MODIFIED: changes in RAG_PDF_Groq route

README.md

@@ -73,7 +73,10 @@ Generative AI, powered by advanced machine learning models, enables the creation
 - **Route:** `/text_summarizer_groq`
 - **Description:**  Dive into a realm of creativity with our text summarization endpoint, where the model mixtral-8x7b-32768 crafts concise summaries from your input text, delivering insights at the speed of thought.
 
+### 12. RAG Using Groq
 
+- **Route:** `/RAG_PDF_Groq`
+- **Description:** This endpoint uses the pdf and give the answer based on the prompt provided using Groq,with a default model input of llama2-70b-4096, but offering alternatives like mixtral-8x7b-32768 and gemma-7b-it.
 
 ## Usage
 

api.py

@@ -289,10 +289,12 @@ async def groq_text_summary(input_text: str = Form(...)):
     except Exception as e:
         return ResponseText(response=f"Error: {str(e)}")
 
-@app.post("/RAG_PDF_Groq",description="The endpoint uses the pdf and give the answer based on the prompt provided using groq")
-async def talk_pd_groq(pdf: UploadFile = File(...),prompt: str = Form(...)):
+@app.post("/RAG_PDF_Groq",description="The endpoint uses the pdf and give the answer based on the prompt provided using groq\
+          In model input default is mixtral-8x7b-32768 but you can choose llama2-70b-4096 and gemma-7b-it.")
+async def talk_pd_groq(pdf: UploadFile = File(...),prompt: str = Form(...),
+                       model: Optional[str] = Form('llama2-70b-4096')):
     try:
-        rag_chain = groq_pdf(pdf.file)
+        rag_chain = groq_pdf(pdf.file,model)
         out = rag_chain.invoke(prompt)
         db = MongoDB()
         payload = {

helper_functions.py

@@ -188,10 +188,10 @@ def questions_generator(doc):
     ques = ques_gen_chain.run(document_ques_gen)
     return ques
 
-def groq_pdf(pdf):
+def groq_pdf(pdf,model):
     llm = ChatGroq(
             api_key=os.environ['GROQ_API_KEY'],
-            model_name='mixtral-8x7b-32768'
+            model_name=model
     )
     text = "".join(page.extract_text() for page in PdfReader(pdf).pages)
     text_splitter = RecursiveCharacterTextSplitter(chunk_size=10000, chunk_overlap=1000)