023
null
null
null
null
Creating Large Language Models on Your Laptop
How to run an LLM on your laptop
https://www.technologyreview.com/2025/07/17/1120391/how-to-run-an-llm-on-your-laptop/
It's now possible to run useful models from the safety and comfort of your own computer. Here's how.
CLONE: Customizing LLMs for Efficient Latency-Aware Inference at the Edge
2506.02847v1
ExeGPT
\cite{ExeGPT}
ExeGPT: Constraint-Aware Resource Scheduling for LLM Inference
http://arxiv.org/abs/2404.07947v1
This paper presents ExeGPT, a distributed system designed for constraint-aware LLM inference. ExeGPT finds and runs with an optimal execution schedule to maximize inference throughput while satisfying a given latency constraint. By leveraging the distribution of input and output sequences, it effectively allocates reso...
true
true
Hyungjun Oh and Kihong Kim and Jaemin Kim and Sungkyun Kim and Junyeol Lee and Du{-}seong Chang and Jiwon Seo
2,024
null
https://doi.org/10.1145/3620665.3640383
10.1145/3620665.3640383
null
ExeGPT: Constraint-Aware Resource Scheduling for LLM Inference
ASPLOS'24 - Lightning Talks - Session 2D - ExeGPT: Constraint ...
https://www.youtube.com/watch?v=UhBwDpY4hV4
... Inference Systems Paper Title: ExeGPT: Constraint-Aware Resource Scheduling for LLM Inference Authors: Hyungjun Oh, Kihong Kim, Jaemin Kim
CLONE: Customizing LLMs for Efficient Latency-Aware Inference at the Edge
2506.02847v1
Splitwise
\cite{Splitwise}
Splitwise: Efficient generative LLM inference using phase splitting
http://arxiv.org/abs/2311.18677v2
Recent innovations in generative large language models (LLMs) have made their applications and use-cases ubiquitous. This has led to large-scale deployments of these models, using complex, expensive, and power-hungry AI accelerators, most commonly GPUs. These developments make LLM inference efficiency an important chal...
true
true
Pratyush Patel and Esha Choukse and Chaojie Zhang and {\'{I}}{\~{n}}igo Goiri and Aashaka Shah and Saeed Maleki and Ricardo Bianchini
2,023
null
https://doi.org/10.48550/arXiv.2311.18677
10.48550/ARXIV.2311.18677
CoRR
Splitwise: Efficient generative LLM inference using phase splitting
Splitwise: Efficient generative LLM inference using phase splitting
http://arxiv.org/pdf/2311.18677v2
Recent innovations in generative large language models (LLMs) have made their applications and use-cases ubiquitous. This has led to large-scale deployments of these models, using complex, expensive, and power-hungry AI accelerators, most commonly GPUs. These developments make LLM inference efficiency an important chal...
CLONE: Customizing LLMs for Efficient Latency-Aware Inference at the Edge
2506.02847v1
PagedAttention
\cite{PagedAttention}
Efficient Memory Management for Large Language Model Serving with PagedAttention
http://arxiv.org/abs/2309.06180v1
High throughput serving of large language models (LLMs) requires batching sufficiently many requests at a time. However, existing systems struggle because the key-value cache (KV cache) memory for each request is huge and grows and shrinks dynamically. When managed inefficiently, this memory can be significantly wasted...
true
true
Woosuk Kwon and Zhuohan Li and Siyuan Zhuang and Ying Sheng and Lianmin Zheng and Cody Hao Yu and Joseph Gonzalez and Hao Zhang and Ion Stoica
2,023
null
https://doi.org/10.1145/3600006.3613165
10.1145/3600006.3613165
null
Efficient Memory Management for Large Language Model Serving with PagedAttention
Efficient Memory Management for Large Language Model ...
https://arxiv.org/pdf/2309.06180
Efficient Memory Management for Large Language Model Serving with PagedAttention Woosuk Kwon 1,∗ Zhuohan Li 1,∗ Siyuan Zhuang 1 Ying Sheng 1,2 Lianmin Zheng 1 Cody Hao Yu 3 Joseph E. Gonzalez 1 Hao Zhang 4 Ion Stoica 1 1 UC Berkeley 2Stanford University 3Independent Researcher 4UC San Diego Abstract High throughput ser...
CLONE: Customizing LLMs for Efficient Latency-Aware Inference at the Edge
2506.02847v1
Just-in-time
\cite{Just-in-time}
Just-in-time Quantization with Processing-In-Memory for Efficient ML Training
http://arxiv.org/abs/2311.05034v1
Data format innovations have been critical for machine learning (ML) scaling, which in turn fuels ground-breaking ML capabilities. However, even in the presence of low-precision formats, model weights are often stored in both high-precision and low-precision during training. Furthermore, with emerging directional data ...
true
true
Mohamed Assem Ibrahim and Shaizeen Aga and Ada Li and Suchita Pati and Mahzabeen Islam
2,023
null
https://arxiv.org/abs/2311.05034
null
null
Just-in-time Quantization with Processing-In-Memory for Efficient ML Training
Just-in-time Quantization with Processing-In-Memory for Efficient ML ...
https://arxiv.org/abs/2311.05034
We explore just-in-time quantization (JIT-Q) where we only store high-precision weights in memory and generate low-precision weights only when needed.
