Spaces:
Sleeping
Sleeping
Create app.py
Browse files
app.py
ADDED
@@ -0,0 +1,366 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import gradio as gr
|
2 |
+
import requests
|
3 |
+
import os
|
4 |
+
import json
|
5 |
+
from datetime import datetime
|
6 |
+
from typing import List, Tuple
|
7 |
+
|
8 |
+
# Configuration
|
9 |
+
OPENROUTER_API_KEY = os.getenv("OPENROUTER_API_KEY", "")
|
10 |
+
OPENROUTER_BASE_URL = "https://openrouter.ai/api/v1"
|
11 |
+
|
12 |
+
# Available models with details
|
13 |
+
MODELS = {
|
14 |
+
"OpenAI": {
|
15 |
+
"gpt-4-turbo-preview": "Latest GPT-4 Turbo",
|
16 |
+
"gpt-4": "GPT-4 Standard",
|
17 |
+
"gpt-3.5-turbo": "GPT-3.5 Turbo (Fastest)",
|
18 |
+
"gpt-3.5-turbo-16k": "GPT-3.5 Turbo 16K"
|
19 |
+
},
|
20 |
+
"Anthropic": {
|
21 |
+
"anthropic/claude-3-opus-20240229": "Claude 3 Opus (Most Powerful)",
|
22 |
+
"anthropic/claude-3-sonnet-20240229": "Claude 3 Sonnet (Balanced)",
|
23 |
+
"anthropic/claude-3-haiku-20240307": "Claude 3 Haiku (Fastest)",
|
24 |
+
"anthropic/claude-2.1": "Claude 2.1",
|
25 |
+
"anthropic/claude-instant-1.2": "Claude Instant"
|
26 |
+
},
|
27 |
+
"Google": {
|
28 |
+
"google/gemini-pro": "Gemini Pro",
|
29 |
+
"google/palm-2-chat-bison": "PaLM 2 Chat"
|
30 |
+
},
|
31 |
+
"Meta": {
|
32 |
+
"meta-llama/llama-3-70b-instruct": "Llama 3 70B",
|
33 |
+
"meta-llama/llama-3-8b-instruct": "Llama 3 8B",
|
34 |
+
"meta-llama/llama-2-70b-chat": "Llama 2 70B"
|
35 |
+
},
|
36 |
+
"Mistral": {
|
37 |
+
"mistralai/mistral-7b-instruct": "Mistral 7B",
|
38 |
+
"mistralai/mixtral-8x7b-instruct": "Mixtral 8x7B"
|
39 |
+
},
|
40 |
+
"Others": {
|
41 |
+
"databricks/dbrx-instruct": "DBRX Instruct",
|
42 |
+
"cohere/command-r-plus": "Command R+",
|
43 |
+
"zero-one-ai/yi-34b-chat": "Yi 34B Chat"
|
44 |
+
}
|
45 |
+
}
|
46 |
+
|
47 |
+
# Flatten models for dropdown
|
48 |
+
MODEL_OPTIONS = []
|
49 |
+
for provider, models in MODELS.items():
|
50 |
+
for model_id, model_name in models.items():
|
51 |
+
MODEL_OPTIONS.append(f"{provider}: {model_name}")
|
52 |
+
|
53 |
+
# Custom CSS
|
54 |
+
custom_css = """
|
55 |
+
.gradio-container {
|
56 |
+
font-family: 'Inter', sans-serif;
|
57 |
+
}
|
58 |
+
|
59 |
+
.main-container {
|
60 |
+
max-width: 1200px;
|
61 |
+
margin: 0 auto;
|
62 |
+
}
|
63 |
+
|
64 |
+
/* Chat container styling */
|
65 |
+
.chat-container {
|
66 |
+
height: 600px !important;
|
67 |
+
overflow-y: auto;
|
68 |
+
border-radius: 10px;
|
69 |
+
padding: 20px;
|
70 |
+
background-color: var(--background-fill-secondary);
|
71 |
+
}
|
72 |
+
|
73 |
+
/* Message styling */
|
74 |
+
.message {
|
75 |
+
margin: 10px 0;
|
76 |
+
padding: 15px;
|
77 |
+
border-radius: 10px;
|
78 |
+
animation: fadeIn 0.3s ease-in;
|
79 |
+
}
|
80 |
+
|
81 |
+
.user-message {
|
82 |
+
background-color: var(--primary-500);
|
83 |
+
color: white;
|
84 |
+
margin-left: 20%;
|
85 |
+
}
|
86 |
+
|
87 |
+
.assistant-message {
|
88 |
+
background-color: var(--neutral-100);
