Dataset Viewer
Auto-converted to Parquet Duplicate
openreview_id
stringlengths
10
10
arxiv_id
stringlengths
10
10
Model
listlengths
0
6
type
stringclasses
3 values
GitHub
listlengths
0
2
abstract_original
stringlengths
446
3.07k
project_page
stringclasses
2 values
Space
listlengths
0
2
Dataset
listlengths
0
3
title
stringlengths
15
138
authors
listlengths
1
35
arxiv_id_original
stringlengths
0
10
id
stringlengths
10
10
OpenReview
stringlengths
42
42
categories
stringlengths
5
56
update_date
timestamp[ns]date
2023-05-02 00:00:00
2025-06-13 00:00:00
title_clean
stringlengths
15
140
abstract
stringlengths
2.65k
7.53k
citation_count
float64
0
2.46k
__index_level_0__
int64
0
1.73k
1pSL2cXWoz
2402.17888
[]
Poster
[]
Post-hoc out-of-distribution (OOD) detection has garnered intensive attention in reliable machine learning. Many efforts have been dedicated to deriving score functions based on logits, distances, or rigorous data distribution assumptions to identify low-scoring OOD samples. Nevertheless, these estimate scores may fail...
[]
[]
ConjNorm: Tractable Density Estimation for Out-of-Distribution Detection
[ "Bo Peng", "Yadan Luo", "Yonggang Zhang", "Yixuan Li", "Zhen Fang" ]
2402.17888
1pSL2cXWoz
https://openreview.net/forum?id=1pSL2cXWoz
cs.LG cs.AI
2024-02-29T00:00:00
conjnorm: tractable density estimation for out-of-distribution detection
The paper introduces **CONJNORM**, a novel method for **Out-of-Distribution (OOD) detection** that focuses on **tractable density estimation**. It addresses the limitations of existing density-based OOD methods, which often rely on strong distributional assumptions (e.g., Gaussian) or struggle with the computational in...
14
0
jjiOHEcS2c
2310.01840
[]
Poster
[ "https://github.com/cszhilu1998/SelfHDR" ]
Merging multi-exposure images is a common approach for obtaining high dynamic range (HDR) images, with the primary challenge being the avoidance of ghosting artifacts in dynamic scenes. Recent methods have proposed using deep neural networks for deghosting. However, the methods typically rely on sufficient data with HD...
[]
[]
Self-Supervised High Dynamic Range Imaging with Multi-Exposure Images in Dynamic Scenes
[ "Zhilu Zhang", "Haoyu Wang", "Shuai Liu", "Xiaotao Wang", "LEI LEI", "Wangmeng Zuo" ]
2310.01840
jjiOHEcS2c
https://openreview.net/forum?id=jjiOHEcS2c
cs.CV
2024-02-29T00:00:00
self-supervised high dynamic range imaging with multi-exposure images in dynamic scenes
This paper introduces **SelfHDR**, a novel self-supervised method for High Dynamic Range (HDR) image reconstruction from multi-exposure Low Dynamic Range (LDR) images in dynamic scenes. The primary challenge addressed is the need for costly HDR ground-truth data in supervised methods and the limitations of synthetic da...
3
1
2msbbX3ydD
2310.07704
[]
Spotlight Poster
[ "https://github.com/apple/ml-ferret" ]
We introduce Ferret, a new Multimodal Large Language Model (MLLM) capable of understanding spatial referring of any shape or granularity within an image and accurately grounding open-vocabulary descriptions. To unify referring and grounding in the LLM paradigm, Ferret employs a novel and powerful hybrid region represen...
[]
[]
Ferret: Refer and Ground Anything Anywhere at Any Granularity
[ "Haoxuan You", "Haotian Zhang", "Zhe Gan", "Xianzhi Du", "Bowen Zhang", "Zirui Wang", "Liangliang Cao", "Shih-Fu Chang", "Yinfei Yang" ]
2310.07704
2msbbX3ydD
https://openreview.net/forum?id=2msbbX3ydD
cs.CV cs.CL
2023-10-12T00:00:00
ferret: refer and ground anything anywhere at any granularity
The paper introduces **Ferret**, a novel Multimodal Large Language Model (MLLM) designed to enhance spatial understanding and open-vocabulary grounding capabilities. Ferret addresses the limitations of existing MLLMs by unifying referring and grounding tasks, handling diverse region shapes, and improving robustness aga...
