| from model import get_gemma |
| from rag_utils import generate_context |
|
|
|
|
| def generate_response(history=[], temperature: float=0.0, top_k=None, top_p=None): |
| |
| gemma_model = get_gemma(temperature=temperature, top_k=top_k, top_p=top_p) |
| |
| response = gemma_model.invoke(history).content |
|
|
| return response |
|
|
| def generate_RAG_response(query: str, file_path, history=[]): |
| gemma_model = get_gemma() |
| query, context = generate_context(query, file_path) |
| if len(history) > 1: |
| prompt = history[:-1] |
| prompt = prompt.append({"role" : "user", "content" : f"INSTRUCTION: Answer the query with given context in mind.\nQUERY: {query}\n\nCONTEXT : {context}"}) |
| else: |
| prompt = [{"role" : "user", "content" : f"INSTRUCTION: Answer the query with given context in mind.\nQUERY: {query}\n\nCONTEXT : {context}"}] |
| response = gemma_model.invoke(history).content |
|
|
| return response |
