English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Python is a popular dynamic programming language, evidenced by its ranking asthe second most commonly used language on GitHub. However, its dynamic typesystem can lead to potential type errors, leading researchers to exploreautomatic type inference approaches for Python programs. The rule-based typeinference approaches can ensure the accuracy of predicted variable types, butthey suffer from low coverage problems. Supervised type inference approaches,while feature-agnostic, require large, high-quality annotated datasets and arelimited to pre-defined types. As zero-shot approaches, the cloze-styleapproaches reformulate the type inference problem into a fill-in-the-blankproblem. However, their performance is limited.  This paper introduces TypeGen, a few-shot generative type inference approachthat incorporates static domain knowledge from static analysis. TypeGen createschain-of-thought (COT) prompts by translating the type inference steps ofstatic analysis into prompts based on the type dependency graphs (TDGs),enabling language models to learn from how static analysis infers types. Bycombining COT prompts with code slices and type hints, TypeGen constructsexample prompts from human annotations. TypeGen only requires very fewannotated examples to teach language models to generate similar COT prompts viain-context learning. Moreover, TypeGen enhances the interpretability of resultsthrough the use of the input-explanation-output strategy. Experiments show thatTypeGen outperforms the best baseline Type4Py by 10.0% for argument typeprediction and 22.5% in return value type prediction in terms of top-1 ExactMatch by using only five examples. Furthermore, TypeGen achieves substantialimprovements of 27% to 84% compared to the zero-shot performance of largelanguage models with parameter sizes ranging from 1.3B to 175B in terms oftop-1 Exact Match.

Generative Type Inference for Python