Upload 4 files

gradio/.gradio/certificate.pem

@@ -0,0 +1,31 @@
+-----BEGIN CERTIFICATE-----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=
+-----END CERTIFICATE-----

gradio/app.py

@@ -0,0 +1,504 @@
+import gradio as gr
+import pandas as pd
+import json
+from pathlib import Path
+from typing import Dict, Optional
+
+
+class SpectralLeaderboard:
+    def __init__(self, data_file: str = "../leaderboard_v_1.0.json"):
+        # 获取当前脚本的目录
+        current_dir = Path(__file__).parent
+        # 构建正确的数据文件路径
+        if data_file.startswith("../"):
+            self.data_file = current_dir.parent / data_file[3:]
+        else:
+            self.data_file = Path(data_file)
+
+        print(f"🔍 Looking for data file at: {self.data_file}")
+        print(f"📂 Current working directory: {Path.cwd()}")
+        print(f"📄 Script location: {Path(__file__).parent}")
+        print(f"✅ Data file exists: {self.data_file.exists()}")
+
+        self.data = self._load_data()
+
+    def _load_data(self) -> Dict:
+        """加载排行榜数据"""
+        try:
+            with open(self.data_file, "r", encoding="utf-8") as f:
+                data = json.load(f)
+                print(f"✅ Successfully loaded {data['leaderboard_info']['total_models']} models from {self.data_file}")
+                return data
+        except FileNotFoundError:
+            print(f"❌ Data file {self.data_file} not found. Creating empty leaderboard.")
+            return {"leaderboard_info": {"total_models": 0}, "models": []}
+        except Exception as e:
+            print(f"❌ Error loading data: {e}")
+            return {"leaderboard_info": {"total_models": 0}, "models": []}
+
+    def _format_accuracy(self, accuracy: Optional[float]) -> str:
+        """格式化准确率显示"""
+        if accuracy is None:
+            return "-"
+        return f"{accuracy:.1f}"
+
+    def _calculate_average(self, results: Dict) -> Optional[float]:
+        """计算平均准确率，使用overall_accuracy字段"""
+        return results.get("overall_accuracy")
+
+    def _get_model_type_icon(self, model_type: str) -> str:
+        """获取模型类型图标"""
+        icons = {"open_source": "🔓", "proprietary": "🔒", "baseline": "📊"}
+        return icons.get(model_type, "❓")
+
+    def _get_multimodal_icon(self, is_multimodal: bool) -> str:
+        """获取多模态图标"""
+        return "👁️" if is_multimodal else "📝"
+
+    def _get_rank_display(self, rank: int) -> str:
+        """获取排名显示，前三名显示奖牌"""
+        medals = {1: "🥇", 2: "🥈", 3: "🥉"}
+        return medals.get(rank, str(rank))
+
+    def _create_link(self, text: str, url: str) -> str:
+        """创建HTML链接"""
+        if url and url.strip():
+            return f'<a href="{url}" target="_blank" style="text-decoration: none; color: inherit;">{text}</a>'
+        return text
+
+    def get_leaderboard_df(
+        self,
+        model_type_filter: str = "All",
+        multimodal_filter: str = "All",
+        sort_by: str = "Overall",
+        ascending: bool = False,
+    ) -> pd.DataFrame:
+        """生成排行榜DataFrame"""
+
+        models = self.data.get("models", [])
+        print(f"📊 Processing {len(models)} models")
+
+        # 筛选模型
+        filtered_models = []
+        for model in models:
+            # 模型类型筛选
+            if model_type_filter != "All" and model.get("model_type", "") != model_type_filter:
+                continue
+
+            # 多模态筛选
+            if multimodal_filter == "Multimodal Only" and not model.get("is_multimodal", False):
+                continue
+            elif multimodal_filter == "Text Only" and model.get("is_multimodal", False):
+                continue
+
+            filtered_models.append(model)
+
+        print(f"🔍 After filtering: {len(filtered_models)} models")
+
+        # 构建DataFrame数据
+        data = []
+        for model in filtered_models:
+            try:
+                results = model.get("results", {})
+
+                # 获取各项准确率
+                overall_accuracy = self._calculate_average(results)
+                signal_acc = results.get("Signal", {}).get("accuracy")
+                perception_acc = results.get("Perception", {}).get("accuracy")
+                semantic_acc = results.get("Semantic", {}).get("accuracy")
+                generation_acc = results.get("Generation", {}).get("accuracy")
+
+                # 创建带链接的模型名和提交者
+                model_name_display = self._create_link(model.get("name", "Unknown"), model.get("name_link", ""))
+                submitter_display = self._create_link(
+                    model.get("submitter", "Unknown"), model.get("submitter_link", "")
+                )
+
+                row = {
+                    "Type": self._get_model_type_icon(model.get("model_type", "unknown")),
+                    "Model": model_name_display,
+                    "Size": model.get("model_size", "Unknown"),
+                    "MM": self._get_multimodal_icon(model.get("is_multimodal", False)),
