Workshop
Third Workshop on Efficient Natural Language and Speech Processing (ENLSP-III): Towards the Future of Large Language Models and their Emerging Descendants
Mehdi Rezagholizadeh · Peyman Passban · Yue Dong · Yu Cheng · Soheila Samiee · Lili Mou · Qun Liu · Boxing Chen
Room 206 - 207
Sat 16 Dec, 6:15 a.m. PST
The third version of the Efficient Natural Language and Speech Processing (ENLSP-III) workshop will focus on the future of large language and speech foundation models; and how to make them more efficient in terms of Data, Model, Training, and Inference for real-world applications as well as academic research. The workshop program offers an interactive platform for gathering different experts and talents from academia and industry through invited talks, panel discussion, paper submissions, reviews, interactive posters, oral presentations and a mentorship program. This will be a unique opportunity to discuss and share challenging problems, build connections, exchange ideas and brainstorm solutions, and foster future collaborations. The topics of this workshop can be of interest for people working on general machine learning, deep learning, optimization, theory and NLP & Speech applications.
Schedule
Sat 6:15 a.m. - 6:20 a.m.
|
Breakfast
|
🔗 |
Sat 6:16 a.m. - 6:20 a.m.
|
Opening Speech
(
Opening
)
>
link
SlidesLive Video |
Mehdi Rezagholizadeh 🔗 |
Sat 6:20 a.m. - 6:45 a.m.
|
Deploying efficient translation at every level of the stack
(
KeyNote Talk
)
>
SlidesLive Video |
Kenneth Heafield 🔗 |
Sat 6:45 a.m. - 7:30 a.m.
|
Simple and efficient self-training approaches for speech recognition
(
KeyNote Talk
)
>
SlidesLive Video |
Tatiana Likhomanenko · Samy Bengio 🔗 |
Sat 7:30 a.m. - 7:36 a.m.
|
[Paper-Oral 1] Query-Dependent Prompt Evaluation and Optimization with Offline Inverse RL
(
Oral
)
>
link
SlidesLive Video |
Hao Sun · Alihan Hüyük · Mihaela van der Schaar 🔗 |
Sat 7:36 a.m. - 7:42 a.m.
|
[Paper-Oral 2] MatFormer: Nested Transformer for Elastic Inference
(
Oral
)
>
SlidesLive Video |
11 presentersFnu Devvrit · Sneha Kudugunta · Aditya Kusupati · Tim Dettmers · Kaifeng Chen · Inderjit Dhillon · Yulia Tsvetkov · Hanna Hajishirzi · Sham Kakade · Ali Farhadi · Prateek Jain |
Sat 7:42 a.m. - 7:48 a.m.
|
[Paper-Oral 3] Decoding Data Quality via Synthetic Corruptions: Embedding-guided Pruning of Code Data
(
Oral
)
>
SlidesLive Video |
Yu Yang · Aaditya Singh · Mostafa Elhoushi · Anas Mahmoud · Kushal Tirumala · Fabian Gloeckle · Baptiste Roziere · Carole-Jean Wu · Ari Morcos · Newsha Ardalani 🔗 |
Sat 7:48 a.m. - 7:54 a.m.
|
[Paper-Oral 4] FlashFFTConv: Efficient Convolutions for Long Sequences with Tensor Cores
(
Oral
)
>
SlidesLive Video |
Dan Fu · Hermann Kumbong · Eric Nguyen · Christopher Ré 🔗 |
Sat 7:54 a.m. - 8:00 a.m.
|
[Paper-Oral 5] Ensemble of low-rank adapters for large language model fine-tuning
(
Oral
)
>
SlidesLive Video |
Xi Wang · Laurence Aitchison · Maja Rudolph 🔗 |
Sat 8:00 a.m. - 8:30 a.m.
|
Morning Break and Poster Setup
|
🔗 |
Sat 8:30 a.m. - 9:00 a.m.
|
Branch-Train-Merge: Embarrassingly Parallel Training of Expert Language Models
(
KeyNote Talk
)
>
SlidesLive Video |
Luke Zettlemoyer 🔗 |
Sat 9:00 a.m. - 9:30 a.m.
|
Knowledge Consolidation and Utilization (In)Ability of Large Language Models
(
KeyNote Talk
)
>
SlidesLive Video |
Sarath Chandar 🔗 |
Sat 9:30 a.m. - 9:36 a.m.
