Spaces:
Build error
Build error
Added more verbose to data setup.
Browse files- chatbot/data.py +7 -0
chatbot/data.py
CHANGED
@@ -23,6 +23,8 @@ def download_test_data():
|
|
23 |
def load_data():
|
24 |
with st.spinner(text="Loading and indexing the provided dataset – hang tight! This may take a few seconds."):
|
25 |
documents = SimpleDirectoryReader(input_dir="./data", recursive=True).load_data()
|
|
|
|
|
26 |
llm = AzureOpenAI(
|
27 |
model="gpt-3.5-turbo",
|
28 |
engine=st.secrets["ENGINE"],
|
@@ -36,6 +38,8 @@ def load_data():
|
|
36 |
"André's research. Keep your answers technical and based on facts;"
|
37 |
" do not hallucinate features.",
|
38 |
)
|
|
|
|
|
39 |
# You need to deploy your own embedding model as well as your own chat completion model
|
40 |
embed_model = OpenAIEmbedding(
|
41 |
model="text-embedding-ada-002",
|
@@ -44,7 +48,10 @@ def load_data():
|
|
44 |
api_base=st.secrets["OPENAI_API_BASE"],
|
45 |
api_type="azure",
|
46 |
api_version=st.secrets["OPENAI_API_VERSION"],
|
|
|
47 |
)
|
|
|
|
|
48 |
service_context = ServiceContext.from_defaults(llm=llm, embed_model=embed_model)
|
49 |
set_global_service_context(service_context)
|
50 |
index = VectorStoreIndex.from_documents(documents) # , service_context=service_context)
|
|
|
23 |
def load_data():
|
24 |
with st.spinner(text="Loading and indexing the provided dataset – hang tight! This may take a few seconds."):
|
25 |
documents = SimpleDirectoryReader(input_dir="./data", recursive=True).load_data()
|
26 |
+
|
27 |
+
with st.spinner(text="Setting up Azure OpenAI..."):
|
28 |
llm = AzureOpenAI(
|
29 |
model="gpt-3.5-turbo",
|
30 |
engine=st.secrets["ENGINE"],
|
|
|
38 |
"André's research. Keep your answers technical and based on facts;"
|
39 |
" do not hallucinate features.",
|
40 |
)
|
41 |
+
|
42 |
+
with st.spinner(text="Setting up OpenAI Embedding..."):
|
43 |
# You need to deploy your own embedding model as well as your own chat completion model
|
44 |
embed_model = OpenAIEmbedding(
|
45 |
model="text-embedding-ada-002",
|
|
|
48 |
api_base=st.secrets["OPENAI_API_BASE"],
|
49 |
api_type="azure",
|
50 |
api_version=st.secrets["OPENAI_API_VERSION"],
|
51 |
+
embed_batch_size=10, # set to one to reduce rate limit -> may degrade response runtime
|
52 |
)
|
53 |
+
|
54 |
+
with st.spinner(text="Setting up Vector Store Index..."):
|
55 |
service_context = ServiceContext.from_defaults(llm=llm, embed_model=embed_model)
|
56 |
set_global_service_context(service_context)
|
57 |
index = VectorStoreIndex.from_documents(documents) # , service_context=service_context)
|