Spaces:
Sleeping
Sleeping
Commit
·
5f0033b
1
Parent(s):
68d4a44
Update space
Browse files
app.py
CHANGED
@@ -15,28 +15,26 @@ def respond(
|
|
15 |
temperature,
|
16 |
top_p,
|
17 |
):
|
18 |
-
|
19 |
-
|
20 |
for val in history:
|
21 |
if val[0]:
|
22 |
-
|
23 |
if val[1]:
|
24 |
-
|
25 |
-
|
26 |
-
messages.append({"role": "user", "content": message})
|
27 |
|
28 |
response = ""
|
29 |
-
|
30 |
-
|
31 |
-
|
32 |
-
|
|
|
33 |
stream=True,
|
34 |
temperature=temperature,
|
35 |
top_p=top_p,
|
36 |
):
|
37 |
-
|
38 |
-
|
39 |
-
response += token
|
40 |
yield response
|
41 |
|
42 |
|
|
|
15 |
temperature,
|
16 |
top_p,
|
17 |
):
|
18 |
+
# Format the conversation history into a prompt
|
19 |
+
prompt = f"{system_message}\n\n"
|
20 |
for val in history:
|
21 |
if val[0]:
|
22 |
+
prompt += f"User: {val[0]}\n"
|
23 |
if val[1]:
|
24 |
+
prompt += f"Assistant: {val[1]}\n"
|
25 |
+
prompt += f"User: {message}\nAssistant:"
|
|
|
26 |
|
27 |
response = ""
|
28 |
+
|
29 |
+
# Use text generation instead of chat completion
|
30 |
+
for message in client.text_generation(
|
31 |
+
prompt,
|
32 |
+
max_new_tokens=max_tokens,
|
33 |
stream=True,
|
34 |
temperature=temperature,
|
35 |
top_p=top_p,
|
36 |
):
|
37 |
+
response += message
|
|
|
|
|
38 |
yield response
|
39 |
|
40 |
|