feat: added all the files

Dockerfile

@@ -0,0 +1,18 @@
+FROM python:3.9
+
+RUN useradd -m -u 1000 user
+USER user
+ENV PATH="/home/user/.local/bin:$PATH"
+
+WORKDIR /app
+
+COPY --chown=user ./requirements.txt requirements.txt
+RUN pip install --no-cache-dir --upgrade -r requirements.txt
+
+COPY --chown=user . /app
+
+# Install DVC if needed (uncomment if using DVC in container)
+RUN pip install --no-cache-dir dvc
+
+CMD ["dvc","repro"]
+# CMD ["uvicorn", "main:app", "--host", "0.0.0.0", "--port", "7860"]

components/model_ingestion.py

@@ -0,0 +1,14 @@
+from transformers import RobertaTokenizerFast, RobertaForSequenceClassification
+from utils.commons import HUGGINGFACE_MODEL_PATH, MODEL_SAVE_PATH
+import os
+
+class ModelIngestor:
+    @staticmethod
+    def download_model():
+        """Download and save the model"""
+        if not os.path.exists(MODEL_SAVE_PATH):
+            os.makedirs(MODEL_SAVE_PATH)
+            
+        RobertaTokenizerFast.from_pretrained(HUGGINGFACE_MODEL_PATH).save_pretrained(MODEL_SAVE_PATH)
+        RobertaForSequenceClassification.from_pretrained(HUGGINGFACE_MODEL_PATH).save_pretrained(MODEL_SAVE_PATH)
+        print(f"Model saved to {MODEL_SAVE_PATH}")

components/model_loader.py

@@ -0,0 +1,13 @@
+from transformers import RobertaTokenizerFast, RobertaForSequenceClassification
+from utils.commons import MODEL_SAVE_PATH
+
+class ModelLoader:
+    def __init__(self):
+        self.tokenizer = None
+        self.model = None
+        
+    def load_model(self):
+        """Load model from local path"""
+        self.tokenizer = RobertaTokenizerFast.from_pretrained(MODEL_SAVE_PATH)
+        self.model = RobertaForSequenceClassification.from_pretrained(MODEL_SAVE_PATH)
+        return self.tokenizer, self.model

components/pipeline_preparer.py

@@ -0,0 +1,11 @@
+from transformers import TextClassificationPipeline
+
+class PipelinePreparer:
+    @staticmethod
+    def prepare_pipeline(tokenizer, model):
+        """Create text classification pipeline"""
+        return TextClassificationPipeline(
+            tokenizer=tokenizer,
+            model=model,
+            return_all_scores=True
+        )

components/predictor.py

@@ -0,0 +1,12 @@
+class Predictor:
+    def __init__(self, pipeline):
+        self.pipeline = pipeline
+        
+    def predict(self, text):
+        """Make prediction"""
+        preds_list = self.pipeline(text)
+        best_pred = max(preds_list[0], key=lambda x: x["score"])
+        return {
+            "predicted_intent": best_pred["label"],
+            "confidence": best_pred["score"]
+        }

dvc.yaml

@@ -0,0 +1,17 @@
+stages:
+  model_ingestion:
+    cmd: python -c "from components.model_ingestion import ModelIngestor; ModelIngestor.download_model()"
+    deps:
+      - components/model_ingestion.py
+      - utils/commons.py
+    outs:
+      - models/
+      
+  serve_model:
+    cmd: python main.py
+    deps:
+      - main.py
+      - components/
+      - utils/
+    outs:
+      - logs/

main.py

@@ -0,0 +1,38 @@
+from fastapi import FastAPI
+from pydantic import BaseModel
+from components.model_loader import ModelLoader
+from components.pipeline_preparer import PipelinePreparer
+from components.predictor import Predictor
+import uvicorn
+
+app = FastAPI()
+
+class PredictionRequest(BaseModel):
+    sentence: str
+
+@app.on_event("startup")
+async def startup_event():
+    print("Initializing model...")
+    try:
+        # Model initialization
+        loader = ModelLoader()
+        tokenizer, model = loader.load_model()
+        pipeline = PipelinePreparer.prepare_pipeline(tokenizer, model)
+        app.state.predictor = Predictor(pipeline)
+        print("Model initialized successfully")
+    except Exception as e:
+        print(f"Error initializing model: {e}")
+        app.state.predictor = None
+
+@app.get("/")
+def health_check():
+    return {"Message": "Service is healthy", "Status": "OK"}
+
+@app.post("/predict")
+def predict(request: PredictionRequest):
+    if not app.state.predictor:
+        return {"error": "Model not initialized"}
+    return app.state.predictor.predict(request.sentence)
+
+if __name__ == "__main__":
+    uvicorn.run(app, host="0.0.0.0", port=7860)

utils/commons.py

@@ -0,0 +1,3 @@
+# Commonly used variables
+HUGGINGFACE_MODEL_PATH = "bespin-global/klue-roberta-small-3i4k-intent-classification"
+MODEL_SAVE_PATH = "./models"