1 Issue closed by 1 person

• Why cat 255 tokens to seg_token_mask?

  #172 closed Jan 24, 2025

2 Issues opened by 2 people

• Inference produces blank segmentation masks

  #174 opened Jan 29, 2025

• can't load LISA++

  #173 opened Jan 21, 2025

6 Unresolved conversations

Sometimes conversations happen on old items that aren’t yet closed. Here is a list of all the Issues and Pull Requests with unresolved conversations.

• > > LLaVA-1.5 uses 336px image resolution, so you should change the clip model and control max context length. Also, the image token length is set to 256 by default, but when the resolution is changed to 336, the image token length should be set to 576. Overall, some implementation details need further consideration to adapt to llava-1.5. You should check that in detail.

  #163 commented on Jan 6, 2025 • 0 new comments

• gt_mask shape is (0, x, y) while pred_mask shape is (some number, x, y), which raise assertion error for vqa and reason seg datasets

  #120 commented on Jan 6, 2025 • 0 new comments

• AttributeError: 'NoneType' object has no attribute 'shape'

  #117 commented on Jan 7, 2025 • 0 new comments

• Generated several bin files instead of only one bin file using zero_to_fp32.py

  #155 commented on Jan 14, 2025 • 0 new comments

• Can't reproduce the results of lisa-7b

  #135 commented on Jan 22, 2025 • 0 new comments

• Mask image seems to be nothing at all

  #160 commented on Jan 28, 2025 • 0 new comments