Model Testing for EOS81EW

[pauline-banye]

Model Test

Hi @GemmaTuron

MODEL TEST FOR EOS81EW

I tested the model on my local environment and on the Ersilia CLI with the --repo_path option.

Result

• [x] Model eos81ew fetched, served and ran successfully on the CLI.
• [X] Model was tested using a list of smiles and a single smile string eos81ew_fetch.log eos81ew_list_run.csv eos81ew_run.csv

[GemmaTuron]

Hi @Femme-js and @DhanshreeA the model is ready for testing!

[AhmedYusuff]

Hi @GemmaTuron.

I ran the model several times on my Local environment, which is Ubuntu 22.04.

While fetching the model i got a KeyError: 'Input' error message.

I have attached the log file below

my.log

[GemmaTuron]

Hi @AhmedYusuff

Thanks for testing. While the final output of the log file is indeed KeyError: Input, you need to look through the log file to locate the real source of the error. In your case, that seems to be socket.timeout: The read operation timed out which points to an internet connection issue? try to do some googling / reading on that! Fyi and @pauline-banye , I've been able to successfully fetch and run the model in my Ubuntu 20.04

[AhmedYusuff]

Hello @GemmaTuron, I've been able to resolve the KeyError: Input error.

But I seem to have encountered another error Model API eos81ew:run did not produce an outputLoading PAMPA graph convolutional neural network model.

My log file is below new.log

[pauline-banye]

Hello @GemmaTuron, I've been able to resolve the KeyError: Input error.

But I seem to have encountered another error Model API eos81ew:run did not produce an outputLoading PAMPA graph convolutional neural network model.

My log file is below new.log

Hi @AhmedYusuff, I tested the model and it fetched successfully. eos81ew_fetch.log. Could you try running the fetch command again?

*Probably unrelated but its best you delete models that are not in use. run ersilia delete <name-of-model>

[AhmedYusuff]

Hi @pauline-banye, Looking through the error log, it seems to be having trouble calculating the Metadata on my end here.

Looking through the model repo

https://github.com/ersilia-os/eos81ew

The slug or model identifier I'm getting while fetching Pampa5 seems to be different from what you have parallel-artificial-membrane-permeability-assay-5

Could that be a problem?

[pauline-banye]

Hi @pauline-banye, Looking through the error log, it seems to be having trouble calculating the Metadata on my end here.

Looking through the model repo

https://github.com/ersilia-os/eos81ew

The slug or model identifier I'm getting while fetching Pampa5 seems to be different from what you have parallel-artificial-membrane-permeability-assay-5

Could that be a problem?

It could be actually. Let me make a PR to change this, then we can test the model again.

[GemmaTuron]

Hi @pauline-banye

I am working on reformatting the metadata for all the NCATS models. Please do not edit it as we are changing the airtable - metadata updates to fix a bug

[pauline-banye]

@AhmedYusuff I've made the PR, you can test again once @GemmaTuron merges it.

[GemmaTuron]

Hi @AhmedYusuff

Your error is at loading the model checkpoints. PermissionError: [Errno 13] Permission denied: '../../checkpoints'

Can you try cloning the model in your system, and then running the fetch command but adding the flag at the end --repo-path <path_to_cloned_repo>

@pauline-banye can provide more guidance on how to do this. In any case I've updated the metadata files as well

Thanks!

[pauline-banye]

Hi @pauline-banye

I am working on reformatting the metadata for all the NCATS models. Please do not edit it as we are changing the airtable - metadata updates to fix a bug

Sure @GemmaTuron, I understand.

[AhmedYusuff]

Hello @pauline-banye , Should I change the directory to the Model or I should just specify the path in --repo_path ?

[GemmaTuron]

Hello @pauline-banye , Should I change the directory to the Model or I should just specify the path in --repo_path ?

This might help @AhmedYusuff: docs

[AhmedYusuff]

Many thanks, @GemmaTuron.

From what I can see, it seems that I don't have the necessary permission to open some files. That's why I keep getting this error message.

@pauline-banye, Can you please make the "gccn_model.pt" file in the Checkpoints directory Executable for other users and groups, I did that with the repo clone on my system, but I'm still getting the same error.

[AhmedYusuff]

Hi @GemmaTuron and @pauline-banye . This problem seems related to Ubuntu 22.04.

I tested the Model on Google Colab and it was successful. I was able to Fetch, serve and Predict. Attach is the link.

https://colab.research.google.com/drive/177w3Vfsby2sjMn8bzo821rheb9hL_U7V?usp=sharing

[pauline-banye]

Many thanks, @GemmaTuron.

From what I can see, it seems that I don't have the necessary permission to open some files. That's why I keep getting this error message.

@pauline-banye, Can you please make the "gccn_model.pt" file in the Checkpoints directory Executable for other users and groups, I did that with the repo clone on my system, but I'm still getting the same error.

Hi @AhmedYusuff, the gc n_model.pt is the model data itself. Could you clarify why it needs to be executable?

[pauline-banye]

Hello @pauline-banye , Should I change the directory to the Model or I should just specify the path in --repo_path ?

Hi @AhmedYusuff could you share the command you specified to run the fetch command with the repo path flag?

[AhmedYusuff]

This is the command

Hello @pauline-banye , Should I change the directory to the Model or I should just specify the path in --repo_path ?

Hi @AhmedYusuff could you share the command you specified to run the fetch command with the repo path flag?

I used this command. ersilia -v fetch eos81ew --repo_path /home/control/eos81ew/

[AhmedYusuff]

Many thanks, @GemmaTuron. From what I can see, it seems that I don't have the necessary permission to open some files. That's why I keep getting this error message. @pauline-banye, Can you please make the "gccn_model.pt" file in the Checkpoints directory Executable for other users and groups, I did that with the repo clone on my system, but I'm still getting the same error.

Hi @AhmedYusuff, the gc n_model.pt is the model data itself. Could you clarify why it needs to be executable?

@pauline-banye , The permission denied error messages pointed me to ../../checkpoints

[AhmedYusuff]

Hello @pauline-banye , @GemmaTuron.

The issue has been resolved and Model Fetched and tested Successfully on Ubuntu 22.04.

The Model was trying to create a directory, But the User had insufficient permission, So I had to grant the user Privileges.

This is the fetch log. eos81ewfetch.log

The Model was tested with a simple smile string. eos81ewtest.csv

[pauline-banye]

Hello @pauline-banye , @GemmaTuron.

The issue has been resolved and Model Fetched and tested Successfully.

The Model was trying to create a directory, But the User had insufficient permission, So I had to grant the user Privileges.

This is the fetch log. eos81ewfetch.log

The Model was tested with a simple smile string. eos81ewtest.csv

That's good to hear @AhmedYusuff 😊. Good job debugging 👍🏻.

[AhmedYusuff]

Hello @pauline-banye , @GemmaTuron. The issue has been resolved and Model Fetched and tested Successfully. The Model was trying to create a directory, But the User had insufficient permission, So I had to grant the user Privileges. This is the fetch log. eos81ewfetch.log The Model was tested with a simple smile string. eos81ewtest.csv

That's good to hear @AhmedYusuff 😊. Good job debugging 👍🏻.

Many thanks @pauline-banye . I'm happy to be of help

[samuelmaina]

Used the 2 files(run and the list_run) provided by @pauline-banye . The model run successfully and provided the following out put. run_output.csv and list_run_output.csv

[GemmaTuron]

Perfect,

Thanks @AhmedYusuff and @samuelmaina for the work! I'll mark this issue as completed and the model as functional in our database 🎉