Search

Recently, the demand for psychological counseling has significantly increased as more individuals express concerns about their mental health. This surge has accelerated efforts to improve the accessibility of counseling by using large language models (LLMs) as counselors. To ensure client privacy, training open-source LLMs faces a key challenge: the absence of realistic counseling datasets. To address this, we introduce Cactus, a multi-turn dialogue dataset that emulates real-life interactions using the goal-oriented and structured approach of Cognitive Behavioral Therapy (CBT). We create a diverse and realistic dataset by designing clients with varied, specific personas, and having counselors systematically apply CBT techniques in their interactions. To assess the quality of our data, we benchmark against established psychological criteria used to evaluate real counseling sessions, ensuring alignment with expert evaluations. Experimental results demonstrate that Camel, a model trained with Cactus, outperforms other models in counseling skills, highlighting its effectiveness and potential as a counseling agent. We make our data, model, and code publicly available., Comment: Under Review

Multi-agent trajectory forecasting in autonomous driving requires an agent to accurately anticipate the behaviors of the surrounding vehicles and pedestrians, for safe and reliable decision-making. Due to partial observability in these dynamical scenes, directly obtaining the posterior distribution over future agent trajectories remains a challenging problem. In realistic embodied environments, each agent's future trajectories should be both diverse since multiple plausible sequences of actions can be used to reach its intended goals, and admissible since they must obey physical constraints and stay in drivable areas. In this paper, we propose a model that synthesizes multiple input signals from the multimodal world|the environment's scene context and interactions between multiple surrounding agents|to best model all diverse and admissible trajectories. We compare our model with strong baselines and ablations across two public datasets and show a significant performance improvement over previous state-of-the-art methods. Lastly, we offer new metrics incorporating admissibility criteria to further study and evaluate the diversity of predictions. Codes are at: https://github.com/kami93/CMU-DATF., Comment: ECCV 2020

This paper describes the formation of, and initial results for, a new FLUXNET coordination network for ecosystem-scale methane (CH4) measurements at 60 sites globally, organized by the Global Carbon Project in partnership with other initiatives and regional flux tower networks. The objectives of the effort are presented along with an overview of the coverage of eddy covariance (EC) CH4 flux measurements globally, initial results comparing CH4 fluxes across the sites, and future research directions and needs. Annual estimates of net CH4 fluxes across sites ranged from −0.2 ± 0.02 g C m(exp -2) yr(exp -1) for an upland forest site to 114.9 ± 13.4 g C m(exp -2) yr(exp -1) for an estuarine freshwater marsh, with fluxes exceeding 40 g C m(exp -2) yr(exp -1) at multiple sites. Average annual soil and air temperatures were found to be the strongest predictor of annual CH4 flux across wetland sites globally. Water table position was positively correlated with annual CH4 emissions, although only for wetland sites that were not consistently inundated throughout the year. The ratio of annual CH4 fluxes to ecosystem respiration increased significantly with mean site temperature. Uncertainties in annual CH4 estimates due to gap-filling and random errors were on average ±1.6 g C m(exp -2) yr(exp -1) at 95% confidence, with the relative error decreasing exponentially with increasing flux magnitude across sites. Through the analysis and synthesis of a growing EC CH4 flux database, the controls on ecosystem CH4 fluxes can be better understood, used to inform and validate Earth system models, and reconcile differences between land surface model- and atmospheric-based estimates of CH4 emissions.

The lack of a standardized database of eddy covariance observations has been an obstacle for data-driven estimation of terrestrial carbon dioxide fluxes in Asia. In this study, we developed such a standardized database using 54 sites from various databases by applying consistent postprocessing for data-driven estimation of gross primary productivity (GPP) and net ecosystem carbon dioxide exchange (NEE). Data-driven estimation was conducted by using a machine learning algorithm: support vector regression (SVR), with remote sensing data for 2000 to 2015 period. Site-level evaluation of the estimated carbon dioxide fluxes shows that although performance varies in different vegetation and climate classifications, GPP and NEE at 8 days are reproduced (e.g., r (exp 2) =0.73 and 0.42 for 8 day GPP and NEE). Evaluation of spatially estimated GPP with Global Ozone Monitoring Experiment 2 sensor-based Sun-induced chlorophyll fluorescence shows that monthly GPP variations at subcontinental scale were reproduced by SVR (r (exp 2)=1.00, 0.94, 0.91, and 0.89 for Siberia, East Asia, South Asia, and Southeast Asia, respectively). Evaluation of spatially estimated NEE with net atmosphere-land carbon dioxide fluxes of Greenhouse Gases Observing Satellite (GOSAT) Level 4A product shows that monthly variations of these data were consistent in Siberia and East Asia; meanwhile, inconsistency was found in South Asia and Southeast Asia. Furthermore, differences in the land carbon dioxide fluxes from SVR-NEE and GOSAT Level 4A were partially explained by accounting for the differences in the definition of land carbon dioxide fluxes. These data-driven estimates can provide a new opportunity to assess carbon dioxide fluxes in Asia and evaluate and constrain terrestrial ecosystem models.