|
[Paper-Oral 6] LoDA: Low-Dimensional Adaptation of Large Language Models
(
Oral
)
>
SlidesLive Video |
Jing Liu · Toshiaki Koike-Akino · Perry Wang · Matthew Brand · Ye Wang · Kieran Parsons 🔗 |
Sat 9:36 a.m. - 9:42 a.m.
|
[Paper-Oral 7] MultiPrompter: Cooperative Prompt Optimization with Multi-Agent Reinforcement Learning
(
Oral
)
>
SlidesLive Video |
Dong-Ki Kim · Sungryull Sohn · Lajanugen Logeswaran · Dongsub Shim · Honglak Lee 🔗 |
Sat 9:42 a.m. - 9:48 a.m.
|
[Paper-Oral 8] LoftQ: LoRA-Fine-Tuning-Aware Quantization for Large Language Models
(
Oral
)
>
SlidesLive Video |
Yixiao Li · Yifan Yu · Chen Liang · Nikos Karampatziakis · Pengcheng He · Weizhu Chen · Tuo Zhao 🔗 |
Sat 9:48 a.m. - 9:54 a.m.
|
[Paper-Oral 9] Improving Linear Attention via Softmax Mimicry
(
Oral
)
>
SlidesLive Video |
Michael Zhang · Kush Bhatia · Hermann Kumbong · Christopher Ré 🔗 |
Sat 9:54 a.m. - 10:00 a.m.
|
[Paper-Oral 10] PaSS: Parallel Speculative Sampling
(
Oral
)
>
SlidesLive Video |
Giovanni Monea · Armand Joulin · Edouard Grave 🔗 |
Sat 10:00 a.m. - 11:00 a.m.
|
Lunch Break
|
🔗 |
Sat 11:00 a.m. - 12:00 p.m.
|
Poster Session 1 (Paper IDs:# 1-45) ( Break and Poster Session ) > link | 🔗 |
Sat 12:00 p.m. - 12:30 p.m.
|
LLMs for Protein Design: A Research Journey
(
KeyNote Talk
)
>
SlidesLive Video |
Ali Madani 🔗 |
Sat 12:30 p.m. - 1:00 p.m.
|
End-to-End Speech Recognition: The Journey from Research to Production
(
KeyNote Talk
)
>
SlidesLive Video |
Tara Sainath 🔗 |
Sat 1:00 p.m. - 1:20 p.m.
|
Break and Poster Setup
|
🔗 |
Sat 1:20 p.m. - 2:10 p.m.
|
Interactive Panel Discussion
(
Panel
)
>
SlidesLive Video |
Nazneen Rajani · Tim Dettmers · Minjia Zhang 🔗 |
Sat 2:10 p.m. - 2:15 p.m.
|
Best Paper and Poster Awards
(
Closing remark
)
>
SlidesLive Video |
Mehdi Rezagholizadeh 🔗 |
Sat 2:15 p.m. - 3:15 p.m.
|
Poster Session 2 (Paper IDs:# 46-96) ( Poster ) > link | 🔗 |
-
|
What is Lost in Knowledge Distillation?
