Spaces:
Runtime error
Runtime error
Update app.py
Browse files
app.py
CHANGED
|
@@ -4,13 +4,11 @@ import os
|
|
| 4 |
|
| 5 |
MODELS = {
|
| 6 |
"Zephyr 7B Beta": "HuggingFaceH4/zephyr-7b-beta",
|
| 7 |
-
"DeepSeek Coder V2": "deepseek-ai/DeepSeek-Coder-V2-Instruct",
|
| 8 |
"Meta Llama 3.1 8B": "meta-llama/Meta-Llama-3.1-8B-Instruct",
|
| 9 |
"Meta-Llama 3.1 70B-Instruct": "meta-llama/Meta-Llama-3.1-70B-Instruct",
|
| 10 |
"Microsoft": "microsoft/Phi-3-mini-4k-instruct",
|
| 11 |
"Mixtral 8x7B": "mistralai/Mistral-7B-Instruct-v0.3",
|
| 12 |
"Mixtral Nous-Hermes": "NousResearch/Nous-Hermes-2-Mixtral-8x7B-DPO",
|
| 13 |
-
"Cohere Command R+": "CohereForAI/c4ai-command-r-plus",
|
| 14 |
"Aya-23-35B": "CohereForAI/aya-23-35B"
|
| 15 |
}
|
| 16 |
|
|
@@ -18,7 +16,7 @@ def get_client(model_name):
|
|
| 18 |
model_id = MODELS[model_name]
|
| 19 |
hf_token = os.getenv("HF_TOKEN")
|
| 20 |
if not hf_token:
|
| 21 |
-
raise ValueError("HF_TOKEN
|
| 22 |
return InferenceClient(model_id, token=hf_token)
|
| 23 |
|
| 24 |
def respond(
|
|
@@ -73,28 +71,102 @@ def respond(
|
|
| 73 |
chat_history.append((message, partial_message))
|
| 74 |
yield chat_history
|
| 75 |
except Exception as e:
|
| 76 |
-
error_message = f"
|
| 77 |
chat_history.append((message, error_message))
|
| 78 |
yield chat_history
|
| 79 |
|
| 80 |
def clear_conversation():
|
| 81 |
return []
|
| 82 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 83 |
with gr.Blocks() as demo:
|
| 84 |
gr.Markdown("# Prompting AI Chatbot")
|
| 85 |
gr.Markdown("언어모델별 프롬프트 테스트 챗봇입니다.")
|
| 86 |
-
|
| 87 |
-
with gr.
|
| 88 |
-
with gr.
|
| 89 |
-
|
| 90 |
-
|
| 91 |
-
|
| 92 |
-
|
| 93 |
-
|
| 94 |
-
|
| 95 |
-
|
| 96 |
-
|
| 97 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 98 |
value="""반드시 한글로 답변할 것.
|
| 99 |
너는 최고의 비서이다.
|
| 100 |
내가 요구하는것들을 최대한 자세하고 정확하게 답변하라.
|
|
@@ -102,17 +174,33 @@ with gr.Blocks() as demo:
|
|
| 102 |
label="System Message",
|
| 103 |
lines=3
|
| 104 |
)
|
| 105 |
-
|
| 106 |
-
|
| 107 |
-
|
| 108 |
-
|
| 109 |
-
|
| 110 |
-
|
| 111 |
-
|
| 112 |
-
|
| 113 |
-
|
| 114 |
-
|
| 115 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 116 |
|
| 117 |
if __name__ == "__main__":
|
| 118 |
-
demo.launch()
|
|
|
|
| 4 |
|
| 5 |
MODELS = {
|
| 6 |
"Zephyr 7B Beta": "HuggingFaceH4/zephyr-7b-beta",
|
|
|
|
| 7 |
"Meta Llama 3.1 8B": "meta-llama/Meta-Llama-3.1-8B-Instruct",
|
| 8 |
"Meta-Llama 3.1 70B-Instruct": "meta-llama/Meta-Llama-3.1-70B-Instruct",
|
| 9 |
"Microsoft": "microsoft/Phi-3-mini-4k-instruct",
|
| 10 |
"Mixtral 8x7B": "mistralai/Mistral-7B-Instruct-v0.3",
|
| 11 |
"Mixtral Nous-Hermes": "NousResearch/Nous-Hermes-2-Mixtral-8x7B-DPO",
|
|
|
|
| 12 |
"Aya-23-35B": "CohereForAI/aya-23-35B"
|
| 13 |
}
|
| 14 |
|
|
|
|
| 16 |
model_id = MODELS[model_name]
|
| 17 |
hf_token = os.getenv("HF_TOKEN")
|
| 18 |
if not hf_token:
|
| 19 |
+
raise ValueError("HF_TOKEN 환경 변수가 필요합니다.")
