Groups

Visual Commonsense Reasoning

Contact owners and managers

1–30 of 44

welcome to the VCR google group! Here is the place to discuss the dataset, ask/answer questions, and share awesome things that you find.

0 selected

3/31/25

Starting the Project kindly help

I have cloned the github repository for the project and downloaded the dataset. How to go to the next

unread,

Starting the Project kindly help

I have cloned the github repository for the project and downloaded the dataset. How to go to the next

3/31/25

6/11/24

Tag Information

Hi Authors, I am wondering if there is a way to replicate the tags that is shown in the paper.

unread,

Tag Information

Hi Authors, I am wondering if there is a way to replicate the tags that is shown in the paper.

6/11/24

Xuejiao Tang, … xinyan wang4

11/29/21

Error an torchvision

Hi,do you solve the problem yet？I have the same problem，and my pytorch is 1.0.1,torchvision is 0.2.0

unread,

Error an torchvision

Hi,do you solve the problem yet？I have the same problem，and my pytorch is 1.0.1,torchvision is 0.2.0

11/29/21

amingf...@gmail.com, … Rowan Zellers4

9/10/21

Sure! Just make up dummy predictions for the rationales task and submit those. You'll then need

unread,

Sure! Just make up dummy predictions for the rationales task and submit those. You'll then need

9/10/21

Ziwei Qin, Rowan Zellers2

9/10/21

Questions about the performance of pretrained multi-modal models.

hi Ziwei, sorry for the delay :) My hunch is that since all of these models are finetuned, they can

unread,

Questions about the performance of pretrained multi-modal models.

hi Ziwei, sorry for the delay :) My hunch is that since all of these models are finetuned, they can

9/10/21

Reza Esfandiarpoor, Rowan Zellers3

5/11/21

Replicating Q-A results with the shared checkpoints

Thanks. I check the splits again and also try with this specific version of PyTorch. Thanks, Reza On

unread,

Replicating Q-A results with the shared checkpoints

Thanks. I check the splits again and also try with this specific version of PyTorch. Thanks, Reza On

5/11/21

牛国成, Rowan Zellers2

5/10/21

what is the relationship between http://visualcommonsense.com/explore/?im=2518 and https://leaderboard.allenai.org/vcr/submissions/get-started

hey! Sorry for the confusion, visualcommonsense.com/leaderboard is the official VCR leaderboard, and

unread,

what is the relationship between http://visualcommonsense.com/explore/?im=2518 and https://leaderboard.allenai.org/vcr/submissions/get-started

hey! Sorry for the confusion, visualcommonsense.com/leaderboard is the official VCR leaderboard, and

5/10/21

S OB, Rowan Zellers3

5/5/21

setting up conda env in google colab?

Hi, I didn't realise you could use something other than conda :) thanks for pointing that out.

unread,

setting up conda env in google colab?

Hi, I didn't realise you could use something other than conda :) thanks for pointing that out.

5/5/21

Arjun Singh, Rowan Zellers3

4/21/21

Unused variables: q_obj_reps, a_obj_reps

Hi Rowan, Thank you for your prompt response. Kind regards, Arjun On Wednesday, April 21, 2021 at 4:

unread,

Unused variables: q_obj_reps, a_obj_reps

Hi Rowan, Thank you for your prompt response. Kind regards, Arjun On Wednesday, April 21, 2021 at 4:

4/21/21

Usha Varshini Saravanan

4/6/21

Checkpoint Usage

Hi! Could you please tell me how to use your checkpoints on Ubuntu OS? Thank you!

unread,

Checkpoint Usage

Hi! Could you please tell me how to use your checkpoints on Ubuntu OS? Thank you!

4/6/21

3/11/21

Running out of memory whilst precomputing representations

Hi, I have been trying to generate the precomputed representations for the training data using the

unread,

Running out of memory whilst precomputing representations

Hi, I have been trying to generate the precomputed representations for the training data using the

3/11/21

Alessandro Suglia, Rowan Zellers2

1/15/21

Code for adversarial matching procedure

thanks Alessandro! See here for a starting point however I'm not quite sure adversarial matching

unread,

Code for adversarial matching procedure

thanks Alessandro! See here for a starting point however I'm not quite sure adversarial matching

1/15/21

lsci...@gmail.com, Rowan Zellers4

12/11/20

A image is destroyed

I just redownloaded / unzipped all the images to double check, but was able to find the file (and the

unread,

A image is destroyed

I just redownloaded / unzipped all the images to double check, but was able to find the file (and the

12/11/20

Chao Chen, Rowan Zellers2

12/8/20

Data generation script

Sorry for the delay, the model is "X-152-32x8d-FPN-IN5k" from the Detectron model zoo:

unread,

Data generation script

Sorry for the delay, the model is "X-152-32x8d-FPN-IN5k" from the Detectron model zoo:

12/8/20

Jaemin Cho, Rowan Zellers4

11/9/20

Submission format

I see. Thanks for the clarification! On Monday, November 9, 2020 at 4:59:45 PM UTC-5 Rowan Zellers

unread,

Submission format

I see. Thanks for the clarification! On Monday, November 9, 2020 at 4:59:45 PM UTC-5 Rowan Zellers

11/9/20

bonny...@gmail.com, Rowan Zellers3

11/3/20

About Adversarial Matching Details

Hello, Your reply and code are very helpful. Thank you. 2020년 11월 4일 수요일 오전 4시 10분 4초 UTC+9에 Rowan

unread,

About Adversarial Matching Details

Hello, Your reply and code are very helpful. Thank you. 2020년 11월 4일 수요일 오전 4시 10분 4초 UTC+9에 Rowan