CLONE: Customizing LLMs for Efficient Latency-Aware Inference at the Edge
2506.02847v1
llm_flash
\cite{llm_flash}
LLM in a flash: Efficient Large Language Model Inference with Limited Memory
http://arxiv.org/abs/2312.11514v3
Large language models (LLMs) are central to modern natural language processing, delivering exceptional performance in various tasks. However, their substantial computational and memory requirements present challenges, especially for devices with limited DRAM capacity. This paper tackles the challenge of efficiently run...
true
true
Keivan Alizadeh and Iman Mirzadeh and Dmitry Belenko and Karen Khatamifard and Minsik Cho and Carlo C. Del Mundo and Mohammad Rastegari and Mehrdad Farajtabar
2,023
null
https://doi.org/10.48550/arXiv.2312.11514
10.48550/ARXIV.2312.11514
CoRR
LLM in a flash: Efficient Large Language Model Inference with Limited Memory
LLM in a Flash: Efficient Large Language Model Inference with ...
https://machinelearning.apple.com/research/efficient-large-language
# LLM in a Flash: Efficient Large Language Model Inference with Limited Memory This paper tackles the challenge of efficiently running LLMs that exceed the available DRAM capacity by storing the model parameters in flash memory, but bringing them on demand to DRAM. Our method involves constructing an inference cost mod...
CLONE: Customizing LLMs for Efficient Latency-Aware Inference at the Edge
2506.02847v1
jawahar2023llm
\cite{jawahar2023llm}
LLM Performance Predictors are good initializers for Architecture Search
http://arxiv.org/abs/2310.16712v2
In this work, we utilize Large Language Models (LLMs) for a novel use case: constructing Performance Predictors (PP) that estimate the performance of specific deep neural network architectures on downstream tasks. We create PP prompts for LLMs, comprising (i) role descriptions, (ii) instructions for the LLM, (iii) hype...
true
true
Jawahar, Ganesh and Abdul-Mageed, Muhammad and Lakshmanan, Laks VS and Ding, Dujian
2,023
null
null
null
arXiv preprint arXiv:2310.16712
LLM Performance Predictors are good initializers for Architecture Search
LLM Performance Predictors are good initializers for Architecture Search
http://arxiv.org/pdf/2310.16712v2
In this work, we utilize Large Language Models (LLMs) for a novel use case: constructing Performance Predictors (PP) that estimate the performance of specific deep neural network architectures on downstream tasks. We create PP prompts for LLMs, comprising (i) role descriptions, (ii) instructions for the LLM, (iii) hype...
CLONE: Customizing LLMs for Efficient Latency-Aware Inference at the Edge
2506.02847v1
huang2024new
\cite{huang2024new}
New Solutions on LLM Acceleration, Optimization, and Application
http://arxiv.org/abs/2406.10903v1
Large Language Models (LLMs) have become extremely potent instruments with exceptional capacities for comprehending and producing human-like text in a wide range of applications. However, the increasing size and complexity of LLMs present significant challenges in both training and deployment, leading to substantial co...
true
true
Huang, Yingbing and Wan, Lily Jiaxin and Ye, Hanchen and Jha, Manvi and Wang, Jinghua and Li, Yuhong and Zhang, Xiaofan and Chen, Deming
2,024
null
null
null
null
New Solutions on LLM Acceleration, Optimization, and Application
New Solutions on LLM Acceleration, Optimization, and Application
http://arxiv.org/pdf/2406.10903v1
Large Language Models (LLMs) have become extremely potent instruments with exceptional capacities for comprehending and producing human-like text in a wide range of applications. However, the increasing size and complexity of LLMs present significant challenges in both training and deployment, leading to substantial co...
CLONE: Customizing LLMs for Efficient Latency-Aware Inference at the Edge
2506.02847v1
liu2024optimizing
\cite{liu2024optimizing}
Optimizing LLM Queries in Relational Data Analytics Workloads
http://arxiv.org/abs/2403.05821v2
Batch data analytics is a growing application for Large Language Models (LLMs). LLMs enable users to perform a wide range of natural language tasks, such as classification, entity extraction, and translation, over large datasets. However, LLM inference is highly costly and slow: for example, an NVIDIA L4 GPU running Ll...
true
true
Liu, Shu and Biswal, Asim and Cheng, Audrey and Mo, Xiangxi and Cao, Shiyi and Gonzalez, Joseph E and Stoica, Ion and Zaharia, Matei
2,024
null
null
null
arXiv preprint arXiv:2403.05821
Optimizing LLM Queries in Relational Data Analytics Workloads
Optimizing LLM Queries in Relational Data Analytics Workloads
http://arxiv.org/pdf/2403.05821v2
Batch data analytics is a growing application for Large Language Models (LLMs). LLMs enable users to perform a wide range of natural language tasks, such as classification, entity extraction, and translation, over large datasets. However, LLM inference is highly costly and slow: for example, an NVIDIA L4 GPU running Ll...
