Open JohannHalley opened 3 months ago
Instruction Tuning for Large Language Models: A Survey https://arxiv.org/pdf/2308.10792
FINETUNED LANGUAGE MODELS ARE ZERO-SHOT LEARNERS https://arxiv.org/pdf/2109.01652
Lisa Grossman Liu et al., “A Deep Database of Medical Abbreviations and Acronyms for Natural Language Processing,” Scientific Data 8, no. 1 (June 2, 2021): 149, https://doi.org/10.1038/s41597-021-00929-4.
This repository contains the live version of the Medical Abbreviation and Acronym Meta-Inventory, a large-scale American English database of medical abbreviations and their possible senses. https://github.com/lisavirginia/clinical-abbreviations
Manda Hosseini, Mandana Hosseini, and Reza Javidan, “Leveraging Large Language Models for Clinical Abbreviation Disambiguation,” Journal of Medical Systems, n.d.
F = BERT (x) (1) H2 = L2 (T (F )) (2) P = W3 ∗ H2 + b3 (3) Where P represents the output of the method, which corresponds to the final prediction for disambiguating the abbreviation’s sense. It is obtained by applying the final linear layer, represented by weights W3 and biases b3, to the output H2 of the transformer layer. The output H2 is the result of passing the output F of the BERT model through a transformer layer T. The transformer layer T captures long-range dependencies in the input data and enhances the model’s ability to understand the context and relationships between words.
BioGPT is used for data augmentation, which addresses the problem of data imbalance by generating more training instances, especially for those with limited number of instances with abbreviated meaning.
demo
MedAL
https://arc.net/l/quote/dizfzncf
chatgpt 4o
Parathyroid hormonerelated protein
可以用Levenshtein距离
llama3
https://huggingface.co/spaces/ysharma/Chat_with_Meta_llama3_8b