Search Results for author:
Found 8 papers, 4 papers with code
Extrinsically-Focused Evaluation of Omissions in Medical Summarization
In safety-critical domains such as medicine, more rigorous evaluation is required, especially given the potential for LLMs to omit important information in the resulting summary.
Generating medically-accurate summaries of patient-provider dialogue: A multi-stage approach using large language models
A medical provider's summary of a patient visit serves several critical purposes, including clinical decision-making, facilitating hand-offs between providers, and as a reference for the patient.
CONSCENDI: A Contrastive and Scenario-Guided Distillation Approach to Guardrail Models for Virtual Assistants
We use CONSCENDI to exhaustively generate training data with two key LLM-powered components: scenario-augmented generation and contrastive training examples.
Quantitative analysis of collagen remodeling in pancreatic lesions using computationally translated collagen images derived from brightfield microscopy images
However, a technical approach that can perform a quantitative analysis of fibrillar collagen directly on standard slides stained with H&E can (i) discard the need for specialized and costly equipment or labels, (ii) further supplement the conventional histopathological insights and, (iii) potentially be integrated within the framework of standard histopathology workflow.
DERA: Enhancing Large Language Model Completions with Dialog-Enabled Resolving Agents
Large language models (LLMs) have emerged as valuable tools for many natural language understanding tasks.
Training Neural Networks with Fixed Sparse Masks
In this paper, we show that it is possible to induce a fixed sparse mask on the model's parameters that selects a subset to update over many iterations.
Adding more data does not always help: A study in medical conversation summarization with PEGASUS
Summarized conversations are used to facilitate patient hand-offs between physicians, and as part of providing care in the future.
RealMix: Towards Realistic Semi-Supervised Deep Learning Algorithms
Notably, RealMix achieves an error rate of 9. 79% on CIFAR10 with 250 labels and is the only SSL method tested able to surpass baseline performance when there is significant mismatch in the labeled and unlabeled data distributions.