+                    "Overall": self._format_accuracy(overall_accuracy),
+                    "Signal": self._format_accuracy(signal_acc),
+                    "Perception": self._format_accuracy(perception_acc),
+                    "Semantic": self._format_accuracy(semantic_acc),
+                    "Generation": self._format_accuracy(generation_acc),
+                    "Submitter": submitter_display,
+                    "Date": (model.get("submission_time", "")[:10] if model.get("submission_time") else "-"),
+                    # 用于排序的数值列
+                    "overall_val": overall_accuracy or 0,
+                    "signal_val": signal_acc or 0,
+                    "perception_val": perception_acc or 0,
+                    "semantic_val": semantic_acc or 0,
+                    "generation_val": generation_acc or 0,
+                }
+                data.append(row)
+            except Exception as e:
+                print(f"⚠️ Error processing model {model.get('name', 'Unknown')}: {e}")
+                continue
+
+        df = pd.DataFrame(data)
+        print(f"📋 Created DataFrame with {len(df)} rows")
+
+        if len(df) == 0:
+            print("📋 Empty DataFrame, returning empty table")
+            return pd.DataFrame(
+                columns=[
+                    "Rank",
+                    "Type",
+                    "Model",
+                    "Size",
+                    "MM",
+                    "Overall",
+                    "Signal",
+                    "Perception",
+                    "Semantic",
+                    "Generation",
+                    "Submitter",
+                    "Date",
+                ]
+            )
+
+        # 排序
+        sort_mapping = {
+            "Overall": "overall_val",
+            "Signal": "signal_val",
+            "Perception": "perception_val",
+            "Semantic": "semantic_val",
+            "Generation": "generation_val",
+            "Model": "Model",
+            "Date": "Date",
+        }
+
+        sort_col = sort_mapping.get(sort_by, "overall_val")
+        df = df.sort_values(by=sort_col, ascending=ascending)
+
+        # 添加带奖牌的排名
+        ranks = []
+        for i in range(len(df)):
+            rank_num = i + 1
+            ranks.append(self._get_rank_display(rank_num))
+
+        df.insert(0, "Rank", ranks)
+
+        # 移除用于排序的辅助列
+        display_columns = [
+            "Rank",
+            "Type",
+            "Model",
+            "Size",
+            "MM",
+            "Overall",
+            "Signal",
+            "Perception",
+            "Semantic",
+            "Generation",
+            "Submitter",
+            "Date",
+        ]
+        result_df = df[display_columns]
+        print(f"✅ Returning DataFrame with {len(result_df)} rows")
+        return result_df
+
+    def get_subcategory_details(self, model_name: str) -> pd.DataFrame:
+        """获取模型的子类别详细结果"""
+        # 移除HTML标签进行匹配
+        clean_model_name = model_name
+        if "<a href=" in model_name:
+            # 提取链接中的文本
+            import re
+
+            match = re.search(r">([^<]+)<", model_name)
+            if match:
+                clean_model_name = match.group(1)
+
+        for model in self.data.get("models", []):
+            if model.get("name") == clean_model_name:
+                data = []
+                results = model.get("results", {})
+                for level, level_data in results.items():
+                    if level == "overall_accuracy":  # 跳过总体准确率字段
+                        continue
+
+                    subcategories = level_data.get("subcategories", {})
+                    for subcat, subcat_data in subcategories.items():
+                        data.append(
+                            {
+                                "Level": level,
+                                "Subcategory": subcat,
+                                "Accuracy": self._format_accuracy(subcat_data.get("accuracy")),
+                            }
+                        )
+                return pd.DataFrame(data)
+        return pd.DataFrame()
+
+
+def create_leaderboard():
+    """创建排行榜Gradio界面"""
+
+    leaderboard = SpectralLeaderboard()
+
+    with gr.Blocks(
+        title="🔬 SpectrumLab Leaderboard",
+        theme=gr.themes.Default(),
+        css="""
+        .gradio-container {
+            max-width: 1400px !important;
+        }
+        .dataframe table {
+            border-collapse: collapse !important;
+        }
+        .dataframe td, .dataframe th {
+            padding: 8px 12px !important;
+            border: 1px solid #e1e5e9 !important;
+        }
+        .dataframe th {
+            background-color: #f8f9fa !important;
+            font-weight: 600 !important;
+        }
+        .dataframe tr:nth-child(even) {
+            background-color: #f8f9fa !important;
+        }
+        .dataframe tr:hover {
