RL Group Meeting, Hsinchu city, National Yang Ming Chiao Tung University, Hsinchu city, Taiwan
To reduce the limitation of the large language model, the existing work enhances pre-trained LLMs using grounded knowledge.
Course paper presentation, Hsinchu city, National Yang Ming Chiao Tung University, Hsinchu city, Taiwan
To improve Reader′ s and Writer′ s for multi-dimensional emotion regression tasks with EMOBANK, this paper proposes an Adversarial Attention Network.
Course paper presentation, Hsinchu city, National Yang Ming Chiao Tung University, Hsinchu city, Taiwan
Empathetic dialogue generation task aims at generating empathetic responses, based on perceived emotions instead of definite annotated emotions.
Lab Seminar, Hsinchu city, National Yang Ming Chiao Tung University, Hsinchu city, Taiwan
This paper utilizes a prompt pool to leverage task-specific knowledge and generate instance-specific prompts using attention mechanisms.
RL Group Meeting, Hsinchu city, National Yang Ming Chiao Tung University, Hsinchu city, Taiwan
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.
RL Group Meeting, Hsinchu city, National Yang Ming Chiao Tung University, Hsinchu city, Taiwan
In this research, the authors address challenges in empathetic dialogue generation.
RL Group Meeting, Hsinchu city, National Yang Ming Chiao Tung University, Hsinchu city, Taiwan
For language models, many methods using teacher forcing (TF) to train.
RL Group Meeting, Hsinchu city, National Yang Ming Chiao Tung University, Hsinchu city, Taiwan
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.
Lab Seminar, Hsinchu city, National Yang Ming Chiao Tung University, Hsinchu city, Taiwan
This paper uses a formal treatment of retrieval-based models to characterize their performance via a novel statistical perspective.
RL Group Meeting, Hsinchu city, National Yang Ming Chiao Tung University, Hsinchu city, Taiwan
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.
Lab Seminar, Hsinchu city, National Yang Ming Chiao Tung University, Hsinchu city, Taiwan
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.
RL Group Meeting, Hsinchu city, National Yang Ming Chiao Tung University, Hsinchu city, Taiwan
This paper proposes LLaMA-Adapter, a lightweight adaption method to efficiently finetune LLaMA into an instruction-following model.
RL Group Meeting, Hsinchu city, National Yang Ming Chiao Tung University, Hsinchu city, Taiwan
Most existing retrieval-augmented LMs employ a retrieve-and-generate setup that only retrieves information once based on the input.
RL Group Meeting, Hsinchu city, National Yang Ming Chiao Tung University, Hsinchu city, Taiwan
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.
NLG Group Meeting, Hsinchu city, National Yang Ming Chiao Tung University, Hsinchu city, Taiwan
In this paper, they present a comprehensive evaluation of parameter efficient learning methods (PERMs) for generation tasks in natural language processing.
Course paper presentation, Hsinchu city, National Yang Ming Chiao Tung University, Hsinchu city, Taiwan
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.
Lab Seminar, Hsinchu city, National Yang Ming Chiao Tung University, Hsinchu city, Taiwan
In this paper, they introduce RL-based DM using a novel mixture of expert language model (MoE-LM) that consists of
RL Group Meeting, Hsinchu city, National Yang Ming Chiao Tung University, Hsinchu city, Taiwan
In this paper, they first introduce an open-source modular library RL4LMs, for optimizing language generators with RL.
RL Group Meeting, Hsinchu city, National Yang Ming Chiao Tung University, Hsinchu city, Taiwan
This paper proposes HyperPrompt, a novel architecture for prompt-based task-conditioning of self-attention in Transformers.
RL Group Meeting, Hsinchu city, National Yang Ming Chiao Tung University, Hsinchu city, Taiwan
This paper show a method to align language models with user intent on a wide range of tasks by fine-tuning with human feedback.
RL Group Meeting, Hsinchu city, National Yang Ming Chiao Tung University, Hsinchu city, Taiwan
This paper present SECOND THOUGHTS, a new learning paradigm that enables language models (LMs) to re-align with human values.
Course paper presentation, Hsinchu city, National Yang Ming Chiao Tung University, Hsinchu city, Taiwan
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.
Lab Seminar, Hsinchu city, National Yang Ming Chiao Tung University, Hsinchu city, Taiwan
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.
RL Group Meeting, Hsinchu city, National Yang Ming Chiao Tung University, Hsinchu city, Taiwan
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.
RL Group Meeting, Hsinchu City, National Yang Ming Chiao Tung University, Hsinchu city, Taiwan
In value-based reinforcement learning methods, function approximation errors are known to lead to overestimated value estimates and sub-optimal policies.
Course paper presentation, Taipei city, National Taipei Tech., Taipei city, Taiwan
Paper presentation and implemenation for the course called Machine Learning in National Taipei Tech.
Course paper presentation, Taipei city, National Taipei Tech., Taipei city, Taiwan
Paper presentation for the course called File and Storage system in National Taipei Tech.
Group Meeting, Taipei city, National Taipei Tech., Taipei city, Taiwan
All of my slides which were presented while I was in Taipei Tech.
Course paper presentation, Taoyuan city, Chang Gung Uiversity, Taoyuan city, Taiwan
Paper presentation for the course called Computer Security in Chang Gung University