(
Poster
)
>
|
Manas Ranjan Mohanty · Tanya Roosta · Peyman Passban 🔗 |
-
|
NLLB-CLIP - train performant multilingual image retrieval model on a budget
(
Poster
)
>
|
Alexander Visheratin 🔗 |
-
|
DePT: Decomposed Prompt Tuning for Parameter-Efficient Fine-tuning
(
Poster
)
>
|
Zhengxiang Shi · Aldo Lipani 🔗 |
-
|
LLM-MQ: Mixed-precision Quantization for Efficient LLM Deployment
(
Poster
)
>
|
Shiyao Li · Xuefei Ning · Ke Hong · Tengxuan Liu · Luning Wang · Xiuhong Li · Kai Zhong · Guohao Dai · Huazhong Yang · Yu Wang 🔗 |
-
|
Transfer Learning for Structured Pruning under Limited Task Data
(
Poster
)
>
|
Lucio M Dery · Awni Hannun · David Grangier 🔗 |
-
|
Embedding User-Generated Content using Structural Supervision and Generative Models
(
Poster
)
>
|
Vinay Shukla · Yang Yang · Siddarth Malreddy · Jinoo Baek · Dale Johnson · Wenfei Zou · Karthik Lakshmanan · Mark Williams · Minh Pham 🔗 |
-
|
Parameter Efficient Finetuning for Reducing Activation Density in Transformers
(
Poster
)
>
|
Bharat Runwal · Tejaswini Pedapati · Pin-Yu Chen 🔗 |
-
|
GQKVA: Efficient Pre-training of Transformers by Grouping Queries, Keys, and Values
(
Poster
)
>
|
Farnoosh Javadi · Walid Ahmed · Habib Hajimolahoseini · Foozhan Ataiefard · Mohammad Hassanpour · Saina Asani · Austin Wen · Omar Mohamed Awad · Kangling Liu · Yang Liu 🔗 |
-
|
Structure Discovery in Prompted Weak Supervision
(
Poster
)
>
|
Jinyan Su · Peilin Yu · Jieyu Zhang · Stephen Bach 🔗 |
-
|
SPEED: Speculative Pipelined Execution for Efficient Decoding
(
Poster
)
>
|
Coleman Hooper · Sehoon Kim · Hiva Mohammadzadeh · Hasan Genc · Kurt Keutzer · Amir Gholami · Sophia Shao 🔗 |
-
|
Efficiently Adapting Pretrained Language Models to New Languages
(
Poster
)
>
|
Zoltan Csaki · Pian Pawakapan · Urmish Thakker · Qiantong Xu 🔗 |
-
|
Efficient LLM Inference on CPUs
(
Poster
)
>
|
Haihao Shen · Hanwen Chang · Bo Dong · Hengyu Meng · Yu Luo 🔗 |
-
|
Efficient Long-Range Transformers: You Need to Attend More, but Not Necessarily at Every Layer
(
Poster
)
>
|
Qingru Zhang · Dhananjay Ram · Cole Hawkins · Sheng Zha · Tuo Zhao 🔗 |
-
|
IceFormer: Accelerated Inference with Long-Sequence Transformers on CPUs
(
Poster
)
>
|
Yuzhen Mao · Martin Ester · Ke Li 🔗 |
-
|
On the Zero-Shot Generalization of Machine-Generated Text Detectors
(
Poster
)
>
|
Xiao Pu · Jingyu Zhang · Xiaochuang Han · Yulia Tsvetkov · Tianxing He 🔗 |
-
|
Intra-Class Similarity-Guided Feature Distillation
(
Poster
)
>
|
Khouloud Saadi · Jelena Mitrović · Michael Granitzer 🔗 |
-
|
Less is More! A slim architecture, optimal for language tasks
(
Poster
)
>
|
Luca Herranz-Celotti · Ermal Rrapaj 🔗 |
-
|
Comprehensive Bench-marking of Entropy and Margin Based Scoring Metrics for Data Selection
(
Poster
)
>
|
Anusha Sabbineni · Nikhil Anand · Maria Minakova 🔗 |
-
|
Lightweight Retrieval Tuning for Black-Box Language Models
(
Poster
)
>
|
Xiao-Wen Yang · Hong-Jie You · Pengxiao Song · Hao-Ran Hao · Jie-Jing Shao · Yu-Feng Li 🔗 |
-
|
Skeleton-of-Thought: Large Language Models Can Do Parallel Decoding
(
Poster
)
>
|
Xuefei Ning · Zinan Lin · Zixuan Zhou · Zifu Wang · Huazhong Yang · Yu Wang 🔗 |
-
|
