| from langchain.chains import RetrievalQA | |
| from langchain.prompts import PromptTemplate | |
| from langchain_together import Together | |
| from langchain_community.vectorstores import Qdrant | |
| from main import get_repo_name | |
| import qdrant_client | |
| import os | |
| from langchain_community.embeddings.fastembed import FastEmbedEmbeddings | |
| embeddings = FastEmbedEmbeddings(model_name="BAAI/bge-small-en-v1.5") | |
| client = qdrant_client.QdrantClient( | |
| os.getenv("QDRANT_HOST"), | |
| api_key=os.getenv("QDRANT_API_KEY") | |
| ) | |
| B_INST, E_INST = "[INST]", "[/INST]" | |
| B_SYS, E_SYS = "<<SYS>>\n", "\n<</SYS>>\n\n" | |
| def get_prompt(instruction, new_system_prompt ): | |
| SYSTEM_PROMPT = B_SYS + new_system_prompt + E_SYS | |
| prompt_template = B_INST + SYSTEM_PROMPT + instruction + E_INST | |
| return prompt_template | |
| sys_prompt = """You are a helpful, smart and intelligent coding assistant. Always answer as helpfully as possible using the context code provided. Your answers should only answer the question once, you can provide code snippets but make sure you explain them thoroughly | |
| If a question does not make any sense, or is not factually coherent, explain why instead of answering something not correct. If you don't know the answer to a question, please don't share false information. """ | |
| instruction = """CONTEXT CODE:/n/n {context}/n | |
| Question: {question}""" | |
| prompt_template = get_prompt(instruction, sys_prompt) | |
| llama_prompt = PromptTemplate( | |
| template=prompt_template, input_variables=["context", "question"] | |
| ) | |
| llama2_llm = Together( | |
| model="togethercomputer/llama-2-70b-chat", | |
| temperature=0.7, | |
| max_tokens=1024, | |
| together_api_key="d8ec7106bd0c268bf4672dba83272b86054fbe849eba82f3f75ceb17e6d57eb0" | |
| ) | |
| def process_llm_response(llm_response): | |
| response = " " | |
| response += llm_response['result'] + "\n\nSources\n" | |
| for source in llm_response['source_documents']: | |
| response +="Source - "+source.metadata['source'] +"\n" | |
| return response | |
| def answer_query(query,url): | |
| vectorstore = Qdrant( | |
| client=client, | |
| collection_name=get_repo_name(url), | |
| embeddings=embeddings | |
| ) | |
| qa_chain = RetrievalQA.from_chain_type(llm= llama2_llm, chain_type_kwargs = {"prompt": llama_prompt},chain_type="stuff",retriever= vectorstore.as_retriever(),return_source_documents = True) | |
| return process_llm_response(qa_chain(query)) | |