Spaces:
Build error
Build error
Update app.py
Browse filesadded deug statements to observe code execution
app.py
CHANGED
|
@@ -26,20 +26,21 @@ def store_document(text):
|
|
| 26 |
print("storing document")
|
| 27 |
|
| 28 |
embedding = embedding_model.encode([text])
|
|
|
|
| 29 |
index.add(np.array(embedding, dtype=np.float32))
|
| 30 |
documents.append(text)
|
| 31 |
|
| 32 |
-
print(f"your document has been stored
|
| 33 |
|
| 34 |
return "Document stored!"
|
| 35 |
|
| 36 |
def retrieve_document(query):
|
| 37 |
-
print(f"retrieving doc based on {query}")
|
| 38 |
|
| 39 |
query_embedding = embedding_model.encode([query])
|
| 40 |
_, closest_idx = index.search(np.array(query_embedding, dtype=np.float32), 1)
|
| 41 |
|
| 42 |
-
print(f"retrieved: {documents[closest_idx[0][0]]}")
|
| 43 |
|
| 44 |
return documents[closest_idx[0][0]]
|
| 45 |
|
|
@@ -87,15 +88,18 @@ def chatbot(pdf_file, user_question):
|
|
| 87 |
return f"Error retrieving document relevant to the query: {user_question} \n{e}"
|
| 88 |
|
| 89 |
if doc:
|
|
|
|
| 90 |
# Split into smaller chunks
|
| 91 |
chunks = split_text(doc)
|
| 92 |
|
| 93 |
# Use only the first chunk (to optimize token usage)
|
| 94 |
prompt = f"Based on this document, answer the question:\n\nDocument:\n{chunks[0]}\n\nQuestion: {user_question}"
|
|
|
|
| 95 |
else:
|
| 96 |
prompt=user_question
|
| 97 |
|
| 98 |
try:
|
|
|
|
| 99 |
response = together.Completion.create(
|
| 100 |
model="mistralai/Mistral-7B-Instruct-v0.1",
|
| 101 |
prompt=prompt,
|
|
|
|
| 26 |
print("storing document")
|
| 27 |
|
| 28 |
embedding = embedding_model.encode([text])
|
| 29 |
+
print(f"embedding: \n{embedding}")
|
| 30 |
index.add(np.array(embedding, dtype=np.float32))
|
| 31 |
documents.append(text)
|
| 32 |
|
| 33 |
+
print(f"your document has been stored")
|
| 34 |
|
| 35 |
return "Document stored!"
|
| 36 |
|
| 37 |
def retrieve_document(query):
|
| 38 |
+
print(f"retrieving doc based on: \n{query}")
|
| 39 |
|
| 40 |
query_embedding = embedding_model.encode([query])
|
| 41 |
_, closest_idx = index.search(np.array(query_embedding, dtype=np.float32), 1)
|
| 42 |
|
| 43 |
+
print(f"retrieved: \n{documents[closest_idx[0][0]]}")
|
| 44 |
|
| 45 |
return documents[closest_idx[0][0]]
|
| 46 |
|
|
|
|
| 88 |
return f"Error retrieving document relevant to the query: {user_question} \n{e}"
|
| 89 |
|
| 90 |
if doc:
|
| 91 |
+
print("found doc")
|
| 92 |
# Split into smaller chunks
|
| 93 |
chunks = split_text(doc)
|
| 94 |
|
| 95 |
# Use only the first chunk (to optimize token usage)
|
| 96 |
prompt = f"Based on this document, answer the question:\n\nDocument:\n{chunks[0]}\n\nQuestion: {user_question}"
|
| 97 |
+
print(f"prompt: \n{prompt}")
|
| 98 |
else:
|
| 99 |
prompt=user_question
|
| 100 |
|
| 101 |
try:
|
| 102 |
+
print("asking")
|
| 103 |
response = together.Completion.create(
|
| 104 |
model="mistralai/Mistral-7B-Instruct-v0.1",
|
| 105 |
prompt=prompt,
|