Investigating the Impact of Compression on Parametric Knowledge in Language Models
(
Poster
)
>
|
Satya Sai Srinath Namburi · Makesh Narsimhan Sreedhar · Srinath Srinivasan · Frederic Sala 🔗 |
-
|
Get more for less: Principled Data Selection for Warming Up Fine-Tuning in LLMs
(
Poster
)
>
|
Feiyang Kang · Hoang Anh Just · Himanshu Jahagirdar · Yifan Sun · Yuanzhi Zhang · Rongxing Du · Anit Kumar Sahu · Ruoxi Jia 🔗 |
-
|
Exploiting Transformer Activation Sparsity with Dynamic Inference
(
Poster
)
>
|
Mikołaj Piórczyński · Filip Szatkowski · Klaudia Bałazy · Bartosz Wójcik 🔗 |
-
|
Retrieval Augmented Generation for Dialog Modeling
(
Poster
)
>
|
Lilly Kumari · Usama Bin Shafqat · Nikhil Sarda 🔗 |
-
|
TCNCA: Temporal Convolution Network with Chunked Attention for Scalable Sequence Processing
(
Poster
)
>
|
Aleksandar Terzic · Michael Hersche · Geethan Karunaratne · Luca Benini · Abu Sebastian · Abbas Rahimi 🔗 |
-
|
Sorted LLaMA: Unlocking the Potential of Intermediate Layers of Large Language Models for Dynamic Inference Using Sorted Fine-Tuning (SoFT)
(
Poster
)
>
|
Parsa Kavehzadeh · Mojtaba Valipour · Marzieh Tahaei · Ali Ghodsi · Boxing Chen · Mehdi Rezaghoizadeh 🔗 |
-
|
Sheared LLaMA: Accelerating Language Model Pre-training via Structured Pruning
(
Poster
)
>
|
Mengzhou Xia · Tianyu Gao · Zhiyuan Zeng · Danqi Chen 🔗 |
-
|
Automatic Construction of a Korean Toxic Query Dataset for Ethical Tuning of Large Language Models
(
Poster
)
>
|
SungJoo Byun · Dongjun Jang · Hyemi Jo · HYOPIL SHIN 🔗 |
-
|
BTLM-3B-8K: 7B Parameter Performance in a 3B Parameter Model
(
Poster
)
>
|
13 presentersNolan Dey · Daria Soboleva · Faisal Al-Khateeb · Bowen Yang · Ribhu Pathria · Hemant Khachane · Shaheer Muhammad · Zhiming (Charles) Chen · Robert Myers · Jacob Robert Steeves · Natalia Vassilieva · Marvin Tom · Joel Hestness |
-
|
Sparse Fine-Tuning for Inference Acceleration of Large Language Models
(
Poster
)
>
|
Eldar Kurtic · Denis Kuznedelev · Elias Frantar · Michael Goin · Dan Alistarh 🔗 |
-
|
Model Tells You What to Discard: Adaptive KV Cache Compression for LLMs
(
Poster
)
>
|
Suyu Ge · Yunan Zhang · Liyuan Liu · Minjia Zhang · Jiawei Han · Jianfeng Gao 🔗 |
-
|
MUX-PLMs: Data Multiplexing for High-throughput Language Models
(
Poster
)
>
|
Vishvak Murahari · Ameet Deshpande · Carlos Jimenez · Izhak Shafran · Mingqiu Wang · Yuan Cao · Karthik Narasimhan 🔗 |
-
|
Towards End-to-end 4-Bit Inference on Generative Large Language Models
(
Poster
)
>
|
Saleh Ashkboos · Ilia Markov · Elias Frantar · Tingxuan Zhong · Xincheng Wang · Jie Ren · Torsten Hoefler · Dan Alistarh 🔗 |
-
|
SortedNet, a Place for Every Network and Every Network in its Place
(
Poster
)
>
|
Mojtaba Valipour · Mehdi Rezaghoizadeh · Hossein Rajabzadeh · Marzieh Tahaei · Boxing Chen · Ali Ghodsi 🔗 |
-
|
FineQuant: Unlocking Efficiency with Fine-Grained Weight-Only Quantization for LLMs
(
Poster
)
>
|
Young Jin Kim · Rawn Henry · Raffy Fahim · Hany Awadalla 🔗 |
-
|
KronA: Parameter Efficient Tuning with Kronecker Adapter
(
Poster
)
>
|
Ali Edalati · Marzieh Tahaei · Ivan Kobyzev · Vahid Partovi Nia · James J. Clark · Mehdi Rezaghoizadeh 🔗 |
