Spaces:
Running
on
Zero
Running
on
Zero
Update app.py
Browse files
app.py
CHANGED
@@ -3,6 +3,7 @@ from transformers import AutoModelForCausalLM, AutoTokenizer
|
|
3 |
import torch
|
4 |
import time
|
5 |
import spaces
|
|
|
6 |
|
7 |
# Model configurations
|
8 |
MODELS = {
|
@@ -44,10 +45,9 @@ def generate_response(model_id, conversation, user_message, max_length=512, temp
|
|
44 |
)
|
45 |
messages.append({"role": "system", "content": system_prompt})
|
46 |
|
47 |
-
# Add conversation history
|
48 |
for msg in conversation:
|
49 |
-
|
50 |
-
messages.append({"role": msg["role"], "content": msg["content"]})
|
51 |
|
52 |
# Add current user message
|
53 |
messages.append({"role": "user", "content": user_message})
|
@@ -76,33 +76,61 @@ def generate_response(model_id, conversation, user_message, max_length=512, temp
|
|
76 |
outputs[0][inputs['input_ids'].shape[-1]:],
|
77 |
skip_special_tokens=True
|
78 |
).strip()
|
|
|
79 |
return response, load_time, generation_time
|
80 |
|
81 |
-
def
|
82 |
-
"""
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
83 |
if not message.strip():
|
84 |
-
|
85 |
-
|
86 |
-
model_id = MODELS.get(model_name, MODELS["Athena-R3X
|
87 |
try:
|
88 |
-
# Format history for Athena
|
89 |
-
formatted_history = []
|
90 |
-
for i in range(0, len(history), 2):
|
91 |
-
if i < len(history):
|
92 |
-
user_msg = history[i][1] if history[i][0] == "user" else ""
|
93 |
-
assistant_msg = history[i+1][1] if i+1 < len(history) and history[i+1][0] == "assistant" else ""
|
94 |
-
if user_msg:
|
95 |
-
formatted_history.append({"role": "user", "content": user_msg})
|
96 |
-
if assistant_msg:
|
97 |
-
formatted_history.append({"role": "assistant", "content": assistant_msg})
|
98 |
response, load_time, generation_time = generate_response(
|
99 |
-
model_id,
|
100 |
)
|
101 |
-
|
102 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
103 |
except Exception as e:
|
104 |
-
|
105 |
-
|
|
|
106 |
|
107 |
css = """
|
108 |
.message {
|
@@ -110,22 +138,73 @@ css = """
|
|
110 |
margin: 5px;
|
111 |
border-radius: 10px;
|
112 |
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
113 |
"""
|
114 |
|
115 |
theme = gr.themes.Monochrome()
|
116 |
|
117 |
-
with gr.Blocks(title="Athena Playground Chat", css=css, theme=theme) as demo:
|
118 |
gr.Markdown("# 🚀 Athena Playground Chat")
|
119 |
gr.Markdown("*Powered by HuggingFace ZeroGPU*")
|
120 |
|
121 |
-
|
122 |
-
|
123 |
-
|
|
|
|
|
124 |
with gr.Row():
|
125 |
-
user_input = gr.Textbox(label="Your message", scale=8, autofocus=True)
|
126 |
-
send_btn = gr.Button(value="Send", scale=1)
|
127 |
|
128 |
-
#
|
129 |
gr.Markdown("### ⚙️ Model & Generation Settings")
|
130 |
with gr.Row():
|
131 |
model_choice = gr.Dropdown(
|
@@ -135,7 +214,7 @@ with gr.Blocks(title="Athena Playground Chat", css=css, theme=theme) as demo:
|
|
135 |
info="Select which Athena model to use"
|
136 |
)
|
137 |
max_length = gr.Slider(
|
138 |
-
32,
|
139 |
label="📝 Max Tokens",
|
140 |
info="Maximum number of tokens to generate"
|
141 |
)
|
@@ -145,14 +224,35 @@ with gr.Blocks(title="Athena Playground Chat", css=css, theme=theme) as demo:
|
|
145 |
info="Higher values = more creative responses"
|
146 |
)
|
147 |
|
148 |
-
|
149 |
-
|
150 |
-
|
|
|
|
|
|
|
|
|
151 |
send_btn.click(
|
152 |
chat_submit,
|
153 |
-
inputs=[
|
154 |
-
outputs=[chatbot,
|
155 |
)
|
156 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
157 |
if __name__ == "__main__":
|
158 |
demo.launch()
|
|
|
3 |
import torch
|
4 |
import time
|
5 |
import spaces
|
6 |
+
import re
|
7 |
|
8 |