328
2
PKICZXVY9M
2401.15914
[]
Poster
[]
Existing vision-language models exhibit strong generalization on a variety of visual domains and tasks. However, such models mainly perform zero-shot recognition in a closed-set manner, and thus struggle to handle open-domain visual concepts by design. There are recent finetuning methods, such as prompt learning, that ...
[]
[]
Overcoming the Pitfalls of Vision-Language Model Finetuning for OOD Generalization
[ "Yuhang Zang", "Hanlin Goh", "Joshua M. Susskind", "Chen Huang" ]
2401.15914
PKICZXVY9M
https://openreview.net/forum?id=PKICZXVY9M
cs.CV cs.AI
2024-04-17T00:00:00
overcoming the pitfalls of vision-language model finetuning for ood generalization
This paper, "Overcoming the Pitfalls of Vision-Language Model Finetuning for OOD Generalization," addresses a critical issue in the application of large-scale vision-language models (VLMs) like CLIP: their tendency to overfit known (in-distribution, ID) classes during finetuning, leading to degraded performance on unkn...
13
3
Lvf7GnaLru
2312.16313
[]
Poster
[]
Supervised learning datasets may contain multiple cues that explain the training set equally well, i.e., learning any of them would lead to the correct predictions on the training data. However, many of them can be spurious, i.e., lose their predictive power under a distribution shift and consequently fail to generaliz...
[]
[]
Unraveling the Key Components of OOD Generalization via Diversification
[ "Harold Luc Benoit", "Liangze Jiang", "Andrei Atanov", "Oguzhan Fatih Kar", "Mattia Rigotti", "Amir Zamir" ]
2312.16313
Lvf7GnaLru
https://openreview.net/forum?id=Lvf7GnaLru
cs.LG
2024-04-23T00:00:00
unraveling the key components of ood generalization via diversification
This paper, "UNRAVELING THE KEY COMPONENTS OF OOD GENERALIZATION VIA DIVERSIFICATION," investigates the factors influencing the Out-of-Distribution (OOD) generalization abilities of recently developed "diversification" methods. These methods, exemplified by DivDis (Lee et al., 2023) and D-BAT (Pagliardini et al., 2023)...
2
4
iIT02bAKzv
2310.02998
[]
Poster
[]
Large Vision-Language Models (LVLMs) can understand the world comprehensively by integrating rich information from different modalities, achieving remarkable performance improvements on various multimodal downstream tasks. However, deploying LVLMs is often problematic due to their massive computational/energy costs and...
[]
[]
ECoFLaP: Efficient Coarse-to-Fine Layer-Wise Pruning for Vision-Language Models
[ "Yi-Lin Sung", "Jaehong Yoon", "Mohit Bansal" ]
2310.02998
iIT02bAKzv
https://openreview.net/forum?id=iIT02bAKzv
cs.CV cs.AI cs.CL cs.LG
2024-01-29T00:00:00
ecoflap: efficient coarse-to-fine layer-wise pruning for vision-language models
The paper introduces **ECoFLaP (Efficient Coarse-to-Fine Layer-Wise Pruning)**, a novel two-stage weight pruning approach designed for Large Vision-Language Models (LVLMs). It addresses the significant computational and memory costs associated with deploying LVLMs, as well as the limitations of existing pruning methods...
14
5
eFWG9Cy3WK
2310.01334
[]
Spotlight Poster
[]
Sparsely activated Mixture-of-Experts (SMoE) has shown promise to scale up the learning capacity of neural networks, however, they have issues like: ($a$) $\textit{High Memory Usage,}$ due to duplication of the network layers into multiple copies as experts; and ($b$) $\textit{Redundancy in Experts,}$ as common learnin...
