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

Large languages models (LLMs) trained on datasets of publicly availablesource code have established a new state-of-the-art in code completion.However, these models are mostly unaware of the code that already exists withina specific project, preventing the models from making good use of existingAPIs. Instead, LLMs often invent, or "hallucinate", non-existent APIs orproduce variants of already existing code. Although the API information isavailable to IDEs, the input size limit of LLMs prevents code completiontechniques from including all relevant context into the prompt. This paperpresents De-Hallucinator, an LLM-based code completion technique that groundsthe predictions of a model through a novel combination of retrieving suitableAPI references and iteratively querying the model with increasingly suitablecontext information in the prompt. The approach exploits the observation thatLLMs often predict code that resembles the desired completion, but that failsto correctly refer to already existing APIs. De-Hallucinator automaticallyidentifies project-specific API references related to the code prefix and tothe model's initial predictions and adds these references into the prompt. Ourevaluation applies the approach to the task of predicting API usages inopen-source Python projects. We show that De-Hallucinator consistently improvesthe predicted code across four state-of-the-art LLMs compared to querying themodel only with the code before the cursor. In particular, the approachimproves the edit distance of the predicted code by 23-51% and the recall ofcorrectly predicted API usages by 24-61% relative to the baseline.

De-Hallucinator: Iterative Grounding for LLM-Based Code Completion