Spaces:
Running
Running
SkyNetWalker
commited on
Update app.py
Browse files
app.py
CHANGED
@@ -16,17 +16,29 @@ client = OpenAI(
|
|
16 |
|
17 |
print("Client initialized.")
|
18 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
19 |
def respond(
|
20 |
message,
|
21 |
history: list[tuple[str, str]],
|
22 |
-
|
|
|
23 |
max_tokens,
|
24 |
temperature,
|
25 |
top_p,
|
26 |
-
model_name,
|
27 |
):
|
28 |
print(f"Received message: {message}")
|
29 |
print(f"History: {history}")
|
|
|
|
|
|
|
|
|
30 |
print(f"System message: {system_message}")
|
31 |
print(f"Max tokens: {max_tokens}, Temperature: {temperature}, Top-P: {top_p}")
|
32 |
print(f"Selected model: {model_name}")
|
@@ -47,7 +59,7 @@ def respond(
|
|
47 |
print("Sending request to OpenAI API.")
|
48 |
|
49 |
for message in client.chat.completions.create(
|
50 |
-
model=model_name,
|
51 |
max_tokens=max_tokens,
|
52 |
stream=True,
|
53 |
temperature=temperature,
|
@@ -60,40 +72,68 @@ def respond(
|
|
60 |
yield response
|
61 |
|
62 |
print("Completed response generation.")
|
63 |
-
|
64 |
chatbot = gr.Chatbot(height=400)
|
65 |
|
66 |
print("Chatbot interface created.")
|
67 |
|
68 |
-
# Define the list of models
|
69 |
models = [
|
70 |
-
"PowerInfer/SmallThinker-3B-Preview",
|
71 |
-
"Qwen/QwQ-32B-Preview",
|
72 |
-
"Qwen/Qwen2.5-Coder-32B-Instruct",
|
73 |
-
"meta-llama/Llama-3.2-3B-Instruct",
|
74 |
-
|
75 |
-
"microsoft/Phi-3-mini-128k-instruct", #fail
|
76 |
-
#"microsoft/Phi-3-medium-128k-instruct", #fail
|
77 |
-
#"microsoft/phi-4", #fail, too large to be loaded automatically (29GB > 10GB)
|
78 |
-
#"meta-llama/Llama-3.3-70B-Instruct", #fail, need HF Pro subscription
|
79 |
]
|
80 |
|
81 |
-
# Add a title and move the model dropdown to the top
|
82 |
with gr.Blocks() as demo:
|
83 |
-
gr.Markdown("# LLM Test")
|
84 |
|
85 |
-
|
86 |
-
|
|
|
|
|
|
|
|
|
|
|
87 |
|
88 |
-
#
|
89 |
-
gr.ChatInterface(
|
90 |
respond,
|
91 |
additional_inputs=[
|
92 |
-
|
93 |
-
gr.
|
94 |
-
|
95 |
-
|
96 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
97 |
],
|
98 |
fill_height=True,
|
99 |
chatbot=chatbot,
|
|
|
16 |
|
17 |
print("Client initialized.")
|
18 |
|
19 |
+
# Pre-set system prompts
|
20 |
+
SYSTEM_PROMPTS = {
|
21 |
+
"General Assistant": "You are a helpful, respectful and honest assistant. Always provide accurate information and admit when you're not sure about something.",
|
22 |
+
"Code Helper": "You are a programming assistant. Help users with coding questions, debugging, and best practices. Provide clear explanations and code examples when appropriate.",
|
23 |
+
"Creative Writer": "You are a creative writing assistant. Help users with storytelling, character development, and creative writing techniques. Be imaginative and encouraging."
|
24 |
+
}
|
25 |
+
|
26 |
def respond(
|
27 |
message,
|
28 |
history: list[tuple[str, str]],
|
29 |
+
preset_prompt,
|
30 |
+
custom_prompt,
|
31 |
max_tokens,
|
32 |
temperature,
|
33 |
top_p,
|
34 |
+
model_name,
|
35 |
):
|
36 |
print(f"Received message: {message}")
|
37 |
print(f"History: {history}")
|
38 |
+
|
39 |
+
# Use custom prompt if provided, otherwise use selected preset
|
40 |
+
system_message = custom_prompt if custom_prompt.strip() else SYSTEM_PROMPTS[preset_prompt]
|
41 |
+
|
42 |
print(f"System message: {system_message}")
|
43 |
print(f"Max tokens: {max_tokens}, Temperature: {temperature}, Top-P: {top_p}")
|
44 |
print(f"Selected model: {model_name}")
|
|
|
59 |
print("Sending request to OpenAI API.")
|
60 |
|
61 |
for message in client.chat.completions.create(
|
62 |
+
model=model_name,
|
63 |
max_tokens=max_tokens,
|
64 |
stream=True,
|
65 |
temperature=temperature,
|
|
|
72 |
yield response
|
73 |
|
74 |
print("Completed response generation.")
|
75 |
+
|
76 |
chatbot = gr.Chatbot(height=400)
|
77 |
|
78 |
print("Chatbot interface created.")
|
79 |
|
|
|
80 |
models = [
|
81 |
+
"PowerInfer/SmallThinker-3B-Preview",
|
82 |
+
"Qwen/QwQ-32B-Preview",
|
83 |
+
"Qwen/Qwen2.5-Coder-32B-Instruct",
|
84 |
+
"meta-llama/Llama-3.2-3B-Instruct",
|
85 |
+
"microsoft/Phi-3-mini-128k-instruct",
|
|
|
|
|
|
|
|
|
86 |
]
|
87 |
|
|
|
88 |
with gr.Blocks() as demo:
|
89 |
+
gr.Markdown("# LLM Test")
|
90 |
|
91 |
+
with gr.Row():
|
92 |
+
# Model selection at the top
|
93 |
+
model_dropdown = gr.Dropdown(
|
94 |
+
choices=models,
|
95 |
+
value=models[0],
|
96 |
+
label="Select Model:"
|
97 |
+
)
|
98 |
|
99 |
+
# Chatbot interface
|
100 |
+
chat_interface = gr.ChatInterface(
|
101 |
respond,
|
102 |
additional_inputs=[
|
103 |
+
# Prompt selections
|
104 |
+
gr.Dropdown(
|
105 |
+
choices=list(SYSTEM_PROMPTS.keys()),
|
106 |
+
value=list(SYSTEM_PROMPTS.keys())[0],
|
107 |
+
label="Select System Prompt:"
|
108 |
+
),
|
109 |
+
gr.Textbox(
|
110 |
+
value="",
|
111 |
+
label="Custom System Prompt (leaves blank to use preset):",
|
112 |
+
lines=2
|
113 |
+
),
|
114 |
+
# Other parameters
|
115 |
+
gr.Slider(
|
116 |
+
minimum=1,
|
117 |
+
maximum=4096,
|
118 |
+
value=1024,
|
119 |
+
step=1,
|
120 |
+
label="Max new tokens:"
|
121 |
+
),
|
122 |
+
gr.Slider(
|
123 |
+
minimum=0.1,
|
124 |
+
maximum=1.0,
|
125 |
+
value=0.3,
|
126 |
+
step=0.1,
|
127 |
+
label="Temperature:"
|
128 |
+
),
|
129 |
+
gr.Slider(
|
130 |
+
minimum=0.1,
|
131 |
+
maximum=1.0,
|
132 |
+
value=0.95,
|
133 |
+
step=0.05,
|
134 |
+
label="Top-P:"
|
135 |
+
),
|
136 |
+
model_dropdown,
|
137 |
],
|
138 |
fill_height=True,
|
139 |
chatbot=chatbot,
|