Picture for Boris Ginsburg

Boris Ginsburg

Training and Inference Efficiency of Encoder-Decoder Speech Models

Add code
Mar 07, 2025
Viaarxiv icon

Scoring Verifiers: Evaluating Synthetic Verification in Code and Reasoning

Add code
Feb 19, 2025
Viaarxiv icon

Star Attention: Efficient LLM Inference over Long Sequences

Add code
Nov 26, 2024
Viaarxiv icon

NeKo: Toward Post Recognition Generative Correction Large Language Models with Task-Oriented Experts

Add code
Nov 08, 2024
Viaarxiv icon

Anticipating Future with Large Language Model for Simultaneous Machine Translation

Add code
Oct 29, 2024
Figure 1 for Anticipating Future with Large Language Model for Simultaneous Machine Translation
Figure 2 for Anticipating Future with Large Language Model for Simultaneous Machine Translation
Figure 3 for Anticipating Future with Large Language Model for Simultaneous Machine Translation
Figure 4 for Anticipating Future with Large Language Model for Simultaneous Machine Translation
Viaarxiv icon

VoiceTextBlender: Augmenting Large Language Models with Speech Capabilities via Single-Stage Joint Speech-Text Supervised Fine-Tuning

Add code
Oct 23, 2024
Viaarxiv icon

Three-in-One: Fast and Accurate Transducer for Hybrid-Autoregressive ASR

Add code
Oct 03, 2024
Viaarxiv icon

nGPT: Normalized Transformer with Representation Learning on the Hypersphere

Add code
Oct 01, 2024
Viaarxiv icon

Developing Instruction-Following Speech Language Model Without Speech Instruction-Tuning Data

Add code
Sep 30, 2024
Figure 1 for Developing Instruction-Following Speech Language Model Without Speech Instruction-Tuning Data
Figure 2 for Developing Instruction-Following Speech Language Model Without Speech Instruction-Tuning Data
Figure 3 for Developing Instruction-Following Speech Language Model Without Speech Instruction-Tuning Data
Figure 4 for Developing Instruction-Following Speech Language Model Without Speech Instruction-Tuning Data
Viaarxiv icon

Large Language Model Based Generative Error Correction: A Challenge and Baselines for Speech Recognition, Speaker Tagging, and Emotion Recognition

Add code
Sep 17, 2024
Figure 1 for Large Language Model Based Generative Error Correction: A Challenge and Baselines for Speech Recognition, Speaker Tagging, and Emotion Recognition
Figure 2 for Large Language Model Based Generative Error Correction: A Challenge and Baselines for Speech Recognition, Speaker Tagging, and Emotion Recognition
Figure 3 for Large Language Model Based Generative Error Correction: A Challenge and Baselines for Speech Recognition, Speaker Tagging, and Emotion Recognition
Figure 4 for Large Language Model Based Generative Error Correction: A Challenge and Baselines for Speech Recognition, Speaker Tagging, and Emotion Recognition
Viaarxiv icon