+            background-color: #e8f4f8 !important;
+        }
+        """,
+    ) as demo:
+        gr.Markdown(
+            """
+            # ���� SpectrumLab Leaderboard
+            
+            A comprehensive benchmark for evaluating large language models on **spectroscopic analysis tasks**.
+            
+            📊 **Evaluation Levels**: Signal Processing, Perception, Semantic Understanding, Generation  
+            🔬 **Domains**: IR, NMR, UV-Vis, Mass Spectrometry and more  
+            🌟 **Multimodal**: Support for both text-only and vision-language models
+            """
+        )
+
+        with gr.Row():
+            info = leaderboard.data.get("leaderboard_info", {"total_models": 0})
+            gr.Markdown(
+                f"""
+                **📈 Stats**: {info["total_models"]} models evaluated  
+                **🏅 Rankings**: 🥇🥈🥉 medals for top performers  
+                **🔗 Submit**: Send evaluation results to contribute your model!
+                """
+            )
+
+        with gr.Row():
+            with gr.Column(scale=2):
+                model_type_filter = gr.Dropdown(
+                    choices=["All", "open_source", "proprietary", "baseline"],
+                    value="All",
+                    label="🏷️ Model Type",
+                )
+
+            with gr.Column(scale=2):
+                multimodal_filter = gr.Dropdown(
+                    choices=["All", "Multimodal Only", "Text Only"],
+                    value="All",
+                    label="👁️ Modality",
+                )
+
+            with gr.Column(scale=2):
+                sort_by = gr.Dropdown(
+                    choices=[
+                        "Overall",
+                        "Signal",
+                        "Perception",
+                        "Semantic",
+                        "Generation",
+                        "Model",
+                        "Date",
+                    ],
+                    value="Overall",
+                    label="📊 Sort By",
+                )
+
+            with gr.Column(scale=1):
+                ascending = gr.Checkbox(value=False, label="⬆️ Ascending")
+
+            with gr.Column(scale=1):
+                refresh_btn = gr.Button("🔄 Refresh", variant="secondary")
+
+        # 主排行榜表格
+        initial_df = leaderboard.get_leaderboard_df()
+        leaderboard_table = gr.Dataframe(
+            value=initial_df,
+            interactive=False,
+            wrap=True,
+            datatype=["html"] * len(initial_df.columns) if len(initial_df.columns) > 0 else ["html"] * 12,
+            column_widths=(
+                [
+                    "6%",
+                    "5%",
+                    "18%",
+                    "8%",
+                    "5%",
+                    "10%",
+                    "10%",
+                    "10%",
+                    "10%",
+                    "10%",
+                    "16%",
+                    "10%",
+                ]
+                if len(initial_df.columns) > 0
+                else None
+            ),
+            label="🏆 Model Rankings",
+        )
+
+        # 模型详细信息
+        with gr.Accordion("📋 Model Details", open=False):
+            model_choices = [model.get("name", "Unknown") for model in leaderboard.data.get("models", [])]
+            model_select = gr.Dropdown(
+                choices=model_choices,
+                label="Select Model for Details",
+            )
+
+            with gr.Row():
+                with gr.Column():
+                    subcategory_table = gr.Dataframe(label="📊 Subcategory Results")
+
+                with gr.Column():
+                    model_info = gr.Markdown(label="ℹ️ Model Information")
+
+        # 图例说明
+        with gr.Accordion("📖 Legend & Info", open=False):
+            gr.Markdown(
+                """
+                ### 🔍 Column Explanations
+                
+                - **Rank**: 🥇 1st place, 🥈 2nd place, 🥉 3rd place, then numbers
+                - **Type**: 🔓 Open Source, 🔒 Proprietary, 📊 Baseline
+                - **MM**: 👁️ Multimodal, 📝 Text-only  
+                - **Overall**: Average accuracy across all evaluated levels
+                - **Signal**: Low-level signal processing tasks
+                - **Perception**: Mid-level feature extraction tasks
+                - **Semantic**: High-level understanding tasks
+                - **Generation**: Spectrum generation tasks
+                
+                ### 📝 Notes
+                - "-" indicates the model was not evaluated on that benchmark
+                - Rankings are based on overall performance across all evaluated tasks
+                - Multimodal models can process both text and spectroscopic images
+                - Click on model names and submitters to visit their pages
+                
+                ### 📊 Task Categories
+                
+                **Signal Level:**