CLONE: Customizing LLMs for Efficient Latency-Aware Inference at the Edge
2506.02847v1
hubara2018quantized
\cite{hubara2018quantized}
Quantized neural networks: Training neural networks with low precision weights and activations
null
null
true
false
Hubara, Itay and Courbariaux, Matthieu and Soudry, Daniel and El-Yaniv, Ran and Bengio, Yoshua
2,018
null
null
null
Journal of Machine Learning Research
Quantized neural networks: Training neural networks with low precision weights and activations
Quantized Neural Networks: Training Neural Networks with Low Precision Weights and Activations
http://arxiv.org/pdf/1609.07061v1
We introduce a method to train Quantized Neural Networks (QNNs) --- neural networks with extremely low precision (e.g., 1-bit) weights and activations, at run-time. At train-time the quantized weights and activations are used for computing the parameter gradients. During the forward pass, QNNs drastically reduce memory...
CLONE: Customizing LLMs for Efficient Latency-Aware Inference at the Edge
2506.02847v1
GPT3.int8
\cite{GPT3.int8}
LLM.int8(): 8-bit Matrix Multiplication for Transformers at Scale
null
null
true
false
Tim Dettmers and Mike Lewis and Younes Belkada and Luke Zettlemoyer
2,022
null
http://papers.nips.cc/paper\_files/paper/2022/hash/c3ba4962c05c49636d4c6206a97e9c8a-Abstract-Conference.html
null
null
LLM.int8(): 8-bit Matrix Multiplication for Transformers at Scale
LLM.int8(): 8-bit Matrix Multiplication for Transformers at Scale
http://arxiv.org/pdf/2208.07339v2
Large language models have been widely adopted but require significant GPU memory for inference. We develop a procedure for Int8 matrix multiplication for feed-forward and attention projection layers in transformers, which cut the memory needed for inference by half while retaining full precision performance. With our ...
CLONE: Customizing LLMs for Efficient Latency-Aware Inference at the Edge
2506.02847v1
Deja
\cite{Deja}
Deja Vu: Contextual Sparsity for Efficient LLMs at Inference Time
http://arxiv.org/abs/2310.17157v1
Large language models (LLMs) with hundreds of billions of parameters have sparked a new wave of exciting AI applications. However, they are computationally expensive at inference time. Sparsity is a natural approach to reduce this cost, but existing methods either require costly retraining, have to forgo LLM's in-conte...
true
true
Zichang Liu and Jue Wang and Tri Dao and Tianyi Zhou and Binhang Yuan and Zhao Song and Anshumali Shrivastava and Ce Zhang and Yuandong Tian and Christopher R...
2,023
null
https://proceedings.mlr.press/v202/liu23am.html
null
null
Deja Vu: Contextual Sparsity for Efficient LLMs at Inference Time
Deja Vu: Contextual Sparsity for Efficient LLMs at Inference Time
http://arxiv.org/pdf/2310.17157v1
Large language models (LLMs) with hundreds of billions of parameters have sparked a new wave of exciting AI applications. However, they are computationally expensive at inference time. Sparsity is a natural approach to reduce this cost, but existing methods either require costly retraining, have to forgo LLM's in-conte...
CLONE: Customizing LLMs for Efficient Latency-Aware Inference at the Edge
2506.02847v1
SmoothQuant
\cite{SmoothQuant}
SmoothQuant: Accurate and Efficient Post-Training Quantization for Large Language Models
http://arxiv.org/abs/2211.10438v7
Large language models (LLMs) show excellent performance but are compute- and memory-intensive. Quantization can reduce memory and accelerate inference. However, existing methods cannot maintain accuracy and hardware efficiency at the same time. We propose SmoothQuant, a training-free, accuracy-preserving, and general-p...
true
true
Guangxuan Xiao and Ji Lin and Micka{\"{e}}l Seznec and Hao Wu and Julien Demouth and Song Han
2,023
null
https://proceedings.mlr.press/v202/xiao23c.html
null
null
SmoothQuant: Accurate and Efficient Post-Training Quantization for Large Language Models
SmoothQuant: Accurate and Efficient Post-Training Quantization for ...
https://arxiv.org/abs/2211.10438
Image 2: arxiv logo>cs> arXiv:2211.10438 **arXiv:2211.10438** (cs) View a PDF of the paper titled SmoothQuant: Accurate and Efficient Post-Training Quantization for Large Language Models, by Guangxuan Xiao and 5 other authors View a PDF of the paper titled SmoothQuant: Accurate and Efficient Post-Training Quantizatio...