add submit

app.py

@@ -35,7 +35,13 @@ from src.display.utils import (
 )
 from src.envs import API, EVAL_REQUESTS_PATH, EVAL_RESULTS_PATH, DYNAMIC_INFO_REPO, DYNAMIC_INFO_FILE_PATH, DYNAMIC_INFO_PATH, IS_PUBLIC, QUEUE_REPO, REPO_ID, RESULTS_REPO, TOKEN
 from src.populate import get_evaluation_queue_df, get_leaderboard_df, get_leaderboard_df_quota
-from src.submission.submit import add_new_eval
+from src.submission.submit import (
+    add_new_eval,
+    submit_model_info,
+    submit_api_info,
+    submit_inference_info,
+    submit_eval_complete
+)
 from src.scripts.update_all_request_files import update_dynamic_files
 from src.tools.collections import update_collections
 from src.tools.datastatics import get_statics
@@ -429,7 +435,100 @@ with demo:
                     queue=True,
                 )
             gr.Markdown(EVALUATION_METRIC_TEXT, elem_classes="markdown-text")
-        with gr.TabItem("📝 About", elem_id="llm-benchmark-tab-table", id=2):
+        with gr.TabItem("📤 Submit Model", elem_id="submit-model-tab", id=2):
+            gr.Markdown("## 提交模型进行评估", elem_classes="markdown-text")
+            
+            with gr.Row():
+                with gr.Column(scale=1):
+                    gr.Markdown("### 1. 提交模型信息")
+                    model_name = gr.Textbox(
+                        label="Model Name",
+                        placeholder="请输入模型名称，格式：用户名/模型名",
+                        info="例如：microsoft/DialoGPT-medium"
+                    )
+                    revision_commit = gr.Textbox(
+                        label="Revision commit",
+                        placeholder="请输入版本信息，默认为 main",
+                        value="main"
+                    )
+                    
+                    # 模型信息验证按钮
+                    model_info_btn = gr.Button("验证模型信息", variant="secondary")
+                    model_info_output = gr.Textbox(label="验证结果", interactive=False)
+                    
+                with gr.Column(scale=1):
+                    gr.Markdown("### 2. 提交API信息 (API only)")
+                    model_api_url = gr.Textbox(
+                        label="Model online api url",
+                        placeholder="请输入API URL，例如：https://api.openai.com/v1/chat/completions"
+                    )
+                    model_api_key = gr.Textbox(
+                        label="Model online api key",
+                        placeholder="请输入API密钥",
+                        type="password"
+                    )
+                    online_api_model_name = gr.Textbox(
+                        label="Online api model name",
+                        placeholder="请输入API模型名称，例如：gpt-3.5-turbo"
+                    )
+                    
+                    # API信息验证按钮
+                    api_info_btn = gr.Button("验证API信息", variant="secondary")
+                    api_info_output = gr.Textbox(label="验证结果", interactive=False)
+            
+            with gr.Row():
+                with gr.Column(scale=1):
+                    gr.Markdown("### 3. 提交推理信息 (inference only)")
+                    runsh_file = gr.File(
+                        label="upload run.sh file",
+                        file_types=[".sh"],
+                        file_count="single"
+                    )
+                    adapter_file = gr.File(
+                        label="upload model_adapter.py file", 
+                        file_types=[".py"],
+                        file_count="single"
+                    )
+                    
+                    # 推理信息验证按钮
+                    inference_info_btn = gr.Button("验证推理文件", variant="secondary")
+                    inference_info_output = gr.Textbox(label="验证结果", interactive=False)
+            
+            with gr.Row():
+                with gr.Column():
+                    gr.Markdown("### 4. 提交评估")
+                    submit_btn = gr.Button("Submit Eval", variant="primary", size="lg")
+                    submit_output = gr.Textbox(label="提交结果", interactive=False)
+            
+            # 绑定事件
+            model_info_btn.click(
+                fn=lambda name, rev: submit_model_info(name, rev),
+                inputs=[model_name, revision_commit],
+                outputs=model_info_output
+            )
+            
+            api_info_btn.click(
+                fn=lambda url, key, name: submit_api_info(url, key, name),
+                inputs=[model_api_url, model_api_key, online_api_model_name],
+                outputs=api_info_output
+            )
+            
+            inference_info_btn.click(
+                fn=lambda runsh, adapter: submit_inference_info(runsh, adapter),
+                inputs=[runsh_file, adapter_file],
+                outputs=inference_info_output
+            )
+            
+            submit_btn.click(
+                fn=lambda name, rev, url, key, api_name, runsh, adapter: submit_eval_complete(
+                    name, rev, url, key, api_name, runsh, adapter
+                ),
+                inputs=[model_name, revision_commit, model_api_url, model_api_key, 
+                       online_api_model_name, runsh_file, adapter_file],
+                outputs=submit_output
+            )
+
+        with gr.TabItem("📝 About", elem_id="llm-benchmark-tab-table", id=3):
             gr.Markdown(LLM_BENCHMARKS_TEXT, elem_classes="markdown-text")
             gr.HTML(TABLE_TEXT)
             gr.Markdown(LLM_BENCHMARKS_TEXT2, elem_classes="markdown-text")

src/submission/submit.py

@@ -16,6 +16,148 @@ from src.submission.check_validity import (
 REQUESTED_MODELS = None
 USERS_TO_SUBMISSION_DATES = None
 
