Search for: All records

Recent work learns contextual representations of source code by reconstructing tokens from their context. For downstream semantic understanding tasks like code clone detection, these representations should ideally capture program functionality. However, we show that the popular reconstruction-based RoBERTa model is sensitive to source code edits, even when the edits preserve semantics. We propose ContraCode: a contrastive pre-training task that learns code functionality, not form. ContraCode pre-trains a neural network to identify functionally similar variants of a program among many non-equivalent distractors. We scalably generate these variants using an automated source-to-source compiler as a form of data augmentation. Contrastive pre-training outperforms RoBERTa on an adversarial code clone detection benchmark by 39% AUROC. Surprisingly, improved adversarial robustness translates to better accuracy over natural code; ContraCode improves summarization and TypeScript type inference accuracy by 2 to 13 percentage points over competitive baselines. All source is available at https://github.com/parasj/contracode. 

Due to the theragnostic potential of mesoporous silica nanoparticles (MSNs), these were extensively investigated as a novel approach to improve clinical outcomes. Boasting an impressive array of formulations and modifications, MSNs demonstrate significant in vivo efficacy when used to identify or treat myriad malignant diseases in preclinical models. As MSNs continue transitioning into clinical trials, a thorough understanding of the characteristics of effective MSNs is necessary. This review highlights recent discoveries and advances in MSN understanding and technology. Specific focus is given to cancer theragnostic approaches using MSNs. Characteristics of MSNs such as size, shape, and surface properties are discussed in relation to effective nanomedicine practice and projected clinical efficacy. Additionally, tumor-targeting options used with MSNs are presented with extensive discussion on active-targeting molecules. Methods for decreasing MSN toxicity, improving site-specific delivery, and controlling release of loaded molecules are further explained. Challenges facing the field and translation to clinical environments are presented alongside potential avenues for continuing investigations.