Upload app.py

Revised commit

app.py

@@ -32,6 +32,8 @@ checkpoint_1 = "Highway/SubCat"
 
 checkpoint_2 = "Highway/ExtraOver"
 
+checkpoint_3 = "Highway/Conversion"
+
 
 @st.cache(allow_output_mutation=True)
 def load_model_1():
@@ -53,6 +55,16 @@ def load_tokenizer_2():
     return AutoTokenizer.from_pretrained(checkpoint_2)
 
 
+@st.cache(allow_output_mutation=True)
+def load_model_3():
+    return AutoModelForSequenceClassification.from_pretrained(checkpoint_3)
+
+
+@st.cache(allow_output_mutation=True)
+def load_tokenizer_3():
+    return AutoTokenizer.from_pretrained(checkpoint_3)
+
+
 st.set_page_config(
     page_title="Cost Data Classifier", layout= "wide", initial_sidebar_state="auto", page_icon="💷"
 )
@@ -160,7 +172,7 @@ if submitted:
             ),
             autosize=False,
             width=800,
-            height=1000,
+            height=500,
             xaxis_title="Likelihood of SubCatName",
             yaxis_title="SubCatNames",
             # legend_title="Topics"
@@ -247,3 +259,73 @@ if submitted:
         Prediction_confidence_2 = st.metric("Prediction confidence", (str(round(sorted_preds_2[0][1]*100, 1))+"%"))
 
         st.success("Great! ExtraOver successfully predicted. ", icon="✅")
+
+
+
+    # Third prediction
+
+    label_list_3 = ['0.04', '0.045', '0.05', '0.1', '0.15', '0.2', '1.0', '7.0', '166.67', 'Others']
+
+    joined_clean_sents = prep_text(Text_entry)
+
+    # tokenize
+    tokenizer_3 = load_tokenizer_3()
+    tokenized_text_3 = tokenizer_3(joined_clean_sents, return_tensors="pt")
+
+    # predict
+    model_3 = load_model_3()
+    text_logits_3 = model_3(**tokenized_text_3).logits
+    predictions_3 = torch.softmax(text_logits_3, dim=1).tolist()[0]
+    predictions_3 = [round(a_, 3) for a_ in predictions_3]
+
+    # dictionary with label as key and percentage as value
+    pred_dict_3 = (dict(zip(label_list_3, predictions_3)))
+
+    # sort 'pred_dict' by value and index the highest at [0]
+    sorted_preds_3 = sorted(pred_dict_3.items(), key=lambda x: x[1], reverse=True)
+
+    # Make dataframe for plotly bar chart
+    u_3, v_3 = zip(*sorted_preds_3)
+    x_3 = list(u_3)
+    y_3 = list(v_3)
+    df4 = pd.DataFrame()
+    df4['Conversion_factor'] = x_3
+    df4['Likelihood'] = y_3
+
+    e1, e2, e3 = st.columns([1.5, 0.5, 1])
+
+    with e1:
+        st.header("Conversion_factor")
+        # plot graph of predictions
+        fig = px.bar(df4, x="Likelihood", y="Conversion_factor", orientation="h")
+
+        fig.update_layout(
+            # barmode='stack',
+            template='ggplot2',
+            font=dict(
+                family="Arial",
+                size=14,
+                color="black"
+            ),
+            autosize=False,
+            width=800,
+            height=800,
+            xaxis_title="Likelihood of Conversion_factor",
+            yaxis_title="Conversion_factor",
+            # legend_title="Topics"
+        )
+
+        fig.update_xaxes(tickangle=0, tickfont=dict(family='Arial', color='black', size=14))
+        fig.update_yaxes(tickangle=0, tickfont=dict(family='Arial', color='black', size=14))
+        fig.update_annotations(font_size=14)  # this changes y_axis, x_axis and subplot title font sizes
+
+        # Plot
+        st.plotly_chart(fig, use_container_width=False)
+
+    with e3:
+        st.header("")
+        predicted_3 = st.metric("Predicted ExtraOver", sorted_preds_3[0][0])
+        Prediction_confidence_3 = st.metric("Prediction confidence",
+                                            (str(round(sorted_preds_3[0][1] * 100, 1)) + "%"))
+
+        st.success("Great! Conversion_factor successfully predicted. ", icon="✅")