Media Summary: Speaker: Zhiyuan Liu, Associate Professor, Tsinghua University Over the past years, large-scale Speaker: Subho Mukherjee, Senior Researcher, Microsoft Presented by Danqi Chen (Princeton University) on April 29, 2022 Abstract: Large

Research Talk Knowledgeable Pre Trained - Detailed Analysis & Overview

Speaker: Zhiyuan Liu, Associate Professor, Tsinghua University Over the past years, large-scale Speaker: Subho Mukherjee, Senior Researcher, Microsoft Presented by Danqi Chen (Princeton University) on April 29, 2022 Abstract: Large TL;DR: Fine-tune less than 1% of GPT-2 language model parameters to decrease its prejudice. Debiasing Your persona-eval pipeline rates an Alexander Hamilton simulation at 80% personality fidelity. It is also rating a Hamilton who ... Performance scaling and power efficiency with traditional computing architectures becomes increasingly challenging as next ...

Photo Gallery

Research talk: Knowledgeable pre-trained language models
Research talk: Resource-efficient learning for large pretrained models
How do I give a great research talk? | Cell Mentor
How to Give a Great Research Talk
LTI Colloquium: Surprising Findings About Language Model Pre-training
On the Role of Pre-trained Language Models in Word Ordering: A Case Study with BART
“Usable knowledge” and linking research to practice
Who Can Be A Researcher? | Merle Massie | TEDxUniversityofSaskatchewan
Things about a PhD nobody told you about | Laura Valadez-Martinez | TEDxLoughboroughU
Debiasing Pre-Trained Language Models via Efficient Fine-Tuning
The Miranda Hypothesis: How Hamilton Poisoned Persona Evals - Jacob E. Thomas, Results Gen
BERT_SE: A Pre-Trained Language Representation Model for Software Engineering
View Detailed Profile
Research talk: Knowledgeable pre-trained language models

Research talk: Knowledgeable pre-trained language models

Speaker: Zhiyuan Liu, Associate Professor, Tsinghua University Over the past years, large-scale

Research talk: Resource-efficient learning for large pretrained models

Research talk: Resource-efficient learning for large pretrained models

Speaker: Subho Mukherjee, Senior Researcher, Microsoft

How do I give a great research talk? | Cell Mentor

How do I give a great research talk? | Cell Mentor

It's not enough to publish your

How to Give a Great Research Talk

How to Give a Great Research Talk

Writing papers and giving

LTI Colloquium: Surprising Findings About Language Model Pre-training

LTI Colloquium: Surprising Findings About Language Model Pre-training

Presented by Danqi Chen (Princeton University) on April 29, 2022 Abstract: Large

On the Role of Pre-trained Language Models in Word Ordering: A Case Study with BART

On the Role of Pre-trained Language Models in Word Ordering: A Case Study with BART

The short COLING 2022 oral

“Usable knowledge” and linking research to practice

“Usable knowledge” and linking research to practice

Ellen Condliffe Lagemann on “usable

Who Can Be A Researcher? | Merle Massie | TEDxUniversityofSaskatchewan

Who Can Be A Researcher? | Merle Massie | TEDxUniversityofSaskatchewan

Research

Things about a PhD nobody told you about | Laura Valadez-Martinez | TEDxLoughboroughU

Things about a PhD nobody told you about | Laura Valadez-Martinez | TEDxLoughboroughU

This

Debiasing Pre-Trained Language Models via Efficient Fine-Tuning

Debiasing Pre-Trained Language Models via Efficient Fine-Tuning

TL;DR: Fine-tune less than 1% of GPT-2 language model parameters to decrease its prejudice. Debiasing

The Miranda Hypothesis: How Hamilton Poisoned Persona Evals - Jacob E. Thomas, Results Gen

The Miranda Hypothesis: How Hamilton Poisoned Persona Evals - Jacob E. Thomas, Results Gen

Your persona-eval pipeline rates an Alexander Hamilton simulation at 80% personality fidelity. It is also rating a Hamilton who ...

BERT_SE: A Pre-Trained Language Representation Model for Software Engineering

BERT_SE: A Pre-Trained Language Representation Model for Software Engineering

BERT_SE: A

Research Talk (ISSCC - F1): Balancing Hardware Flexibility and Efficiency for Deep Learning

Research Talk (ISSCC - F1): Balancing Hardware Flexibility and Efficiency for Deep Learning

Performance scaling and power efficiency with traditional computing architectures becomes increasingly challenging as next ...