3 citations · 5 across the 13 of their papers we have counts for
14 papers
Grow, Don't Overwrite: Fine-tuning Without Forgetting
Dyah Adila, Hanna Mazzawi, Benoit Dherin +1
Adapting pre-trained models to specialized tasks often leads to catastrophic forgetting, where new knowledge overwrites foundational capabilities. Existing methods either compromis…
Weight Updates as Activation Shifts: A Principled Framework for Steering
Dyah Adila, John Cooper, Alexander Yun +2
Activation steering promises to be an extremely parameter-efficient form of adaptation, but its effectiveness depends on critical design choices -- such as intervention location an…
SlopCodeBench: Benchmarking How Coding Agents Degrade Over Long-Horizon Iterative Tasks
Gabriel Orlanski, Devjeet Roy, Alexander Yun +7
Software development is iterative, yet agentic coding benchmarks hide design issues through their single-shot setup. Recent iterative benchmarks attempt to remedy this but heavily…
CrEst: Credibility Estimation for Contexts in LLMs via Weak Supervision
Dyah Adila, Shuai Zhang, Boran Han +2
The integration of contextual information has significantly enhanced the performance of large language models (LLMs) on knowledge-intensive tasks. However, existing methods often o…
Personalize Your LLM: Fake it then Align it
Yijing Zhang, Dyah Adila, Changho Shin +1
Personalizing large language models (LLMs) is essential for delivering tailored interactions that improve user experience. Many existing personalization methods require fine-tuning…
Discovering Bias in Latent Space: An Unsupervised Debiasing Approach
Dyah Adila, Shuai Zhang, Boran Han +1
The question-answering (QA) capabilities of foundation models are highly sensitive to prompt variations, rendering their performance susceptible to superficial, non-meaning-alterin…
Is Free Self-Alignment Possible?
Dyah Adila, Changho Shin, Yijing Zhang +1
Aligning pretrained language models (LMs) often requires large-scale preference data and substantial computational resources. These costs become even more prohibitive for multi-obj…
Multimodal Data Curation via Object Detection and Filter Ensembles
Tzu-Heng Huang, Changho Shin, Sui Jiet Tay +2
We propose an approach for curating multimodal data that we used for our entry in the 2023 DataComp competition filtering track. Our technique combines object detection and weak su…
Zero-Shot Robustification of Zero-Shot Models
Dyah Adila, Changho Shin, Linrong Cai +1
Zero-shot inference is a powerful paradigm that enables the use of large pretrained models for downstream classification tasks without further training. However, these models are v…
Geometry-Aware Adaptation for Pretrained Models
Nicholas Roberts, Xintong Li, Dyah Adila +4
Machine learning models -- including prominent zero-shot models -- are often trained on datasets whose labels are only a small proportion of a larger label space. Such spaces are c…
Mitigating Source Bias for Fairer Weak Supervision
Changho Shin, Sonia Cromp, Dyah Adila +1
Weak supervision enables efficient development of training sets by reducing the need for ground truth labels. However, the techniques that make weak supervision attractive -- such…
AutoWS-Bench-101: Benchmarking Automated Weak Supervision with 100 Labels
Nicholas Roberts, Xintong Li, Tzu-Heng Huang +7
Weak supervision (WS) is a powerful method to build labeled datasets for training supervised models in the face of little-to-no labeled data. It replaces hand-labeling data with ag…
Shoring Up the Foundations: Fusing Model Embeddings and Weak Supervision
Mayee F. Chen, Daniel Y. Fu, Dyah Adila +4
Foundation models offer an exciting new paradigm for constructing models with out-of-the-box embeddings and a few labeled examples. However, it is not clear how to best apply found…
Understanding Out-of-distribution: A Perspective of Data Dynamics
Dyah Adila, Dongyeop Kang
Despite machine learning models' success in Natural Language Processing (NLP) tasks, predictions from these models frequently fail on out-of-distribution (OOD) samples. Prior works…