|
| 20 |
return InferenceClient(model_id, token=hf_token)
|
| 21 |
|
| 22 |
def respond(
|
|
|
|
| 71 |
chat_history.append((message, partial_message))
|
| 72 |
yield chat_history
|
| 73 |
except Exception as e:
|
| 74 |
+
error_message = f"오류가 발생했습니다: {str(e)}"
|
| 75 |
chat_history.append((message, error_message))
|
| 76 |
yield chat_history
|
| 77 |
|
| 78 |
def clear_conversation():
|
| 79 |
return []
|
| 80 |
|
| 81 |
+
# Cohere Command R+ 전용 응답 함수
|
| 82 |
+
from openai import OpenAI
|
| 83 |
+
|
| 84 |
+
ACCESS_TOKEN = os.getenv("HF_TOKEN")
|
| 85 |
+
|
| 86 |
+
cohere_client = OpenAI(
|
| 87 |
+
base_url="https://api-inference.huggingface.co/v1/",
|
| 88 |
+
api_key=ACCESS_TOKEN,
|
| 89 |
+
)
|
| 90 |
+
|
| 91 |
+
def cohere_respond(
|
| 92 |
+
message,
|
| 93 |
+
chat_history,
|
| 94 |
+
system_message,
|
| 95 |
+
max_tokens,
|
| 96 |
+
temperature,
|
| 97 |
+
top_p,
|
| 98 |
+
):
|
| 99 |
+
messages = [{"role": "system", "content": system_message}]
|
| 100 |
+
|
| 101 |
+
for human, assistant in chat_history:
|
| 102 |
+
if human:
|
| 103 |
+
messages.append({"role": "user", "content": human})
|
| 104 |
+
if assistant:
|
| 105 |
+
messages.append({"role": "assistant", "content": assistant})
|
| 106 |
+
|
| 107 |
+
messages.append({"role": "user", "content": message})
|
| 108 |
+
|
| 109 |
+
response = ""
|
| 110 |
+
|
| 111 |
+
try:
|
| 112 |
+
for msg in cohere_client.chat.completions.create(
|
| 113 |
+
model="CohereForAI/c4ai-command-r-plus-08-2024",
|
| 114 |
+
max_tokens=max_tokens,
|
| 115 |
+
stream=True,
|
| 116 |
+
temperature=temperature,
|
| 117 |
+
top_p=top_p,
|
| 118 |
+
messages=messages,
|
| 119 |
+
):
|
| 120 |
+
token = msg.choices[0].delta.content
|
| 121 |
+
response += token
|
| 122 |
+
if len(chat_history) > 0 and chat_history[-1][0] == message:
|
| 123 |
+
chat_history[-1] = (message, response)
|
| 124 |
+
else:
|
| 125 |
+
chat_history.append((message, response))
|
| 126 |
+
yield chat_history
|
| 127 |
+
except Exception as e:
|
| 128 |
+
error_message = f"오류가 발생했습니다: {str(e)}"
|
| 129 |
+
chat_history.append((message, error_message))
|
| 130 |
+
yield chat_history
|
| 131 |
+
|
| 132 |
with gr.Blocks() as demo:
|
| 133 |
gr.Markdown("# Prompting AI Chatbot")
|
| 134 |
gr.Markdown("언어모델별 프롬프트 테스트 챗봇입니다.")
