· Event Recap
Theory Meets Practice: five workshops with dida
From explainable AI for LLMs to tumor tissue profiling – in five hands-on workshops with dida, researchers and engineers showed how current methods become working code. Three sessions are on YouTube.
In January, we started Theory Meets Practice, a workshop series with the Berlin machine learning company dida. In each session, experts explain a current method, and you code along on your own laptop.
- Opening the Black Box: Explainable AI for LLMs (7 January) – Reduan Achtibat (Fraunhofer HHI) on layer-wise relevance propagation for transformers. Watch the recording.
- Giving Eyes to your AI: Multimodal Agents with Haystack (21 January) – Bilge Yücel (deepset, the makers of Haystack) on agents that understand images and text.
- Embedded AI: Deep Learning Models on €5 Microcontrollers (29 April) – Elias Trommer (Infineon) on quantization, inference engines and deploying models to embedded hardware.
- Explainable Scientific Discovery with AI Memory (27 May) – Lazar Obradović (cognee) on turning scientific papers into knowledge graphs to evaluate new hypotheses. Watch the recording.
- AI for Precision Medicine (24 June) – Lukas Ruff, Christina Embacher, Jonas Dippel and Maaike Galama (Aignostics) on pathology foundation models and tumor tissue profiling. Watch the recording.
Thank you to dida for co-hosting the series and to all speakers! All workshops are listed on the workshops page.
