Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -18,25 +18,6 @@ models = {
|
|
| 18 |
def get_client(model):
|
| 19 |
return InferenceClient(model=model, token=hf_token)
|
| 20 |
|
| 21 |
-
# 히스토리 관리 클래스
|
| 22 |
-
class PromptHistory:
|
| 23 |
-
def __init__(self):
|
| 24 |
-
self.history = []
|
| 25 |
-
|
| 26 |
-
def add_entry(self, prompt, response, model, system_message):
|
| 27 |
-
self.history.append({
|
| 28 |
-
"prompt": prompt,
|
| 29 |
-
"response": response,
|
| 30 |
-
"model": model,
|
| 31 |
-
"system_message": system_message
|
| 32 |
-
})
|
| 33 |
-
|
| 34 |
-
def get_history(self):
|
| 35 |
-
return self.history
|
| 36 |
-
|
| 37 |
-
# 히스토리 인스턴스 생성
|
| 38 |
-
prompt_history = PromptHistory()
|
| 39 |
-
|
| 40 |
# 응답 생성 함수
|
| 41 |
def respond(prompt, system_message, max_tokens, temperature, top_p, selected_model):
|
| 42 |
stop_event.clear()
|
|
@@ -50,7 +31,6 @@ def respond(prompt, system_message, max_tokens, temperature, top_p, selected_mod
|
|
| 50 |
|
| 51 |
try:
|
| 52 |
response = ""
|
| 53 |
-
total_tokens_used = 0 # 사용된 토큰 수 추적
|
| 54 |
|
| 55 |
# 모델에서 응답을 스트리밍
|
| 56 |
for chunk in client.text_generation(
|
|
@@ -64,31 +44,16 @@ def respond(prompt, system_message, max_tokens, temperature, top_p, selected_mod
|
|
| 64 |
break
|
| 65 |
if chunk:
|
| 66 |
response += chunk
|
| 67 |
-
total_tokens_used += len(chunk.split()) # 청크당 사용된 토큰 수 추산
|
| 68 |
|
| 69 |
-
#
|
| 70 |
-
prompt_history.add_entry(prompt, response, selected_model, system_message)
|
| 71 |
-
|
| 72 |
-
return response
|
| 73 |
|
| 74 |
except Exception as e:
|
| 75 |
-
return f"오류 발생: {str(e)}"
|
| 76 |
-
|
| 77 |
-
# 대화 계속 함수
|
| 78 |
-
def continue_writing(chatbot, system_message, max_tokens, temperature, top_p, selected_model):
|
| 79 |
-
last_message = chatbot[-1][1] if chatbot else ""
|
| 80 |
-
return respond(last_message, system_message, max_tokens, temperature, top_p, selected_model)
|
| 81 |
|
| 82 |
# 응답 중단 함수
|
| 83 |
def stop_generation():
|
| 84 |
stop_event.set()
|
| 85 |
|
| 86 |
-
# 히스토리 확인 함수
|
| 87 |
-
def view_history():
|
| 88 |
-
history = prompt_history.get_history()
|
| 89 |
-
history_str = "\n\n".join([f"모델: {entry['model']}\n프롬프트: {entry['prompt']}\n응답: {entry['response']}\n시스템 메시지: {entry['system_message']}" for entry in history])
|
| 90 |
-
return history_str if history_str else "히스토리가 없습니다."
|
| 91 |
-
|
| 92 |
# Gradio UI 구성
|
| 93 |
with gr.Blocks() as demo:
|
| 94 |
gr.Markdown("# 프롬프트 최적화 Playground")
|
|
@@ -124,20 +89,14 @@ with gr.Blocks() as demo:
|
|
| 124 |
|
| 125 |
with gr.Row():
|
| 126 |
send = gr.Button("전송")
|
| 127 |
-
continue_btn = gr.Button("계속 작성")
|
| 128 |
stop = gr.Button("🛑 생성 중단")
|
| 129 |
clear = gr.Button("🗑️ 대화 내역 지우기")
|
| 130 |
-
view_history_btn = gr.Button("🗂️ 히스토리 보기")
|
| 131 |
|
| 132 |
# Event handlers
|
| 133 |
send.click(respond, inputs=[prompt, system_message, max_tokens, temperature, top_p, model], outputs=[chatbot])
|
| 134 |
prompt.submit(respond, inputs=[prompt, system_message, max_tokens, temperature, top_p, model], outputs=[chatbot])
|
| 135 |
-
continue_btn.click(continue_writing,
|
| 136 |
-
inputs=[chatbot, system_message, max_tokens, temperature, top_p, model],
|
| 137 |
-
outputs=[chatbot])
|
| 138 |
stop.click(stop_generation)
|
| 139 |
clear.click(lambda: None, outputs=[chatbot])
|
| 140 |
-
view_history_btn.click(view_history, outputs=[gr.Textbox(label="히스토리")])
|
| 141 |
|
| 142 |
# UI 실행
|
| 143 |
demo.launch()
|
|
|
|
| 18 |
def get_client(model):
|
| 19 |
return InferenceClient(model=model, token=hf_token)
|
| 20 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 21 |
# 응답 생성 함수
|
| 22 |
def respond(prompt, system_message, max_tokens, temperature, top_p, selected_model):
|
| 23 |
stop_event.clear()
|
|
|
|
| 31 |
|
| 32 |
try:
|
| 33 |
response = ""
|
|
|
|
| 34 |
|
| 35 |
# 모델에서 응답을 스트리밍
|
| 36 |
for chunk in client.text_generation(
|
|
|
|
| 44 |
break
|
| 45 |
if chunk:
|
| 46 |
response += chunk
|
|
|
|
| 47 |
|
| 48 |
+
return [(prompt, response)] # 대화창에 프롬프트와 응답을 표시
|
|
|
|
|
|
|
|
|
|
| 49 |
|
| 50 |
except Exception as e:
|
| 51 |
+
return [(prompt, f"오류 발생: {str(e)}")]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 52 |
|
| 53 |
# 응답 중단 함수
|
| 54 |
def stop_generation():
|
| 55 |
stop_event.set()
|
| 56 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 57 |
# Gradio UI 구성
|
| 58 |
with gr.Blocks() as demo:
|
| 59 |
gr.Markdown("# 프롬프트 최적화 Playground")
|
|
|
|
| 89 |
|
| 90 |
with gr.Row():
|
| 91 |
send = gr.Button("전송")
|
|
|
|
| 92 |
stop = gr.Button("🛑 생성 중단")
|
| 93 |
clear = gr.Button("🗑️ 대화 내역 지우기")
|
|
|
|
| 94 |
|
| 95 |
# Event handlers
|
| 96 |
send.click(respond, inputs=[prompt, system_message, max_tokens, temperature, top_p, model], outputs=[chatbot])
|
| 97 |
prompt.submit(respond, inputs=[prompt, system_message, max_tokens, temperature, top_p, model], outputs=[chatbot])
|
|
|
|
|
|
|
|
|
|
| 98 |
stop.click(stop_generation)
|
| 99 |
clear.click(lambda: None, outputs=[chatbot])
|
|
|
|
| 100 |
|
| 101 |
# UI 실행
|
| 102 |
demo.launch()
|