Activities ▸ Lab Meeting

Lab members work on a variety of research projects. The Lab Meeting provides a weekly opportunity to share their work with other lab members and get their feedback.

Logistics

1. Lab-wide announcement (15 min.)
• Everyone’s submission plan
• Brief recap of conference deadlines
• Discussion about lab management
2. Paper skimming (weekly)
• 3 people presents their favorite papers (15 min. x 3)
• Talk: 5 min. / QA : 10 min.
3. Poster presentation (monthly)
• 4-5 people present their work by a poster
4. Spotlight presentation(s) (on demand)
• 1-2 people present their relatively complete work in oral presentation

Meetings Log

‣

2026

Date	Type	Presenters	Topic
02/26/2026	Paper Skimming	SHI,YutingHinata Tezuka	1. [Buzeta et al.] Seeing to Generalize: How Visual Data Corrects Binding Shortcuts /2. [Lindsey] Emergent Introspective Awareness in Large Language Models
02/19/2026	Spotlight Oral	N Naohiro Kaide S Shohei Kitajima	NLP2026 Poster Presentation/Feedback (JAPANESE SESSION)
02/12/2026	Paper Skimming	Mizuki koyamagawa	1. [Zhong] DUET: Distilled LLM Unlearning from an Efficiently Contextualized Teacher /2. [Shi et al.] Vision Function Layer in Multimodal LLMs
02/05/2026	Spotlight Oral	P Peter_Obule A Akira Ishii	Masters Defense Practice Presentation
01/29/2026	Poster	K Kim Pothong	Cognitive Alignment in Question Generation
01/22/2026	Paper Skimming	Koga Kobayashi Hinata Tezuka	1. [Zhang et al.] From Reasoning to Answer: Empirical, Attention-Based and Mechanistic Insights into Distilled DeepSeek R1 Models / 2. [Han et al.] Token-Budget-Aware LLM Reasoning / 3. [He et al.] Towards Global-level Mechanistic Interpretability: A Perspective of Modular Circuits
01/15/2026	Paper Skimming	T Tien Dang HuuJIN, Tao	1. [Wei et al.] AdaDecode: Accelerating LLM Decoding with Adaptive Layer Parallelism / 2. A note on model immunization

