Create app.py

app.py

@@ -0,0 +1,43 @@
+from fastapi import FastAPI, Query
+from transformers import CLIPModel, CLIPProcessor
+import torch
+
+# Initialize FastAPI
+app = FastAPI()
+
+# Load CLIP model and processor from Hugging Face
+model = CLIPModel.from_pretrained("openai/clip-vit-base-patch32")
+processor = CLIPProcessor.from_pretrained("openai/clip-vit-base-patch32")
+
+# Load and process document
+with open("test.txt", "r", encoding="utf-8") as f:
+    sentences = [line.strip() for line in f if line.strip()]
+
+# Encode document sentences
+with torch.no_grad():
+    sentence_inputs = processor(text=sentences, return_tensors="pt", padding=True, truncation=True)
+    sentence_embeddings = model.get_text_features(**sentence_inputs)
+
+@app.get("/")
+def welcome():
+    return {"message": "CLIP-based Document Retrieval Service is Running!"}
+
+@app.get("/search")
+def search(text: str = Query(..., description="Enter your query"), top_k: int = 5):
+    with torch.no_grad():
+        query_inputs = processor(text=[text], return_tensors="pt", padding=True, truncation=True)
+        query_embedding = model.get_text_features(**query_inputs)
+
+        # Compute cosine similarity
+        scores = torch.nn.functional.cosine_similarity(query_embedding, sentence_embeddings)[0]
+        top_indices = torch.topk(scores, k=top_k).indices
+
+        results = [{
+            "matched_sentence": sentences[i],
+            "similarity_score": round(scores[i].item(), 3)
+        } for i in top_indices]
+
+    return {
+        "query": text,
+        "top_matches": results
+    }