|
|
|
@ -36,24 +36,24 @@ from langchain.tools import tool
|
|
|
|
|
ROOT_DIR = "./data/"
|
|
|
|
|
|
|
|
|
|
# ---------- Tools ----------
|
|
|
|
|
openai_api_key = os.environ["OPENAI_API_KEY"]
|
|
|
|
|
llm = ChatOpenAI(model_name="gpt-4", temperature=1.0, openai_api_key=openai_api_key)
|
|
|
|
|
|
|
|
|
|
worker_tools = [
|
|
|
|
|
DuckDuckGoSearchRun(),
|
|
|
|
|
WriteFileTool(root_dir=ROOT_DIR),
|
|
|
|
|
ReadFileTool(root_dir=ROOT_DIR),
|
|
|
|
|
process_csv,
|
|
|
|
|
# openai_api_key = os.environ["OPENAI_API_KEY"]
|
|
|
|
|
# llm = ChatOpenAI(model_name="gpt-4", temperature=1.0, openai_api_key=openai_api_key)
|
|
|
|
|
|
|
|
|
|
# worker_tools = [
|
|
|
|
|
# DuckDuckGoSearchRun(),
|
|
|
|
|
# WriteFileTool(root_dir=ROOT_DIR),
|
|
|
|
|
# ReadFileTool(root_dir=ROOT_DIR),
|
|
|
|
|
# process_csv,
|
|
|
|
|
|
|
|
|
|
WebpageQATool(qa_chain=load_qa_with_sources_chain(llm)),
|
|
|
|
|
|
|
|
|
|
# Tool(name='terminal', func=Terminal.execute, description='Operates a terminal'),
|
|
|
|
|
# Tool(name='code_writer', func=CodeWriter(), description='Writes code'),
|
|
|
|
|
# Tool(name='code_editor', func=CodeEditor(), description='Edits code'),#
|
|
|
|
|
]
|
|
|
|
|
|
|
|
|
|
# ---------- Vector Store ----------
|
|
|
|
|
embeddings_model = OpenAIEmbeddings()
|
|
|
|
|
embedding_size = 1536
|
|
|
|
|
index = faiss.IndexFlatL2(embedding_size)
|
|
|
|
|
vectorstore = FAISS(embeddings_model.embed_query, index, InMemoryDocstore({}), {})
|
|
|
|
|
# WebpageQATool(qa_chain=load_qa_with_sources_chain(llm)),
|
|
|
|
|
|
|
|
|
|
# # Tool(name='terminal', func=Terminal.execute, description='Operates a terminal'),
|
|
|
|
|
# # Tool(name='code_writer', func=CodeWriter(), description='Writes code'),
|
|
|
|
|
# # Tool(name='code_editor', func=CodeEditor(), description='Edits code'),#
|
|
|
|
|
# ]
|
|
|
|
|
|
|
|
|
|
# # ---------- Vector Store ----------
|
|
|
|
|
# embeddings_model = OpenAIEmbeddings()
|
|
|
|
|
# embedding_size = 1536
|
|
|
|
|
# index = faiss.IndexFlatL2(embedding_size)
|
|
|
|
|
# vectorstore = FAISS(embeddings_model.embed_query, index, InMemoryDocstore({}), {})
|