Update app.py
Browse files
app.py
CHANGED
|
@@ -24,57 +24,6 @@ def generate_conversation_id():
|
|
| 24 |
return str(uuid.uuid4())[:8]
|
| 25 |
|
| 26 |
|
| 27 |
-
'''def generate_response(user_message, max_tokens, temperature, top_p, history_state):
|
| 28 |
-
if not user_message.strip():
|
| 29 |
-
return history_state, history_state
|
| 30 |
-
|
| 31 |
-
start_tag = "<|im_start|>"
|
| 32 |
-
sep_tag = "<|im_sep|>"
|
| 33 |
-
end_tag = "<|im_end|>"
|
| 34 |
-
|
| 35 |
-
system_message = "Your role as an assistant..."
|
| 36 |
-
messages = [{"role": "system", "content": system_message}]
|
| 37 |
-
for message in history_state:
|
| 38 |
-
messages.append({"role": message["role"], "content": message["content"]})
|
| 39 |
-
messages.append({"role": "user", "content": user_message})
|
| 40 |
-
|
| 41 |
-
try:
|
| 42 |
-
response = client.chat.completions.create(
|
| 43 |
-
model="tgi",
|
| 44 |
-
messages=messages,
|
| 45 |
-
max_tokens=int(max_tokens),
|
| 46 |
-
temperature=temperature,
|
| 47 |
-
top_p=top_p,
|
| 48 |
-
stream=True,
|
| 49 |
-
extra_body={"max_new_tokens": int(max_tokens)}
|
| 50 |
-
)
|
| 51 |
-
except Exception as e:
|
| 52 |
-
print(f"[ERROR] OpenAI API call failed: {e}")
|
| 53 |
-
yield history_state + [{"role": "user", "content": user_message},
|
| 54 |
-
{"role": "assistant", "content": "⚠️ Generation failed."}], history_state
|
| 55 |
-
return
|
| 56 |
-
|
| 57 |
-
|
| 58 |
-
assistant_response = ""
|
| 59 |
-
new_history = history_state + [
|
| 60 |
-
{"role": "user", "content": user_message},
|
| 61 |
-
{"role": "assistant", "content": ""}
|
| 62 |
-
]
|
| 63 |
-
|
| 64 |
-
try:
|
| 65 |
-
for chunk in response:
|
| 66 |
-
if not chunk.choices or not chunk.choices[0].delta or not chunk.choices[0].delta.content:
|
| 67 |
-
continue
|
| 68 |
-
token = chunk.choices[0].delta.content
|
| 69 |
-
assistant_response += token
|
| 70 |
-
new_history[-1]["content"] = assistant_response.strip()
|
| 71 |
-
yield new_history, new_history
|
| 72 |
-
except Exception:
|
| 73 |
-
pass
|
| 74 |
-
|
| 75 |
-
yield new_history, new_history'''
|
| 76 |
-
|
| 77 |
-
|
| 78 |
import tiktoken
|
| 79 |
enc = tiktoken.encoding_for_model("gpt-3.5-turbo") # any OpenAI encoding works
|
| 80 |
|
|
@@ -272,4 +221,14 @@ with gr.Blocks(theme=gr.themes.Soft()) as demo:
|
|
| 272 |
example3_button.click(fn=lambda: gr.update(value=example_messages["JEE Main 2025 Probability & Statistics"]), inputs=None, outputs=user_input)
|
| 273 |
example4_button.click(fn=lambda: gr.update(value=example_messages["JEE Main 2025 Laws of Motion"]), inputs=None, outputs=user_input)
|
| 274 |
|
| 275 |
-
demo.launch(share=True, ssr_mode=False)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 24 |
return str(uuid.uuid4())[:8]
|
| 25 |
|
| 26 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 27 |
import tiktoken
|
| 28 |
enc = tiktoken.encoding_for_model("gpt-3.5-turbo") # any OpenAI encoding works
|
| 29 |
|
|
|
|
| 221 |
example3_button.click(fn=lambda: gr.update(value=example_messages["JEE Main 2025 Probability & Statistics"]), inputs=None, outputs=user_input)
|
| 222 |
example4_button.click(fn=lambda: gr.update(value=example_messages["JEE Main 2025 Laws of Motion"]), inputs=None, outputs=user_input)
|
| 223 |
|
| 224 |
+
#demo.launch(share=True, ssr_mode=False)
|
| 225 |
+
|
| 226 |
+
if __name__ == "__main__":
|
| 227 |
+
demo.queue( # turn the queue on
|
| 228 |
+
concurrency_count = 8, # how many jobs run simultaneously
|
| 229 |
+
max_size = 40, # 40-6 = 34 jobs can wait in line
|
| 230 |
+
api_open = False # no public /queue/status endpoint
|
| 231 |
+
).launch(
|
| 232 |
+
share = True,
|
| 233 |
+
ssr_mode = False
|
| 234 |
+
)
|