Spaces:
Running
Running
| VIDORE_DATASETS_KEYWORDS = [ | |
| "arxivqa", | |
| "docvqa", | |
| "infovqa", | |
| "tabfquad", | |
| "tatdqa", | |
| "shift", | |
| "artificial_intelligence", | |
| "energy", | |
| "government_reports", | |
| "healthcare_industry", | |
| ] | |
| VIDORE_2_DATASETS_KEYWORDS = [ | |
| "restaurant_esg", | |
| "rse_restaurant", | |
| "axa", | |
| "mit_biomedical", | |
| "economics_macro", | |
| ] | |
| def get_datasets_nickname(dataset_name) -> str: | |
| if "arxivqa" in dataset_name: | |
| return "ArxivQA" | |
| elif "docvqa" in dataset_name: | |
| return "DocVQA" | |
| elif "infovqa" in dataset_name: | |
| return "InfoVQA" | |
| elif "tabfquad" in dataset_name: | |
| return "TabFQuad" | |
| elif "tatdqa" in dataset_name: | |
| return "TAT-DQA" | |
| elif "shift" in dataset_name: | |
| return "Shift Project" | |
| elif "artificial_intelligence" in dataset_name: | |
| return "Artificial Intelligence" | |
| elif "energy" in dataset_name: | |
| return "Energy" | |
| elif "government_reports" in dataset_name: | |
| return "Government Reports" | |
| elif "healthcare_industry" in dataset_name: | |
| return "Healthcare Industry" | |
| elif "restaurant_esg" in dataset_name: | |
| return "ESG Restaurant Human" | |
| elif "rse_restaurant" in dataset_name and "multilingual" in dataset_name: | |
| return "ESG Restaurant Synthetic Multilingual" | |
| elif "rse_restaurant" in dataset_name: | |
| return "ESG Restaurant Synthetic" | |
| elif "axa" in dataset_name and "multilingual" in dataset_name: | |
| return "AXA Multilingual" | |
| elif "axa" in dataset_name: | |
| return "AXA" | |
| elif "mit_biomedical" in dataset_name and "multilingual" in dataset_name: | |
| return "MIT Biomedical Multilingual" | |
| elif "mit_biomedical" in dataset_name: | |
| return "MIT Biomedical" | |
| elif "economics_macro" in dataset_name and "multilingual" in dataset_name: | |
| return "Economics Macro Multilingual" | |
| elif "economics_macro" in dataset_name: | |
| return "Economics Macro" | |
| else: | |
| raise ValueError(f"Dataset {dataset_name} not found in ViDoRe") | |