‣

2025

Date	Type	Presenters	Topic
12/25/2025	Poster	P Peter_Obule S Shohei Kitajima	1. Hallucination Control Using Representation Engineering with Uncertainty Aware Steering / 2. Comparative Analysis of Methods for Inserting Unknown Knowledge into LLMs
12/18/2025	Paper Skimming	kenkenSHI,Yuting A Akira Ishii	1. [ Podolak et al.] Read Your Own Mind: Reasoning Helps Surface Self-Confidence Signals in LLMs /2. [Assouel et al.] Visual Symbolic Mechanisms: Emergent Symbol Processing in Vision Language Models /3. [Tan et al.] Cascading Large Langugage Models for Salient Event Graph Generation
12/11/2025	Paper Skimming	houjing T TRAN, Thu Thi Anh	1. [Basile et al.]Head Pursuit: Probing Attention Specialization in Multimodal Transformers/ 2. [Wang et al.] Rehearsal-free Continual Language Learning via Efficient Parameter Isolation - ACL Anthology
12/04/2025	Spotlight Oral	Naoya Inoue	RRI (Responsible Research and Innovation) Seminar
11/27/2025	Poster	N NGUYEN Phuong MinhMariko Kato	1. Attention-aware Intervention/ 2. [Kato et al.] Affinity and Diversity: A Unified Metric for Demonstration Selection via Internal Representations
11/20/2025	Paper Skimming	Raka S Shohei KitajimaMizuki koyamagawa	1. [Zhang et al.] TAdaRAG: Task Adaptive Retrieval-Augmented Generation via On-the-Fly Knowledge Graph Construction / 2. [Cheng et al.] xRAG: Extreme Context Compression for Retrieval-augmented Generation with One Token / 3. [Zhang et al.] Catastrophic Failure of LLM Unlearning via Quantization | OpenReviewBas
11/13/2025	Paper Skimming	Hakaze Cho T Tien Dang Huu P Peter_Obule	1. [Anonymous] Language Models are Injective and Hence Invertible / 2. [Chen et al.] Unlearning Isn't Invisible: Detecting Unlearning Traces in LLMs from Model Outputs / 3. [Zhao et al.] Steering Knowledge Selection Behaviours in LLMs via SAE-Based Representation Engineering
11/06/2025	Paper Skimming	houjingKoga Kobayashi K Kim Pothong	1. [Jian et al.]  Vision Transformers Don't Need Trained Registers 2. [Yue et al.] Does Reinforcement Learning Really Incentivize Reasoning Capacity in LLMs Beyond the Base Model? 3. [Baes et al.] LSC-Eval: A General Framework to Evaluate Methods for Assessing Dimensions of Lexical Semantic Change Using LLM-Generated Synthetic Data
10/30/2025	Poster	T TRAN, Thu Thi Anh	On Subspace Orthogonality in Continual Learning of LLMs with LoRA
10/23/2025	Paper Skimming	SHI,YutingRaka	1. [Li et al.] Lost in Embeddings: Information Loss in Vision-Language Models 2. [Nguyen et al.] Enhancing Retrieval Augmented Generation with Hierarchical Text Segmentation Chunking
10/16/2025	Paper Skimming	S Shohei KitajimaMizuki koyamagawaHinata Tezuka	1. [Yamada et al.] Dynamic Injection of Entity Knowledge into Dense Retrievers 2. [Marks et al.] Sparse Feature Circuits: Discovering and Editing Interpretable Casual Graphs in Language Models 3. [Barbero et al.] Why do LLMs attend to the first token?
10/10/2025	Kick-off	Naoya Inoue	Kick-off meeting
09/03/2025	Spotlight Oral	A Akira IshiiMariko Kato	Mid-Term Presentation Practice
09/02/2025	Spotlight Oral	Hinata Tezuka P Peter_Obule	Mid-Term Presentation Practice
08/08/2025	Spotlight Oral	T Tien Dang Huu	MS Defense Practice
07/25/2025	Paper Skimming	A Akira Ishiihoujing	1. [Kang et al.] See What You Are Told: Visual Attention Sink in Large Multimodal Models | OpenReview 2. [Mu et al.] A Causal Approach for Counterfactual Reasoning in Narratives
07/18/2025	Paper Skimming	K Kim Pothong N NGUYEN Phuong MinhSHI,Yuting	1. [Liu et al.] DoRA: Weight-Decomposed Low Rank Adaptation 2. [Nikankin et al.] Same Task, Different Circuits: Disentangling Modality-Specific Mechanisms in VLMs 3. [Xu et al.] INSTRUCTSCORE: Towards Explainable Text Generation Evaluation with Automatic Feedback
