changed

Dockerfile

@@ -0,0 +1,22 @@
+FROM python:3.9
+
+WORKDIR /app
+COPY . /app
+
+ENV HF_HOME=/app/.cache
+
+RUN mkdir -p /app/.cache/huggingface/hub && \
+    chmod -R 777 /app/.cache && \
+    chmod -R 777 /app/.cache/huggingface
+
+
+
+RUN pip install --upgrade pip
+RUN pip install --no-cache-dir -r requirements.txt
+
+COPY --chown=user ./requirements.txt requirements.txt
+RUN pip install --no-cache-dir --upgrade -r requirements.txt
+
+EXPOSE 7860
+
+CMD ["uvicorn", "main:app", "--host", "0.0.0.0", "--port", "7860"]

main.py

@@ -0,0 +1,100 @@
+from fastapi import FastAPI, HTTPException
+from pydantic import BaseModel
+from sentence_transformers import SentenceTransformer
+import numpy as np
+import pandas as pd
+
+# Initialize FastAPI
+app = FastAPI()
+
+# Load the sentence transformer model
+try:
+    model = SentenceTransformer("sentence-transformers/all-MiniLM-L6-v2", trust_remote_code=True)
+    print("✅ Model loaded successfully")
+except Exception as e:
+    raise RuntimeError(f"❌ Failed to load model: {str(e)}")
+
+# Define request schemas
+class CosineSimilarityInput(BaseModel):
+    text1: str
+    text2: str
+
+class MessageInput(BaseModel):
+    message: str
+
+# Load SMS dataset from Excel
+file_path = "sms_process_data_main.xlsx"
+df = pd.read_excel(file_path)
+
+# Precompute embeddings
+transactional_examples = df[df['label'] == 'Transaction']['MessageText'].tolist()
+offer_examples = df[df['label'] == 'Offer']['MessageText'].tolist()
+
+transactional_embeddings = [model.encode(msg, convert_to_tensor=True).cpu().numpy() for msg in transactional_examples]
+offer_embeddings = [model.encode(msg, convert_to_tensor=True).cpu().numpy() for msg in offer_examples]
+
+# Function to compute cosine similarity
+def cosine_similarity(vec1, vec2):
+    norm1 = np.linalg.norm(vec1)
+    norm2 = np.linalg.norm(vec2)
+    if norm1 == 0 or norm2 == 0:
+        return 0.0  # Prevent division by zero
+    return np.dot(vec1, vec2) / (norm1 * norm2)
+
+# 🚀 1️⃣ Homepage Endpoint
+@app.get("/")
+async def home():
+    return {"message": "Welcome to Classification of SMS"}
+
+# 🔢 2️⃣ Cosine Similarity Endpoint
+@app.post("/cosine_similarity")
+async def compute_similarity(input_data: CosineSimilarityInput):
+    """
+    Compute cosine similarity between two input texts.
+    """
+    try:
+        emb1 = model.encode(input_data.text1, convert_to_tensor=True).cpu().numpy()
+        emb2 = model.encode(input_data.text2, convert_to_tensor=True).cpu().numpy()
+        similarity = cosine_similarity(emb1, emb2)
+        return {"cosine_similarity": round(float(similarity), 4)}
+    except Exception as e:
+        raise HTTPException(status_code=500, detail=f"Error computing similarity: {str(e)}")
+
+# 📩 3️⃣ SMS Classification Endpoint
+@app.post("/predict_label/")
+async def classify_message(input_data: MessageInput):
+    """
+    Classify an SMS as either 'Transaction' or 'Offer'.
+    """
+    try:
+        # Validate input
+        text_input = input_data.message.strip()
+        if not text_input:
+            raise HTTPException(status_code=400, detail="Input message cannot be empty")
+
+        # Encode input text
+        input_embedding = model.encode(text_input, convert_to_tensor=True).cpu().numpy()
+
+        # Compute similarity scores
+        transactional_scores = [cosine_similarity(input_embedding, emb) for emb in transactional_embeddings]
+        offer_scores = [cosine_similarity(input_embedding, emb) for emb in offer_embeddings]
+
+        # Get max similarity
+        max_transactional = max(transactional_scores, default=0)
+        max_offer = max(offer_scores, default=0)
+
+        # Determine label and probability
+        if max_transactional > max_offer:
+            label = "Transaction"
+
+        else:
+            label = "Offer"
+
+        return {
+            "label": label
+        }
+
+    except Exception as e:
+        raise HTTPException(status_code=500, detail=f"Unexpected error: {str(e)}")
+
+

requirements.txt

@@ -0,0 +1,8 @@
+fastapi
+pandas
+scikit-learn
+joblib
+uvicorn
+sentence-transformers
+numpy
+openpyxl