BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Program Synthesis and Understanding with Pretrained Language Model
 s - Ignacio Iacobacci\, Huawei 
DTSTART:20221110T110000Z
DTEND:20221110T120000Z
UID:TALK192710@talks.cam.ac.uk
CONTACT:Panagiotis Fytas
DESCRIPTION:In the last few years there has been a tremendous growth in th
 e topic of understanding and generation using NLP-grounded deep learning m
 odels. While earlier approaches were able to deal with just the simplest t
 asks\, the recent application of Pretrained Language Models (PLMs)\, speci
 fically trained with code snippets\, has brought new capabilities\, especi
 ally for the task of text-to-code generation or program synthesis. This ta
 lk will discuss the reasons of the recent growth of interest on this topic
 . We will discuss the main differences between working on natural language
  and programming language. We will provide an overview of the latest appro
 aches\, their intended use and limitations. Among other models we will int
 roduce PanguCoder\, our brand new Huawei in-house model for code synthesis
 \, which constitutes the building block of the AI-assisted tool for code g
 eneration from Huawei. We will cover existing datasets and benchmarks are 
 useful to make a fair comparison among different approaches. We will menti
 on CodeXGlue\, a benchmark that cover most common code-oriented tasks\, an
 d HumanEval which is\, at this time\, the de facto benchmark for text-to-c
 ode generation. Finally\, we will show some applications in the real word 
 and future perspectives in the area
LOCATION:GR04\, English Faculty Building\, 9 West Road\, Sidgwick Site
END:VEVENT
END:VCALENDAR