|
89 |
+
margin-right: 20%;
|
90 |
+
}
|
91 |
+
|
92 |
+
/* Input area styling */
|
93 |
+
.input-container {
|
94 |
+
border-top: 2px solid var(--border-color-primary);
|
95 |
+
padding-top: 20px;
|
96 |
+
margin-top: 20px;
|
97 |
+
}
|
98 |
+
|
99 |
+
/* Model selector styling */
|
100 |
+
.model-selector {
|
101 |
+
background-color: var(--background-fill-secondary);
|
102 |
+
padding: 10px;
|
103 |
+
border-radius: 8px;
|
104 |
+
margin-bottom: 10px;
|
105 |
+
}
|
106 |
+
|
107 |
+
/* Animation */
|
108 |
+
@keyframes fadeIn {
|
109 |
+
from { opacity: 0; transform: translateY(10px); }
|
110 |
+
to { opacity: 1; transform: translateY(0); }
|
111 |
+
}
|
112 |
+
|
113 |
+
/* Status indicator */
|
114 |
+
.status-indicator {
|
115 |
+
display: inline-block;
|
116 |
+
width: 10px;
|
117 |
+
height: 10px;
|
118 |
+
border-radius: 50%;
|
119 |
+
margin-right: 5px;
|
120 |
+
}
|
121 |
+
|
122 |
+
.status-online { background-color: #10b981; }
|
123 |
+
.status-offline { background-color: #ef4444; }
|
124 |
+
.status-loading { background-color: #f59e0b; }
|
125 |
+
"""
|
126 |
+
|
127 |
+
def get_model_id_from_name(model_name: str) -> str:
|
128 |
+
"""Extract model ID from display name"""
|
129 |
+
for provider, models in MODELS.items():
|
130 |
+
for model_id, name in models.items():
|
131 |
+
if f"{provider}: {name}" == model_name:
|
132 |
+
return model_id
|
133 |
+
return "gpt-3.5-turbo" # fallback
|
134 |
+
|
135 |
+
def check_api_status() -> bool:
|
136 |
+
"""Check if API key is set and valid"""
|
137 |
+
return bool(OPENROUTER_API_KEY)
|
138 |
+
|
139 |
+
def format_message_history(history: List[Tuple[str, str]]) -> List[dict]:
|
140 |
+
"""Format chat history for API request"""
|
141 |
+
messages = []
|
142 |
+
for user_msg, assistant_msg in history:
|
143 |
+
if user_msg:
|
144 |
+
messages.append({"role": "user", "content": user_msg})
|
145 |
+
if assistant_msg:
|
146 |
+
messages.append({"role": "assistant", "content": assistant_msg})
|
147 |
+
return messages
|
148 |
+
|
149 |
+
def call_openrouter_api(
|
150 |
+
message: str,
|
151 |
+
history: List[Tuple[str, str]],
|
152 |
+
model_name: str,
|
153 |
+
temperature: float,
|
154 |
+
max_tokens: int,
|
155 |
+
system_prompt: str
|
156 |
+
) -> str:
|
157 |
+
"""Call OpenRouter API"""
|
158 |
+
|
159 |
+
if not check_api_status():
|
160 |
+
return "β API Key not set. Please set OPENROUTER_API_KEY in Space Settings > Variables and Secrets"
|
161 |
+
|
162 |
+
model_id = get_model_id_from_name(model_name)
|
163 |
+
|
164 |
+
headers = {
|
165 |
+
"Authorization": f"Bearer {OPENROUTER_API_KEY}",
|
166 |
+
"Content-Type": "application/json",
|
167 |
+
"HTTP-Referer": "https://huggingface.co",
|
168 |
+
"X-Title": "HuggingFace Space"
|
169 |
+
}
|
170 |
+
|
171 |
+
# Build messages
|
172 |
+
messages = []
|
173 |
+
if system_prompt:
|
174 |
+
messages.append({"role": "system", "content": system_prompt})
|
175 |
+
|
176 |
+
