Requires detailed test procedure instructions

[yangt1013]

Hi author, I am glad about your work. But I am a beginner in LLM, your code confuses me a lot.

1. First, in the captions_generations folder, execute python main.py --gpu DEVICE_ID --dataset paco_lvis_v1_test --batch_size BATCH_SIZE. Errors will often be reported. │ /ssd1/yangt/FG-OVD/code/FG-OVD-main/captions_generation/fill_categories.py:233 in │
   │ create_hardnegatives_object │
   │ │
   │ 230 │ │ # if there are no possible hardnegative for this object, we remove it from the l │
   │ 231 │ │ has_attributes[object_to_hn] = possible_hn != [[],[],[],[]] │
   │ 232 │ │
   │ ❱ 233 │ assert len(hardnegative_captions) > 0, "zero hardnegatives generated, the caption is │
   │ 234 │ return positive_caption, hardnegative_captions │
   │ 235 │
   │ 236 def create_hardnegatives(objects, n_hardnegatives): │
   ╰────────────────────────────────────────────────────────────────────────────────────────────────────╯
   AssertionError: zero hardnegatives generated, the caption is A
   I don't know why. Can you check carefully and make sure there are no errors before publishing?
2. When I skipped generating the dataset and ran the test files in the detector_inference folder, I found that the locations of some files were not displayed, and the parameter settings in args were very confusing. I don't know how to operate them.
   Traceback (most recent call last):
   File "/ssd1/yangt/FG-OVD/code/FG-OVD-main/detectors_inferences/cora_inferences.py", line 15, in
   import util.misc as utils
   ModuleNotFoundError: No module named 'util'
3. Please check all the files carefully, list the detailed running instructions and the detailed location and meaning of the args file, otherwise it is really confusing.
   Now we are very interested in this work, but after trying to run it for a week, we found various errors that made it impossible to solve. Can you help us solve these problems? Looking forward to your reply