+                - Spectrum Type Classification (TC)
+                - Spectrum Quality Assessment (QE)
+                - Basic Feature Extraction (FE)
+                - Impurity Peak Detection (ID)
+                
+                **Perception Level:**
+                - Functional Group Recognition (GR)
+                - Elemental Compositional Prediction (EP)
+                - Peak Assignment (PA)
+                - Basic Property Prediction (PP)
+                
+                **Semantic Level:**
+                - Molecular Structure Elucidation (SE)
+                - Fusing Spectroscopic Modalities (FM)
+                - Multimodal Molecular Reasoning (MR)
+                
+                **Generation Level:**
+                - Forward Problems (FP)
+                - Inverse Problems (IP)
+                - De Novo Generation (DnG)
+                """
+            )
+
+        def update_leaderboard(model_type, multimodal, sort_by_val, asc):
+            """更新排行榜"""
+            print(f"🔄 Updating leaderboard with filters: {model_type}, {multimodal}, {sort_by_val}, {asc}")
+            return leaderboard.get_leaderboard_df(
+                model_type_filter=model_type,
+                multimodal_filter=multimodal,
+                sort_by=sort_by_val,
+                ascending=asc,
+            )
+
+        def update_model_details(model_name):
+            """更新模型详细信息"""
+            if not model_name:
+                return pd.DataFrame(), ""
+
+            # 获取子类别详情
+            subcategory_df = leaderboard.get_subcategory_details(model_name)
+
+            # 获取模型基本信息
+            for model in leaderboard.data.get("models", []):
+                if model.get("name") == model_name:
+                    # 处理链接显示
+                    def format_link(name, url):
+                        if url and url.strip():
+                            return f"[{name}]({url})"
+                        return "Not provided"
+
+                    model_info_dict = model.get("model_info", {})
+                    results = model.get("results", {})
+
+                    info_md = f"""
+                    ### {model.get("name", "Unknown")}
+                    
+                    **👤 Submitter**: {model.get("submitter", "Unknown")}  
+                    **📅 Submission**: {model.get("submission_time", "")[:10] if model.get("submission_time") else "Unknown"}  
+                    **🏷️ Type**: {model.get("model_type", "Unknown")}  
+                    **📏 Size**: {model.get("model_size", "Unknown")}  
+                    **👁️ Multimodal**: {"Yes" if model.get("is_multimodal", False) else "No"}  
+                    
+                    **📝 Description**: {model_info_dict.get("description", "") or "No description provided"}
+                    
+                    **🔗 Links**:  
+                    - **Homepage**: {format_link("Visit", model_info_dict.get("homepage", ""))}
+                    - **Paper**: {format_link("Read", model_info_dict.get("paper", ""))}  
+                    - **Code**: {format_link("View", model_info_dict.get("code", ""))}
+                    
+                    **📊 Performance Summary**:
+                    - **Overall**: {leaderboard._format_accuracy(results.get("overall_accuracy"))}%
+                    - **Signal**: {leaderboard._format_accuracy(results.get("Signal", {}).get("accuracy"))}%
+                    - **Perception**: {leaderboard._format_accuracy(results.get("Perception", {}).get("accuracy"))}%
+                    - **Semantic**: {leaderboard._format_accuracy(results.get("Semantic", {}).get("accuracy"))}%
+                    - **Generation**: {leaderboard._format_accuracy(results.get("Generation", {}).get("accuracy"))}%
+                    """
+                    return subcategory_df, info_md
+
+            return pd.DataFrame(), ""
+
+        # 事件绑定
+        for component in [model_type_filter, multimodal_filter, sort_by, ascending]:
+            component.change(
+                fn=update_leaderboard,
+                inputs=[model_type_filter, multimodal_filter, sort_by, ascending],
+                outputs=[leaderboard_table],
+            )
+
+        refresh_btn.click(
+            fn=update_leaderboard,
+            inputs=[model_type_filter, multimodal_filter, sort_by, ascending],
+            outputs=[leaderboard_table],
+        )
+
+        model_select.change(
+            fn=update_model_details,
+            inputs=[model_select],
+            outputs=[subcategory_table, model_info],
+        )
+
+    return demo
+
+
+if __name__ == "__main__":
+    app = create_leaderboard()
+    print("🚀 Starting SpectrumLab Leaderboard...")
+    app.launch(
+        server_name="0.0.0.0",
+        share=True,
+        show_api=False,
+        inbrowser=True,
+    )

gradio/requirements.txt

@@ -0,0 +1 @@
+gradio==5.35.0

leaderboard_v_1.0.json

@@ -0,0 +1,1465 @@
+{
