paper-with-me

홈 › Papers

Alexa Teacher Model: Pretraining and Distilling Multi-Billion-Parameter Encoders for Natural Language Understanding Systems

2022-06-15 · Jack FitzGerald, Shankar Ananthakrishnan, Konstantine Arkoudas, Davide Bernardi, Abhishek Bhagia, Claudio Delli Bovi, Jin Cao, Rakesh Chada, Amit Chauhan, Luoxin Chen, Anurag Dwarakanath, Satyam Dwivedi, Turan Gojayev, Karthik Gopalakrishnan, Thomas Gueudre, Dilek Hakkani-Tur, Wael Hamza, Jonathan Hueser, Kevin Martin Jose, Haidar Khan, Beiye Liu, Jianhua Lu, Alessandro Manzotti, Pradeep Natarajan, Karolina Owczarzak, Gokmen Oz, Enrico Palumbo, Charith Peris, Chandana Satya Prakash, Stephen Rawls, Andy Rosenbaum, Anjali Shenoy, Saleh Soltan, Mukund Harakere Sridhar, Liz Tan, Fabian Triefenbach, Pan Wei, Haiyang Yu, Shuai Zheng, Gokhan Tur, Prem Natarajan

We present results from a large-scale experiment on pretraining encoders with non-embedding parameter counts ranging from 700M to 9.3B, their subsequent distillation into smaller models ranging from 17M-170M parameters, and their application to the Natural Language Understanding (NLU) component of a virtual assistant system. Though we train using 70% spoken-form data, our teacher models perform comparably to XLM-R and mT5 when evaluated on the written-form Cross-lingual Natural Language Inference (XNLI) corpus. We perform a second stage of pretraining on our teacher models using in-domain data from our system, improving error rates by 3.86% relative for intent classification and 7.01% relative for slot filling. We find that even a 170M-parameter model distilled from our Stage 2 teacher model has 2.88% better intent classification and 7.69% better slot filling error rates when compared to the 2.3B-parameter teacher trained only on public data (Stage 1), emphasizing the importance of in-domain data for pretraining. When evaluated offline using labeled NLU data, our 17M-parameter Stage 2 distilled model outperforms both XLM-R Base (85M params) and DistillBERT (42M params) by 4.23% to 6.14%, respectively. Finally, we present results from a full virtual assistant experimentation platform, where we find that models trained using our pretraining and distillation pipeline outperform models distilled from 85M-parameter teachers by 3.74%-4.91% on an automatic measurement of full-system user dissatisfaction.

📄 PDF Abstract BibTeX arXiv:2206.07808

Code (0)

등록된 구현이 없습니다.

Tasks

Cross-Lingual Natural Language Inferenceintent-classificationIntent ClassificationNatural Language InferenceNatural Language Understandingslot-fillingSlot FillingXLM-R

Methods 이 논문이 사용한 방법론

Multi-Head Attention 설명 없음
Attention 설명 없음
XLM-R XLM-R
Linear Layer A Linear Layer is a projection $\mathbf{XW + b}$.
BPE Byte Pair Encoding, or BPE, is a subword segmentation algorithm that encodes rare and unknown words as sequences of subword units. The intuition is that various word…
Layer Normalization Unlike batch normalization, Layer Normalization directly estimates the normalization statistics from the summed inputs…
Inverse Square Root Schedule Inverse Square Root is a learning rate schedule 1 / $\sqrt{\max\left(n, k\right)}$ where $n$ is the current training iteration and $k$ is the number of warm-up steps. This…
Gated Linear Unit A Gated Linear Unit, or GLU computes: $$ \mathrm{GLU}(a, b) = a \otimes \sigma(b) $$ It is used in natural language processing architectures, for example the Gated CNN,…

Similar Papers 제목 키워드 기반

AlexaTM 20B: Few-Shot Learning Using a Large-Scale Multilingual Seq2Seq Model

2022-08-02 · Saleh Soltan, Shankar Ananthakrishnan, Jack FitzGerald, Rahul Gupta 외

In this work, we demonstrate that multilingual large-scale sequence-to-sequence (seq2seq) models, pre-trained on a mixture of denoising and Causal Language Modeling (CLM) tasks, are more efficient few-shot learners than …

Causal Language ModelingCommon Sense ReasoningCoreference ResolutionDecoder+8

DistillPath: An Efficient 22M Distilled Pathology Encoder Approaching Large Foundation Model Performance

2026-08-18 · Ramon Kaspar, Andrey Ignatov, Valentina Boeva arxiv

Many high-performing pathology tile encoders are now foundation models with hundreds of millions to over a billion parameters. Encoding and storing the thousands of tiles in each whole-slide image with such models is cos…

DreamTeacher: Pretraining Image Backbones with Deep Generative Models

2023-07-14 · ICCV 2023 1 · Daiqing Li, Huan Ling, Amlan Kar, David Acuna 외

In this work, we introduce a self-supervised feature representation learning framework DreamTeacher that utilizes generative networks for pre-training downstream image backbones. We propose to distill knowledge from a tr…

Knowledge DistillationRepresentation Learning

Chorus: Multi-Teacher Pretraining for Holistic 3D Gaussian Scene Encoding

2025-12-19 · Yue Li, Qi Ma, Runyi Yang, Mengjiao Ma 외 arxiv

While 3DGS has emerged as a high-fidelity scene representation, encoding rich, general-purpose features directly from its primitives remains under-explored. We address this gap by introducing Chorus, a multi-teacher pret…

Instance SegmentationPoint Clouds

Sparse Distillation: Speeding Up Text Classification by Using Bigger Student Models

2021-10-16 · NAACL 2022 7 · Qinyuan Ye, Madian Khabsa, Mike Lewis, Sinong Wang 외

Distilling state-of-the-art transformer models into lightweight student models is an effective way to reduce computation cost at inference time. The student models are typically compact transformers with fewer parameters…

ClassificationDomain GeneralizationPrivacy PreservingSentence+3