arXiv subjects

Takashi Kodama

Publications and source records attributed to Takashi Kodama.

13 recordsLinked to original sources

arXiv2026

Measuring Task-Agnostic Training Data Influence Across Language Model Pretraining

Measuring training data influence consistently across language model pretraining is challenging. It is difficult to select downstream tasks or validation sets representative of a model's general capabilities, and reliance on task performance at intermediate checkpoints complicates comparisons across training. We propose a measure of training data influence that does not require selecting a downstream task or validation set as the attribution target. Specifically, we define an example's influence by how much its gradient update reduces the squared distance to the final parameters of a given pretraining run, and estimate this quantity from intermediate checkpoints without retraining. Applying the method to 18 configurations from the Pythia and PolyPythia suites, we find systematic temporal changes in influential data. Early in training, literature-related data are more strongly aligned with the trajectory toward the final parameters, whereas STEM data become more strongly aligned in later stages. This qualitative crossover is broadly consistent across model configurations. Our results provide a tractable trajectory-level view of how influential data change throughout pretraining, complementing influence analyses defined with respect to specific downstream tasks or validation sets.

cs.CL↗

arXiv2026

Investigating Learner-Aware Design of LLM-Generated Educational Feedback

Although large language models (LLMs) show promise for generating educational feedback, it remains unclear how feedback should be designed (e.g., tone and information coverage) to support answer revision and learner acceptance across diverse learner profiles. We define six feedback designs for multiple-choice biology questions, including a baseline design and variants with additional feedback elements, and conduct an empirical study with high school students. We evaluate feedback using immediate revision performance and six subjective evaluation criteria, and analyze how feedback preferences vary across learner profiles based on personality traits. Our results show that feedback with clear and comprehensive guidance improves revision performance and receives favorable evaluations across learner profiles, whereas informational novelty and affective framing vary across profiles. These findings suggest that learner profiles should be considered when designing LLM-generated feedback.

cs.CL↗

arXiv2026

Dango: A Strictly L1-Only Large Language Model for Studying Second Language Acquisition

We introduce Dango, a 1.8B-parameter large language model designed for controlled studies of L1-to-L2 (Japanese-to-English) transfer in second language acquisition (SLA). While previous studies have explored SLA in language models, they have predominantly relied on smaller or non-decoder models, limiting their ability to generate open-ended text and reducing their suitability as practical L2 simulators. We identify a key challenge when scaling models to this size: L2 contamination within the "monolingual" pretraining corpus used for L1 acquisition. To address this, we propose a filtering method to reduce premature exposure to English while preserving realistic, minimal exposure. We then fine-tune the model on LLM-generated L2-learning lessons to simulate the L2 acquisition process. Our evaluations confirm that Dango develops human-like L2 production patterns, outperforming both unfiltered and standard multilingual baselines. We release the model, data, and code to facilitate reproducible computational SLA research and learner-facing applications.

cs.CL↗

arXiv2025

Are Checklists Really Useful for Automatic Evaluation of Generative Tasks?

Automatic evaluation of generative tasks using large language models faces challenges due to ambiguous criteria. Although automatic checklist generation is a potentially promising approach, its usefulness remains underexplored. We investigate whether checklists should be used for all questions or selectively, generate them using six methods, evaluate their effectiveness across eight model sizes, and identify checklist items that correlate with human evaluations. Through experiments on pairwise comparison and direct scoring tasks, we find that selective checklist use tends to improve evaluation performance in pairwise settings, while its benefits are less consistent in direct scoring. Our analysis also shows that even checklist items with low correlation to human scores often reflect human-written criteria, indicating potential inconsistencies in human evaluation. These findings highlight the need to more clearly define objective evaluation criteria to guide both human and automatic evaluations. \footnote{Our code is available at~https://github.com/momo0817/checklist-effectiveness-study

cs.CL↗

arXiv2025

AnswerCarefully: A Dataset for Improving the Safety of Japanese LLM Output

In this paper we present AnswerCarefully, a dataset for promoting the safety and appropriateness of Japanese LLM outputs. The dataset consists of 1,800 pairs of questions and reference answers, where the questions require special attention in answering. It covers a wide range of risk categories established in prior English-language datasets, but the data samples are original in that they are manually created to reflect the socio-cultural context of LLM usage in Japan. We show that using this dataset for instruction to fine-tune a Japanese LLM led to improved output safety without compromising the utility of general responses. We also report the results of a safety evaluation of 12 Japanese LLMs using this dataset as a benchmark. Finally, we describe the latest update on the dataset which provides English translations and annotations of the questions, aimed at facilitating the derivation of similar datasets in different languages and regions.