-
|
ReLU Strikes Back: Exploiting Activation Sparsity in Large Language Models
(
Poster
)
>
|
Seyed Iman Mirzadeh · Keivan Alizadeh-Vahid · Sachin Mehta · Carlo C Del Mundo · Oncel Tuzel · Golnoosh Samei · Mohammad Rastegari · Mehrdad Farajtabar 🔗 |
-
|
SwiftLearn: A Data-Efficient Training Method of Deep Learning Models using Importance Sampling
(
Poster
)
>
|
11 presentersHabib Hajimolahoseini · Omar Mohamed Awad · Walid Ahmed · Austin Wen · Saina Asani · Mohammad Hassanpour · Farnoosh Javadi · Mehdi Ahmadi · Foozhan Ataiefard · Kangling Liu · Yang Liu |
-
|
Efficient Stagewise Pretraining via Progressive Subnetworks
(
Poster
)
>
|
Abhishek Panigrahi · Nikunj Saunshi · Kaifeng Lyu · Sobhan Miryoosefi · Sashank Reddi · Satyen Kale · Sanjiv Kumar 🔗 |
-
|
Herd: Using multiple, smaller LLMs to match the performances of proprietary, large LLMs via an intelligent composer
(
Poster
)
>
|
Surya Narayanan Hari · Matt Thomson 🔗 |
-
|
Efficient Online Data Mixing For Language Model Pre-Training
(
Poster
)
>
|
Alon Albalak · Liang-Ming Pan · Colin Raffel · William Yang Wang 🔗 |
-
|
Student as an Inherent Denoiser of Noisy Teacher
(
Poster
)
>
|
Jiachen Zhao 🔗 |
-
|
UT5: Pretraining Non autoregressive T5 with unrolled denoising
(
Poster
)
>
|
Mahmoud Salem · Jiayu Ye · Frederick Liu · Chu-Cheng Lin 🔗 |
-
|
LatticeGen: A Cooperative Framework Which Hides Generated Text in A Lattice For Privacy-Aware Generation on Cloud
(
Poster
)
>
|
Zhang · Tianxing He · Tianle Wang · Lu Mi · Niloofar Mireshghallah · Binyi Chen · Hao Wang · Yulia Tsvetkov 🔗 |
-
|
Measuring and Improving Recall in Convolutional Language Models
(
Poster
)
>
|
Evan Sabri Eyuboglu · Simran Arora · Aman Timalsina · Isys Johnson · Michael Poli · James Zou · Atri Rudra · Christopher Ré 🔗 |
-
|
Multimodal Multi-Hop Question Answering Through a Conversation Between Tools and Efficiently Finetuned Large Language Models
(
Poster
)
>
|
Hossein Rajabzadeh · Suyuchen Wang · HYOCK JU KWON · Bang Liu 🔗 |
-
|
Beyond Chinchilla-Optimal: Accounting for Inference in Language Model Scaling Laws
(
Poster
)
>
|
Nikhil Sardana · Jonathan Frankle 🔗 |
-
|
Continual Pre-Training of Large Language Models: How to (re)warm your model?
(
Poster
)
>
|
Kshitij Gupta · Benjamin Thérien · Adam Ibrahim · Mats L Richter · Quentin Anthony · Eugene Belilovsky · Irina Rish · Timothee Lesort 🔗 |
-
|
Improving Natural Language Understanding with Computation-Efficient Retrieval Representation Fusion
(
Poster
)
>
|
Shangyu Wu · Ying Xiong · Yufei CUI · Xue (Steve) Liu · Buzhou Tang · Tei-Wei Kuo · Chun Jason XUE 🔗 |
-
|
Mixture of Quantized Experts (MoQE): Complementary Effect of Low-bit Quantization and Robustness
(
Poster
)
>
|
Young Jin Kim · Raffy Fahim · Hany Awadalla 🔗 |
-
|
DiffTune: A Diffusion-Based Approach to Diverse Instruction-Tuning Data Generation
(
Poster
)
>
|
Suyuchen Wang · Bang Liu 🔗 |
-
|
QDyLoRA: Quantized Dynamic Low-Rank Adaptation for Efficient Large Language Model Tuning
(
Poster
)
>
|
Hossein Rajabzadeh · Mojtaba Valipour · Marzieh Tahaei · HYOCK JU KWON · Ali Ghodsi · Boxing Chen · Mehdi Rezaghoizadeh 🔗 |
-
|
