try model

app.py

@@ -1,9 +1,15 @@
-from transformers import pipeline
+## SETUP #####################################################################################################################
+from transformers import AutoTokenizer, AutoModelForSequenceClassification
+import torch
 import streamlit as st
 import pandas as pd
 from PIL import Image
 import os
 
+# load model
+model = AutoModelForSequenceClassification.from_pretrained("amandakonet/climatebert-fact-checking")
+tokenizer = AutoTokenizer.from_pretrained("amandakonet/climatebert-fact-checking")
+
 # read in example
 ex_claims = ['Global warming is driving polar bears toward extinction',
  'Global warming is driving polar bears toward extinction',
@@ -85,6 +91,9 @@ ex_labels = ['NOT_ENOUGH_INFO',
 
 ex_df = pd.DataFrame({'claim' : ex_claims, 'evidence' : ex_evidence, 'label': ex_labels})
 
+
+## TEXT ######################################################################################################################
+
 # title
 st.title('Combatting Climate Change Misinformation with Transformers')
 
@@ -165,8 +174,9 @@ st.markdown("* evidence *refutes* (contradicts) claim")
 st.markdown("* evidence *does not provide enough info to support or refute* (neutral) claim")
 st.markdown("For this project, I fine-tune ClimateBERT (4) on the text entailment task")
 
+## EXAMPLE ###################################################################################################################
+
 st.markdown("## Try it out!")
-txt_class = pipeline('text-classification', model='amandakonet/climatebert-fact-checking', use_auth_token = os.environ["hf_token"])
 
 # select climate claim
 option_claim = st.selectbox('Select a climate claim to test', ex_df['claim'].unique())
@@ -181,6 +191,21 @@ st.write('You selected:', option_evidence)
 
 st.write(type(option))
 
+st.markdown("Now, we can use your selected (claim,evidence) pair in the fine-tuned transformer!")
+
+
+# tokenize
+features = tokenizer(option_claim, option_evidence,  
+                   padding='max_length', truncation=True, return_tensors="pt", max_length=512)
+
+# set model into eval
+model.eval()
+with torch.no_grad():
+   scores = model(**features).logits
+   label_mapping = ['contradiction', 'entailment', 'neutral']
+   labels = [label_mapping[score_max] for score_max in scores.argmax(dim=1)]
+   print("The claim: ", option_claim, "is ", labels ", by ", option_evidence)
+
 # section 6: analysis
 st.markdown("## Critical Analysis")
 st.markdown("What else could we do?")