Spaces:
Running
Running
Commit
•
6ab04f4
1
Parent(s):
728fb17
Update app.py
Browse files
app.py
CHANGED
|
@@ -18,7 +18,7 @@ def respond(
|
|
| 18 |
message,
|
| 19 |
history: list[tuple[str, str]],
|
| 20 |
system_message="AI Assistant Role",
|
| 21 |
-
max_tokens=
|
| 22 |
temperature=0.7,
|
| 23 |
top_p=0.95,
|
| 24 |
):
|
|
@@ -30,7 +30,6 @@ def respond(
|
|
| 30 |
memory.append((message, None))
|
| 31 |
|
| 32 |
messages = [{"role": "system", "content": full_system_message}]
|
| 33 |
-
|
| 34 |
# 메모리에서 대화 기록을 가져와 메시지 목록에 추가
|
| 35 |
for val in memory:
|
| 36 |
if val[0]:
|
|
@@ -42,7 +41,6 @@ def respond(
|
|
| 42 |
"Authorization": f"Bearer {TOKEN}",
|
| 43 |
"Content-Type": "application/json"
|
| 44 |
}
|
| 45 |
-
|
| 46 |
payload = {
|
| 47 |
"model": "meta-llama/Meta-Llama-3.1-405B-Instruct",
|
| 48 |
"max_tokens": max_tokens,
|
|
@@ -50,7 +48,6 @@ def respond(
|
|
| 50 |
"top_p": top_p,
|
| 51 |
"messages": messages
|
| 52 |
}
|
| 53 |
-
|
| 54 |
response = requests.post("https://api-inference.huggingface.co/v1/chat/completions", headers=headers, json=payload, stream=True)
|
| 55 |
|
| 56 |
# Stream 방식으로 데이터를 출력
|
|
@@ -79,9 +76,8 @@ demo = gr.ChatInterface(
|
|
| 79 |
gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
|
| 80 |
gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
|
| 81 |
gr.Slider(minimum=0.1, maximum=1.0, value=0.95, step=0.05, label="Top-p (nucleus sampling)"),
|
| 82 |
-
]
|
| 83 |
-
streaming=True # 스트리밍 모드 활성화
|
| 84 |
)
|
| 85 |
|
| 86 |
if __name__ == "__main__":
|
| 87 |
-
demo.queue().launch(max_threads=20)
|
|
|
|
| 18 |
message,
|
| 19 |
history: list[tuple[str, str]],
|
| 20 |
system_message="AI Assistant Role",
|
| 21 |
+
max_tokens=512,
|
| 22 |
temperature=0.7,
|
| 23 |
top_p=0.95,
|
| 24 |
):
|
|
|
|
| 30 |
memory.append((message, None))
|
| 31 |
|
| 32 |
messages = [{"role": "system", "content": full_system_message}]
|
|
|
|
| 33 |
# 메모리에서 대화 기록을 가져와 메시지 목록에 추가
|
| 34 |
for val in memory:
|
| 35 |
if val[0]:
|
|
|
|
| 41 |
"Authorization": f"Bearer {TOKEN}",
|
| 42 |
"Content-Type": "application/json"
|
| 43 |
}
|
|
|
|
| 44 |
payload = {
|
| 45 |
"model": "meta-llama/Meta-Llama-3.1-405B-Instruct",
|
| 46 |
"max_tokens": max_tokens,
|
|
|
|
| 48 |
"top_p": top_p,
|
| 49 |
"messages": messages
|
| 50 |
}
|
|
|
|
| 51 |
response = requests.post("https://api-inference.huggingface.co/v1/chat/completions", headers=headers, json=payload, stream=True)
|
| 52 |
|
| 53 |
# Stream 방식으로 데이터를 출력
|
|
|
|
| 76 |
gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
|
| 77 |
gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
|
| 78 |
gr.Slider(minimum=0.1, maximum=1.0, value=0.95, step=0.05, label="Top-p (nucleus sampling)"),
|
| 79 |
+
]
|
|
|
|
| 80 |
)
|
| 81 |
|
| 82 |
if __name__ == "__main__":
|
| 83 |
+
demo.queue().launch(max_threads=20)
|