jayleicn/singularity

Issues

• Corruption Issue with ssv2 videos

  #31 opened a year ago by Hritikbansal
  1

• anet_qa acc is 0

  #29 opened a year ago by fake-warrior8
  0

• Probable GPU Memory Leak In Finetuning

  #30 opened a year ago by israwal
  0

• Discrepancy of 0.04% in ActivityNet-QA evaluation code

  #28 opened 2 years ago by israwal
  0

• Why add 1e-4 in softmax

  #27 opened 2 years ago by Andy1621
  4

• Preprocessing for DiDeMo

  #26 opened 2 years ago by Andy1621
  2

• Why do you train "msrvtt_mc" with "msrvtt_ret"?

  #23 opened 2 years ago by wonjiny
  1

• Downloading timeout

  #25 opened 2 years ago by gengyuanmax
  1

• Why does the text encoder forward/backward twice for VTM/VTC and MLM loss?

  #24 opened 2 years ago by Andy1621
  4

• ValueError: invalid literal for int() with base 10: 'attention'

  #22 opened 2 years ago by wonjiny
  0

• Data/model server offline

  #20 opened 2 years ago by YiLi-Intel
  7

• Could there be an unfair comparision?

  #21 opened 2 years ago by SCZwangxiao
  1

• Zero-shot performance on SSv2

  #8 opened 2 years ago by vateye
  1

• Different accuracy after evaluating on test set of MSRVTT-QA

  #17 opened 2 years ago by soumyasj
  7

• How many files were available during pre-training on WebVid+CC3M?

  #19 opened 2 years ago by prote376
  2

• State_dicts mismatch when trying to reproduce evaluation

  #18 opened 2 years ago by RealAntonVoronov
  2

• How to train it on Chinese video-text dataset?

  #6 opened 2 years ago by skyz8421
  1

• Results minival vs minival_emb

  #9 opened 2 years ago by avinashsai
  1

• Will using more frames in training be helpful

  #15 opened 2 years ago by FangmingZhou
  2

• About "Rerank dual-encoder results with cross-encoder“

  #14 opened 2 years ago by ruiyan1995
  4

• How to run Open-ended QA

  #16 opened 2 years ago by ShuangLI59
  4

• Apply to new dataset

  #11 opened 2 years ago by nikky4D
  4

• Is msrvtt_ret_train9k.json used anywhere?

  #13 opened 2 years ago by TheShadow29
  1

• Training Error

  #7 opened 2 years ago by yuanze-lin
  5

• Data pre-processing

  #10 opened 2 years ago by yuanze-lin
  3

• MSRVTT-MC Training and Evaluation

  #5 opened 2 years ago by vateye
  3

• typos in preprint paper

  #4 opened 2 years ago by FangmingZhou
  1

• Any configs for pretraining on 4 frames setting?

  #2 opened 2 years ago by vateye
  1

• High MLM probability during pretraining

  #1 opened 2 years ago by vateye
  1

• A question about whether to try other data sets.

  #3 opened 2 years ago by Decem-Y
  2