messages.extend(format_message_history(history))
|
177 |
+
messages.append({"role": "user", "content": message})
|
178 |
+
|
179 |
+
data = {
|
180 |
+
"model": model_id,
|
181 |
+
"messages": messages,
|
182 |
+
"temperature": temperature,
|
183 |
+
"max_tokens": max_tokens,
|
184 |
+
"stream": True
|
185 |
+
}
|
186 |
+
|
187 |
+
try:
|
188 |
+
response = requests.post(
|
189 |
+
f"{OPENROUTER_BASE_URL}/chat/completions",
|
190 |
+
headers=headers,
|
191 |
+
json=data,
|
192 |
+
stream=True
|
193 |
+
)
|
194 |
+
|
195 |
+
if response.status_code != 200:
|
196 |
+
return f"β Error: {response.status_code} - {response.text}"
|
197 |
+
|
198 |
+
# Handle streaming response
|
199 |
+
full_response = ""
|
200 |
+
for line in response.iter_lines():
|
201 |
+
if line:
|
202 |
+
line_str = line.decode('utf-8')
|
203 |
+
if line_str.startswith("data: "):
|
204 |
+
if line_str == "data: [DONE]":
|
205 |
+
break
|
206 |
+
try:
|
207 |
+
chunk = json.loads(line_str[6:])
|
208 |
+
if 'choices' in chunk and chunk['choices']:
|
209 |
+
delta = chunk['choices'][0].get('delta', {})
|
210 |
+
content = delta.get('content', '')
|
211 |
+
full_response += content
|
212 |
+
yield full_response
|
213 |
+
except json.JSONDecodeError:
|
214 |
+
continue
|
215 |
+
|
216 |
+
except requests.exceptions.RequestException as e:
|
217 |
+
yield f"β Connection error: {str(e)}"
|
218 |
+
except Exception as e:
|
219 |
+
yield f"β Unexpected error: {str(e)}"
|
220 |
+
|
221 |
+
# Create Gradio interface
|
222 |
+
with gr.Blocks(css=custom_css, theme=gr.themes.Soft()) as demo:
|
223 |
+
# Header
|
224 |
+
gr.Markdown(
|
225 |
+
"""
|
226 |
+
# π OpenRouter Multi-Model Chat
|
227 |
+
|
228 |
+
<div style='text-align: center; margin-bottom: 20px;'>
|
229 |
+
<p>Chat with various AI models including GPT-4, Claude 3, Gemini, and more!</p>
|
230 |
+
<p>Powered by <a href='https://openrouter.ai' target='_blank'>OpenRouter</a></p>
|
231 |
+
</div>
|
232 |
+
"""
|
233 |
+
)
|
234 |
+
|
235 |
+
# API Status
|
236 |
+
with gr.Row():
|
237 |
+
with gr.Column(scale=1):
|
238 |
+
api_status = gr.Markdown(
|
239 |
+
f"**API Status:** {'π’ Connected' if check_api_status() else 'π΄ Not Connected'}"
|
240 |
+
)
|
241 |
+
|
242 |
+
# Main chat interface
|
243 |
+
with gr.Row():
|
244 |
+
# Left sidebar - Settings
|
245 |
+
with gr.Column(scale=3):
|
246 |
+
model_dropdown = gr.Dropdown(
|
247 |
+
choices=MODEL_OPTIONS,
|
248 |
+
value="OpenAI: GPT-3.5 Turbo (Fastest)",
|
249 |
+
label="π€ Select Model",
|
250 |
+
interactive=True
|
251 |
+
)
|
252 |
+
|
253 |
+
with gr.Accordion("βοΈ Advanced Settings", open=False):
|
254 |
+
system_prompt = gr.Textbox(
|
255 |
+
label="System Prompt",
|
256 |
+
placeholder="You are a helpful assistant...",
|
257 |
+
lines=3
|
258 |
+
)
|
259 |
+
|
260 |
+
temperature = gr.Slider(
|
261 |
+
minimum=0,
|
262 |
+
maximum=2,
|
263 |
+
value=0.7,
|
264 |
+
step=0.1,
|
265 |
+
label="Temperature (Creativity)"
|
266 |
+
)
|
267 |