cs.CL↗

arXiv2024

LLM-jp: A Cross-organizational Project for the Research and Development of Fully Open Japanese LLMs

This paper introduces LLM-jp, a cross-organizational project for the research and development of Japanese large language models (LLMs). LLM-jp aims to develop open-source and strong Japanese LLMs, and as of this writing, more than 1,500 participants from academia and industry are working together for this purpose. This paper presents the background of the establishment of LLM-jp, summaries of its activities, and technical reports on the LLMs developed by LLM-jp. For the latest activities, visit https://llm-jp.nii.ac.jp/en/.

cs.CL↗

arXiv2024

RecMind: Japanese Movie Recommendation Dialogue with Seeker's Internal State

Humans pay careful attention to the interlocutor's internal state in dialogues. For example, in recommendation dialogues, we make recommendations while estimating the seeker's internal state, such as his/her level of knowledge and interest. Since there are no existing annotated resources for the analysis, we constructed RecMind, a Japanese movie recommendation dialogue dataset with annotations of the seeker's internal state at the entity level. Each entity has a subjective label annotated by the seeker and an objective label annotated by the recommender. RecMind also features engaging dialogues with long seeker's utterances, enabling a detailed analysis of the seeker's internal state. Our analysis based on RecMind reveals that entities that the seeker has no knowledge about but has an interest in contribute to recommendation success. We also propose a response generation framework that explicitly considers the seeker's internal state, utilizing the chain-of-thought prompting. The human evaluation results show that our proposed method outperforms the baseline method in both consistency and the success of recommendations.

cs.CL↗

arXiv2021

Modulation of interfacial thermal transport between fumed silica nanoparticles by surface chemical functionalization for advanced thermal insulation

Since solid-state heat transport in a highly porous nanocomposite strongly depends on the thermal boundary conductance (TBC) between constituent nanomaterials, further suppression of the TBC is important for improving performance of thermal insulators. Here, targeting a nanocomposite fabricated by stamping fumed silica nanoparticles, we perform a wide variety of surface functionalization on fumed silica nanoparticles by silane coupling method and investigate the impact on the thermal conductivity (Km). The Km of the silica nanocomposite is approximately 20 and 9 mW/m/K under atmospheric and vacuum condition at the material density of 0.2 g/cm3 without surface functionalization, respectively, and the experimental results indicate that the Km can be modulated depending on the chemical structure of molecules. The surface modification with a linear alkyl chain of optimal length significantly suppresses Km by approximately 30%, and the suppression can be further enhanced to approximately 50% with the infrared opacifier. The magnitude of suppression was found to sensitively depend on the length of terminal chain. The magnitude is also related to the number of reactive silanol groups in the chemical structure, where the surface modification with fluorocarbon gives the largest suppression. The surface hydrophobization merits thermal insulation through significant suppression of the TBC, presumably by reducing the water molecules that otherwise would serve as heat conduction channels at the interface. On the other hand, when the chain length is long, the suppression is counteracted by the enhanced phonon transmission through the silane coupling molecules that grows with the chain length. This is supported by the analytical model and present simulation results, leading to predict the optimal chemical structure for better thermal insulation.

physics.app-ph↗

arXiv2021

Anisotropic Thermal Conductivity Measurement of Organic Thin Film with Bidirectional 3omega Method

Organic thin film materials with molecular ordering are gaining attention as they exhibit semiconductor characteristics. When using them for electronics, the thermal management becomes important, where heat dissipation is directional owning to the anisotropic thermal conductivity arising from the molecular ordering. However, it is difficult to evaluate the anisotropy by simultaneously measuring in-plane and cross-plane thermal conductivities of the film on a substrate, because the film is typically as thin as tens to hundreds of nanometers and its in-plane thermal conductivity is low. Here, we develop a novel bidirectional 3ω system that measures the anisotropic thermal conductivity of thin films by patterning two metal wires with different widths and preparing the films on top, and extracting the in-plane and cross-plane thermal conductivities using the difference in their sensitivities to the metal-wire width. Using the developed system, the thermal conductivity of spin-coated poly(3,4-ethylenedioxythiophene) olystyrene sulfonate (PEDOT:PSS) with thickness of 70 nm was successfully measured. The measured in-plane thermal conductivity of PEDOT:PSS film was as high as 2.9 W m-1 K-1 presumably due to the high structural ordering, giving anisotropy of 10. The calculations of measurement sensitivity to the film thickness and thermal conductivities suggest that the device can be applied to much thinner films by utilizing metal wires with smaller width.