07/11/2025	Poster	houjingHinata Tezuka	1. Counting First or Detection First? An Investigation on Counting Objects in LVLMs 2. The Transfer Neurons Hypothesis: An Underlying Mechanism for Language Subspace Transitions in Multilingual LLMs
07/04/2025	Paper Skimming	kenken P Peter_ObuleHinata Tezuka	1. [Chuang et al.] DoLa: Decoding by Contrasting Layers Improves Factuality in Large Language Models | OpenReview 2. [Kallini et al.] Mission: Impossible Language Models - ACL Anthology 3. [Yu et al.] Neuron-Level Knowledge Attribution in Large Language Models - ACL Anthology
06/27/2025	Paper Skimming	Mariko KatoHakaze ChoNaoya Inoue	1. [Hu et al.] Understanding In-context Learning of Addition via Activation Subspaces 2. [Modell et al.] The Origins of Representation Manifolds in Large Language Models 3. [Ortu et al.] Competition of Mechanisms: Tracing How Language Models Handle Facts and Counterfactuals
06/20/2025	Paper Skimming	A Akira Ishii T TRAN, Thu Thi AnhKoga Kobayashi	1. [Lyu et al.] FACTTRACK: Time-Aware World State Tracking in Story Outlines 2. [Huang et al.] Mitigating Catastrophic Forgetting in Large Language Models with Self-Synthesized Rehearsal 3. [Busbridge et al.] Distillation Scaling Laws
06/13/2025	Poster	Lukas Hofbauer T Tien Dang Huu	1.On Effects of Steering Latent Representation for Large Language Model Unlearning; 2.Step-wise Decomposition Improves Calibration for Answering Multi-Hop Questions
05/30/2025	Paper Skimming	P Peter_ObuleSHI,Yuting	1. [Feng et al.] LEGEND: Leveraging Representation Engineering to Annotate Safety Margin for Preference Datasets 2. [Tjandrasuwita et al.] Understanding the Emergence of Multimodal Representation Alignment
05/23/2025	Paper Skimming	Mariko KatoHinata Tezuka	1. [Ethelo et al.] Inferring Functionality of Attention Heads from their Parameters 2. [Kaplan et al.] From Tokens to Words: On the Inner Lexicon of LLMs
05/16/2025	Paper Skimming	kenkenhoujingHakaze Cho	1. [Jiang et al.] Interpreting and Editing Vision-Language Representations to Mitigate Hallucinations 2. [Stolfo et al.] Confidence Regulation in LLMs 3. [Jin et al.] Massive Values in Self-Attention Modules are the Key to Contextual Knowledge Understanding
05/09/2025	Poster	Hakaze Cho K Kim Pothong SHI,Yuting	1. Mechanism Interpretability on ICL 2. FOCUS Benchmark 3. Understanding Compositional Visual Reasoning
04/25/2025	Paper Skimming	N NGUYEN Phuong Minh K Kim Pothong Lukas Hofbauer	1. [Ye et al.] SATLM: Satisfiability-Aided Language Models Using Declarative Prompting 2. [Tanneru et al.] Quantifying Uncertainty in Natural Language Explanations of Large Language Models 3. [Celikylimaz et al.] Evaluation of Text Generation: A Survey
04/18/2025	Paper Skimming	T TRAN, Thu Thi AnhHinata Tezuka P Peter_Obule	1. [Bi et al.] Iterative Refinement of Project-Level Code Generation 2. [Schug et al.] Attention as a HyperNetwork 3. [Li et al.] Inference Time Intervention: Eliciting Truthful Answer from a Language Model
04/11/2025	Paper Skimming	Naoya Inoue T Tien Dang Huu A Akira Ishii	1. [Stolfo et al.] Confidence Regulation Neurons in LMs 2. [Tien Dang Huu] Unlearning in LLMs: An Introduction 3. [Hatzel et al.] Story Embeddings—Narrative-Focused Representations of Fictional Stories
04/04/2025	Kick-off	Naoya Inoue	Kick-off Meeting
03/03/2025		K Kim Pothong Hinata TezukaMariko Kato	NLP2025 Practice Presentation
02/20/2025		houjing	NLP 2025 Practice Presentation
02/13/2025		Hakaze Cho	Revisiting In-context Learning Inference Circuit in Large Lan- guage Models (ICLR 2025)
02/06/2025		houjing	Final defense practice
01/30/2025		S Shotaro Kitamura	Quick walkthrough: running Llama on Kagayaki
01/23/2025		kenken	Brainstorming on Logo Design
01/16/2025		Naoya Inoue	General Discussions
01/01/2025

