arXiv subjects

Justin Sun

Publications and source records attributed to Justin Sun.

3 recordsLinked to original sources

arXiv2026

Estimating Earth's Temperature Response with Transformed and Augmented OLS

The long-term relationship between radiative forcing and surface temperature is central to predicting the impacts of climate change. This study employs multicointegration to characterize this relationship and provides the first application of the Transformed and Augmented Ordinary Least Squares (TAOLS) estimator to estimate the multicointegration model. The main objective is to estimate the Equilibrium Climate Sensitivity (ECS), defined as the global mean surface temperature increase following a doubling of atmospheric carbon dioxide. Diagnostic tests reveal that radiative forcing innovations are strongly non-Gaussian, providing a key motivation for applying semiparametric TAOLS rather than the parametric Gaussian maximum likelihood method. TAOLS is also robust to misspecification of the short-run dynamics. Using the three data pairings of Bruns et al. (2020), we obtain TAOLS estimates of ECS ranging from $1.81^{\circ}$C to $2.49^{\circ}$C, consistently below their main maximum likelihood estimate of $2.80^{\circ}$C. Because the model we use is linear and is based on global-mean data, it cannot capture state-dependent feedbacks or regional differences; extending TAOLS along either dimension is a natural next step.

econ.EM↗

arXiv2024

MAGID: An Automated Pipeline for Generating Synthetic Multi-modal Datasets

Development of multimodal interactive systems is hindered by the lack of rich, multimodal (text, images) conversational data, which is needed in large quantities for LLMs. Previous approaches augment textual dialogues with retrieved images, posing privacy, diversity, and quality constraints. In this work, we introduce Multimodal Augmented Generative Images Dialogues (MAGID), a framework to augment text-only dialogues with diverse and high-quality images. Subsequently, a diffusion model is applied to craft corresponding images, ensuring alignment with the identified text. Finally, MAGID incorporates an innovative feedback loop between an image description generation module (textual LLM) and image quality modules (addressing aesthetics, image-text matching, and safety), that work in tandem to generate high-quality and multi-modal dialogues. We compare MAGID to other SOTA baselines on three dialogue datasets, using automated and human evaluation. Our results show that MAGID is comparable to or better than baselines, with significant improvements in human evaluation, especially against retrieval baselines where the image database is small.

cs.CL↗

arXiv2024

Bootstrapping LLM-based Task-Oriented Dialogue Agents via Self-Talk

Large language models (LLMs) are powerful dialogue agents, but specializing them towards fulfilling a specific function can be challenging. Instructing tuning, i.e. tuning models on instruction and sample responses generated by humans (Ouyang et al., 2022), has proven as an effective method to do so, yet requires a number of data samples that a) might not be available or b) costly to generate. Furthermore, this cost increases when the goal is to make the LLM follow a specific workflow within a dialogue instead of single instructions. Inspired by the self-play technique in reinforcement learning and the use of LLMs to simulate human agents, we propose a more effective method for data collection through LLMs engaging in a conversation in various roles. This approach generates a training data via "self-talk" of LLMs that can be refined and utilized for supervised fine-tuning. We introduce an automated way to measure the (partial) success of a dialogue. This metric is used to filter the generated conversational data that is fed back in LLM for training. Based on our automated and human evaluations of conversation quality, we demonstrate that such self-talk data improves results. In addition, we examine the various characteristics that showcase the quality of generated dialogues and how they can be connected to their potential utility as training data.

cs.CL↗