[]
[]
Merge, Then Compress: Demystify Efficient SMoE with Hints from Its Routing Policy
[ "Pingzhi Li", "Zhenyu Zhang", "Prateek Yadav", "Yi-Lin Sung", "Yu Cheng", "Mohit Bansal", "Tianlong Chen" ]
2310.01334
eFWG9Cy3WK
https://openreview.net/forum?id=eFWG9Cy3WK
cs.LG cs.AI cs.CL
2024-03-15T00:00:00
merge, then compress: demystify efficient smoe with hints from its routing policy
This paper, "MERGE, THEN COMPRESS: DEMYSTIFY EFFICIENT SMOE WITH HINTS FROM ITS ROUTING POLICY," addresses the significant memory usage and expert redundancy issues prevalent in conventional Sparsely activated Mixture-of-Experts (SMoE) models. The authors propose a novel two-stage approach: **M-SMOE (Merge SMOE)** foll...
39
6
JnRStoIuTe
2305.18424
[]
Poster
[]
Methods for carefully selecting or generating a small set of training data to learn from, i.e., data pruning, coreset selection, and dataset distillation, have been shown to be effective in reducing the ever-increasing cost of training neural networks. Behind this success are rigorously designed, yet expensive, strateg...
[]
[]
Repeated Random Sampling for Minimizing the Time-to-Accuracy of Learning
[ "Patrik Okanovic", "Roger Waleffe", "Vasilis Mageirakos", "Konstantinos Nikolakakis", "Amin Karbasi", "Dionysios Kalogerias", "Nezihe Merve Gürel", "Theodoros Rekatsinas" ]
2305.18424
JnRStoIuTe
https://openreview.net/forum?id=JnRStoIuTe
cs.LG cs.CV
2023-05-31T00:00:00
repeated random sampling for minimizing the time-to-accuracy of learning
The paper introduces **Repeated Sampling of Random Subsets (RS2)**, a novel yet simple method aimed at minimizing the "time-to-accuracy" in deep neural network training, addressing the high computational costs associated with large datasets. **Method Used (RS2):** Instead of relying on complex and computationally expe...
14
7
vE1e1mLJ0U
2306.16922
[]
Poster
[]
Biological cortical neurons are remarkably sophisticated computational devices,temporally integrating their vast synaptic input over an intricate dendritic tree,subject to complex, nonlinearly interacting internal biological processes. A recentstudy proposed to characterize this complexity by fitting accurate surrogate...
[]
[]
The Expressive Leaky Memory Neuron: an Efficient and Expressive Phenomenological Neuron Model Can Solve Long-Horizon Tasks.
[ "Aaron Spieler", "Nasim Rahaman", "Georg Martius", "Bernhard Schölkopf", "Anna Levina" ]
2306.16922
vE1e1mLJ0U
https://openreview.net/forum?id=vE1e1mLJ0U
cs.NE cs.AI cs.LG q-bio.NC
2024-03-19T00:00:00
the expressive leaky memory neuron: an efficient and expressive phenomenological neuron model can solve long-horizon tasks
This paper introduces the **Expressive Leaky Memory (ELM) neuron**, a biologically inspired, phenomenological neuron model designed to efficiently capture the complex input-output (I/O) relationships of cortical neurons and solve long-horizon temporal tasks. **Method Used:** The ELM neuron is structured as a recurren...
5
8
xwKt6bUkXj
2309.12927
[]
Poster
[]
Recurrent neural networks (RNNs) in the brain and in silico excel at solving tasks with intricate temporal dependencies. Long timescales required for solving such tasks can arise from properties of individual neurons (single-neuron timescale, $\tau$, e.g., membrane time constant in biological neurons) or recurrent inte...
