Jason Phang
Member Of Technical Staff at EleutherAI
San Francisco, California, United States
Summary
🤩
Rockstar🎓
Top SchoolJason Phang is a PhD candidate at NYU and researcher at EleutherAI with 11 years of experience engineering large-scale NLP and generative models. He contributes to high-profile open-source projects like Hugging Face Transformers and EleutherAI’s GPT‑NeoX, including adding DeBERTaV2 model support and writing scripts to merge and reconfigure 20B‑parameter model weights to run on a single GPU. His work spans model development, evaluation, and deployment — from integrating GPT‑3 into an evaluation harness to building TensorFlow and image-wise implementations for a breast cancer screening classifier — combining applied medical ML with foundational language-model engineering. Pragmatic about reproducibility and resource constraints, he focuses on making cutting-edge models practical for real-world research and deployment.
12 years of coding experience
2 years of employment as a software developer
New York University
Stackoverflow
36reputation
2kreached
1answer
0questions
Github Skills (32)
transformers10
pytorch10
openai-api10
language-model10
gpt-310
parallelization10
python10
classification10
model-driven10
machine-learning10
language-models10
inference10
nlpjs10
model-building10
pretrained-model10
Programming languages (6)
Github contributions (5)
nyu-mll/jiant
Mar 2019 - Oct 2022
jiant is an nlp toolkit
Role in this project:
Back-end Developer & ML Engineer
Back-end Developer & ML EngineerContributions:1 release, 71 reviews, 145 commits in 3 years 8 months
Contributions summary:Jason primarily contributed to the inference interface and data processing for a natural language processing (NLP) toolkit. They developed a REPL (read-eval-print loop) and corpus processing capabilities within the `cola_inference.py` file. Further contributions included adding features for evaluating model performance using labeled data. Their work directly aligns with the toolkit's purpose to support NLP tasks such as CoLA.
nlptransformersmultitask-learningsentence-representationbert
nyukat/breast_cancer_classifier
Apr 2019 - Feb 2020
Deep Neural Networks Improve Radiologists' Performance in Breast Cancer Screening
Role in this project:ML Engineer
ML EngineerContributions:8 commits, 10 pushes, 2 branches in 10 months
Contributions summary:Jason primarily contributed to the implementation and refinement of a deep learning model for breast cancer classification. Their work included fixing issues related to output column swapping and adjusting assertions for variable batch sizes. The user also added a TensorFlow implementation of the model, indicating a focus on model development and potentially deployment. Furthermore, the addition of image-wise models and supporting notebooks suggests contributions to model evaluation and analysis.
breast-cancerbreast-cancer-diagnosistensorflowclassificationscreening
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.