11/3/20

Ali TT, rowan....@gmail.com2

6/19/20

hey Xuejiao, I think the issue might be if you use one GPU (rather than the three I used), your batch

unread,

hey Xuejiao, I think the issue might be if you use one GPU (rather than the three I used), your batch

6/19/20

yinxiy...@gmail.com, Rowan Zellers6

4/15/20

How to test after the training

Thank you for your reply. The data for the test is calculated in the paper, and there is test.jsonl

unread,

How to test after the training

Thank you for your reply. The data for the test is calculated in the paper, and there is test.jsonl

4/15/20

Ali TT, Rowan Zellers2

3/30/20

Hi there, See the "Reasoning" paragraph in the VCR paper. We max-pool the three

unread,

Hi there, See the "Reasoning" paragraph in the VCR paper. We max-pool the three

3/30/20

deanpl...@gmail.com, … rowan....@gmail.com7

2/13/20

Input embedding for the ablation experiments

Sure! try * https://s3-us-west-2.amazonaws.com/ai2-rowanz/r2c/bert_endingonly_answer_test.h5 * https:

unread,

Input embedding for the ablation experiments

Sure! try * https://s3-us-west-2.amazonaws.com/ai2-rowanz/r2c/bert_endingonly_answer_test.h5 * https:

2/13/20

2/3/20

pip install git+git://github.com/pytorch/vision.git@24577864e92b72f7066e1ed16e978e873e19d13d install slow and file multiatt/default.json not found

hi ,author I so sorry to disturb you... When I start , I find if I want to use torchvision (from the

unread,

pip install git+git://github.com/pytorch/vision.git@24577864e92b72f7066e1ed16e978e873e19d13d install slow and file multiatt/default.json not found

hi ,author I so sorry to disturb you... When I start , I find if I want to use torchvision (from the

2/3/20

yinxiy...@gmail.com, Rowan Zellers2

1/21/20

I cannot install torchvision by readme method. (Windows os)

hi there! Sorry, I don't have a windows OS (and this is a torchvision problem) so I'm sure

unread,

I cannot install torchvision by readme method. (Windows os)

hi there! Sorry, I don't have a windows OS (and this is a torchvision problem) so I'm sure

1/21/20

Xuejiao Tang, Rowan Zellers2

1/16/20

question about contextualization and reasoning code

Hi Xuejiao, Yeah, the qa_similarity is the r_{i}Wq_{i}. The weight W is in the span_attention module.

unread,

question about contextualization and reasoning code

Hi Xuejiao, Yeah, the qa_similarity is the r_{i}Wq_{i}. The weight W is in the span_attention module.

1/16/20

MUGE KURAL, Rowan Zellers2

1/7/20

Stored image features

Hi Muge, We don't, sorry! The model I used is built off faster rcnn, which requires localized

unread,

Stored image features

Hi Muge, We don't, sorry! The model I used is built off faster rcnn, which requires localized

1/7/20

Xuejiao Tang, Rowan Zellers2

1/6/20

question about QA-R

hi Xuejiao, Does it mean for QA-R task, the input is question+one correct answer+4 rationale choices?

unread,

question about QA-R

hi Xuejiao, Does it mean for QA-R task, the input is question+one correct answer+4 rationale choices?

1/6/20

Xuejiao Tang, Rowan Zellers3

12/16/19

training answers for about 50 hours...it's still running

Hi Rowan, thanks for your answer :):):):):) Xuejiao

unread,

training answers for about 50 hours...it's still running

Hi Rowan, thanks for your answer :):):):):) Xuejiao

12/16/19

Xuejiao Tang, Jize Cao5

12/6/19

fid = h5f.open(name, flags, fapl=fapl) File "h5py/_objects.pyx", line 54, in h5py._objects.with_phil.wrapper File "h5py/_objects.pyx", line 55, in h5py._objects.with_phil.wrapper File "h5py/h5f.pyx", line 85, in h5py.h5f.open OSError: Unable to

Thanks for your reply, I tried to re-download several times before, but it doesn't work. I will

unread,

fid = h5f.open(name, flags, fapl=fapl) File "h5py/_objects.pyx", line 54, in h5py._objects.with_phil.wrapper File "h5py/_objects.pyx", line 55, in h5py._objects.with_phil.wrapper File "h5py/h5f.pyx", line 85, in h5py.h5f.open OSError: Unable to

Thanks for your reply, I tried to re-download several times before, but it doesn't work. I will

12/6/19

11/21/19

Question about allennlp.common.util.ScatterableList issue

Problem solved with handling dictionary key 'metadata' Does this happens with only 1 GPU

unread,

Question about allennlp.common.util.ScatterableList issue

Problem solved with handling dictionary key 'metadata' Does this happens with only 1 GPU

11/21/19

woosik Yang, rowan....@gmail.com3

11/13/19

Question about running the code

Thanks now it works! :)

unread,

Question about running the code

Thanks now it works! :)

11/13/19

apo...@iiitd.ac.in, rowan....@gmail.com3

11/5/19

Evaluating models on leaderboard

Hey, great. Yes, I will work with val set as test set is unlabelled.

unread,

Evaluating models on leaderboard

Hey, great. Yes, I will work with val set as test set is unlabelled.

11/5/19

Search

Clear search

Close search

Google apps

Main menu