[]
[]
Emergent mechanisms for long timescales depend on training curriculum and affect performance in memory tasks
[ "Sina Khajehabdollahi", "Roxana Zeraati", "Emmanouil Giannakakis", "Tim Jakob Schäfer", "Georg Martius", "Anna Levina" ]
2309.12927
xwKt6bUkXj
https://openreview.net/forum?id=xwKt6bUkXj
cs.NE q-bio.NC
2024-10-31T00:00:00
emergent mechanisms for long timescales depend on training curriculum and affect performance in memory tasks
This paper investigates how recurrent neural networks (RNNs) develop long timescales to solve memory-dependent tasks, focusing on the interplay between single-neuron intrinsic timescales ($\tau$) and network-mediated timescales ($\tau_{net}$). A key aspect of their methodology is the comparison of different training cu...
5
9
sY5N0zY5Od
2310.03714
[]
Spotlight Poster
[]
The ML community is rapidly exploring techniques for prompting language models (LMs) and for stacking them into pipelines that solve complex tasks. Unfortunately, existing LM pipelines are typically implemented using hard-coded “prompt templates”, i.e. lengthy strings discovered via trial and error. Toward a more syste...
[]
[]
DSPy: Compiling Declarative Language Model Calls into State-of-the-Art Pipelines
[ "Omar Khattab", "Arnav Singhvi", "Paridhi Maheshwari", "Zhiyuan Zhang", "Keshav Santhanam", "Sri Vardhamanan A", "Saiful Haq", "Ashutosh Sharma", "Thomas T. Joshi", "Hanna Moazam", "Heather Miller", "Matei Zaharia", "Christopher Potts" ]
2310.03714
sY5N0zY5Od
https://openreview.net/forum?id=sY5N0zY5Od
cs.CL cs.AI cs.IR cs.LG
2023-10-06T00:00:00
dspy: compiling declarative language model calls into self-improving pipelines
The paper introduces **DSPy**, a new programming model and compiler designed to address the challenges of building and optimizing Language Model (LM) pipelines, particularly the brittleness and unscalability of traditional "hard-coded prompt templates." **Method Used:** DSPy abstracts LM pipelines as **text transform...
287
10
1YPfmglNRU
2403.00694
[]
Poster
[]
Decision-makers are often experts of their domain and take actions based on their domain knowledge. Doctors, for instance, may prescribe treatments by predicting the likely outcome of each available treatment. Actions of an expert thus naturally encode part of their domain knowledge, and can help make inferences within...
[]
[]
Defining Expertise: Applications to Treatment Effect Estimation
[ "Alihan Hüyük", "Qiyao Wei", "Alicia Curth", "Mihaela van der Schaar" ]
2403.00694
1YPfmglNRU
https://openreview.net/forum?id=1YPfmglNRU
stat.ML cs.AI cs.LG stat.ME
2024-03-04T00:00:00
defining expertise: applications to treatment effect estimation
This paper, "Defining Expertise: Applications to Treatment Effect Estimation," introduces a novel perspective on leveraging decision-makers' expertise as an inductive bias in treatment effect estimation, rather than solely viewing it as a source of confounding. **Core Concepts and Definitions:** The authors formally ...
2
11
fq1wNrC2ai
2404.12648
[]
Poster
[]
We study infinite-horizon average-reward Markov decision processes (AMDPs) in the context of general function approximation. Specifically, we propose a novel algorithmic framework named Fixed-Point Local Optimization (FLOP), which incorporates both model-based and value-based incarnations. In particular, FLOP features ...
[]
[]
Sample-efficient Learning of Infinite-horizon Average-reward MDPs with General Function Approximation
[ "Jianliang He", "Han Zhong", "Zhuoran Yang" ]
2404.12648
fq1wNrC2ai
https://openreview.net/forum?id=fq1wNrC2ai
cs.LG stat.ML
2024-04-22T00:00:00
sample-efficient learning of infinite-horizon average-reward mdps with general function approximation
This paper addresses the challenging problem of sample-efficient learning in **infinite-horizon average-reward Markov Decision Processes (AMDPs)**, particularly in the context of **general function approximation**. The authors propose a novel algorithmic framework called **Local-fitted Optimization with OPtimism (LOOP)...
