arXiv subjects

Diego Garcia

Publications and source records attributed to Diego Garcia.

4 recordsLinked to original sources

arXiv2026

Knowledge Distillation for Automated AI Tutor Evaluation

The rapid integration of Large Language Models (LLMs) into K-12 and higher education has outpaced the development of reliable methods for evaluating their pedagogical quality. As the research community starts to explore the space of automating evaluation of AI tutors, we introduce FATE (FLC AI Tutor Evaluator), a specialized 8B-parameter language model designed to evaluate AI tutors. Aligned with the four core evaluation tracks from the BEA 2025 Shared Task, our model assesses pedagogical ability across Mistake Identification, Mistake Location, Guidance, and Actionability. Because pedagogical evaluation is a specialized task with limited labeled data, we leverage knowledge distillation from a frontier LLM to generate additional supervision, yielding absolute performance gains up to 22.63 percentage points. Finally, we demonstrate FATE's utility as an automated evaluator by benchmarking instructional responses generated by popular commercial models, including ChatGPT, Claude, Gemini, and DeepSeek. On average, we have found that Gemini 2.5 Flash perfomed best (82.88%), then ChatGPT 5.5 Instant (80.75%), followed by DeepSeek V4 Flash (80.13%) and Claude Sonnet 4.6 (74.00%).

cs.CL↗

arXiv2021

Governance for Security, Risks, Competition and Cooperation: Mapping the knowledge

The study aims to generate a map of the knowledge based on the research on topics related to governance and security, risks, competition and cooperation for the FDDI (Fudan Development Institute) proceedings publishing project: 'Reflections on Governance: Security and Risks, Competition and Cooperation.' That mapping exercise would enable a broader audience to delve into the current state, and interdisciplinary pathways of the research published worldwide for addressing complex problems of governance. Following this introduction, the second section presents the bibliometric methods used and the results' interpretation. The third section presents the results, followed by the fourth and fifth sections of discussion and conclusion, respectively.

cs.DL↗

arXiv2007

21-cm synthesis observations of VIRGOHI 21 - a possible dark galaxy in the Virgo Cluster

Many observations indicate that dark matter dominates the extra-galactic Universe, yet no totally dark structure of galactic proportions has ever been convincingly identified. Previously we have suggested that VIRGOHI 21, a 21-cm source we found in the Virgo Cluster using Jodrell Bank, was a possible dark galaxy because of its broad line-width (~200 km/s) unaccompanied by any visible gravitational source to account for it. We have now imaged VIRGOHI 21 in the neutral-hydrogen line and find what could be a dark, edge-on, spinning disk with the mass and diameter of a typical spiral galaxy. Moreover, VIRGOHI 21 has unquestionably been involved in an interaction with NGC 4254, a luminous spiral with an odd one-armed morphology, but lacking the massive interactor normally linked with such a feature. Numerical models of NGC 4254 call for a close interaction ~10^8 years ago with a perturber of ~10^11 solar masses. This we take as additional evidence for the massive nature of VIRGOHI 21 as there does not appear to be any other viable candidate. We have also used the Hubble Space Telescope to search for stars associated with the HI and find none down to an I band surface brightness limit of 31.1 +/- 0.2 mag/sq. arcsec.

astro-ph↗

arXiv2005

A Dark Hydrogen Cloud in the Virgo Cluster

VIRGOHI21 is an HI source detected in the Virgo Cluster survey of Davies et al. (2004) which has a neutral hydrogen mass of 10^8 M_solar and a velocity width of Delta V_20 = 220 km/s. From the Tully-Fisher relation, a galaxy with this velocity width would be expected to be 12th magnitude or brighter; however deep CCD imaging has failed to turn up a counterpart down to a surface-brightness level of 27.5 B mag/sq. arcsec. The HI observations show that it is extended over at least 16 kpc which, if the system is bound, gives it a minimum dynamical mass of ~10^11 M_solar and a mass to light ratio of M_dyn/L_B > 500 M_solar/L_solar. If it is tidal debris then the putative parents have vanished; the remaining viable explanation is that VIRGOHI21 is a dark halo that does not contain the expected bright galaxy. This object was found because of the low column density limit of our survey, a limit much lower than that achieved by all-sky surveys such as HIPASS. Further such sensitive surveys might turn up a significant number of the dark matter halos predicted by Dark Matter models.

astro-ph↗