paper-with-me

Papers

KS-Lottery: Finding Certified Lottery Tickets for Multilingual Language Models

2024-02-05 · Fei Yuan, Chang Ma, Shuai Yuan, Qiushi Sun, Lei LI

The lottery ticket hypothesis posits the existence of ``winning tickets'' within a randomly initialized neural network. Do winning tickets exist for LLMs in fine-tuning scenarios? How can we find such winning tickets? In this paper, we propose KS-Lottery, a method to identify a small subset of LLM parameters highly effective in multilingual fine-tuning. Our key idea is to use Kolmogorov-Smirnov Test to analyze the distribution shift of parameters before and after fine-tuning. We further theoretically prove that KS-Lottery can find the certified winning tickets in the embedding layer, fine-tuning on the found parameters is guaranteed to perform as well as full fine-tuning. Comparing KS-Lottery with other parameter-efficient tuning algorithms on translation tasks, the experimental results show that KS-Lottery finds a much smaller set of parameters for fine-tuning while achieving the comparable performance as full fine-tuning LLM. Surprisingly, we find that fine-tuning 18 tokens' embedding of LLaMA suffices to reach the fine-tuning translation performance~\footnote{https://github.com/CONE-MT/KS-Lottery.}.

📄 PDF Abstract BibTeX arXiv:2402.02801

Code (0)

등록된 구현이 없습니다.

Tasks

Translation

Methods 이 논문이 사용한 방법론

SET Dynamic Sparse Training method where weight mask is updated randomly periodically

Similar Papers 제목 키워드 기반

Rare Gems: Finding Lottery Tickets at Initialization

2022-02-24 · Kartik Sreenivasan, Jy-yong Sohn, Liu Yang, Matthew Grinde 외

Large neural networks can be pruned to a small fraction of their original size, with little loss in accuracy, by following a time-consuming "train, prune, re-train" approach. Frankle & Carbin conjecture that we can avoid…

Bespoke vs. Prêt-à-Porter Lottery Tickets: Exploiting Mask Similarity for Trainable Sub-Network Finding

2020-07-06 · Michela Paganini, Jessica Zosa Forde

The observation of sparse trainable sub-networks within over-parametrized networks - also known as Lottery Tickets (LTs) - has prompted inquiries around their trainability, scaling, uniqueness, and generalization propert…

Denoisingimage-classificationImage Classification

When Layers Play the Lottery, all Tickets Win at Initialization

2023-01-25 · Artur Jordao, George Correa de Araujo, Helena de Almeida Maia, Helio Pedrini

Pruning is a standard technique for reducing the computational cost of deep networks. Many advances in pruning leverage concepts from the Lottery Ticket Hypothesis (LTH). LTH reveals that inside a trained dense network e…

All

Winning Lottery Tickets in Deep Generative Models

2020-10-05 · Neha Mukund Kalibhat, Yogesh Balaji, Soheil Feizi

The lottery ticket hypothesis suggests that sparse, sub-networks of a given neural network, if initialized properly, can be trained to reach comparable or even better performance to that of the original network. Prior wo…

LOFT: Finding Lottery Tickets through Filter-wise Training

2022-10-28 · Qihan Wang, Chen Dun, Fangshuo Liao, Chris Jermaine 외

Recent work on the Lottery Ticket Hypothesis (LTH) shows that there exist ``\textit{winning tickets}'' in large neural networks. These tickets represent ``sparse'' versions of the full model that can be trained independe…