6
12
5hAMmCU0bK
2310.12955
[]
Spotlight Poster
[]
Offline reinforcement learning (RL) presents a promising approach for learning reinforced policies from offline datasets without the need for costly or unsafe interactions with the environment. However, datasets collected by humans in real-world environments are often noisy and may even be maliciously corrupted, which ...
[]
[]
Towards Robust Offline Reinforcement Learning under Diverse Data Corruption
[ "Rui Yang", "Han Zhong", "Jiawei Xu", "Amy Zhang", "Chongjie Zhang", "Lei Han", "Tong Zhang" ]
2310.12955
5hAMmCU0bK
https://openreview.net/forum?id=5hAMmCU0bK
cs.LG cs.AI
2024-03-12T00:00:00
towards robust offline reinforcement learning under diverse data corruption
This paper, "Towards Robust Offline Reinforcement Learning Under Diverse Data Corruption," investigates and addresses the challenges of data corruption in offline Reinforcement Learning (RL). **Problem:** Offline RL aims to learn policies from static datasets without costly or unsafe online interactions. However, real...
18
13
9Cu8MRmhq2
2401.16702
[]
Oral
[]
Existing video-language studies mainly focus on learning short video clips, leaving long-term temporal dependencies rarely explored due to over-high computational cost of modeling long videos. To address this issue, one feasible solution is learning the correspondence between video clips and captions, which however ine...
[]
[]
Multi-granularity Correspondence Learning from Long-term Noisy Videos
[ "Yijie Lin", "Jie Zhang", "Zhenyu Huang", "Jia Liu", "zujie wen", "Xi Peng" ]
2401.16702
9Cu8MRmhq2
https://openreview.net/forum?id=9Cu8MRmhq2
cs.CV
2024-01-31T00:00:00
multi-granularity correspondence learning from long-term noisy videos
This paper introduces **Norton (NOise Robust Temporal Optimal traNsport)**, a novel framework for multi-granularity correspondence learning in long-term noisy videos. The core problem addressed is "Multi-Granularity Noisy Correspondence (MNC)," which arises when attempting to learn long-term temporal dependencies by br...
22
14
h922Qhkmx1
2302.02257
[]
Oral
[]
In this work, we define a diffusion-based generative model capable of both music generation and source separation by learning the score of the joint probability density of sources sharing a context. Alongside the classic total inference tasks (i.e., generating a mixture, separating the sources), we also introduce and e...
[]
[]
Multi-Source Diffusion Models for Simultaneous Music Generation and Separation
[ "Giorgio Mariani", "Irene Tallini", "Emilian Postolache", "Michele Mancusi", "Luca Cosmo", "Emanuele Rodolà" ]
2302.02257
h922Qhkmx1
https://openreview.net/forum?id=h922Qhkmx1
cs.SD cs.LG eess.AS
2024-03-19T00:00:00
multi-source diffusion models for simultaneous music generation and separation
This paper introduces a novel **Multi-Source Diffusion Model (MSDM)** designed for the simultaneous tasks of music generation and source separation. Unlike previous approaches that typically address these tasks separately, MSDM learns the **joint probability density of individual sources** (e.g., bass, drums, guitar, p...
43
15
GSBHKiw19c
2310.05422
[]
Spotlight Poster
[]
Learning a precise dynamics model can be crucial for offline reinforcement learning, which, unfortunately, has been found to be quite challenging. Dynamics models that are learned by fitting historical transitions often struggle to generalize to unseen transitions. In this study, we identify a hidden but pivotal factor...
[]
[]
Reward-Consistent Dynamics Models are Strongly Generalizable for Offline Reinforcement Learning
[ "Fan-Ming Luo", "Tian Xu", "Xingchen Cao", "Yang Yu" ]
2310.05422
GSBHKiw19c
https://openreview.net/forum?id=GSBHKiw19c
cs.LG
2023-10-10T00:00:00
reward-consistent dynamics models are strongly generalizable for offline reinforcement learning
This paper introduces **MOREC (Model-based Offline reinforcement learning with Reward Consistency)**, a novel approach to improve the generalization ability of dynamics models in offline Reinforcement Learning (RL). **The Core Problem:** Offline RL relies on learning accurate dynamics models from a fixed dataset of hi...