+    "leaderboard_info": {
+        "total_models": 18
+    },
+    "models": [
+        {
+            "name": "Claude-3.7-Sonnet",
+            "name_link": "https://www.anthropic.com/claude",
+            "submitter": "Anthropic Team",
+            "submitter_link": "mailto:support@anthropic.com",
+            "submission_time": "2025-08-01T17:09:29.917540Z",
+            "model_type": "proprietary",
+            "model_size": "Unknown",
+            "is_multimodal": true,
+            "results": {
+                "Signal": {
+                    "accuracy": 75.78,
+                    "subcategories": {
+                        "Spectrum Type Classification": {
+                            "accuracy": 96.36
+                        },
+                        "Spectrum Quality Assessment": {
+                            "accuracy": 38.33
+                        },
+                        "Basic Feature Extraction": {
+                            "accuracy": 86.27
+                        },
+                        "Impurity Peak Detection": {
+                            "accuracy": 82.14
+                        }
+                    }
+                },
+                "Perception": {
+                    "accuracy": 79.9,
+                    "subcategories": {
+                        "Functional Group Recognition": {
+                            "accuracy": 71.43
+                        },
+                        "Elemental Compositional Prediction": {
+                            "accuracy": 88.89
+                        },
+                        "Peak Assignment": {
+                            "accuracy": 71.05
+                        },
+                        "Basic Property Prediction": {
+                            "accuracy": 88.24
+                        }
+                    }
+                },
+                "Semantic": {
+                    "accuracy": 81.94,
+                    "subcategories": {
+                        "Molecular Structure Elucidation": {
+                            "accuracy": 82.28
+                        },
+                        "Fusing Spectroscopic Modalities": {
+                            "accuracy": 74.36
+                        },
+                        "Multimodal Molecular Reasoning": {
+                            "accuracy": 89.19
+                        }
+                    }
+                },
+                "Generation": {
+                    "accuracy": 8.42,
+                    "subcategories": {
+                        "Forward Problems": {
+                            "accuracy": 20.0
+                        },
+                        "Inverse Problems": {
+                            "accuracy": 0
+                        },
+                        "De Novo Generation": {
+                            "accuracy": 5.26
+                        }
+                    }
+                },
+                "overall_accuracy": 61.51
+            },
+            "model_info": {
+                "homepage": "https://www.anthropic.com/claude",
+                "paper": "",
+                "code": "",
+                "description": "Claude 3.7 Sonnet with enhanced multimodal capabilities"
+            }
+        },
+        {
+            "name": "Doubao-1.5-Vision-Pro-Thinking",
+            "name_link": "https://www.volcengine.com/product/doubao",
+            "submitter": "ByteDance Team",
+            "submitter_link": "https://www.volcengine.com/",
+            "submission_time": "2025-08-01T17:09:29.934647Z",
+            "model_type": "proprietary",
+            "model_size": "Unknown",
+            "is_multimodal": true,
+            "results": {
+                "Signal": {
+                    "accuracy": 69.41,
+                    "subcategories": {
+                        "Spectrum Type Classification": {
+                            "accuracy": 96.36
+                        },
+                        "Spectrum Quality Assessment": {
+                            "accuracy": 35.0
+                        },
+                        "Basic Feature Extraction": {
+                            "accuracy": 78.43
+                        },
+                        "Impurity Peak Detection": {
+                            "accuracy": 67.86
+                        }
+                    }
+                },
+                "Perception": {
+                    "accuracy": 74.69,
+                    "subcategories": {
+                        "Functional Group Recognition": {
+                            "accuracy": 53.33
+                        },
+                        "Elemental Compositional Prediction": {
+                            "accuracy": 80.56
+                        },
+                        "Peak Assignment": {
+                            "accuracy": 73.68
+                        },