# Model configurations
|
9 |
MODELS = {
|
|
|
45 |
)
|
46 |
messages.append({"role": "system", "content": system_prompt})
|
47 |
|
48 |
+
# Add conversation history
|
49 |
for msg in conversation:
|
50 |
+
messages.append(msg)
|
|
|
51 |
|
52 |
# Add current user message
|
53 |
messages.append({"role": "user", "content": user_message})
|
|
|
76 |
outputs[0][inputs['input_ids'].shape[-1]:],
|
77 |
skip_special_tokens=True
|
78 |
).strip()
|
79 |
+
print(f"Generation time: {generation_time:.2f}s")
|
80 |
return response, load_time, generation_time
|
81 |
|
82 |
+
def format_response_with_thinking(response):
|
83 |
+
"""Format response to handle <think></think> tags"""
|
84 |
+
# Check if response contains thinking tags
|
85 |
+
if '<think>' in response and '</think>' in response:
|
86 |
+
# Split the response into parts
|
87 |
+
pattern = r'(.*?)(<think>(.*?)</think>)(.*)'
|
88 |
+
match = re.search(pattern, response, re.DOTALL)
|
89 |
+
|
90 |
+
if match:
|
91 |
+
before_thinking = match.group(1).strip()
|
92 |
+
thinking_content = match.group(3).strip()
|
93 |
+
after_thinking = match.group(4).strip()
|
94 |
+
|
95 |
+
# Create HTML with collapsible thinking section
|
96 |
+
html = f"{before_thinking}\n"
|
97 |
+
html += f'<div class="thinking-container">'
|
98 |
+
html += f'<button class="thinking-toggle" onclick="this.nextElementSibling.classList.toggle(\'hidden\'); this.textContent = this.textContent === \'Show reasoning\' ? \'Hide reasoning\' : \'Show reasoning\'">Show reasoning</button>'
|
99 |
+
html += f'<div class="thinking-content hidden">{thinking_content}</div>'
|
100 |
+
html += f'</div>\n'
|
101 |
+
html += after_thinking
|
102 |
+
|
103 |
+
return html
|
104 |
+
|
105 |
+
# If no thinking tags, return the original response
|
106 |
+
return response
|
107 |
+
|
108 |
+
def chat_submit(message, chat_history, conversation_state, model_name, max_length, temperature):
|
109 |
+
"""Process a new message and update the chat history"""
|
110 |
if not message.strip():
|
111 |
+
return "", chat_history, conversation_state
|
112 |
+
|
113 |
+
model_id = MODELS.get(model_name, MODELS["Athena-R3X 4B"])
|
114 |
try:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
115 |
response, load_time, generation_time = generate_response(
|
116 |
+
model_id, conversation_state, message, max_length, temperature
|
117 |
)
|
118 |
+
|
119 |
+
# Update the conversation state with the raw response
|
120 |
+
conversation_state.append({"role": "user", "content": message})
|
121 |
+
conversation_state.append({"role": "assistant", "content": response})
|
122 |
+
|
123 |
+
# Format the response for display
|
124 |
+
formatted_response = format_response_with_thinking(response)
|
125 |
+
|
126 |
+
# Update the visible chat history
|
127 |
+
chat_history.append((message, formatted_response))
|
128 |
+
|
129 |
+
return "", chat_history, conversation_state
|
130 |
except Exception as e:
|
131 |
+
error_message = f"Error: {str(e)}"
|
132 |
+
chat_history.append((message, error_message))
|
133 |
+
return "", chat_history, conversation_state
|
134 |
|
135 |
css = """
|
136 |
.message {
|
|
|
138 |
margin: 5px;
|
139 |
border-radius: 10px;
|
140 |
}
|
141 |
+
|
142 |
+
.thinking-container {
|
143 |
+
margin: 10px 0;
|
144 |
+
}
|
145 |
+
|
146 |
+
.thinking-toggle {
|
147 |
+
background-color: #f1f1f1;
|
148 |
+
border: 1px solid #ddd;
|
149 |
+
border-radius: 4px;
|
150 |
+
padding: 5px 10px;
|
151 |
+
cursor: pointer;
|
152 |
+
font-size: 0.9em;
|
153 |
+
margin-bottom: 5px;
|
154 |
+
color: #555;
|
155 |
+
}
|
156 |
+
|
