Llama 2: Early Adopters' Utilization of Meta's New Open-Source Pretrained Model
The rapidly evolving field of artificial intelligence (AI) continues to witness the introduction of
innovative open-source pre-trained models, fostering advancements in various applications …
innovative open-source pre-trained models, fostering advancements in various applications …
Accurate and structured pruning for efficient automatic speech recognition
Automatic Speech Recognition (ASR) has seen remarkable advancements with deep neural
networks, such as Transformer and Conformer. However, these models typically have large …
networks, such as Transformer and Conformer. However, these models typically have large …
Continual Domain Adaptation through Pruning-aided Domain-specific Weight Modulation
In this paper, we propose to develop a method to address unsupervised domain adaptation
(UDA) in a practical setting of continual learning (CL). The goal is to update the model on …
(UDA) in a practical setting of continual learning (CL). The goal is to update the model on …
Continual Domain Adaptation through Pruning-aided Domain-specific Weight Modulation
In this paper, we propose to develop a method to address unsupervised domain adaptation
(UDA) in a practical setting of continual learning (CL). The goal is to update the model on …
(UDA) in a practical setting of continual learning (CL). The goal is to update the model on …
Outlier Reduction with Gated Attention for Improved Post-training Quantization in Large Sequence-to-sequence Speech Foundation Models
This paper explores the improvement of post-training quantization (PTQ) after knowledge
distillation in the Whisper speech foundation model family. We address the challenge of …
distillation in the Whisper speech foundation model family. We address the challenge of …
One-pass Multiple Conformer and Foundation Speech Systems Compression and Quantization Using An All-in-one Neural Model
We propose a novel one-pass multiple ASR systems joint compression and quantization
approach using an all-in-one neural model. A single compression cycle allows multiple …
approach using an all-in-one neural model. A single compression cycle allows multiple …
Automatic Data Augmentation for Domain Adapted Fine-Tuning of Self-Supervised Speech Representations
Self-Supervised Learning (SSL) has allowed leveraging large amounts of unlabeled speech
data to improve the performance of speech recognition models even with small annotated …
data to improve the performance of speech recognition models even with small annotated …
[PDF][PDF] Joint On-Demand Pruning and Online Distillation in Automatic Speech Recognition Language Model Optimization.
Automatic speech recognition (ASR) systems have emerged as indispensable tools across a
wide spectrum of applications, ranging from transcription services to voice-activated …
wide spectrum of applications, ranging from transcription services to voice-activated …
Stable Distillation: Regularizing Continued Pre-Training for Low-Resource Automatic Speech Recognition
Continued self-supervised (SSL) pre-training for adapting existing SSL models to the target
domain has shown to be extremely effective for low-resource Automatic Speech Recognition …
domain has shown to be extremely effective for low-resource Automatic Speech Recognition …
[PDF][PDF] Mitigating Overfitting in Structured Pruning of ASR Models with Gradient-Guided Parameter Regularization
Recent advancements in automatic speech recognition such as Wav2vec 2.0 and Whisper,
confront deployment challenges due to their substantial model parameters. Model …
confront deployment challenges due to their substantial model parameters. Model …