Closed GemmaTuron closed 1 year ago
Following the path mentioned in the error: output/root/eos/repository/eos5505/....../run_predict.sh
. I can see that in the run_predict file that the python path is being assigned wrong.
I attach the logs: Test in google colab: Fail to fetch Log fetch: Google_colab_log_fetch_eos5505.txt
Test in the CLI. Fetch successfully, serve and predict. Log fetch: log_fetch_eos5505.txt Log predict: log_predict_eos5505.log Output predictions: input file: "eml_canonicals.csv eos5505_output.csv
Hi @carcablop and @pauline-banye
The issue reported here and for the other models is due to this function from @carcablop pull request https://github.com/ersilia-os/ersilia/pull/525 def get_python_path_env(self, environment): if not self.exists(environment): raise Exception("{0} environment does not exist".format(environment)) python_path = subprocess.check_output("which python", shell=True).strip() index_env = python_path.decode("utf-8").find("envs/") + 5 python_path_to_envs = python_path.decode("utf-8")[0:index_env] return "{0}{1}/bin/python".format(python_path_to_envs, environment)
As you can see the slash is probably cut at python_path.decode("utf-8")[0:index_env]
.
We are now working on reformatting it.
We have modified the function here: https://github.com/ersilia-os/ersilia/commit/70bcf5469d912b86b469a3db9e2978f34ff7a1fe
The models are now working on Colab, can you try again?
Hola @carcablopy @pauline-banye
El problema informado aquí y para los otros modelos se debe a esta función de @carcablopsolicitud de extracción ersilia-os/ersilia#525 def get_python_path_env(yo, entorno): si no self.exists (entorno): generar excepción ("{0} entorno no existe". formato (entorno)) python_path = subprocess.check_output("cuál python", shell=True).strip() index_env = python_path.decode("utf-8").find("envs/") + 5 python_path_to_envs = python_path.decode("utf-8")[0:index_env] devolver "{0}{1}/bin/python".format(python_path_to_envs, entorno)
Como puede ver, la barra probablemente esté cortada en
python_path.decode("utf-8")[0:index_env]
. Ahora estamos trabajando en reformatearla.
Hello @GemmaTuron . Thank you very much for the information. I'm so sorry this is causing the error.
Update: Test in google colab: Fetch successfully, serve and predict. Log fetch: log_fetch_colab_eos5505.txt Output in colab: eos5505_output.csv
The model works on CLI and Colab. Artifacts:
CLI Output: eos5505_output.csv
Colab link: https://colab.research.google.com/drive/1jD2qTBAA3GUqD1-Iw5S_GuFmqC-pNjAN?usp=sharing
Perfect I'll mark this model as completed
Hi @GemmaTuron. Testing the model in Google Colab. I have the same error like other models (eos935d, eos4q1a). I think the path is being assigned wrong somewhere when fetching the model, the path should be *
/usr/local/envs/eos5505/bin/python
.