Posts by Collection

In value-based reinforcement learning methods, function approximation errors are known to lead to overestimated value estimates and sub-optimal policies.

Offline Reinforcement Learning: Tutorial, Review, and Perspectives on Open Problems

Published: November 01, 2022

This paper provide the reader with the conceptual tools needed to get started on research on offline reinforcement learning algorithms:
reinforcement learning algorithms that utilize previously collected data, without additional online data collection.

Efficient Off-Policy Meta-Reinforcement Learning via Probabilistic Context Variables

Published: December 01, 2022

Previous methods rely heavily on on-policy experience, limiting their sample efficiency.

They also lack mechanisms to reason about task uncertainty when adapting to new tasks, limiting their effectiveness in sparse reward problems.

This paper developing an off-policy meta-RL algorithm that disentangles task inference and control.

Achieving excellent sample efficiency during meta-training, enables fast adaptation by accumulating experience online
Performing structured exploration by reasoning about uncertainty over tasks

BEIT:BERT Pre-training of Image Transformers

Published: December 05, 2022

Motivated by BERT, they turn to the denoising auto-encoding idea to pretrain vision transformers, which has not been well studied by the vision community.

Second Thoughts are Best: Learning to Re-Align With Human Values from Text Edits

Published: December 06, 2022

This paper present SECOND THOUGHTS, a new learning paradigm that enables language models (LMs) to re-align with human values.

Training language models to follow instructions with human feedback

Published: January 10, 2023

This paper show a method to align language models with user intent on a wide range of tasks by fine-tuning with human feedback.

HyperPrompt:Prompt-based Task-Conditioning of Transformers

Published: March 07, 2023

This paper proposes HyperPrompt, a novel architecture for prompt-based task-conditioning of self-attention in Transformers.

Is Reinforcement Learning (Not) for Natural Language Processing

Published: April 18, 2023

In this paper, they first introduce an open-source modular library RL4LMs, for optimizing language generators with RL.

A Mixture-of-Expert Approach to RL-based Dialogue Management

Published: April 27, 2023

In this paper, they introduce RL-based DM using a novel mixture of expert language model (MoE-LM) that consists of

Off-Policy Deep Reinforcement Learning without Exploration

Published: May 08, 2023

This paper proposes a new algorithm for off-policy reinforcement learning that combines state-of-the-art deep Q-learning algorithms with a state-conditioned generative model for producing only previously seen actions.

Evaluating Parameter Efficient Learning for Generation

Published: May 09, 2023

In this paper, they present a comprehensive evaluation of parameter efficient learning methods (PERMs) for generation tasks in natural language processing.

Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks

Published: May 30, 2023

This study proposes a fine-tuning recipe for retrieval-augmented generation (RAG) models that combine pre-trained parametric and non-parametric memory for language generation.

Active Retrieval Augmented Generation

Published: June 27, 2023

Most existing retrieval-augmented LMs employ a retrieve-and-generate setup that only retrieves information once based on the input.

LLaMA-Adapter: Efficient Fine-tuning of Language Models with Zero-init Attention

Published: July 25, 2023

This paper proposes LLaMA-Adapter, a lightweight adaption method to efficiently finetune LLaMA into an instruction-following model.

AdaMix: Mixture-of-Adaptations for Parameter-efficient Model Tuning

Published: July 27, 2023

This paper proposes a parameter-efficient fine-tuning method called $\texttt{AdaMix}$, a general parameter-efficient fine-tuning (PEFT) techniques that tunes a mixture of adaptation modules.

A Neural Corpus Indexer for Document Retrieval

Published: August 29, 2023

Current SOTA for document retrieval solutions mainly follow an index-retrieve, where the index is hard to be directly optimized for the final retrieval target.

A Statistical Perspective on Retrieval-Based Models

Published: October 12, 2023

This paper uses a formal treatment of retrieval-based models to characterize their performance via a novel statistical perspective.

GPTQ: Accurate Post-Training Quantization for Generative Pre-trained Transformers

Published: October 24, 2023

Nowadays, Generative Pre-trained Transformer models has not only breakthrough performance across complex language modelling tasks, but also by their extremely high computational and storage costs.

On the Effectiveness of Offline RL for Dialogue Response Generation

Published: December 12, 2023

For language models, many methods using teacher forcing (TF) to train.

Empathetic Dialogue Generation via Sensitive Emotion Recognition and Sensible Knowledge Selection

Published: January 30, 2024

In this research, the authors address challenges in empathetic dialogue generation.

Quark: Controllable Text Generation with Reinforced [Un]learning

Published: April 09, 2024

Large language models may generate content that is misaligned with the user’s expectations. For example, generating toxic words, repeated content, and undesired responses for users.

Effective Structured Prompting by Meta-Learning and Representative Verbalizer

Published: April 11, 2024

This paper utilizes a prompt pool to leverage task-specific knowledge and generate instance-specific prompts using attention mechanisms.

E-CORE: Emotion Correlation Enhanced Empathetic Dialogue Generation

Published: May 02, 2024

Empathetic dialogue generation task aims at generating empathetic responses, based on perceived emotions instead of definite annotated emotions.

Po-Chuan Chen

Posts by Collection

experiences

portfolio

publications

talks