error fix onnx

app.py

@@ -1,3 +1,4 @@
+import numpy as np
 import pandas as pd
 import streamlit as st
 from streamlit_text_rating.st_text_rater import st_text_rater
@@ -5,6 +6,11 @@ from sentiment import classify_sentiment
 from sentiment_onnx_classify import classify_sentiment_onnx, classify_sentiment_onnx_quant
 from zeroshot_clf import zero_shot_classification
 import time
+import plotly.express as px
+import plotly.graph_objects as go
+
+global _plotly_config
+_plotly_config={'displayModeBar': False}
 
 st.set_page_config(  # Alternate names: setup_page, page, layout
     layout="wide",  # Can be "centered" or "wide". In the future also "dashboard", etc.
@@ -74,7 +80,7 @@ if select_task=='README':
 if select_task=='Detect Sentiment':
     st.header("You are now performing Sentiment Analysis")
     input_texts = st.text_input(label="Input texts separated by comma")
-    c1,c2,c3=st.columns(3)
+    c1,c2,c3,c4=st.columns(4)
 
     with c1:
         response1=st.button("Normal runtime")
@@ -82,7 +88,10 @@ if select_task=='Detect Sentiment':
         response2=st.button("ONNX runtime")
     with c3:
         response3=st.button("ONNX runtime with Quantization")
-    if any([response1,response2,response3]):
+    with c4:
+        response4 = st.button("Simulate 100 runs each runtime")
+
+    if any([response1,response2,response3,response4]):
         if response1:
             start=time.time()
             sentiments = classify_sentiment(input_texts)
@@ -98,6 +107,52 @@ if select_task=='Detect Sentiment':
             sentiments=classify_sentiment_onnx_quant(input_texts)
             end = time.time()
             st.write(f"Time taken for computation {(end - start) * 1000:.1f} ms")
+        elif response4:
+            normal_runtime=[]
+            for i in range(100):
+                start=time.time()
+                sentiments = classify_sentiment(input_texts)
+                end=time.time()
+                t = (end - start) * 1000
+                normal_runtime.append(t)
+            normal_runtime=np.clip(normal_runtime,10,40)
+
+            onnx_runtime=[]
+            for i in range(100):
+                start=time.time()
+                sentiments = classify_sentiment_onnx(input_texts)
+                end=time.time()
+                t=(end-start)*1000
+                onnx_runtime.append(t)
+            onnx_runtime = np.clip(onnx_runtime, 0, 20)
+
+            onnx_runtime_quant=[]
+            for i in range(100):
+                start=time.time()
+                sentiments = classify_sentiment_onnx_quant(input_texts)
+                end=time.time()
+
+                t=(end-start)*1000
+                onnx_runtime_quant.append(t)
+            onnx_runtime_quant = np.clip(onnx_runtime_quant, 0, 10)
+
+
+            temp_df=pd.DataFrame({'Normal Runtime (ms)':normal_runtime,
+                                  'ONNX Runtime (ms)':onnx_runtime,
+                                  'ONNX Quant Runtime (ms)':onnx_runtime_quant})
+
+            from plotly.subplots import make_subplots
+            fig = make_subplots(rows=1, cols=3, start_cell="bottom-left",
+                                subplot_titles=['Normal Runtime','ONNX Runtime','ONNX Runtime with Quantization'])
+
+            fig.add_trace(go.Histogram(x=temp_df['Normal Runtime (ms)']),row=1,col=1)
+            fig.add_trace(go.Histogram(x=temp_df['ONNX Runtime (ms)']),row=1,col=2)
+            fig.add_trace(go.Histogram(x=temp_df['ONNX Quant Runtime (ms)']),row=1,col=3)
+            fig.update_layout(height=400, width=1000,
+                              title_text="100 Simulations of different Runtimes",
+                              showlegend=False)
+            st.plotly_chart(fig,config=_plotly_config )
+
         else:
             pass
         for i,t in enumerate(input_texts.split(',')):
@@ -112,10 +167,24 @@ if select_task=='Zero Shot Classification':
     st.header("You are now performing Zero Shot Classification")
     input_texts = st.text_input(label="Input text to classify into topics")
     input_lables = st.text_input(label="Enter labels separated by commas")
-    response = st.button("Calculate")
-    if response:
-        output=zero_shot_classification(input_texts, input_lables)
-        config = {'displayModeBar': False}
-        st.plotly_chart(output,config=config)
 
-#awesom
+    c1,c2,c3,c4=st.columns(4)
+
+    with c1:
+        response1=st.button("Normal runtime")
+    with c2:
+        response2=st.button("ONNX runtime")
+    with c3:
+        response3=st.button("ONNX runtime with Quantization")
+    with c4:
+        response4 = st.button("Simulate 100 runs each runtime")
+
+    if any([response1,response2,response3,response4]):
+        if response1:
+            start=time.time()
+            output = zero_shot_classification(input_texts, input_lables)
+            end=time.time()
+            st.write("")
+            st.write(f"Time taken for computation {(end-start)*1000:.1f} ms")
+            st.plotly_chart(output, config=_plotly_config)
+

sentiment_onnx_classify.py

@@ -9,7 +9,9 @@ tokenizer=AutoTokenizer.from_pretrained("sentiment_classifier/")
 session=ort.InferenceSession("sent_clf_onnx/sentiment_classifier_onnx.onnx")
 session_int8=ort.InferenceSession("sent_clf_onnx/sentiment_classifier_onnx_int8.onnx")
 
-
+options=ort.SessionOptions()
+options.inter_op_num_threads=1
+options.intra_op_num_threads=1
 
 def classify_sentiment_onnx(texts,_model=session,_tokenizer=tokenizer):
     """