updated readme

readme.md

@@ -133,4 +133,40 @@ For instance, when looking at the structure of the baseline model, you will real
 
 Also, looking at the systems submitted to the [DCASE Challenge](http://dcase.community) for the urban sound tagging task will probably give you ideas. For instance, you can have a look at the [submission of Bongjun Kim](http://dcase.community/documents/challenge2019/technical_reports/DCASE2019_Kim_107.pdf), which obtained the 3rd best score at the [DCASE 2019 Challenge, task 5](http://dcase.community/challenge2019/task-urban-sound-tagging) using transfer learning from VGGish.
 
-  
+
+## Trouble shooting
+
+- **I have to wait for ages when I ask for a MyDocker environment** 
+
+  Having access to a GPU environment with MyDocker can be difficult when it's not during the scheduled class hours. In general, you will have to wait 20-30 minutes outside of scheduled sessions, but sometimes (especially during teaching hours) all GPU resources are booked and you can simply not access a GPU with MyDocker. In that case, you might want to work on Google Colab.
+
+  To work on Google Colab for this project, you first have to upload the content of the above Gitlab repository in your Google Drive. For instance, in my Google Drive, I put all resources at ```Enseignement/2024-2025/UST-project-google-colab```. 
+
+  Then you can open any notebook in the baseline folder using Google Colab (e.g., ```setup.ipynb```) and add a code cell at the top with the following piece of code (of course, adapt the ```root_path``` variable to you Google Drive organization):
+
+  ``` 
+  from google.colab import drive
+  import os
+
+  drive.mount('/content/drive')
+
+  root_path = "/content/drive/My Drive/Enseignement/2024-2025/UST-project-google-colab"
+
+  os.chdir(root_path)
+
+  # print("----------")
+  # !pwd
+  # print("----------")
+  # !ls
+  # print("----------")
+
+  !pip install -r requirements.txt
+  ```
+
+  Do not forget to also change the variable ```root_path``` in ```paths.py```.
+
+  You will have to set things up (folders creation, dataset download, data preparation, etc.) once, and then you will be able to work on your model. Then, you should be able to alternate between MyDocker and Google Colab by simply transfering your notebook between the two platforms.
+
+- **I received an error saying my GPU is too oldr** 
+
+  If you receive an error message saying that your GPU is too old, try to disconnect and reconnect to MyDocker. There is an old Nvidia GTX 1080 Ti in the pool, normally you should work with a V100 (you can execute ```nvidia-smi``` to display the GPU characteristics).
\ No newline at end of file