+def submit_model_info(
+    model_name: str,
+    revision_commit: str
+):
+    """
+    提交模型信息 - 对应页面上的 "Submit your modelinfos here!" 部分
+    """
+    if not model_name or not model_name.strip():
+        return styled_error("请输入模型名称")
+    
+    if not revision_commit or not revision_commit.strip():
+        revision_commit = "main"
+    
+    # 验证模型名称格式
+    if "/" not in model_name:
+        return styled_error("模型名称格式不正确，请使用 '用户名/模型名' 格式")
+    
+    # 检查模型是否存在于Hugging Face Hub
+    try:
+        model_info = API.model_info(repo_id=model_name, revision=revision_commit)
+        return styled_message(f"模型信息验证成功！模型: {model_name}, 版本: {revision_commit}")
+    except Exception as e:
+        return styled_error(f"无法找到模型 {model_name}，请检查模型名称和版本是否正确")
+
+def submit_api_info(
+    model_api_url: str,
+    model_api_key: str,
+    online_api_model_name: str
+):
+    """
+    提交API信息 - 对应页面上的 "Submit your API infos here! (API only)" 部分
+    """
+    if not model_api_url or not model_api_url.strip():
+        return styled_error("请输入模型在线API URL")
+    
+    if not model_api_key or not model_api_key.strip():
+        return styled_error("请输入模型在线API密钥")
+    
+    if not online_api_model_name or not online_api_model_name.strip():
+        return styled_error("请输入在线API模型名称")
+    
+    # 验证URL格式
+    if not model_api_url.startswith(('http://', 'https://')):
+        return styled_error("API URL格式不正确，请以 http:// 或 https:// 开头")
+    
+    return styled_message(f"API信息验证成功！URL: {model_api_url}, 模型名: {online_api_model_name}")
+
+def submit_inference_info(
+    runsh_file,
+    adapter_file
+):
+    """
+    提交推理信息 - 对应页面上的 "Submit your inference infos here! (inference only)" 部分
+    """
+    if not runsh_file:
+        return styled_error("请上传 run.sh 文件")
+    
+    if not adapter_file:
+        return styled_error("请上传 model_adapter.py 文件")
+    
+    # 检查文件大小
+    max_size = 5 * 1024 * 1024  # 5MB
+    if os.path.getsize(runsh_file.name) > max_size:
+        return styled_error("run.sh 文件大小不能超过 5MB")
+    
+    if os.path.getsize(adapter_file.name) > max_size:
+        return styled_error("model_adapter.py 文件大小不能超过 5MB")
+    
+    # 验证文件内容
+    try:
+        with open(runsh_file.name, "r") as f:
+            runsh_content = f.read()
+        
+        with open(adapter_file.name, "r") as f:
+            adapter_content = f.read()
+        
+        # 检查run.sh是否包含必要的脚本内容
+        if "python" not in runsh_content.lower():
+            return styled_warning("run.sh 文件可能不包含正确的Python执行脚本")
+        
+        # 检查adapter文件是否包含必要的类定义
+        if "class" not in adapter_content and "def" not in adapter_content:
+            return styled_warning("model_adapter.py 文件可能不包含正确的类或函数定义")
+        
+        return styled_message("推理文件验证成功！run.sh 和 model_adapter.py 文件已准备就绪")
+        
+    except Exception as e:
+        return styled_error(f"文件读取失败: {str(e)}")
+
+def submit_eval_complete(
+    model_name: str,
+    revision_commit: str,
+    model_api_url: str,
+    model_api_key: str,
+    online_api_model_name: str,
+    runsh_file,
+    adapter_file
+):
+    """
+    完整的评估提交 - 整合所有三个部分的信息
+    """
+    # 验证模型信息
+    if not model_name or not model_name.strip():
+        return styled_error("请输入模型名称")
+    
+    if not revision_commit or not revision_commit.strip():
+        revision_commit = "main"
+    
+    # 验证API信息（如果提供）
+    if model_api_url and model_api_key and online_api_model_name:
+        if not model_api_url.startswith(('http://', 'https://')):
+            return styled_error("API URL格式不正确，请以 http:// 或 https:// 开头")
+    
+    # 验证推理文件（如果提供）
+    if runsh_file and adapter_file:
+        max_size = 5 * 1024 * 1024  # 5MB
+        if os.path.getsize(runsh_file.name) > max_size:
+            return styled_error("run.sh 文件大小不能超过 5MB")
+        
+        if os.path.getsize(adapter_file.name) > max_size:
+            return styled_error("model_adapter.py 文件大小不能超过 5MB")
+    
+    # 调用原有的add_new_eval函数
+    try:
+        result = add_new_eval(
+            model=model_name,
+            model_api_url=model_api_url or "",
+            model_api_key=model_api_key or "",
+            model_api_name=online_api_model_name or "",
+            base_model="",  # 可以根据需要设置
+            revision=revision_commit,
+            precision="float16",  # 默认精度
+            private="false",
+            weight_type="Original",  # 默认权重类型
+            model_type="",  # 可以根据需要设置
+            runsh=runsh_file,
+            adapter=adapter_file
+        )
+        return result
+    except Exception as e:
+        return styled_error(f"提交失败: {str(e)}")
+
 def add_new_eval(
     model: str,
     model_api_url: str,