Model Fusion through Bayesian Optimization in Language Model Fine-Tuning
(
Poster
)
>
|
Chaeyun Jang · Jungtaek Kim · Hyungi Lee · Juho Lee 🔗 |
-
|
Group Preference Optimization: Few-Shot Alignment of Large Language Models
(
Poster
)
>
|
Siyan Zhao · John Dang · Aditya Grover 🔗 |
-
|
Fast-ELECTRA for Efficient Pre-training
(
Poster
)
>
|
Chengyu Dong · Liyuan Liu · Hao Cheng · Jingbo Shang · Jianfeng Gao · Xiaodong Liu 🔗 |
-
|
Parameter-Efficient Fine-tuning of InstructBLIP for Visual Reasoning Tasks
(
Poster
)
>
|
Sungkyung Kim · Adam Lee · Junyoung Park · Sounho Chung · Jusang Oh · Jay Yoon Lee 🔗 |
-
|
Local LoRA: Memory-Efficient Fine-Tuning of Large Language Models ( Poster ) > link | Oscar Key · Jean Kaddour · Pasquale Minervini 🔗 |
-
|
A Leap Forward in LLMs Post-Training W4A8 Quantization Using Floating-Point Formats
(
Poster
)
>
|
Xiaoxia Wu · Zhewei Yao · Yuxiong He 🔗 |
-
|
Exploring Post-training Quantization in LLMs from Comprehensive Study to Low Rank Compensation
(
Poster
)
>
|
Zhewei Yao · Xiaoxia Wu · Cheng Li · Stephen Youn · Yuxiong He 🔗 |
-
|
DeepSpeed Data Efficiency: Improving Deep Learning Model Quality and Training Efficiency via Efficient Data Sampling and Routing
(
Poster
)
>
|
Conglong Li · Zhewei Yao · Xiaoxia Wu · Minjia Zhang · Connor Holmes · Cheng Li · Yuxiong He 🔗 |
-
|
Arabic Mini-ClimateGPT : A Climate Change and Sustainability Tailored Arabic LLM
(
Poster
)
>
|
Sahal Shaji Mullappilly · Abdelrahman Shaker · Omkar Thawakar · Hisham Cholakkal · Rao Anwer · Salman Khan · Fahad Shahbaz 🔗 |
-
|
Multimodal Data and Resource Efficient Device-directed Speech Detection with Large Foundation Models
(
Poster
)
>
|
Dominik Wagner · Alexander Churchill · Siddharth Sigtia · Panayiotis Georgiou · Matt Mirsamadi · Aarshee Mishra · Erik Marchi 🔗 |
-
|
Representative Subset Selection for Efficient Fine-Tuning in Self-Supervised Speech Recognition
(
Poster
)
>
|
Abdul Hameed Azeemi · Ihsan Ayyub Qazi · Agha Ali Raza 🔗 |
-
|
ASR Data Selection from Multiple Sources: A Practical Approach on Performance Scaling
(
Poster
)
>
|
Hoang Anh Just · I-Fan Chen · Feiyang Kang · Yuanzhi Zhang · Anit Kumar Sahu · Ruoxi Jia 🔗 |
-
|
Fed-EE: Federating Heterogeneous ASR Models using Early-Exit Architectures
(
Poster
)
>
|
Mohamed Nabih Ali Mohamed Nawar · Alessio Brutti · Falavigna Daniele 🔗 |
-
|
Recursive Joint Cross-Attention for Audio-Visual Speaker Verification
(
Poster
)
>
|
Gnana Praveen Rajasekhar · JAHANGIR ALAM 🔗 |
-
|
Efficient infusion of self-supervised representations in Automatic Speech Recognition
(
Poster
)
>
|
Darshan Prabhu · Sai Ganesh Mirishkar · Pankaj Wasnik 🔗 |
-
|
An efficient clustering algorithm for self-supervised speaker recognition
(
Poster
)
>
|
Abderrahim Fathan · Xiaolin Zhu · JAHANGIR ALAM 🔗 |
-
|
HateXplain Space Model: Fusing Robustness with Explainability in Hate Speech Analysis
(
Poster
)
>
|
Md Fahim · Md Shihab Shahriar · Mohammad Ruhul Amin 🔗 |
-
|
Disclosing the Biases in Large Language Models via Reward Based Questioning
(
Poster
)
>
|
Ezgi Korkmaz 🔗 |
-
|
Evaluating task specific finetuning for protein language models ( Poster ) > link | Robert Schmirler 🔗 |