Get our free extension to see links to code for papers anywhere online!

Add to Chrome

Add to Firefox

Picture for Julian Salazar

Julian Salazar

Meta-Learning the Difference: Preparing Large Language Models for Efficient Adaptation

Jul 07, 2022
Zejiang Hou, Julian Salazar, George Polovets

Access Paper or Ask Questions

Unsupervised Bitext Mining and Translation via Self-trained Contextual Embeddings

Oct 15, 2020
Phillip Keung, Julian Salazar, Yichao Lu, Noah A. Smith

Figure 1 for Unsupervised Bitext Mining and Translation via Self-trained Contextual Embeddings

Figure 2 for Unsupervised Bitext Mining and Translation via Self-trained Contextual Embeddings

Figure 3 for Unsupervised Bitext Mining and Translation via Self-trained Contextual Embeddings

Figure 4 for Unsupervised Bitext Mining and Translation via Self-trained Contextual Embeddings

* To appear in the Transactions of the Association for Computational Linguistics

Access Paper or Ask Questions

Attentional Speech Recognition Models Misbehave on Out-of-domain Utterances

Feb 12, 2020
Phillip Keung, Wei Niu, Yichao Lu, Julian Salazar, Vikas Bhardwaj

Figure 1 for Attentional Speech Recognition Models Misbehave on Out-of-domain Utterances

Figure 2 for Attentional Speech Recognition Models Misbehave on Out-of-domain Utterances

Figure 3 for Attentional Speech Recognition Models Misbehave on Out-of-domain Utterances

Figure 4 for Attentional Speech Recognition Models Misbehave on Out-of-domain Utterances

* Artifacts like our filtered Audio BNC dataset can be found at https://github.com/aws-samples/seq2seq-asr-misbehaves

Access Paper or Ask Questions

Deep Contextualized Acoustic Representations For Semi-Supervised Speech Recognition

Dec 03, 2019
Shaoshi Ling, Yuzong Liu, Julian Salazar, Katrin Kirchhoff

Figure 1 for Deep Contextualized Acoustic Representations For Semi-Supervised Speech Recognition

Figure 2 for Deep Contextualized Acoustic Representations For Semi-Supervised Speech Recognition

Figure 3 for Deep Contextualized Acoustic Representations For Semi-Supervised Speech Recognition

Figure 4 for Deep Contextualized Acoustic Representations For Semi-Supervised Speech Recognition

* Submitted to ICASSP 2020

Access Paper or Ask Questions

Pseudolikelihood Reranking with Masked Language Models

Oct 31, 2019
Julian Salazar, Davis Liang, Toan Q. Nguyen, Katrin Kirchhoff

Figure 1 for Pseudolikelihood Reranking with Masked Language Models

Figure 2 for Pseudolikelihood Reranking with Masked Language Models

Figure 3 for Pseudolikelihood Reranking with Masked Language Models

Figure 4 for Pseudolikelihood Reranking with Masked Language Models

* Work in progress. An earlier version will be presented at DeepLo 2019. https://github.com/aws-samples/amazon-transcribe-research-pseudolikelihood

Access Paper or Ask Questions

Transformers without Tears: Improving the Normalization of Self-Attention

Oct 14, 2019
Toan Q. Nguyen, Julian Salazar

Figure 1 for Transformers without Tears: Improving the Normalization of Self-Attention

Figure 2 for Transformers without Tears: Improving the Normalization of Self-Attention

Figure 3 for Transformers without Tears: Improving the Normalization of Self-Attention

Figure 4 for Transformers without Tears: Improving the Normalization of Self-Attention

* Accepted to IWSLT'19

Access Paper or Ask Questions

Contextual Phonetic Pretraining for End-to-end Utterance-level Language and Speaker Recognition

Jun 30, 2019
Shaoshi Ling, Julian Salazar, Katrin Kirchhoff

Figure 1 for Contextual Phonetic Pretraining for End-to-end Utterance-level Language and Speaker Recognition

Figure 2 for Contextual Phonetic Pretraining for End-to-end Utterance-level Language and Speaker Recognition

Figure 3 for Contextual Phonetic Pretraining for End-to-end Utterance-level Language and Speaker Recognition

Figure 4 for Contextual Phonetic Pretraining for End-to-end Utterance-level Language and Speaker Recognition

* submitted to INTERSPEECH 2019

Access Paper or Ask Questions

Self-Attention Networks for Connectionist Temporal Classification in Speech Recognition

Jan 22, 2019
Julian Salazar, Katrin Kirchhoff, Zhiheng Huang

Figure 1 for Self-Attention Networks for Connectionist Temporal Classification in Speech Recognition

Figure 2 for Self-Attention Networks for Connectionist Temporal Classification in Speech Recognition

Figure 3 for Self-Attention Networks for Connectionist Temporal Classification in Speech Recognition

Figure 4 for Self-Attention Networks for Connectionist Temporal Classification in Speech Recognition

* Under review at ICASSP 2019

Access Paper or Ask Questions