Spaces:
Sleeping
Sleeping
Commit
•
4091a1a
1
Parent(s):
2587659
Update app.py
Browse files
app.py
CHANGED
@@ -47,7 +47,7 @@ def load_doc(list_file_path, chunk_size, chunk_overlap):
|
|
47 |
|
48 |
# Create vector database
|
49 |
def create_db(splits, collection_name):
|
50 |
-
embedding = HuggingFaceEmbeddings()
|
51 |
new_client = chromadb.EphemeralClient()
|
52 |
vectordb = Chroma.from_documents(
|
53 |
documents=splits,
|
@@ -61,7 +61,7 @@ def create_db(splits, collection_name):
|
|
61 |
|
62 |
# Load vector database
|
63 |
def load_db():
|
64 |
-
embedding = HuggingFaceEmbeddings()
|
65 |
vectordb = Chroma(
|
66 |
# persist_directory=default_persist_directory,
|
67 |
embedding_function=embedding)
|
@@ -132,8 +132,8 @@ def initialize_database(list_file_obj, chunk_size, chunk_overlap, progress=gr.Pr
|
|
132 |
#file_path = file_obj.name
|
133 |
list_file_path = [x.name for x in list_file_obj if x is not None]
|
134 |
collection_name = Path(list_file_path[0]).stem
|
135 |
-
|
136 |
-
|
137 |
progress(0.25, desc="Loading document...")
|
138 |
# Load document and create splits
|
139 |
doc_splits = load_doc(list_file_path, chunk_size, chunk_overlap)
|
@@ -174,8 +174,7 @@ def conversation(qa_chain, message, history):
|
|
174 |
# Langchain sources are zero-based
|
175 |
response_source1_page = response_sources[0].metadata["page"] + 1
|
176 |
response_source2_page = response_sources[1].metadata["page"] + 1
|
177 |
-
|
178 |
-
# print('DB source', response_sources)
|
179 |
|
180 |
# Append user message and response to chat history
|
181 |
new_history = history + [(message, response_answer)]
|
|
|
47 |
|
48 |
# Create vector database
|
49 |
def create_db(splits, collection_name):
|
50 |
+
embedding = HuggingFaceEmbeddings(model_name="sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2")
|
51 |
new_client = chromadb.EphemeralClient()
|
52 |
vectordb = Chroma.from_documents(
|
53 |
documents=splits,
|
|
|
61 |
|
62 |
# Load vector database
|
63 |
def load_db():
|
64 |
+
embedding = HuggingFaceEmbeddings(model_name="sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2")
|
65 |
vectordb = Chroma(
|
66 |
# persist_directory=default_persist_directory,
|
67 |
embedding_function=embedding)
|
|
|
132 |
#file_path = file_obj.name
|
133 |
list_file_path = [x.name for x in list_file_obj if x is not None]
|
134 |
collection_name = Path(list_file_path[0]).stem
|
135 |
+
print('list_file_path: ', list_file_path)
|
136 |
+
print('Collection name: ', collection_name)
|
137 |
progress(0.25, desc="Loading document...")
|
138 |
# Load document and create splits
|
139 |
doc_splits = load_doc(list_file_path, chunk_size, chunk_overlap)
|
|
|
174 |
# Langchain sources are zero-based
|
175 |
response_source1_page = response_sources[0].metadata["page"] + 1
|
176 |
response_source2_page = response_sources[1].metadata["page"] + 1
|
177 |
+
print ('Response: ', response)
|
|
|
178 |
|
179 |
# Append user message and response to chat history
|
180 |
new_history = history + [(message, response_answer)]
|