10
16
gU58d5QeGv
2306.00637
[ "warp-ai/wuerstchen" ]
Oral
[]
We introduce Würstchen, a novel architecture for text-to-image synthesis that combines competitive performance with unprecedented cost-effectiveness for large-scale text-to-image diffusion models.A key contribution of our work is to develop a latent diffusion technique in which we learn a detailed but extremely compact...
[ "warp-ai/Wuerstchen" ]
[]
Würstchen: An Efficient Architecture for Large-Scale Text-to-Image Diffusion Models
[ "Pablo Pernias", "Dominic Rampas", "Mats Leon Richter", "Christopher Pal", "Marc Aubreville" ]
2306.00637
gU58d5QeGv
https://openreview.net/forum?id=gU58d5QeGv
cs.CV
2024-05-31T00:00:00
wuerstchen: an efficient architecture for large-scale text-to-image diffusion models
Würstchen introduces a novel, highly efficient three-stage architecture for large-scale text-to-image diffusion models, significantly reducing computational costs while maintaining competitive performance. **Method:** The core innovation lies in learning a **detailed but extremely compact semantic image representatio...
45
17
1YO4EE3SPB
2305.04391
[]
Poster
[]
Diffusion models have emerged as a key pillar of foundation models in visual domains. One of their critical applications is to universally solve different downstream inverse tasks via a single diffusion prior without re-training for each task. Most inverse tasks can be formulated as inferring a posterior distribution o...
[]
[]
A Variational Perspective on Solving Inverse Problems with Diffusion Models
[ "Morteza Mardani", "Jiaming Song", "Jan Kautz", "Arash Vahdat" ]
2305.04391
1YO4EE3SPB
https://openreview.net/forum?id=1YO4EE3SPB
cs.LG cs.CV cs.NA math.NA stat.ML
2023-10-03T00:00:00
a variational perspective on solving inverse problems with diffusion models
This paper introduces **RED-diff**, a novel variational approach for solving inverse problems using diffusion models. It frames the problem as a stochastic optimization task, leveraging the denoising diffusion process as a prior. **Method Description:** RED-diff addresses the challenge of intractable posterior distrib...
146
18
1RE0H6mU7M
2403.09859
[]
Poster
[]
Meta-reinforcement learning (meta-RL) is a promising framework for tackling challenging domains requiring efficient exploration. Existing meta-RL algorithms are characterized by low sample efficiency, and mostly focus on low-dimensional task distributions. In parallel, model-based RL methods have been successful in sol...
[]
[]
MAMBA: an Effective World Model Approach for Meta-Reinforcement Learning
[ "Zohar Rimon", "Tom Jurgenson", "Orr Krupnik", "Gilad Adler", "Aviv Tamar" ]
2403.09859
1RE0H6mU7M
https://openreview.net/forum?id=1RE0H6mU7M
cs.LG
2024-03-18T00:00:00
mamba: an effective world model approach for meta-reinforcement learning
This paper introduces **MAMBA (MetA-RL Model-Based Algorithm)**, a novel approach for meta-reinforcement learning (meta-RL) that leverages the strengths of model-based RL, specifically the Dreamer architecture. The authors address the limitations of existing meta-RL methods, which often suffer from low sample efficienc...
10
19
cxfPefbu1s
2311.14688
[]
Spotlight Poster
[]
We reveal and address the frequently overlooked yet important issue of _disguised procedural unfairness_, namely, the potentially inadvertent alterations on the behavior of neutral (i.e., not problematic) aspects of data generating process, and/or the lack of procedural assurance of the greatest benefit of the least ad...
