Hi,
Apologies for the simple question. I am new to Anvil.
I would like to use the Hugging Face Transformers library on the Anvil server so that I can use their pre-trained RobertA models to do sentiment analysis.
I could configure a Jupyter Notebook and use an uplink to Anvil.
The code would be something like …
# Install the dependencies inside the Jupyter Notebook
!pip install torch torchvision torchaudio
!pip install transformers
!pip install anvil-uplink
import anvil.server
anvil.server.connect(‘<ANVIL TOKEN>’)
# To get he ANVIL_TOKEN
# On Anvil [ ] Uplink [Enable]
# Then copy and paste code the string as the Anvil Token
from transformers import pipeline
# download model & use default pipeline
sent_pipeline = pipeline(“sentiment-analysis”)
# run the model
tinput = “I love Anvil”
sent_pipeline(tinput)
I would like to avoid the overhead of using the Jupyter Notebook and just have the Anvil server run the code.
What I have tried so far is:
In the settings - I have selected “Python 3.10 Beta”
and in the base packages I tried both Machine Learning & Data Science
Under “Add package” I tried to add “transformers” and it wasn’t available in the autocomplete.
Questions
- What settings do I need to use in Anvil so that the Anvil server can recognise the transformers library?
- I want to load the model only once when form is initiated. Where do I put this code so that it is only run once
Thanks
Chris