+
|
268 |
+
max_tokens = gr.Slider(
|
269 |
+
minimum=50,
|
270 |
+
maximum=4096,
|
271 |
+
value=1024,
|
272 |
+
step=50,
|
273 |
+
label="Max Tokens"
|
274 |
+
)
|
275 |
+
|
276 |
+
# Model info
|
277 |
+
gr.Markdown(
|
278 |
+
"""
|
279 |
+
### π Model Categories
|
280 |
+
- **OpenAI**: GPT-4, GPT-3.5
|
281 |
+
- **Anthropic**: Claude 3 (Opus, Sonnet, Haiku)
|
282 |
+
- **Google**: Gemini Pro, PaLM 2
|
283 |
+
- **Meta**: Llama 3, Llama 2
|
284 |
+
- **Mistral**: Mistral 7B, Mixtral
|
285 |
+
"""
|
286 |
+
)
|
287 |
+
|
288 |
+
# Right side - Chat
|
289 |
+
with gr.Column(scale=7):
|
290 |
+
chatbot = gr.Chatbot(
|
291 |
+
height=500,
|
292 |
+
label="π¬ Conversation",
|
293 |
+
elem_classes="chat-container"
|
294 |
+
)
|
295 |
+
|
296 |
+
with gr.Row():
|
297 |
+
msg = gr.Textbox(
|
298 |
+
label="Type your message",
|
299 |
+
placeholder="Ask me anything...",
|
300 |
+
lines=2,
|
301 |
+
scale=9
|
302 |
+
)
|
303 |
+
|
304 |
+
with gr.Column(scale=1, min_width=80):
|
305 |
+
send_btn = gr.Button("Send π€", variant="primary")
|
306 |
+
clear_btn = gr.Button("Clear ποΈ")
|
307 |
+
|
308 |
+
# Examples
|
309 |
+
gr.Examples(
|
310 |
+
examples=[
|
311 |
+
"What are the key differences between GPT-4 and Claude 3?",
|
312 |
+
"Write a Python function to calculate fibonacci numbers",
|
313 |
+
"Explain quantum computing in simple terms",
|
314 |
+
"What's the meaning of life?",
|
315 |
+
"Help me write a professional email"
|
316 |
+
],
|
317 |
+
inputs=msg,
|
318 |
+
label="π‘ Example Prompts"
|
319 |
+
)
|
320 |
+
|
321 |
+
# Event handlers
|
322 |
+
def respond(message, chat_history, model, temp, tokens, sys_prompt):
|
323 |
+
bot_message = ""
|
324 |
+
for response in call_openrouter_api(
|
325 |
+
message, chat_history, model, temp, tokens, sys_prompt
|
326 |
+
):
|
327 |
+
bot_message = response
|
328 |
+
yield chat_history + [(message, bot_message)]
|
329 |
+
|
330 |
+
def clear_chat():
|
331 |
+
return None
|
332 |
+
|
333 |
+
# Connect events
|
334 |
+
msg.submit(
|
335 |
+
respond,
|
336 |
+
[msg, chatbot, model_dropdown, temperature, max_tokens, system_prompt],
|
337 |
+
chatbot
|
338 |
+
).then(
|
339 |
+
lambda: "", None, msg
|
340 |
+
)
|
341 |
+
|
342 |
+
send_btn.click(
|
343 |
+
respond,
|
344 |
+
[msg, chatbot, model_dropdown, temperature, max_tokens, system_prompt],
|
345 |
+
chatbot
|
346 |
+
).then(
|
347 |
+
lambda: "", None, msg
|
348 |
+
)
|
349 |
+
|
350 |
+
clear_btn.click(clear_chat, None, chatbot)
|
351 |
+
|
352 |
+
# Footer
|
353 |
+
gr.Markdown(
|
354 |
+
"""
|
355 |
+
---
|
356 |
+
<div style='text-align: center; color: gray;'>
|
357 |
+
Made with β€οΈ using Gradio |
|
358 |
+
<a href='https://github.com/yourusername/yourrepo' target='_blank'>GitHub</a> |
|
359 |
+
<a href='https://openrouter.ai/docs' target='_blank'>API Docs</a>
|
360 |
+
</div>
|
361 |
+
"""
|
362 |
+
)
|
363 |
+
|
364 |
+
# Launch
|
365 |
+
if __name__ == "__main__":
|
366 |
+
demo.launch()
|