MERA Mitap on benchmarks (AI column)

15 April 17.00 MERA and the AI Alliance invite T-Bank to meet at benchmarks in all their forms and manifestations.

Headliners of the program are the authors of the MERA benchmark, the de facto standard for automatic testing of Russian-language LLM, and the creators of the Russian LLM Arena, the main platform for comparing models in real time. 

Mitap organizers promise to talk about: How benchmarks are arranged for text and multimodal models; What to consider when checking the LLM for the quality of writing code; How to compare specialized ML models with each other

After the presentations, the participants of the meeting will be able to ask the speakers about the plans for the development of projects and offer their answers to open questions, which in the field of benchmarks are becoming more and more.

The number of seats is limited, there will be no broadcast, the recording is likely to be too. Anyway, register.