Data Science Stack Exchange
Data Science Stack Exchange

Questions tagged [deployment]

15 questions
5
votes
2 answers

Do model training pipeline should run on dev, staging and production environment?

I know it's a best practice to ship our code from dev to staging to production by including different level tests and validations that will help to confidently deploy on the production environment. But, for the ML models, data scientists will first…
shaik moeed
  • 151
  • 7
2
votes
1 answer

Ways to share Pytorch model without revealing architecture?

We are trying to give a model to collaborators but would like to protect the IP. What are some ways to encrypt/hide/compile the definition when sharing a trained model?
illan
  • 21
  • 1
1
vote
1 answer

What is the best approach to deploy N number of ML models as a scalable service in the Cloud?

I've N (~50) number of sentiment models of different languages, which were fine tuned on HggingFace's transformer models. Each of the models as 2-3 GB in size approx. Now, how can I deploy all these sentiment models as a scalable service in a cloud…
Abdullah Al Imran
  • 129
  • 1
  • 4
1
vote
0 answers

How is model scheduling set up in practice?

I have been working on various machine learning models so far, but never yet on the deployment phase of an ML project. I have vaguely used Apache Airflow and I'm aware that it is a tool for scheduling DAGs, but I never set up such a scheduling on…
lazarea
  • 289
  • 1
  • 11
1
vote
0 answers

FileNotFoundError: Unsuccessful TensorSliceReader constructor

I am trying to deploy my model. I am encountering the following problem: FileNotFoundError: Unsuccessful TensorSliceReader constructor: Failed to find any matching files for ram://a603e930-4fda-4105-8554-7af5e5fc02f5/variables/variables You may be…
Karthik Bhandary
  • 111
  • 5
1
vote
1 answer

How to handle categorical feature engineering in ML production?

I have a classification dataset ,where I have a lot of categorical columns . I have one hot encoded ie. dummy variables in my training . How to handle this in production side of ML. There are cases in which there is drift in data in future datasets…
Shubh
  • 198
  • 6
0
votes
0 answers

Deploying a model with GPU and pay-per-inference

I may have the wrong stack exchange. If that's the case, could someone point me to a stack that could help with this. Anyways... My backend employs a sentence transformer model from HuggingFace. Since the number of requests per day is small,…
AlwaysLearning
  • 131
  • 2
0
votes
0 answers

How to deploy a machine learning model in the cloud with API access?

I want to deploy a machine learning model (more precisely stable diffusion) in the cloud and make it accessible via an API. I am relatively new to this topic, so I am not quite sure what services there are and what might best fit my requirements.…
Fabian Riedlsperger
  • 1
  • 1
0
votes
0 answers

Can accuracy improve when there is evidence of domain-shift between training and deployment?

A model for image analysis was trained using data captured with imaging system A. I then deployed the model on imaging system B. System B has better image contrast than system A. Features in the last fully connected layer indicate a domain-shift…
Dav95
  • 1
0
votes
0 answers

How to use early_stopping_rounds in the Final Model? (CatBoost example with Optuna)

Imagine we have a model in the sklearn pipeline: # pipeline for numerical data num_preprocessing = Pipeline([('num_imputer', SimpleImputer(strategy='mean')), # imputing with mean ('minmaxscaler', MinMaxScaler())]) #…
Freejack
  • 1
  • 1
0
votes
2 answers

Which random_state to use in test_train_split when deploying final model?

I have developed a Random Forest that gives varying results depending on the random state of the test train split. This is normal, because a lot of the values in the data are extreme, without being actual outliers. When saving the model for future…
Nemo_the_scientist
  • 469
  • 11
0
votes
1 answer

After experimentation, how do we learn a final model for deployment?

I have a question regarding learning a "final" model for deployment. Assuming my task is a classification one, my workflow during experimentation is as follows. Get the data: X, y Split the data: X_tr, y_tr (training), X_tu, y_tu (tuning) and X_te,…
Jane Wayne
  • 171
  • 3
0
votes
0 answers

ValueError: dtype='numeric' is not compatible with arrays of bytes/strings.Convert your data to numeric values explicitly instead

I am new to machine learning model deployment. In a sequence, local system rises an error while running "app.py" file, that is: "ValueError: dtype='numeric' is not compatible with arrays of bytes/strings.Convert your data to numeric values…
shrikanth Arunagiri
  • 1
  • 1
  • 1
0
votes
1 answer

Can I use Population Stability Index (PSI) when observations have multiple variables?

I understand from resources like this one that the Population Stability Index (PSI) can be used to test for data drift when a machine learning model is in production. However, the resources I have looked at describe PSI in terms of a single…
Fijoy Vadakkumpadan
  • 113
  • 4
0
votes
0 answers

Shadow deployment vs batch prediction on log?

In shadow deployment we 1) deploy a new model in parallel with the existing one; 2) route every incoming request to both the existing and candidate model; 3) serve the existing model's predictions for users but only log the candidate's for…
Lerner Zhang
  • 496
  • 3
  • 10