Add application file

app.py

@@ -1,20 +1,17 @@
 from fastapi import FastAPI
 from transformers import AutoTokenizer
 from optimum.onnxruntime import ORTModelForSequenceClassification
-from optimum.onnxruntime import ORTQuantizer
-from optimum.onnxruntime.configuration import AutoQuantizationConfig
+from optimum.pipelines import pipeline
 import time
 
-app = FastAPI()
-
-# Load the ONNX model and tokenizer
+# Initialize the tokenizer and ONNX model
 model_path = "./model_onnx/model_quantized.onnx"
-tokenizer_path = "./model_onnx"
-
+tokenizer = AutoTokenizer.from_pretrained("cardiffnlp/twitter-roberta-base-sentiment")
 model = ORTModelForSequenceClassification.from_pretrained(model_path)
-tokenizer = AutoTokenizer.from_pretrained(tokenizer_path)
 pipe = pipeline("text-classification", model=model, tokenizer=tokenizer)
 
+app = FastAPI()
+
 def preprocess_tweet(tweet: str) -> str:
     tweet_words = []
     for word in tweet.split(' '):
@@ -31,34 +28,27 @@ def home():
 
 @app.get("/analyze")
 def analyze_sentiment(tweet: str):
-    # Preprocess the tweet
     tweet_proc = preprocess_tweet(tweet)
-
-    # Measure the time taken for the inference
+    
     start_time = time.time()
     
-    # Use the pipeline to get the sentiment analysis result
     results = pipe(tweet_proc, return_all_scores=True)
     
-    # Calculate the inference time
     inference_time = time.time() - start_time
 
-    # Map the labels to desired names
     label_map = {
         "LABEL_0": "Negative",
         "LABEL_1": "Neutral",
         "LABEL_2": "Positive"
     }
 
-    # Find the label with the highest score
     highest_score_result = max(results[0], key=lambda x: x['score'])
     highest_label = label_map[highest_score_result['label']]
     highest_score = round(highest_score_result['score'], 4)
 
-    # Return the original tweet, the label with the highest score, and the inference time
     return {
         "text": tweet,
         "label": highest_label,
         "score": highest_score,
-        "inference_time": round(inference_time, 4)  # In seconds
+        "inference_time": round(inference_time, 4)
     }

model_quantized.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4fb016b6fed49bb877eee7cec2da0a28e6703d200741d19547c06c59c6993078
+size 1453