Spaces:
Running
Running
Updated app.py
#10
by
subatomicERROR
- opened
app.py
CHANGED
@@ -1,109 +1,106 @@
|
|
1 |
# app.py
|
2 |
|
3 |
import os
|
|
|
4 |
from http.client import HTTPMessage
|
5 |
-
|
6 |
-
os.system('pip install dashscope')
|
7 |
-
|
8 |
-
import gradio as gr
|
9 |
from http import HTTPStatus
|
10 |
-
import
|
11 |
-
from dashscope import Generation
|
12 |
-
from dashscope.api_entities.dashscope_response import
|
13 |
from typing import List, Optional, Tuple, Dict
|
14 |
from urllib.error import HTTPError
|
15 |
|
16 |
-
|
|
|
17 |
|
|
|
18 |
YOUR_API_TOKEN = os.getenv('YOUR_API_TOKEN')
|
|
|
|
|
|
|
|
|
19 |
dashscope.api_key = YOUR_API_TOKEN
|
20 |
|
21 |
History = List[Tuple[str, str]]
|
22 |
Messages = List[Dict[str, str]]
|
23 |
|
|
|
24 |
|
25 |
def clear_session() -> History:
|
26 |
return '', []
|
27 |
|
28 |
-
|
29 |
-
|
30 |
-
if system is None or len(system) == 0:
|
31 |
system = default_system
|
32 |
return system, system, []
|
33 |
|
34 |
-
|
35 |
def history_to_messages(history: History, system: str) -> Messages:
|
36 |
messages = [{'role': Role.SYSTEM, 'content': system}]
|
37 |
-
for
|
38 |
-
messages.append({'role': Role.USER, 'content':
|
39 |
-
messages.append({'role': Role.ASSISTANT, 'content':
|
40 |
return messages
|
41 |
|
42 |
-
|
43 |
def messages_to_history(messages: Messages) -> Tuple[str, History]:
|
44 |
assert messages[0]['role'] == Role.SYSTEM
|
45 |
system = messages[0]['content']
|
46 |
-
history = []
|
47 |
-
for q, r in zip(messages[1::2], messages[2::2]):
|
48 |
-
history.append([q['content'], r['content']])
|
49 |
return system, history
|
50 |
|
51 |
-
|
52 |
-
|
53 |
-
) -> Tuple[str, str, History]:
|
54 |
-
if query is None:
|
55 |
query = ''
|
56 |
-
if history
|
57 |
history = []
|
58 |
messages = history_to_messages(history, system)
|
59 |
messages.append({'role': Role.USER, 'content': query})
|
60 |
label_model = f"qwen2.5-coder-{radio.lower()}-instruct"
|
61 |
-
|
62 |
-
|
63 |
-
|
64 |
-
|
65 |
-
|
66 |
-
|
67 |
-
|
68 |
-
|
69 |
-
|
70 |
-
|
71 |
-
|
72 |
-
|
73 |
-
|
74 |
-
|
75 |
-
response.request_id, response.status_code,
|
76 |
-
|
77 |
-
|
78 |
-
|
79 |
-
|
|
|
|
|
|
|
|
|
80 |
mark_ = gr.Markdown(value=f"<center><font size=8>Qwen2.5-Coder-{radio}-instruct👾</center>")
|
81 |
chatbot = gr.Chatbot(label=f'Qwen2.5-Coder-{radio.lower()}-instruct')
|
82 |
-
|
83 |
-
if system is None or len(system) == 0:
|
84 |
system = default_system
|
85 |
-
|
86 |
return mark_, chatbot, system, system, ""
|
87 |
|
88 |
-
|
89 |
def update_other_radios(value, other_radio1, other_radio2):
|
90 |
-
if value
|
91 |
-
if other_radio1
|
92 |
selected = other_radio1
|
93 |
else:
|
94 |
selected = other_radio2
|
95 |
return selected, other_radio1, other_radio2
|
96 |
return value, "", ""
|
97 |
|
98 |
-
|
99 |
def main():
|
100 |
-
#
|
101 |
with gr.Blocks() as demo:
|
102 |
with gr.Row():
|
103 |
-
options_coder = ["0.5B", "1.5B", "3B", "7B", "14B", "32B"
|
104 |
with gr.Row():
|
105 |
-
radio = gr.Radio(choices=options_coder, label="Qwen2.5-Coder
|
106 |
-
|
107 |
with gr.Row():
|
108 |
with gr.Accordion():
|
109 |