+                        "Basic Property Prediction": {
+                            "accuracy": 91.18
+                        }
+                    }
+                },
+                "Semantic": {
+                    "accuracy": 75.77,
+                    "subcategories": {
+                        "Molecular Structure Elucidation": {
+                            "accuracy": 68.75
+                        },
+                        "Fusing Spectroscopic Modalities": {
+                            "accuracy": 66.67
+                        },
+                        "Multimodal Molecular Reasoning": {
+                            "accuracy": 91.89
+                        }
+                    }
+                },
+                "Generation": {
+                    "accuracy": 25.64,
+                    "subcategories": {
+                        "Forward Problems": {
+                            "accuracy": 66.67
+                        },
+                        "Inverse Problems": {
+                            "accuracy": 5.0
+                        },
+                        "De Novo Generation": {
+                            "accuracy": 5.26
+                        }
+                    }
+                },
+                "overall_accuracy": 61.38
+            },
+            "model_info": {
+                "homepage": "https://www.volcengine.com/product/doubao",
+                "paper": "",
+                "code": "",
+                "description": "Doubao 1.5 Vision Pro with enhanced reasoning capabilities"
+            }
+        },
+        {
+            "name": "Qwen2.5-VL-72B-Instruct",
+            "name_link": "https://qwenlm.github.io/",
+            "submitter": "Alibaba DAMO Academy",
+            "submitter_link": "https://damo.alibaba.com/",
+            "submission_time": "2025-08-01T17:09:29.925741Z",
+            "model_type": "open_source",
+            "model_size": "72B",
+            "is_multimodal": true,
+            "results": {
+                "Signal": {
+                    "accuracy": 78.0,
+                    "subcategories": {
+                        "Spectrum Type Classification": {
+                            "accuracy": 94.55
+                        },
+                        "Spectrum Quality Assessment": {
+                            "accuracy": 38.33
+                        },
+                        "Basic Feature Extraction": {
+                            "accuracy": 86.27
+                        },
+                        "Impurity Peak Detection": {
+                            "accuracy": 92.86
+                        }
+                    }
+                },
+                "Perception": {
+                    "accuracy": 72.49,
+                    "subcategories": {
+                        "Functional Group Recognition": {
+                            "accuracy": 42.22
+                        },
+                        "Elemental Compositional Prediction": {
+                            "accuracy": 80.56
+                        },
+                        "Peak Assignment": {
+                            "accuracy": 78.95
+                        },
+                        "Basic Property Prediction": {
+                            "accuracy": 88.24
+                        }
+                    }
+                },
+                "Semantic": {
+                    "accuracy": 78.35,
+                    "subcategories": {
+                        "Molecular Structure Elucidation": {
+                            "accuracy": 66.25
+                        },
+                        "Fusing Spectroscopic Modalities": {
+                            "accuracy": 76.92
+                        },
+                        "Multimodal Molecular Reasoning": {
+                            "accuracy": 91.89
+                        }
+                    }
+                },
+                "Generation": {
+                    "accuracy": 13.51,
+                    "subcategories": {
+                        "Forward Problems": {
+                            "accuracy": 30.0
+                        },
+                        "Inverse Problems": {
+                            "accuracy": 0
+                        },
+                        "De Novo Generation": {
+                            "accuracy": 10.53
+                        }
+                    }
+                },
+                "overall_accuracy": 60.59
+            },
+            "model_info": {
+                "homepage": "https://qwenlm.github.io/",
+                "paper": "https://arxiv.org/abs/2407.10671",
+                "code": "https://github.com/QwenLM/Qwen2.5",
+                "description": "Qwen2.5-VL-72B large-scale open-source vision-language model"
+            }
+        },
+        {
+            "name": "GPT-4.1",
+            "name_link": "https://openai.com/gpt-4",