‣

2024

Date	Type	Presenters	Topic
12/11/2024		SHI,Yuting	Probing Implicit Multi-Step Reasoning in Visual-Language Models
11/27/2024		Y Yan_zhenzhu	Research Proposal: Traffic Sign Benchmarking
11/20/2024		Hakaze Cho	Light Hearted Mechanistic Interpretability
11/13/2024		JIN, Tao	UI Operation Agent
10/30/2024		K Kim Pothong	Leveraging Socratic Feedback: Using Counter-Arguments to Argument Revision
10/23/2024		T TRAN, Thu Thi Anh	(Master’s Thesis Recap) Towards x86 Instruction Set Emulation in Java via Project-based Text-to-Code Generation using Reinforcement Learning
10/16/2024		S SAKAI Yoshihiro	Speech Title: Improve your Presentation
10/08/2024		Hakaze Cho	Kick-off Meeting
09/24/2024		Mariko Kato	Image Encodings are Informative Tokens for Frozen Text Transformers
09/03/2024		houjing	Practice for Mid-term Presentation
08/27/2024		houjingMariko KatoHakaze Cho	YANS Practice Talk
08/06/2024		Irfan Robbanikenken	Practice Talk for MS Defense
07/16/2024		Irfan Robbani	Flee the Flaw: Annotating the Underlying Logic of Fallacious Arguments Through Templates and Slot-filling
07/09/2024		Naoya Inoue	Special Session: 2nd Kick-off!
07/02/2024		houjing	Probing Visual Prompts in Multimodal Large Language Models (MLLMs)
06/25/2024		K Kim Pothong Hakaze Cho	NL260 Practice Talk x 2
06/18/2024		kenken	Meta Cognitive Knowledge in LLMs
05/28/2024		S Shotaro Kitamura	Towards Making Japanese Implicit Multi-hop Question Benchmarks
05/21/2024		Y Yan_zhenzhu	A Survey of Vision Language Model in Autonomous Driving
05/14/2024		SHI,Yuting	How to Make Explanation Behind Visual Reasoning Response of LVLMs
05/07/2024		K Kim Pothong	FA1A2 - A Dataset of Fallacious/Non-Fallacious Arguments for LLM Probing
04/22/2024		JIN, Tao	A Survey of the Latest GUI Agent for Cross Application Task and Progress Report
04/16/2024		Hakaze Cho	Word Embedding “Geometry”
04/10/2024		T Tien Dang Huu	Machine Unlearning: Setting, Benchmark, Evaluation
04/04/2024		Naoya Inoue	Kick-off Meeting
03/04/2024		SHI,YutingGAO_BowenIrfan Robbani	(NLP2024 poster & presentation practice #2)
02/29/2024		Hakaze ChoNaoya Inoue	NLP2024 poster & presentation practice #1
02/22/2024		Irfan Robbanikenken	Practice talk for mid-term presentation
02/21/2024		S Shotaro Kitamura	Practice talk for mid-term presentation
02/08/2024		SHI,YutingGAO_Bowen	Practice talk for MS thesis defense
02/01/2024		Naoya Inoue	Self-Awareness Improves Reliability of LM as KB
01/25/2024		K Kim Pothong	Fallacies Detection Exploitation Toward Large Language Model
01/18/2024		kenken	Can Data Augmentation with Predicate Logic Extend the Predicate Logic Inference of LLM?
01/11/2024		T Tien Dang Huu	Machine Unlearning: Similarity-based methods

‣

2023

Date	Type	Presenters	Topic
12/21/2023		S SAKAI Yoshihiro	In-Context Learning
12/12/2023		A Akira Ishii	Identify Important Event in Narrative
12/05/2023		houjing	Various Probing on VLMs
11/28/2023		Irfan Robbani	Why My Argument has Fallacy
11/21/2023		Naoya Inoue	(Only Announcement)
11/14/2023		GAO_Bowen	Entailment Tree Construction: Step-by-Step Approach Using Large Language Models
11/07/2023		S Shotaro Kitamura	Towards Making Better Implicit Multi-hop QA Benchmark
10/31/2023		SHI,Yuting	Non-end-to-end Evaluation Approaches for Visual Reasoning
10/17/2023		Hakaze Cho	A Comprehensive (maybe) Survey / Intro about ICL Theory
10/10/2023		Naoya Inoue	Kick-off Meeting
10/02/2023		JIN, Tao	The Rise and Potential of Large Language Model Based Agents: A Survey
09/05/2023		SHI,Yuting GAO_Bowen	Practice Talk for Mid-term Presentation
08/07/2023		houjing	From traditional VL to MLLMs
07/24/2023			(Only Announcement)
07/18/2023		kenken	Controller of LLM for Multi-step Reasoning
07/03/2023		Irfan Robbani	Like Father, Like Son: Large Language Models Empower Smaller Language Models for Zero Shot Fallacy Detection
06/26/2023		Naoya Inoue	Let’s Watch Together: “How to Read/Write an International Conference Paper by Prof. Graham Neubig@CMU”
06/19/2023		SHI,Yuting	Towards Inductive Reasoning from Visual Information
06/14/2023		S Shotaro Kitamura	Research Progress Report: Towards generating Japanese multi-hop QA Datasets
05/29/2023			(Only Announcement)
05/22/2023		GAO_Bowen	[Paper Reading] Plan-and-Solve Prompting: Improving Zero-Shot Chain-of-Thought Reasoning by Large Language Models
05/15/2023		JIN, Tao	Self-Instruct_ Aligning LM with Self Generated Instructions
05/08/2023		H Hoai Linh Luu	Improving Robustness of NLP Models by Explanation-based Feedback
04/12/2023		Naoya Inoue	Kick-off Meeting