Related papers: Linguistic Similarity Within Centralized FLOSS Dev…
Supervised Fine-Tuning (SFT) is essential for aligning Large Language Models (LLMs) with user intent, yet it is believed to suppress generative diversity. Although this reduction is frequently referenced, formal empirical testing of the…
Usability and user experience (UX) issues are often not well emphasized and addressed in open source software (OSS) development. There is an imperative need for supporting OSS communities to collaboratively identify, understand, and fix UX…
Peer production projects such as Wikipedia or open-source software development allow volunteers to collectively create knowledge based products. The inclusive nature of such projects poses difficult challenges for ensuring trustworthiness…
High-quality datasets for task-oriented dialog are crucial for the development of virtual assistants. Yet three of the most relevant large scale dialog datasets suffer from one common flaw: the dialog state update can be tracked, to a great…
State of the art component-based software collections - such as FOSS distributions - are made of up to dozens of thousands components, with complex inter-dependencies and conflicts. Given a particular installation of such a system, each…
Frontier large language models (LLMs) excel as autonomous agents in many domains, yet they remain untested in complex enterprise systems where hidden workflows create cascading effects across interconnected databases. Existing enterprise…
We propose an architecture, Flare, that is a structured and easy way to develop applications rapidly, in a multitude of languages, which make use of online storage of data and management of users. The architecture eliminates the need for…
We propose the Single Conversation Methodology (SCM), a novel and pragmatic approach to software development using large language models (LLMs). In contrast to ad hoc interactions with generative AI, SCM emphasizes a structured and…
Large language models (LLMs) are increasingly shaping creative work and problem-solving; however, prior research suggests that they may diminish unassisted creativity. To address this tension, a coach-like LLM environment was developed that…
Open source software (OSS) sustainability depends not only on code contributions but also on governance structures that define who decides, who acts, and how responsibility is distributed. We lack systematic empirical evidence of how…
Foundation models (FMs), including large language models, have become increasingly popular due to their wide-ranging applicability and ability to understand human-like semantics. While previous research has explored the use of FMs in…
Large-scale software development has become a highly collaborative and geographically distributed endeavour, especially in open-source software development ecosystems and their associated developer communities. It has given rise to modern…
One of the primary factors that encourage developers to contribute to open source software (OSS) projects is the collaborative nature of OSS development. However, the collaborative structure of these communities largely remains unclear,…
Users typically interact with and evaluate language models via single outputs, but each output is just one sample from a broad distribution of possible completions. This interaction hides distributional structure such as modes, uncommon…
Centralization in code hosting and package management in the 2010s created fundamental shifts in the social arrangements of open source ecosystems. In a regime of centralized open source, platform effects can both empower and detract from…
The development of open source software (OSS) is a broad field which requires diverse skill sets. For example, maintainers help lead the project and promote its longevity, technical writers assist with documentation, bug reporters identify…
Continuous integration is a software engineering practice of frequently merging all developer working copies with a shared main branch, e.g., several times a day. With the advent of GitHub, a platform well known for its "social coding"…
Federated Learning (FL) stands as a prominent distributed learning paradigm among multiple clients to achieve a unified global model without privacy leakage. In contrast to FL, Personalized federated learning aims at serving for each client…
Software development environments (IDEs) have not followed the IT industry's inexorable trend towards distribution. They do too little to address the problems raised by today's increasingly distributed projects; neither do they facilitate…
When large language models are fine-tuned to generate persona- or tone-conditioned responses, their output diversity is severely limited--a failure we term Cross-Style Collapse. We trace this collapse to the cross-entropy objective, which…