Some functional suggestions
@DontPlanToEnd Thanks for your efforts to bring us a really useful leaderboard, but I noticed some inconvenient places, I wonder if you noticed them.
There is now an option for Foundation Models Only, can we add an option for only fine-tuned models and only merged models (they are multiple choice questions). Because most of the open source models released by other large companies besides Mistralai are heavily censored (especially those from China), I believe that most of us individual users use fine-tuned models or merged models, so I think this distinction is still quite useful.
Can we add an indicator of repetition? Some models are very repetitive, such as the model published by Mistralai, which has excellent performance, but after a few sentences, it tends to be repetitive. Of course, I don't know how it is really formed, it may be caused by the invalid verbosity of GPT data, but we all don't like them, unfortunately most models have these.
Can we add more filtering options like Excel? For example, I want to filter a fine-tuning model with Intelligence higher than 3, Willingness higher than 6, and UGI higher than 40. If there are these additional options, I can quickly query this model. If it can be realized, it will be very convenient. This is not necessary, after all, we can just download the form and filter it ourselves.
My English level is very poor. I use Google Translate to help me communicate. Please forgive me for the poor writing.