+            "submitter": "OpenAI Team",
+            "submitter_link": "mailto:research@openai.com",
+            "submission_time": "2025-08-01T17:09:29.920918Z",
+            "model_type": "proprietary",
+            "model_size": "Unknown",
+            "is_multimodal": true,
+            "results": {
+                "Signal": {
+                    "accuracy": 73.72,
+                    "subcategories": {
+                        "Spectrum Type Classification": {
+                            "accuracy": 94.55
+                        },
+                        "Spectrum Quality Assessment": {
+                            "accuracy": 28.33
+                        },
+                        "Basic Feature Extraction": {
+                            "accuracy": 86.27
+                        },
+                        "Impurity Peak Detection": {
+                            "accuracy": 85.71
+                        }
+                    }
+                },
+                "Perception": {
+                    "accuracy": 68.42,
+                    "subcategories": {
+                        "Functional Group Recognition": {
+                            "accuracy": 53.33
+                        },
+                        "Elemental Compositional Prediction": {
+                            "accuracy": 77.78
+                        },
+                        "Peak Assignment": {
+                            "accuracy": 63.16
+                        },
+                        "Basic Property Prediction": {
+                            "accuracy": 79.41
+                        }
+                    }
+                },
+                "Semantic": {
+                    "accuracy": 80.35,
+                    "subcategories": {
+                        "Molecular Structure Elucidation": {
+                            "accuracy": 82.5
+                        },
+                        "Fusing Spectroscopic Modalities": {
+                            "accuracy": 66.67
+                        },
+                        "Multimodal Molecular Reasoning": {
+                            "accuracy": 91.89
+                        }
+                    }
+                },
+                "Generation": {
+                    "accuracy": 14.62,
+                    "subcategories": {
+                        "Forward Problems": {
+                            "accuracy": 33.33
+                        },
+                        "Inverse Problems": {
+                            "accuracy": 10.53
+                        },
+                        "De Novo Generation": {
+                            "accuracy": 0
+                        }
+                    }
+                },
+                "overall_accuracy": 59.28
+            },
+            "model_info": {
+                "homepage": "https://openai.com/gpt-4",
+                "paper": "https://arxiv.org/abs/2303.08774",
+                "code": "",
+                "description": "GPT-4.1 with improved reasoning capabilities"
+            }
+        },
+        {
+            "name": "Qwen-VL-Max",
+            "name_link": "https://qwenlm.github.io/",
+            "submitter": "Alibaba DAMO Academy",
+            "submitter_link": "https://damo.alibaba.com/",
+            "submission_time": "2025-08-01T17:09:29.923747Z",
+            "model_type": "proprietary",
+            "model_size": "Unknown",
+            "is_multimodal": true,
+            "results": {
+                "Signal": {
+                    "accuracy": 78.57,
+                    "subcategories": {
+                        "Spectrum Type Classification": {
+                            "accuracy": 94.55
+                        },
+                        "Spectrum Quality Assessment": {
+                            "accuracy": 36.67
+                        },
+                        "Basic Feature Extraction": {
+                            "accuracy": 90.2
+                        },
+                        "Impurity Peak Detection": {
+                            "accuracy": 92.86
+                        }
+                    }
+                },
+                "Perception": {
+                    "accuracy": 76.94,
+                    "subcategories": {
+                        "Functional Group Recognition": {
+                            "accuracy": 60.0
+                        },
+                        "Elemental Compositional Prediction": {
+                            "accuracy": 80.56
+                        },
+                        "Peak Assignment": {
+                            "accuracy": 78.95
+                        },
+                        "Basic Property Prediction": {
+                            "accuracy": 88.24
+                        }
+                    }
+                },
+                "Semantic": {
+                    "accuracy": 65.39,
