ArrowLuo / CLIP4Clip

An official implementation for "CLIP4Clip: An Empirical Study of CLIP for End to End Video Clip Retrieval"

https://arxiv.org/abs/2104.08860

ArrowLuo/CLIP4Clip Issues

About 3D Patch Linear
Updated 3 months ago
How to download "cross_pytorch_model.bin" as pretrained weights used in the project? & problem in DDP
Closed 2 years ago12
Evaluation Procedure for Reporting Performance
Updated 4 months ago
post-pretrain or downstream tasks parameters
Updated 4 months ago
Question about the calculation method of loss when there are multiple gpus
Updated 6 months ago1
What do Pair, L, T stand for in the code?
Updated 7 months ago3
Problem about Multi-GPU-Train architecture
Updated 7 months ago
about the results on MSVD
Closed 2 years ago4
Reproduction on LSMDC DataSet
Updated 7 months ago6
ValueError: Parameter config in `CLIP4Clip(config)` should be an instance of class `PretrainedConfig`. To create a model from a Google pretrained model use `model = CLIP4Clip.from_pretrained(PRETRAINED_MODEL_NAME)`
Closed 7 months ago
Problem about reproducing the model
Closed a year ago1
Video-to-video retrieval?
Updated 10 months ago
AttributeError: Caught AttributeError in DataLoader worker process 0.
Updated 10 months ago
train dataset is shuffled regardless of seed
Updated a year ago
Results on MSRVTT and MSVD
Updated a year ago3
running the project
Closed a year ago1
Use my own videos
Updated a year ago1
How to use this repo to retrieve clips among videos by text?
Updated a year ago
Large performance drop if trained with fp32.
Updated a year ago1
loss NaN when training on MSRVTT
Closed a year ago1
train files
Updated a year ago1
Directly pass the entire config as a argument to a function
Updated a year ago
[PAD_TOKEN] is not used, but just adding 0
Updated a year ago
Evaluation batch
Updated a year ago
About mean_pooling on text sequence
Closed a year ago2
How to train and evaluate the model on the Training-7k split?
Updated a year ago1
run simple inference
Updated a year ago1
Some new issues about nccl : torch.distributed.elastic.multiprocessing.errors.ChildFailedError :
Closed 2 years ago3
About tightTrans and type embedding
Updated a year ago1
loss becomes nan
Closed 2 years ago5
Batch size issue concerned with Contrastive learning
Closed 2 years ago2
train on DiDeMo
Updated 2 years ago1
About Performance
Closed 2 years ago4
Poor performance when reproduce CLIP4clip(meanP) on ActivityNet
Closed 2 years ago14
how long have you been training
Updated 2 years ago1
MSVD Weights
Updated 2 years ago1
A potential bug in retrieval metric computing
Closed 2 years ago3
Data split issue
Closed 2 years ago5
Finetuning on custom dataset
Closed 2 years ago1
VIDIOC_REQBUFS: Inappropriate ioctl for device
Closed 2 years ago2
Log stopping problem with no response.
Closed 2 years ago1
subprocess.CalledProcessError
Closed 2 years ago9
Question about loss and batch size
Closed 2 years ago1
How about taking ViViT as video backbone
Closed 2 years ago2
You mentioned in the paper that the GPU is 8 V100, is it 16G or 32G?
Closed 2 years ago1
question about special token "[MASK]"
Closed 2 years ago1
use my own videos
Closed 2 years ago1
torch.distributed.init_process_group(backend="nccl") error and some other errors
Closed 2 years ago5
Question about video input shape
Closed 2 years ago2
the pretrained CLIP weights (VIT-B-16/32) is a TorchScript archive
Closed 2 years ago