|
| 135 |
+
|
| 136 |
+
with gr.Tab("일반 모델"):
|
| 137 |
+
with gr.Row():
|
| 138 |
+
with gr.Column(scale=1):
|
| 139 |
+
model_name = gr.Radio(
|
| 140 |
+
choices=list(MODELS.keys()),
|
| 141 |
+
label="Language Model",
|
| 142 |
+
value="Zephyr 7B Beta"
|
| 143 |
+
)
|
| 144 |
+
max_tokens = gr.Slider(minimum=0, maximum=2000, value=500, step=100, label="Max Tokens")
|
| 145 |
+
temperature = gr.Slider(minimum=0.1, maximum=2.0, value=0.7, step=0.05, label="Temperature")
|
| 146 |
+
top_p = gr.Slider(minimum=0.1, maximum=1.0, value=0.95, step=0.05, label="Top-p")
|
| 147 |
+
system_message = gr.Textbox(
|
| 148 |
+
value="""반드시 한글로 답변할 것.
|
| 149 |
+
너는 최고의 비서이다.
|
| 150 |
+
내가 요구하는것들을 최대한 자세하고 정확하게 답변하라.
|
| 151 |
+
""",
|
| 152 |
+
label="System Message",
|
| 153 |
+
lines=3
|
| 154 |
+
)
|
| 155 |
+
|
| 156 |
+
with gr.Column(scale=2):
|
| 157 |
+
chatbot = gr.Chatbot()
|
| 158 |
+
msg = gr.Textbox(label="메세지를 입력하세요")
|
| 159 |
+
with gr.Row():
|
| 160 |
+
submit_button = gr.Button("전송")
|
| 161 |
+
clear_button = gr.Button("대화 내역 지우기")
|
| 162 |
+
|
| 163 |
+
msg.submit(respond, [msg, chatbot, model_name, max_tokens, temperature, top_p, system_message], chatbot)
|
| 164 |
+
submit_button.click(respond, [msg, chatbot, model_name, max_tokens, temperature, top_p, system_message], chatbot)
|
| 165 |
+
clear_button.click(clear_conversation, outputs=chatbot, queue=False)
|
| 166 |
+
|
| 167 |
+
with gr.Tab("Cohere Command R+"):
|
| 168 |
+
with gr.Row():
|
| 169 |
+
cohere_system_message = gr.Textbox(
|
| 170 |
value="""반드시 한글로 답변할 것.
|
| 171 |
너는 최고의 비서이다.
|
| 172 |
내가 요구하는것들을 최대한 자세하고 정확하게 답변하라.
|
|
|
|
| 174 |
label="System Message",
|
| 175 |
lines=3
|
| 176 |
)
|
| 177 |
+
cohere_max_tokens = gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens")
|
| 178 |
+
cohere_temperature = gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature")
|
| 179 |
+
cohere_top_p = gr.Slider(
|
| 180 |
+
minimum=0.1,
|
| 181 |
+
maximum=1.0,
|
| 182 |
+
value=0.95,
|
| 183 |
+
step=0.05,
|
| 184 |
+
label="Top-P",
|
| 185 |
+
)
|
| 186 |
+
|
| 187 |
+
cohere_chatbot = gr.Chatbot(height=600)
|
| 188 |
+
cohere_msg = gr.Textbox(label="메세지를 입력하세요")
|
| 189 |
+
with gr.Row():
|
| 190 |
+
cohere_submit_button = gr.Button("전송")
|
| 191 |
+
cohere_clear_button = gr.Button("대화 내역 지우기")
|
| 192 |
+
|
| 193 |
+
cohere_msg.submit(
|
| 194 |
+
cohere_respond,
|
| 195 |
+
[cohere_msg, cohere_chatbot, cohere_system_message, cohere_max_tokens, cohere_temperature, cohere_top_p],
|
| 196 |
+
cohere_chatbot
|
| 197 |
+
)
|
| 198 |
+
cohere_submit_button.click(
|
| 199 |
+
cohere_respond,
|
| 200 |
+
[cohere_msg, cohere_chatbot, cohere_system_message, cohere_max_tokens, cohere_temperature, cohere_top_p],
|
| 201 |
+
cohere_chatbot
|
| 202 |
+
)
|
| 203 |
+
cohere_clear_button.click(clear_conversation, outputs=cohere_chatbot, queue=False)
|
| 204 |
|
| 205 |
if __name__ == "__main__":
|
| 206 |
+
demo.launch()
|