physics.ins-det↗

arXiv2021

Scalable monolayer-functionalized nanointerface for thermal conductivity enhancement in copper/diamond composite

Aiming at developing high thermal conductivity copper/diamond composite, an unconventional approach applying self-assembled monolayer (SAM) prior to the high-temperature sintering of copper/diamond composite was utilized to enhance the thermal boundary conductance (TBC) between copper and diamond. The enhancement was first systematically confirmed on a model interface system by detailed SAM morphology characterization and TBC measurements. TBC significantly depends on the SAM coverage and ordering, and the formation of high-quality SAM promoted the TBC to 73 MW/m^2-K from 27 MW/m^2-K, the value without SAM. With the help of molecular dynamics simulations, the TBC enhancement was identified to be determined by the number of SAM bridges and the overlap of vibrational density of states. The diamond particles of 210 {\micro\metre} in size were simultaneously functionalized by SAM with the condition giving the highest TBC in the model system and sintered together with the copper to fabricate isotropic copper/diamond composite of 50% volume fraction. The measured thermal conductivity marked 711 W/m-K at room temperature, the highest value among the ones with similar diamond-particles volume fraction and size. This work demonstrates a novel strategy to enhance the thermal conductivity of composite materials by SAM functionalization.

cond-mat.mtrl-sci↗

arXiv2020

Modeling and Utilizing User's Internal State in Movie Recommendation Dialogue

Intelligent dialogue systems are expected as a new interface between humans and machines. Such an intelligent dialogue system should estimate the user's internal state (UIS) in dialogues and change its response appropriately according to the estimation result. In this paper, we model the UIS in dialogues, taking movie recommendation dialogues as examples, and construct a dialogue system that changes its response based on the UIS. Based on the dialogue data analysis, we model the UIS as three elements: knowledge, interest, and engagement. We train the UIS estimators on a dialogue corpus with the modeled UIS's annotations. The estimators achieved high estimation accuracy. We also design response change rules that change the system's responses according to each UIS. We confirmed that response changes using the result of the UIS estimators improved the system utterances' naturalness in both dialogue-wise evaluation and utterance-wise evaluation.

cs.CL↗

arXiv2020

A System for Worldwide COVID-19 Information Aggregation

The global pandemic of COVID-19 has made the public pay close attention to related news, covering various domains, such as sanitation, treatment, and effects on education. Meanwhile, the COVID-19 condition is very different among the countries (e.g., policies and development of the epidemic), and thus citizens would be interested in news in foreign countries. We build a system for worldwide COVID-19 information aggregation containing reliable articles from 10 regions in 7 languages sorted by topics. Our reliable COVID-19 related website dataset collected through crowdsourcing ensures the quality of the articles. A neural machine translation module translates articles in other languages into Japanese and English. A BERT-based topic-classifier trained on our article-topic pair dataset helps users find their interested information efficiently by putting articles into different categories.

cs.CL↗

arXiv2019

One-directional thermal transport in densely aligned single-wall carbon nanotube films

Individual carbon nanotubes (CNTs) possess extremely high thermal conductivities. However, the thermal conductivities and their anisotropy of macroscopic assemblies of CNTs have so far remained small. Here, we report results of directional thermal transport measurements on a nearly-perfectly aligned CNT film fabricated via controlled vacuum filtration. We found the thermal conductivity to be 43 +- 2.2 W m^-1 K^-1 with a record-high thermal anisotropy of 500. From the temperature dependence of the thermal conductivity and its agreement with the atomistic phonon transport calculation, we conclude that the effect of intertube thermal resistance on heat conduction in the alignment direction is negligible because of the large contact area between CNTs. These observations thus represent ideal unidirectional thermal transport, i.e., the thermal conductivity of the film is determined solely by that of individual CNTs.

physics.app-ph↗