search

microsoft / UniVL

An official implementation for " UniVL: A Unified Video and Language Pre-Training Model for Multimodal Understanding and Generation"
https://arxiv.org/abs/2002.06353
MIT License
339 stars 54 forks source link

Run Without Distributed #26

Open Maddy12 opened 2 years ago

Maddy12 commented 2 years ago

Hello, I am trying to run your code but I keep running into issues with the distributed learning. Is it possible to run without this?

ArrowLuo commented 2 years ago

Hi @Maddy12, what is your error when you run with distributed launch? It can be run on only one GPU. Otherwise, I think you should modify the code to the none distributed version.

Maddy12 commented 2 years ago

I apologize for wasting your time. There was an error but it was not in the distributed element. There is an error in the metrics.py in the function compute_metrics.

My hack is: 

if isinstance(x, list): 
        x = np.concatenate(x)
sx = np.sort(-x, axis=1)

I do not know if this expected behavior or if I am actually now implementing something incorrect.

ArrowLuo commented 2 years ago

Hi @Maddy12, what is the error? Can you print it here? Or you can test x= np.concatenate(tuple(x), axis=0) as follows,

if isinstance(x, list): 
    x= np.concatenate(tuple(x), axis=0)
sx = np.sort(-x, axis=1)