mark_ = gr.Markdown("""<center><font size=8>Qwen2.5-Coder-32B-Instruct Bot👾</center>""")
|
@@ -118,12 +115,12 @@ def main():
|
|
118 |
|
119 |
with gr.Row():
|
120 |
clear_history = gr.Button("🧹 Clear History")
|
121 |
-
|
122 |
|
123 |
textbox.submit(model_chat,
|
124 |
inputs=[textbox, chatbot, system_state, radio],
|
125 |
outputs=[textbox, chatbot, system_input])
|
126 |
-
|
127 |
inputs=[textbox, chatbot, system_state, radio],
|
128 |
outputs=[textbox, chatbot, system_input],
|
129 |
concurrency_limit=100)
|
@@ -134,13 +131,12 @@ def main():
|
|
134 |
inputs=[system_input],
|
135 |
outputs=[system_state, system_input, chatbot])
|
136 |
|
137 |
-
radio.change(
|
138 |
inputs=[radio, system_input],
|
139 |
outputs=[mark_, chatbot, system_state, system_input, textbox])
|
140 |
|
141 |
demo.queue(api_open=False, default_concurrency_limit=40)
|
142 |
-
demo.launch(max_threads=5)
|
143 |
-
|
144 |
|
145 |
if __name__ == "__main__":
|
146 |
main()
|
|
|
1 |
# app.py
|
2 |
|
3 |
import os
|
4 |
+
import logging
|
5 |
from http.client import HTTPMessage
|
|
|
|
|
|
|
|
|
6 |
from http import HTTPStatus
|
7 |
+
import gradio as gr
|
8 |
+
from dashscope import Generation, Role
|
9 |
+
from dashscope.api_entities.dashscope_response import DashScopeResponse
|
10 |
from typing import List, Optional, Tuple, Dict
|
11 |
from urllib.error import HTTPError
|
12 |
|
13 |
+
# Configure logging
|
14 |
+
logging.basicConfig(level=logging.INFO, format='%(asctime)s - %(levelname)s - %(message)s')
|
15 |
|
16 |
+
# Environment variable for API token
|
17 |
YOUR_API_TOKEN = os.getenv('YOUR_API_TOKEN')
|
18 |
+
if not YOUR_API_TOKEN:
|
19 |
+
logging.error("API token not found in environment variables.")
|
20 |
+
exit(1)
|
21 |
+
|
22 |
dashscope.api_key = YOUR_API_TOKEN
|
23 |
|
24 |
History = List[Tuple[str, str]]
|
25 |
Messages = List[Dict[str, str]]
|
26 |
|
27 |
+
default_system = 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.'
|
28 |
|
29 |
def clear_session() -> History:
|
30 |
return '', []
|
31 |
|
32 |
+
def modify_system_session(system: str) -> Tuple[str, str, History]:
|
33 |
+
if not system:
|
|
|
34 |
system = default_system
|
35 |
return system, system, []
|
36 |
|
|
|
37 |
def history_to_messages(history: History, system: str) -> Messages:
|
38 |
messages = [{'role': Role.SYSTEM, 'content': system}]
|
39 |
+
for user_msg, assistant_msg in history:
|
40 |
+
messages.append({'role': Role.USER, 'content': user_msg})
|
41 |
+
messages.append({'role': Role.ASSISTANT, 'content': assistant_msg})
|
42 |
return messages
|
43 |
|
|
|
44 |
def messages_to_history(messages: Messages) -> Tuple[str, History]:
|
45 |
assert messages[0]['role'] == Role.SYSTEM
|
46 |
system = messages[0]['content']
|
47 |
+
history = [(msg['content'], next_msg['content']) for msg, next_msg in zip(messages[1::2], messages[2::2])]
|
|
|
|
|
48 |
return system, history
|
49 |
|
50 |
+
def model_chat(query: Optional[str], history: Optional[History], system: str, radio: str) -> Tuple[str, str, History]:
|
51 |
+
if not query:
|
|
|
|
|
52 |
query = ''
|
53 |
+
if not history:
|
54 |
history = []
|
55 |
