Spaces:
Paused
Paused
Update app.py
Browse files
app.py
CHANGED
@@ -67,6 +67,7 @@ from huggingface_hub import InferenceClient
|
|
67 |
|
68 |
client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")
|
69 |
|
|
|
70 |
def respond(
|
71 |
message,
|
72 |
history: list[tuple[str, str]],
|
@@ -76,18 +77,22 @@ def respond(
|
|
76 |
top_p,
|
77 |
):
|
78 |
try:
|
|
|
79 |
messages = [{"role": "system", "content": system_message}]
|
80 |
|
81 |
-
|
82 |
-
|
83 |
-
|
84 |
-
|
85 |
-
|
|
|
86 |
|
|
|
87 |
messages.append({"role": "user", "content": message})
|
88 |
|
89 |
response = ""
|
90 |
|
|
|
91 |
for message in client.chat_completion(
|
92 |
messages,
|
93 |
max_tokens=max_tokens,
|
@@ -98,10 +103,12 @@ def respond(
|
|
98 |
token = message.choices[0].delta.content
|
99 |
response += token
|
100 |
yield response
|
101 |
-
except Exception as e:
|
102 |
-
# Catching any errors and displaying a fallback message instead
|
103 |
-
yield "Sorry, something went wrong. Please try again."
|
104 |
|
|
|
|
|
|
|
|
|
|
|
105 |
demo = gr.ChatInterface(
|
106 |
respond,
|
107 |
additional_inputs=[
|
@@ -116,7 +123,11 @@ demo = gr.ChatInterface(
|
|
116 |
label="Top-p (nucleus sampling)",
|
117 |
),
|
118 |
],
|
|
|
|
|
|
|
119 |
)
|
120 |
|
121 |
if __name__ == "__main__":
|
122 |
demo.launch()
|
|
|
|
67 |
|
68 |
client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")
|
69 |
|
70 |
+
# Function to respond to user input while maintaining chat history
|
71 |
def respond(
|
72 |
message,
|
73 |
history: list[tuple[str, str]],
|
|
|
77 |
top_p,
|
78 |
):
|
79 |
try:
|
80 |
+
# Append system message at the start
|
81 |
messages = [{"role": "system", "content": system_message}]
|
82 |
|
83 |
+
# Append the chat history
|
84 |
+
for user_msg, bot_reply in history:
|
85 |
+
if user_msg:
|
86 |
+
messages.append({"role": "user", "content": user_msg})
|
87 |
+
if bot_reply:
|
88 |
+
messages.append({"role": "assistant", "content": bot_reply})
|
89 |
|
90 |
+
# Add the latest user message
|
91 |
messages.append({"role": "user", "content": message})
|
92 |
|
93 |
response = ""
|
94 |
|
95 |
+
# Stream the response token by token to avoid loading delays
|
96 |
for message in client.chat_completion(
|
97 |
messages,
|
98 |
max_tokens=max_tokens,
|
|
|
103 |
token = message.choices[0].delta.content
|
104 |
response += token
|
105 |
yield response
|
|
|
|
|
|
|
106 |
|
107 |
+
except Exception:
|
108 |
+
# Suppress any error and maintain conversation without showing error icon or message
|
109 |
+
yield "Error handled silently." # You can replace this with "" to hide any message
|
110 |
+
|
111 |
+
# Gradio interface with preserved history and customizable chatbot behavior
|
112 |
demo = gr.ChatInterface(
|
113 |
respond,
|
114 |
additional_inputs=[
|
|
|
123 |
label="Top-p (nucleus sampling)",
|
124 |
),
|
125 |
],
|
126 |
+
# Enable to retain history and suppress errors
|
127 |
+
retain_history=True, # This will ensure the chat history is not lost between interactions
|
128 |
+
retry_on_error=False # Prevent showing the error message or retry button
|
129 |
)
|
130 |
|
131 |
if __name__ == "__main__":
|
132 |
demo.launch()
|
133 |
+
|