Update app.py

app.py

@@ -9,16 +9,6 @@ from typing import Optional
 # [기본코드] - 수정/삭제 불가
 #############################
 
-# 제거할 모델들을 MODELS 사전에서 제외
-MODELS = {
-    "Zephyr 7B Beta": "HuggingFaceH4/zephyr-7b-beta",
-    "Meta Llama 3.1 8B": "meta-llama/Meta-Llama-3.1-8B-Instruct",
-    "Meta-Llama 3.1 70B-Instruct": "meta-llama/Meta-Llama-3.1-70B-Instruct",
-    "Microsoft": "microsoft/Phi-3-mini-4k-instruct",
-    "Mixtral 8x7B": "mistralai/Mistral-7B-Instruct-v0.3",
-    "Mixtral Nous-Hermes": "NousResearch/Nous-Hermes-2-Mixtral-8x7B-DPO",
-    "Aya-23-35B": "CohereForAI/aya-23-35B"}
-
 # Cohere Command R+ 모델 ID 정의
 COHERE_MODEL = "CohereForAI/c4ai-command-r-plus-08-2024"
 
@@ -31,51 +21,12 @@ def get_client(model_name):
     if not hf_token:
         raise ValueError("HuggingFace API 토큰이 필요합니다.")
 
-    if model_name in MODELS:
-        model_id = MODELS[model_name]
-    elif model_name == "Cohere Command R+":
+    if model_name == "Cohere Command R+":
         model_id = COHERE_MODEL
     else:
         raise ValueError("유효하지 않은 모델 이름입니다.")
     return InferenceClient(model_id, token=hf_token)
 
-
-def respond_hf_qna(
-    question: str,
-    model_name: str,
-    max_tokens: int,
-    temperature: float,
-    top_p: float,
-    system_message: str
-):
-    """
-    HuggingFace 모델(Zephyr 등)에 대해 한 번의 질문(question)에 대한 답변을 반환하는 함수.
-    """
-    try:
-        client = get_client(model_name)
-    except ValueError as e:
-        return f"오류: {str(e)}"
-
-    messages = [
-        {"role": "system", "content": system_message},
-        {"role": "user", "content": question}
-    ]
-
-    try:
-        response = client.chat_completion(
-            messages,
-            max_tokens=max_tokens,
-            temperature=temperature,
-            top_p=top_p,
-            stream=False,
-        )
-        assistant_message = response.choices[0].message.content
-        return assistant_message
-
-    except Exception as e:
-        return f"오류가 발생했습니다: {str(e)}"
-
-
 def respond_cohere_qna(
     question: str,
     system_message: str,
@@ -109,7 +60,6 @@ def respond_cohere_qna(
     except Exception as e:
         return f"오류가 발생했습니다: {str(e)}"
 
-
 def respond_chatgpt_qna(
     question: str,
     system_message: str,
@@ -144,7 +94,6 @@ def respond_chatgpt_qna(
     except Exception as e:
         return f"오류가 발생했습니다: {str(e)}"
 
-
 def respond_deepseek_qna(
     question: str,
     system_message: str,
@@ -180,7 +129,6 @@ def respond_deepseek_qna(
     except Exception as e:
         return f"오류가 발생했습니다: {str(e)}"
 
-
 def respond_claude_qna(
     question: str,
     system_message: str,
@@ -221,7 +169,6 @@ def respond_claude_qna(
     except Exception as e:
         return f"예상치 못한 오류가 발생했습니다: {str(e)}"
 
-
 #############################
 # [기본코드] UI 부분 - 수정/삭제 불가
 #############################
@@ -229,68 +176,6 @@ def respond_claude_qna(
 with gr.Blocks() as demo:
     gr.Markdown("# LLM 플레이그라운드")
 
-    #################
-    # 일반 모델 탭
-    #################
-    with gr.Tab("일반 모델"):
-        # 모델명 선택
-        model_name = gr.Radio(
-            choices=list(MODELS.keys()),
-            label="Language Model (HuggingFace)",
-            value="Zephyr 7B Beta"
-        )
-
-        # 입력1 ~ 입력5 (세로로 하나씩)
-        input1 = gr.Textbox(label="입력1", lines=1)
-        input2 = gr.Textbox(label="입력2", lines=1)
-        input3 = gr.Textbox(label="입력3", lines=1)
-        input4 = gr.Textbox(label="입력4", lines=1)
-        input5 = gr.Textbox(label="입력5", lines=1)
-
-        # 결과
-        answer_output = gr.Textbox(label="결과", lines=5, interactive=False)
-
-        # 고급 설정 - System Message를 Max Tokens 위로 이동
-        with gr.Accordion("고급 설정 (일반 모델)", open=False):
-            system_message = gr.Textbox(
-                value="""반드시 한글로 답변할 것.
-너는 최고의 비서이다.
-내가 요구하는것들을 최대한 자세하고 정확하게 답변하라.
-""",
-                label="System Message",
-                lines=3
-            )
-            max_tokens = gr.Slider(minimum=0, maximum=2000, value=500, step=100, label="Max Tokens")
-            temperature = gr.Slider(minimum=0.1, maximum=2.0, value=0.7, step=0.05, label="Temperature")
-            top_p = gr.Slider(minimum=0.1, maximum=1.0, value=0.95, step=0.05, label="Top-p")
-
-        submit_button = gr.Button("전송")
-
-        def merge_and_call_hf(i1, i2, i3, i4, i5, m_name, mt, temp, top_p_, sys_msg):
-            # 입력1~5를 공백 기준으로 합쳐서 question 구성
-            question = " ".join([i1, i2, i3, i4, i5])
-            return respond_hf_qna(
-                question=question,
-                model_name=m_name,
-                max_tokens=mt,
-                temperature=temp,
-                top_p=top_p_,
-                system_message=sys_msg
-            )
-
-        submit_button.click(
-            fn=merge_and_call_hf,
-            inputs=[
-                input1, input2, input3, input4, input5,
-                model_name,
-                max_tokens,
-                temperature,
-                top_p,
-                system_message
-            ],
-            outputs=answer_output
-        )
-
     #################
     # Cohere Command R+ 탭
     #################
@@ -508,4 +393,4 @@ with gr.Blocks() as demo:
 # 메인 실행부
 #############################
 if __name__ == "__main__":
-    demo.launch()
+    demo.launch()