-
模型:se-resnext 动态图。 https://github.com/PaddlePaddle/models/tree/develop/dygraph/se_resnext
在第一轮至第二轮训练时候,unique_name会增加1,之后不会增加了。这个情况可能引发内存泄露,需要排查一下原因。
![image](https://user-images.githubuserconte…
-
According to your config at Faster RCNN, A mistake is encountered when I train Cascade RCNN. s0.loss_rpn_reg in the log information always is 0.....
![image](https://user-images.githubusercontent.c…
-
ResNeXt has been recently introduced in the code. Current implementation takes advantage of the similarities with Resnet, by enhancing the Resnet class to support ResNeXt topology as well.
While the …
barrh updated
5 years ago
-
Now that the pull request [128](https://github.com/Cadene/pretrained-models.pytorch/pull/128) been merged, pls add `resnext101_32x4d` and `resnext101_64x4d` in `fastai/vision/models/cadene_models.py`…
-
Hi,
I have been recently trying to load the [birds dataset](http://www.vision.caltech.edu/visipedia/CUB-200-2011.html) using dask.
Here is a [link to download](http://www.vision.caltech.edu/visipe…
-
I have built a retinanet cfg, resnext152 is the backbone that I have used, now when I am using the resnext's pretrained weights, my loss is becoming nan in the first iteration only. So, I think that t…
-
Hi, would you like to share your pre-trained model that can be finetuned for both detection and classification?
Thanks.
-
Your open source content is very good. but I met some questions when I try to run the program according to your instructions,I find that my results are generally different from any experimental resul…
-
Thank you very much for your work @craston. We really appreciate it.
I tested the model of 'RGB_HMDB_51_16f' in HMDB split 1. But I just got the accuracy of 55.7% which is 66.7% in your paper. The…
-
Attempting to save a model with torch.save(), without using a state_dict leads to the following error:
5 torch.save(myModel,'/data/users/raghuraman/logs/quantized_resnext.pt')
6
7 # …