[Benchmark] Add VLRMBench Support (#1259)

* add vlrmbench dataset

* Translate comments to English

* Fix code formatting issues

* Reconstructed the dataset and the evaluation code of vlrmbench

* Refactor and translate comments in VLRMBench dataset code to English

* Consolidate dataset imports in __init__.py

* Add multi_solution evaluation support in VLRMBench dataset

* Add foresight evaluation support in VLRMBench dataset
Updated initialization to warn about supported task types.
Fix and update the md5

---------

Co-authored-by: Xinyu Fang <fangxinyutju202009@126.com>

Author: Winston-Yuan

vlmeval/dataset/__init__.py

@@ -80,6 +80,7 @@
 from .mmifeval import MMIFEval
 from .chartmimic import ChartMimic
 from .m4bench import M4Bench
+from .vlrmbench import VLRMBench
 from .mmhelix import MMHELIX
 from .medqbench_mcq import MedqbenchMCQDataset
 from .medqbench_caption import MedqbenchCaptionDataset
@@ -218,7 +219,7 @@ def evaluate(self, eval_file, **judge_kwargs):
     OmniEarthMCQBench, VisFactor, OSTDataset, OCRBench_v2, TreeBench, CVQA, M4Bench,
     AyaVisionBench, TopViewRS, VLMBias, MMHELIX, MedqbenchMCQDataset,
     MedqbenchPairedDescriptionDataset, MedqbenchCaptionDataset, ChartMuseum, ChartQAPro, ReasonMap_Plus,
-    olmOCRBench, OceanOCRBench, MATBench
+    olmOCRBench, OceanOCRBench, MATBench, VLRMBench
 ]
 
 VIDEO_DATASET = [