Hugging-Face Deployment

app/config/config.json

@@ -1,6 +1,5 @@
 {
   "MODEL_NAME": "llama-3.3-70b-versatile",
-  "GROQ_KEY": "",
   "VALID_API_KEY": "2931609bd36ec1a45cb577b3b831dc711c76ae157b3c6250c564284c93b062ff",
 
   "LLM_CONFIG": {

app/core/models.py

@@ -80,9 +80,9 @@ async def llm_setup(config, url):
         temperature=f"{config.get('TEMPERATURE', 0)}",
         max_tokens=f"{config.get('MAX_TOKENS', 300)}",  # Increased token limit for JSON responses
         max_retries=f"{config.get('MAX_RETRIES', 3)}",
-        api_key=f"{config.get('GROQ_KEY')}",
+        api_key=f"{os.getenv('GROQ_KEY')}",
     )
-    logging.info(f"LLM initialized with model: {config.get('MODEL_NAME')}, api_key: {config.get('GROQ_KEY')}")
+    logging.info(f"LLM initialized with model: {config.get('MODEL_NAME')}, api_key: {os.getenv('GROQ_KEY')}")
 
     # Choose template based on whether we need structured JSON output
     prompt_template = prompt_template_description()

main.py

@@ -5,12 +5,14 @@ import json
 import hashlib
 
 # Import our existing pipeline components
-from contextlib import asynccontextmanager
-from fastapi import FastAPI, HTTPException, Depends, status
 from app.utils.util import verify_api_key
 from app.core.models import llm_response_generator
 from app.schema.schema import QuestionRequest, AnswerResponse
 
+from contextlib import asynccontextmanager
+from fastapi import FastAPI, HTTPException, Depends, status
+from fastapi.responses import FileResponse, RedirectResponse
+
 logging.basicConfig(format='%(asctime)s - %(levelname)s - Line: %(lineno)d - %(message)s', 
                     datefmt='%Y-%m-%d %H:%M:%S', 
                     level=logging.INFO)
@@ -18,6 +20,7 @@ logging.basicConfig(format='%(asctime)s - %(levelname)s - Line: %(lineno)d - %(m
 
 os.environ["TOKENIZERS_PARALLELISM"] = "false"
 
+
 # Load config.json at startup
 @asynccontextmanager
 async def lifespan(app: FastAPI):
@@ -33,6 +36,10 @@ async def lifespan(app: FastAPI):
 # FastAPI app
 app = FastAPI(title="HackRx PDF RAG API", version="1.0.0", lifespan=lifespan)
 
+@app.get("/", include_in_schema=False)
+async def docs_redirect():
+    return RedirectResponse("/docs")
+
 @app.post("/api/v1/hackrx/run", response_model=AnswerResponse)
 async def process_questions(request: QuestionRequest, api_key: str = Depends(verify_api_key)):
     try:
@@ -43,7 +50,7 @@ async def process_questions(request: QuestionRequest, api_key: str = Depends(ver
         logging.info(f"Received {len(questions)} questions for processing. Documents URL: {url}")
 
         # Create cache directory if not exists
-        cache_dir = "redis"
+        cache_dir = "cache"
         os.makedirs(cache_dir, exist_ok=True)
         # Create a cache key from url and questions
         cache_key = hashlib.sha256((url + json.dumps(questions, sort_keys=True)).encode()).hexdigest()