Q-Future/Q-Align

Issues

• dataset for AIGCQA-30K-Image

  #25 opened 9 months ago by cyy-1234
  2

• bug encountered when running HF AutoModelForCausalLM

  #3 opened a year ago by asrlhhh
  4

• when try to use flash_attention_v2, not find `_prepare_4d_causal_attention_mask_for_sdpa`

  #40 opened 4 months ago by dongdk
  0

• how to prepare dataset for fine-tuning

  #21 opened a year ago by rokopi-byte
  2

• prediction score

  #11 opened a year ago by Nastu-Ho
  3

• about the GPU usage

  #16 opened a year ago by Gary-bk
  4

• Full Training from Start：CUDA out of memory.

  #35 opened 5 months ago by YUANMU227
  2

• how use the multi-GPU inference ，can you give me some demo code

  #29 opened 6 months ago by skydh
  4

• How Do I Perform Offline Inference? After a Model Is Downloaded to the Local Computer, the Model Cannot Be Executed?

  #30 opened 6 months ago by hjb25286
  1

• RuntimeError: Error(s) in loading state_dict for MPLUGOwl2LlamaForCausalLM: size mismatch for norm.weight: copying a param with shape torch.Size([768]) from checkpoint, the shape in current model is torch.Size([4096]). You may consider adding `ignore_mismatched_sizes=True` in the model `from_pretrained` method.

  #33 opened 6 months ago by style09131111
  1

• When I try to use koniq-10 for training, I encounter the problem that the loss is always 0.

  #37 opened 5 months ago by style09131111
  1

• 'MPLUGOwl2Config' object has no attribute 'mlp_bias'

  #31 opened 6 months ago by mattiasospetti
  2

• commercial use of q-align

  #39 opened 4 months ago by yasamin93
  0

• 🤗If u want to use transformers >= 4.40.2, follow my opt!

  #38 opened 4 months ago by CarlCloudWang
  0

• About the number of input video frames

  #36 opened 5 months ago by elvindp
  1

• Is this metric still depends on fixed transformers version?

  #34 opened 6 months ago by Luciennnnnnn
  0

• hello,how can we use the multi-GPU inference ?

  #32 opened 6 months ago by learn01one
  0

• About map the scores to the ranks

  #28 opened 8 months ago by w-qhai
  0

• Is it possible to reproduce this project on Florence-2 to reduce memory requirements and speed up inference?

  #27 opened 8 months ago by shiertier
  0

• .

  #26 opened 8 months ago by YUANMU227
  1

• Confused about dataset split.

  #23 opened 10 months ago by ZachL1
  5

• Questions about modeling_llama2

  #24 opened 10 months ago by lbc12345
  0

• Batch inference does not yield consistent results

  #22 opened 10 months ago by theicfire
  2

• _prepare_4d_causal_attention_mask_for_sdpa does not exist

  #20 opened a year ago by theicfire
  3

• about the encode and decode

  #15 opened a year ago by learn01one
  1

• About Training with LoRA

  #18 opened a year ago by Gary-bk
  1

• ddp is too slow

  #19 opened a year ago by lliee1
  1

• Program hangs when training from the start

  #7 opened a year ago by Teddy-Xiong
  8

• output decoding

  #14 opened a year ago by Nastu-Ho
  1

• How to train without huggingface_hub and offline?

  #13 opened a year ago by RollingWang
  1

• AttributeError when doing inference

  #12 opened a year ago by machengcheng2016
  1

• aesthetic assessment is sensitive to the size of the input image

  #10 opened a year ago by Nastu-Ho
  1

• NameError: name 'Cache' is not defined

  #9 opened a year ago by rokopi-byte
  1

• install error

  #6 opened a year ago by Hanbo-Sun
  1

• Scoring each video frame using Q-Align

  #8 opened a year ago by Nastu-Ho
  1

• Missing positional argument: 'layer_idx' when using QAlignVideoScorer

  #4 opened a year ago by johind
  2

• About the precision of the conversion of ava dataset

  #1 opened a year ago by Qiulin-W
  1