[]
[]
Procedural Fairness Through Decoupling Objectionable Data Generating Components
[ "Zeyu Tang", "Jialu Wang", "Yang Liu", "Peter Spirtes", "Kun Zhang" ]
2311.14688
cxfPefbu1s
https://openreview.net/forum?id=cxfPefbu1s
cs.CY cs.AI cs.LG
2024-05-13T00:00:00
procedural fairness through decoupling objectionable data generating components
The paper "PROCEDURAL FAIRNESS THROUGH DECOUPLING OBJECTIONABLE DATA GENERATING COMPONENTS" addresses the issue of "disguised procedural unfairness" in automated decision-making systems. This type of unfairness arises when the data generating process inadvertently alters the behavior of neutral components or fails to e...
2
20
ZzmKEpze8e
2306.08448
[]
Poster
[]
In Online Continual Learning (OCL) a learning system receives a stream of data and sequentially performs prediction and training steps. Important challenges in OCL are concerned with automatic adaptation to the particular non-stationary structure of the data, and with quantification of predictive uncertainty. Motivated...
[]
[]
Kalman Filter for Online Classification of Non-Stationary Data
[ "Michalis Titsias", "Alexandre Galashov", "Amal Rannen-Triki", "Razvan Pascanu", "Yee Whye Teh", "Jorg Bornschein" ]
2306.08448
ZzmKEpze8e
https://openreview.net/forum?id=ZzmKEpze8e
cs.LG cs.AI
2024-11-11T00:00:00
kalman filter for online classification of non-stationary data
This paper introduces a novel probabilistic Bayesian online learning approach, named **Kalman Filter for Online Classification of Non-Stationary Data**, designed for Online Continual Learning (OCL) scenarios where data streams are non-stationary and predictive uncertainty is crucial. **Method Used:** The core of the ...
10
21
vfzRRjumpX
2402.01935
[]
Poster
[]
Recent studies have shown that code language model at scale demonstrate significant performance gains on downstream tasks, i.e., code generation. However, most of the existing works on code representation learning train models at a hundred million parameter scale using very limited pretraining corpora. In this work, we...
[]
[]
CODE REPRESENTATION LEARNING AT SCALE
[ "Dejiao Zhang", "Wasi Uddin Ahmad", "Ming Tan", "Hantian Ding", "Ramesh Nallapati", "Dan Roth", "Xiaofei Ma", "Bing Xiang" ]
2402.01935
vfzRRjumpX
https://openreview.net/forum?id=vfzRRjumpX
cs.CL
2024-02-06T00:00:00
code representation learning at scale
This paper introduces **CODESAGE**, a novel bidirectional encoder model for code representation learning, designed to leverage massive code corpora at scale. The core contribution is a **two-stage pretraining scheme** that addresses limitations of prior work, particularly the suboptimal performance of standard Masked L...
12
22
kC5nZDU5zf
2311.04193
[]
Spotlight Poster
[]
Embodied AI models often employ off the shelf vision backbones like CLIP to encode their visual observations. Although such general purpose representations encode rich syntactic and semantic information about the scene, much of this information is often irrelevant to the specific task at hand. This introduces noise wit...
[]
[]
Selective Visual Representations Improve Convergence and Generalization for Embodied AI
[ "Ainaz Eftekhar", "Kuo-Hao Zeng", "Jiafei Duan", "Ali Farhadi", "Aniruddha Kembhavi", "Ranjay Krishna" ]
2311.04193
kC5nZDU5zf
https://openreview.net/forum?id=kC5nZDU5zf
cs.CV cs.AI
2024-03-12T00:00:00
selective visual representations improve convergence and generalization for embodied ai
This document introduces a novel approach to improve embodied AI agents by making their visual representations more selective and task-relevant, inspired by human selective attention. **Problem:** Current embodied AI models often rely on general-purpose vision backbones (like CLIP) that encode a vast amount of visual ...
17
23
GTk0AdOYLq
2310.01381
[]
Poster
[]
Diffusion models have recently been shown to be relevant for high-quality speech generation. Most work has been focused on generating spectrograms, and as such, they further require a subsequent model to convert the spectrogram to a waveform (i.e., a vocoder). This work proposes a diffusion probabilistic end-to-end mod...
