Daniel Wurgaft Email & Phone Number

Toronto, On, Ca

Supervised by Dr. Roger Grosse. Investigating human-like decision-making biases in large language models. Using GPT-J, GPT-3, and OPT.

Toronto, Ontario, Canada

At the Cognitive Network Modeling Group:-Leading the online arm of a study assessing reward learning-related biases under suicidal ideation.-Computational modeling using a hierarchical bayesian model (the Hierarchical Gaussian Filter; HGF).-Assisting with experimental design and programming for an fMRI project focusing on computational modeling and identification of biomarkers for suicidal risk.-Helped design and programmed (using JsPsych) the experimental paradigm for an online experiment assessing Pavlovian bias in people with suicidal ideation.

-Leading a study on human representation learning. -Using computational models of reinforcement learning and category learning.-Building computational models of behavior using Stan.-Programming online cognitive psychology tasks using jsPsych.

Remote (Online)

-Implementing preprocessing and analysis pipelines using R.-In the process of writing a paper on alexithymia and emotion regulation.-Programming and assisting with the experimental design of online behavioral tasks used to assess emotion regulation and emotion schemas. Experiments were programmed using Jspsych, a JavaScript library for online psychology experiments.-Writing preregistrations. -Implemented fine-tuning of a Word2Vec model and conducted data analysis using the fine-tuned model.

-Supported by the University of Toronto Excellence award for research, led a study on the effects of prior knowledge on spatial learning. Was responsible for all aspects of the study: Devised the experimental design, programmed an online version of the experiment, and analyzed data. - Was responsible for data collection and running experiments for another project.

Tel Aviv, Israel

-Developed a program for in-context translation and syntax analysis of Spanish sentences to English using a deep learning translation engine and other NLP methods.-Using a deep learning translation model, developed a translation program that translates subtitle files from 5 languages into English, and edits them according to the company’s needs.