Update app.py

app.py

@@ -17,18 +17,29 @@ logging.basicConfig(level=logging.INFO)
 logger = logging.getLogger(__name__)
 
 # Hugging Face Inference API endpoint for DeepSeek
-API_URL = "https://api-inference.huggingface.co/models/deepseek-ai/DeepSeek-R1-Distill-Qwen-32B"  # Replace with the correct model ID
+API_URL = "https://api-inference.huggingface.co/models/deepseek-ai/deepseek-model"  # Replace with the correct model ID
 headers = {"Authorization": f"Bearer {token}"}
 
 # Define request body schema
 class TextGenerationRequest(BaseModel):
     prompt: str
-    max_new_tokens: int = 50  # Reduce this for faster responses
-    temperature: float = 0.7  # Lower for more deterministic outputs
-    top_k: int = 50
-    top_p: float = 0.9
+    max_new_tokens: int = 150  # Increased to allow longer responses
+    temperature: float = 0.8  # Slightly increased for more varied responses
+    top_k: int = 100  # Broader vocabulary sampling
+    top_p: float = 0.92  # Increased diversity
+    repetition_penalty: float = 1.2  # Penalizes repetition
     do_sample: bool = True
 
+# Function to remove repetitive content in the output
+def remove_repetition(text: str) -> str:
+    seen = set()
+    result = []
+    for word in text.split():
+        if word.lower() not in seen:
+            result.append(word)
+            seen.add(word.lower())
+    return " ".join(result)
+
 # Define API endpoint
 @app.post("/generate-text")
 async def generate_text(request: TextGenerationRequest):
@@ -43,6 +54,7 @@ async def generate_text(request: TextGenerationRequest):
                 "temperature": request.temperature,
                 "top_k": request.top_k,
                 "top_p": request.top_p,
+                "repetition_penalty": request.repetition_penalty,
                 "do_sample": request.do_sample,
             },
         }
@@ -55,9 +67,10 @@ async def generate_text(request: TextGenerationRequest):
             logger.error(f"API Error: {response.status_code} - {response.text}")
             raise HTTPException(status_code=response.status_code, detail=response.text)
 
-        # Extract the generated text from the response
+        # Extract and process the generated text
         generated_text = response.json()[0]["generated_text"]
-        return {"generated_text": generated_text}
+        cleaned_text = remove_repetition(generated_text)  # Clean up repetition
+        return {"generated_text": cleaned_text}
     except Exception as e:
         logger.error(f"Error generating text: {e}")
         raise HTTPException(status_code=500, detail=str(e))
@@ -70,4 +83,4 @@ async def root():
 # Add a test endpoint
 @app.get("/test")
 async def test():
-    return {"message": "API is running!"}
+    return {"message": "API is running!"}