For the Financial phrasebank dataset eval results, are you able to segregate them by "sentence50agree, sentence75Agree etc? as most were trained on sentence75Agree which is easier than the other configs.
Hey there! Unfortunately, this seems to be more of an issue with the model repositories themselves. All of the model results on the leaderboard for financial_phrasebank are self-reported. And the config doesnt seem to be specified in the self-reported model cards (e.g. https://huggingface.co/Sigma/financial-sentiment-analysis/blob/main/README.md)
You could try a few things:
Configure verified evaluation jobs (selecting the desired config and split) here: https://huggingface.co/spaces/autoevaluate/model-evaluator?dataset=financial_phrasebank. Although, I've noticed that the financial_phrasebank dataset only has a train split, not an eval split. I'm not very familiar with financial_phrasebank, but this seems wrong, so you might want to open an issue on their dataset repo here first: https://huggingface.co/datasets/financial_phrasebank/discussions.
Open issues on the model cards and get the owners of the models to report the config and splits that the results are for