157 |
+
.thinking-content {
|
158 |
+
background-color: #f9f9f9;
|
159 |
+
border-left: 3px solid #ccc;
|
160 |
+
padding: 10px;
|
161 |
+
margin-top: 5px;
|
162 |
+
font-size: 0.95em;
|
163 |
+
color: #555;
|
164 |
+
font-family: monospace;
|
165 |
+
white-space: pre-wrap;
|
166 |
+
overflow-x: auto;
|
167 |
+
}
|
168 |
+
|
169 |
+
.hidden {
|
170 |
+
display: none;
|
171 |
+
}
|
172 |
+
"""
|
173 |
+
|
174 |
+
# Add JavaScript to handle the toggle functionality
|
175 |
+
js = """
|
176 |
+
function setupThinkingToggles() {
|
177 |
+
document.querySelectorAll('.thinking-toggle').forEach(button => {
|
178 |
+
button.addEventListener('click', function() {
|
179 |
+
const content = this.nextElementSibling;
|
180 |
+
content.classList.toggle('hidden');
|
181 |
+
this.textContent = content.classList.contains('hidden') ? 'Show reasoning' : 'Hide reasoning';
|
182 |
+
});
|
183 |
+
});
|
184 |
+
}
|
185 |
+
|
186 |
+
// Run after the page loads and when the chat updates
|
187 |
+
document.addEventListener('DOMContentLoaded', setupThinkingToggles);
|
188 |
+
const observer = new MutationObserver(setupThinkingToggles);
|
189 |
+
observer.observe(document.body, { childList: true, subtree: true });
|
190 |
"""
|
191 |
|
192 |
theme = gr.themes.Monochrome()
|
193 |
|
194 |
+
with gr.Blocks(title="Athena Playground Chat", css=css, theme=theme, js=js) as demo:
|
195 |
gr.Markdown("# 🚀 Athena Playground Chat")
|
196 |
gr.Markdown("*Powered by HuggingFace ZeroGPU*")
|
197 |
|
198 |
+
# State to keep track of the conversation for the model
|
199 |
+
conversation_state = gr.State([])
|
200 |
+
|
201 |
+
chatbot = gr.Chatbot(height=500, label="Athena", render_markdown=True)
|
202 |
+
|
203 |
with gr.Row():
|
204 |
+
user_input = gr.Textbox(label="Your message", scale=8, autofocus=True, placeholder="Type your message here...")
|
205 |
+
send_btn = gr.Button(value="Send", scale=1, variant="primary")
|
206 |
|
207 |
+
# Configuration controls
|
208 |
gr.Markdown("### ⚙️ Model & Generation Settings")
|
209 |
with gr.Row():
|
210 |
model_choice = gr.Dropdown(
|
|
|
214 |
info="Select which Athena model to use"
|
215 |
)
|
216 |
max_length = gr.Slider(
|
217 |
+
32, 8000, value=512,
|
218 |
label="📝 Max Tokens",
|
219 |
info="Maximum number of tokens to generate"
|
220 |
)
|
|
|
224 |
info="Higher values = more creative responses"
|
225 |
)
|
226 |
|
227 |
+
# Connect the interface components
|
228 |
+
submit_event = user_input.submit(
|
229 |
+
chat_submit,
|
230 |
+
inputs=[user_input, chatbot, conversation_state, model_choice, max_length, temperature],
|
231 |
+
outputs=[user_input, chatbot, conversation_state]
|
232 |
+
)
|
233 |
+
|
234 |
send_btn.click(
|
235 |
chat_submit,
|
236 |
+
inputs=[user_input, chatbot, conversation_state, model_choice, max_length, temperature],
|
237 |
+
outputs=[user_input, chatbot, conversation_state]
|
238 |
)
|
239 |
|
240 |
+
# Add examples if desired
|
241 |
+
gr.Examples(
|
242 |
+
examples=[
|
243 |
+
"What is artificial intelligence?",
|
244 |
+
"Can you explain quantum computing?",
|
245 |
+
"Write a short poem about technology",
|
246 |
+
"What are some ethical concerns about AI?"
|
247 |
+
],
|
248 |
+
inputs=[user_input]
|
249 |
+
)
|
250 |
+
|
251 |
+
gr.Markdown("""
|
252 |
+
### About the Thinking Tags
|
253 |
+
Some Athena models (particularly R3X series) include reasoning in `<think></think>` tags.
|
254 |
+
Click "Show reasoning" to see the model's thought process behind its answers.
|
255 |
+
""")
|
256 |
+
|
257 |
if __name__ == "__main__":
|
258 |
demo.launch()
|