vilab-ucsd / geonet Goto Github PK
View Code? Open in Web Editor NEWRepository for accessing and training using GeoNet dataset, published at CVPR 2023.
License: MIT License
Repository for accessing and training using GeoNet dataset, published at CVPR 2023.
License: MIT License
Hi, do you have any plans to release the code for reimplementation of the supervised-pretrained ViT-L results presented in your paper (Fig. 1 (c) and Fig. 7)?
Using the example bash command to run a plain setting. But reported with RuntimeError: Function 'LogSoftmaxBackward0' returned nan values in its 0th output.
Here is the full log:
RUNDIR: runs/plain/plain
2023-10-06 19:24:13,056 INFO Start logging
2023-10-06 19:24:13,056 INFO Namespace(config='configs/plain.yml', source='usa', target='asia', lr_rate=0.003, num_class=205, data_root='/remote-home/GeoNet/', json_dir='/remote-home/GeoNet/geoPlaces_metadata.json', trainer='plain', num_iter=100000, batch_size=32, resume=None, exp_name='plain')
2023-10-06 19:24:19,882 INFO train: 178110
2023-10-06 19:24:23,843 INFO test: 17234
2023-10-06 19:24:28,758 INFO train: 187426
2023-10-06 19:24:32,661 INFO test: 26923
2023-10-06 19:24:34,710 INFO Using sgd optimizer
2023-10-06 19:24:34,711 INFO Start training from iteration 0
2023-10-06 19:24:34,711 INFO Using multiple GPUs
Iteration: 0
Iteration: 1
Iteration: 2
Iteration: 3
Iteration: 4
Iteration: 5
/root/anaconda3/envs/openclip/lib/python3.10/site-packages/torch/autograd/__init__.py:200: UserWarning: Error detected in LogSoftmaxBackward0. Traceback of forward call that caused the error:
File "/root/GeoNet/train.py", line 271, in <module>
main()
File "/root/GeoNet/train.py", line 153, in main
trainer(batch_iterator, model_fe, model_cls, *d_list, opt, it, *criterion_list,
File "/root/GeoNet/UDA_trainer/plain.py", line 16, in train_plain
loss = torch.mean(criterion_cls(output_src, lbl_src).squeeze())
File "/root/anaconda3/envs/openclip/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1501, in _call_impl
return forward_call(*args, **kwargs)
File "/root/GeoNet/losses/loss.py", line 27, in forward
return F.cross_entropy(input, target, weight=weight, reduction=self.reduction)
File "/root/anaconda3/envs/openclip/lib/python3.10/site-packages/torch/nn/functional.py", line 3029, in cross_entropy
return torch._C._nn.cross_entropy_loss(input, target, weight, _Reduction.get_enum(reduction), ignore_index, label_smoothing)
(Triggered internally at ../torch/csrc/autograd/python_anomaly_mode.cpp:114.)
Variable._execution_engine.run_backward( # Calls into the C++ engine to run the backward pass
Traceback (most recent call last):
File "/root/GeoNet/train.py", line 271, in <module>
main()
File "/root/GeoNet/train.py", line 153, in main
trainer(batch_iterator, model_fe, model_cls, *d_list, opt, it, *criterion_list,
File "/root/GeoNet/UDA_trainer/plain.py", line 19, in train_plain
loss.backward()
File "/root/anaconda3/envs/openclip/lib/python3.10/site-packages/torch/_tensor.py", line 487, in backward
torch.autograd.backward(
File "/root/anaconda3/envs/openclip/lib/python3.10/site-packages/torch/autograd/__init__.py", line 200, in backward
Variable._execution_engine.run_backward( # Calls into the C++ engine to run the backward pass
RuntimeError: Function 'LogSoftmaxBackward0' returned nan values in its 0th output.
Hi u, I downloaded GeoPlaces Dataset to test. But I found this dataset has a different number of classes than the number of classes you report on paper (205 classes). Specifically, Asia_train and Usa_train have 204 classes, Asia_test has 203 classes and Usa_test has 201 classes.
Thank you.
A declarative, efficient, and flexible JavaScript library for building user interfaces.
๐ Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. ๐๐๐
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google โค๏ธ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.