K

Hi @julian Jannes We are currently testing our model With the entire dataset using GPU with collab since the Gpu is supported using collab For now with two epochs the MSE Loss is significantly decreasing with the same model which shows model is learning well We are expecting a good score at the end of the training for this As I mentioned the training time is too much even on gpu which is 2 hours per epoch on a Tesla GPU P100 I will inform you by tomorrow the final R2 score we are getting after the training gets over .for submitting on the leaderboard we might have to take all the data which might be time consuming or we might have to change the metrics on the backend to pass the dimensionality error

Hi @stubbi Almost all the errors are resolved We are able to test get a score And even the uploading was working well But later on an error pops up saying dimensionality error due to a small subset of data samples taken during our training The metric or evaluation function backend takes into the whole data And if we take the whole data in our code It takes a lot of time for training

That error has been resolved

Hi @stubbi and @antoine any updates on what can be done

Hi @stubbi and @antoine After a lot of debugging and testing we are able to test and run the code using the aqora test command We are getting the same results which I mentioned above a R2 score of 81 After running the notebook with a subset of the original data which can seen in the image . Given that we took only a subset of data due to less compute power of our machine we were able to get a good score We believe if we take all the data there is a possibility of getting good score for the model but given that the training for the large amount of data can be time consuming which I believe is the issue we faced intially with each epoch taking a lot of time One more thing is that we were able to successfully Run and test our notebook and get a good score But there is still issues while uploading the file to the leaderboard Which we believe is due to evaluation and training metrics and data restrictions on the platform backend which takes into account all the data which is giving us errors like dimensionality mismatch We would like to know your thoughts on how to proceed from here Regards

Hi @stubbi @antoine and @paf We can send our submissions but Due to the extensive computational time required by our model, we attempted to run our code on a subset of the original dataset and successfully obtained the R² value : 81 .However, when we executed the same code using the “aqora test” command in the solution file, we encountered an error indicating a mismatch in the dimensions of the dataset. We understand why this error occurs: the testing command calculates the R² value (which is predefined in the backend) and any other metrics based on the entire dataset. Since our model only utilized a subset of the data, this discrepancy leads to the dimensionality mismatch error. What can be done now Any thing if you can suggest

Yes we did the CD in the directory

Hi @stubbi To explain the entire process We followed the H2 Ground state energy tutorial docs section for the intial testing and upload We successfully installed aqora cli using pip in the VS code We also have preinstalled all the up to date libraries and dependencies Then we tested and verified aqora using aqora help And succesfully logged in We were abled to download the templates and data files successfully Everything is working well at this point .now coming to the submission We are not able to find the directory or the solution notebook template for running and uploading the solution We are getting errors such as directory not found Project.toml not found

These are the errors we are getting . We want to know in detail about the submission Can you please guide us on how to test and upload the files .also how to write our code in the solution.ipynb file template in which cell And how to run it Thank you

Any updates on this @stubbi @paf @antoine