[]
[]
DiffAR: Denoising Diffusion Autoregressive Model for Raw Speech Waveform Generation
[ "Roi Benita", "Michael Elad", "Joseph Keshet" ]
2310.01381
GTk0AdOYLq
https://openreview.net/forum?id=GTk0AdOYLq
cs.SD cs.CL eess.AS
2024-03-12T00:00:00
diffar: denoising diffusion autoregressive model for raw speech waveform generation
The paper introduces **DIFFAR (Denoising Diffusion Autoregressive Model for Raw Speech Waveform Generation)**, an end-to-end generative model designed to directly synthesize raw speech waveforms. Unlike many existing systems that rely on intermediate acoustic features like mel-spectrograms and require a separate vocode...
8
24
jsWCmrsHHs
2211.10936
[]
Poster
[]
Recent studies in using deep reinforcement learning (DRL) to solve Job-shop scheduling problems (JSSP) focus on construction heuristics. However, their performance is still far from optimality, mainly because the underlying graph representation scheme is unsuitable for modelling partial solutions at each construction s...
[]
[]
Deep Reinforcement Learning Guided Improvement Heuristic for Job Shop Scheduling
[ "Cong Zhang", "Zhiguang Cao", "Wen Song", "Yaoxin Wu", "Jie Zhang" ]
2211.10936
jsWCmrsHHs
https://openreview.net/forum?id=jsWCmrsHHs
cs.LG cs.AI
2024-02-15T00:00:00
deep reinforcement learning guided improvement heuristic for job shop scheduling
This paper introduces a novel Deep Reinforcement Learning (DRL)-guided *improvement heuristic* for solving Job-shop Scheduling Problems (JSSP), aiming to overcome limitations of existing DRL methods that primarily focus on *construction heuristics* for partial solutions. **Method Used:** 1. **Problem Formulation:** ...
18
25
GURqUuTebY
2403.14966
[]
Spotlight Poster
[]
Recent progress in text-to-3D generation has been achieved through the utilization of score distillation methods: they make use of the pre-trained text-to-image (T2I) diffusion models by distilling via the diffusion model training objective. However, such an approach inevitably results in the use of random timesteps at...
[]
[]
DreamFlow: High-quality text-to-3D generation by Approximating Probability Flow
[ "Kyungmin Lee", "Kihyuk Sohn", "Jinwoo Shin" ]
2403.14966
GURqUuTebY
https://openreview.net/forum?id=GURqUuTebY
cs.CV
2024-03-25T00:00:00
dreamflow: high-quality text-to-3d generation by approximating probability flow
This paper introduces **DreamFlow**, a novel method for high-quality text-to-3D generation that addresses the limitations of existing score distillation (SDS) techniques. **Problem Addressed:** Current text-to-3D generation methods, primarily based on Score Distillation Sampling (SDS) or Variational Score Distillation...
19
26
mYWsyTuiRp
2302.00456
[]
Spotlight Poster
[]
Given that Transformers are ubiquitous in wide tasks, interpreting their internals is a pivotal issue. Still, their particular components, feed-forward (FF) blocks, have typically been less analyzed despite their substantial parameter amounts.We analyze the input contextualization effects of FF blocks by rendering them...
[]
[]
Analyzing Feed-Forward Blocks in Transformers through the Lens of Attention Map
[ "Goro Kobayashi", "Tatsuki Kuribayashi", "Sho Yokoi", "Kentaro Inui" ]
2302.00456
mYWsyTuiRp
https://openreview.net/forum?id=mYWsyTuiRp
cs.CL
2024-04-16T00:00:00
analyzing feed-forward blocks in transformers through the lens of attention maps
This paper presents a novel method to analyze the input contextualization effects of Feed-Forward (FF) blocks within Transformer layers, a component that has historically received less attention in interpretability studies despite its substantial parameter count. The study extends existing norm-based attention map anal...
16
27
End of preview. Expand in Data Studio
README.md exists but content is empty.
Downloads last month
9