messages = history_to_messages(history, system)
|
56 |
messages.append({'role': Role.USER, 'content': query})
|
57 |
label_model = f"qwen2.5-coder-{radio.lower()}-instruct"
|
58 |
+
try:
|
59 |
+
gen = Generation.call(
|
60 |
+
model=label_model,
|
61 |
+
messages=messages,
|
62 |
+
result_format='message',
|
63 |
+
stream=True
|
64 |
+
)
|
65 |
+
for response in gen:
|
66 |
+
if response.status_code == HTTPStatus.OK:
|
67 |
+
role = response.output.choices[0].message.role
|
68 |
+
content = response.output.choices[0].message.content
|
69 |
+
system, history = messages_to_history(messages + [{'role': role, 'content': content}])
|
70 |
+
yield '', history, system
|
71 |
+
else:
|
72 |
+
raise HTTPError(code=response.status_code, msg=f'Request ID: {response.request_id}, Status Code: {response.status_code}, Error Code: {response.code}, Error Message: {response.message}', hdrs=HTTPMessage(), url='http://example.com', fp=None)
|
73 |
+
except HTTPError as e:
|
74 |
+
logging.error(f"HTTP Error: {e}")
|
75 |
+
yield '', history, system
|
76 |
+
except Exception as e:
|
77 |
+
logging.error(f"Unexpected error: {e}")
|
78 |
+
yield '', history, system
|
79 |
+
|
80 |
+
def choose_radio(radio, system):
|
81 |
mark_ = gr.Markdown(value=f"<center><font size=8>Qwen2.5-Coder-{radio}-instruct👾</center>")
|
82 |
chatbot = gr.Chatbot(label=f'Qwen2.5-Coder-{radio.lower()}-instruct')
|
83 |
+
if not system:
|
|
|
84 |
system = default_system
|
|
|
85 |
return mark_, chatbot, system, system, ""
|
86 |
|
|
|
87 |
def update_other_radios(value, other_radio1, other_radio2):
|
88 |
+
if not value:
|
89 |
+
if other_radio1:
|
90 |
selected = other_radio1
|
91 |
else:
|
92 |
selected = other_radio2
|
93 |
return selected, other_radio1, other_radio2
|
94 |
return value, "", ""
|
95 |
|
|
|
96 |
def main():
|
97 |
+
# Create two tabs
|
98 |
with gr.Blocks() as demo:
|
99 |
with gr.Row():
|
100 |
+
options_coder = ["0.5B", "1.5B", "3B", "7B", "14B", "32B"]
|
101 |
with gr.Row():
|
102 |
+
radio = gr.Radio(choices=options_coder, label="Qwen2.5-Coder:", value="32B")
|
103 |
+
|
104 |
with gr.Row():
|
105 |
with gr.Accordion():
|
106 |
mark_ = gr.Markdown("""<center><font size=8>Qwen2.5-Coder-32B-Instruct Bot👾</center>""")
|
|
|
115 |
|
116 |
with gr.Row():
|
117 |
clear_history = gr.Button("🧹 Clear History")
|
118 |
+
submit = gr.Button("🚀 Send")
|
119 |
|
120 |
textbox.submit(model_chat,
|
121 |
inputs=[textbox, chatbot, system_state, radio],
|
122 |
outputs=[textbox, chatbot, system_input])
|
123 |
+
submit.click(model_chat,
|
124 |
inputs=[textbox, chatbot, system_state, radio],
|
125 |
outputs=[textbox, chatbot, system_input],
|
126 |
concurrency_limit=100)
|
|
|
131 |
inputs=[system_input],
|
132 |
outputs=[system_state, system_input, chatbot])
|
133 |
|
134 |
+
radio.change(choose_radio,
|
135 |
inputs=[radio, system_input],
|
136 |
outputs=[mark_, chatbot, system_state, system_input, textbox])
|
137 |
|
138 |
demo.queue(api_open=False, default_concurrency_limit=40)
|
139 |
+
demo.launch(max_threads=5, server_name="0.0.0.0", server_port=7860, share=True)
|
|
|
140 |
|
141 |
if __name__ == "__main__":
|
142 |
main()
|