eDIF: A European Deep Inference Fabric for Remote Interpretability of LLM
This initiative addresses the need for accessible LLM interpretability infrastructure in Europe, enabling remote experimentation for the research community, though it is incremental as it builds on existing NDIF-compatible frameworks.
The paper presents a feasibility study on deploying a European Deep Inference Fabric (eDIF), a GPU-based infrastructure for remote mechanistic interpretability research on large language models like GPT-2 and DeepSeek-R1-70B, with a pilot study involving 16 researchers showing stable performance and positive reception.
This paper presents a feasibility study on the deployment of a European Deep Inference Fabric (eDIF), an NDIF-compatible infrastructure designed to support mechanistic interpretability research on large language models. The need for widespread accessibility of LLM interpretability infrastructure in Europe drives this initiative to democratize advanced model analysis capabilities for the research community. The project introduces a GPU-based cluster hosted at Ansbach University of Applied Sciences and interconnected with partner institutions, enabling remote model inspection via the NNsight API. A structured pilot study involving 16 researchers from across Europe evaluated the platform's technical performance, usability, and scientific utility. Users conducted interventions such as activation patching, causal tracing, and representation analysis on models including GPT-2 and DeepSeek-R1-70B. The study revealed a gradual increase in user engagement, stable platform performance throughout, and a positive reception of the remote experimentation capabilities. It also marked the starting point for building a user community around the platform. Identified limitations such as prolonged download durations for activation data as well as intermittent execution interruptions are addressed in the roadmap for future development. This initiative marks a significant step towards widespread accessibility of LLM interpretability infrastructure in Europe and lays the groundwork for broader deployment, expanded tooling, and sustained community collaboration in mechanistic interpretability research.