+                    "subcategories": {
+                        "Molecular Structure Elucidation": {
+                            "accuracy": 32.5
+                        },
+                        "Fusing Spectroscopic Modalities": {
+                            "accuracy": 71.79
+                        },
+                        "Multimodal Molecular Reasoning": {
+                            "accuracy": 91.89
+                        }
+                    }
+                },
+                "Generation": {
+                    "accuracy": 16.2,
+                    "subcategories": {
+                        "Forward Problems": {
+                            "accuracy": 43.33
+                        },
+                        "Inverse Problems": {
+                            "accuracy": 0
+                        },
+                        "De Novo Generation": {
+                            "accuracy": 5.26
+                        }
+                    }
+                },
+                "overall_accuracy": 59.27
+            },
+            "model_info": {
+                "homepage": "https://qwenlm.github.io/",
+                "paper": "https://arxiv.org/abs/2308.12966",
+                "code": "https://github.com/QwenLM/Qwen-VL",
+                "description": "Qwen-VL-Max with advanced vision-language understanding"
+            }
+        },
+        {
+            "name": "GPT-4-Vision",
+            "name_link": "https://openai.com/gpt-4",
+            "submitter": "OpenAI Team",
+            "submitter_link": "mailto:research@openai.com",
+            "submission_time": "2025-08-01T17:09:29.921628Z",
+            "model_type": "proprietary",
+            "model_size": "Unknown",
+            "is_multimodal": true,
+            "results": {
+                "Signal": {
+                    "accuracy": 73.32,
+                    "subcategories": {
+                        "Spectrum Type Classification": {
+                            "accuracy": 94.55
+                        },
+                        "Spectrum Quality Assessment": {
+                            "accuracy": 33.33
+                        },
+                        "Basic Feature Extraction": {
+                            "accuracy": 72.55
+                        },
+                        "Impurity Peak Detection": {
+                            "accuracy": 92.86
+                        }
+                    }
+                },
+                "Perception": {
+                    "accuracy": 74.74,
+                    "subcategories": {
+                        "Functional Group Recognition": {
+                            "accuracy": 73.33
+                        },
+                        "Elemental Compositional Prediction": {
+                            "accuracy": 72.22
+                        },
+                        "Peak Assignment": {
+                            "accuracy": 71.05
+                        },
+                        "Basic Property Prediction": {
+                            "accuracy": 82.35
+                        }
+                    }
+                },
+                "Semantic": {
+                    "accuracy": 74.97,
+                    "subcategories": {
+                        "Molecular Structure Elucidation": {
+                            "accuracy": 73.75
+                        },
+                        "Fusing Spectroscopic Modalities": {
+                            "accuracy": 53.85
+                        },
+                        "Multimodal Molecular Reasoning": {
+                            "accuracy": 97.3
+                        }
+                    }
+                },
+                "Generation": {
+                    "accuracy": 9.44,
+                    "subcategories": {
+                        "Forward Problems": {
+                            "accuracy": 23.33
+                        },
+                        "Inverse Problems": {
+                            "accuracy": 5.0
+                        },
+                        "De Novo Generation": {
+                            "accuracy": 0
+                        }
+                    }
+                },
+                "overall_accuracy": 58.12
+            },
+            "model_info": {
+                "homepage": "https://openai.com/gpt-4",
+                "paper": "https://arxiv.org/abs/2303.08774",
+                "code": "",
+                "description": "GPT-4 with vision capabilities for multimodal understanding"
+            }
+        },
+        {
+            "name": "Claude-3.5-Sonnet",
+            "name_link": "https://www.anthropic.com/claude",
+            "submitter": "Anthropic Team",
+            "submitter_link": "mailto:support@anthropic.com",
+            "submission_time": "2025-08-01T17:09:29.916821Z",
+            "model_type": "proprietary",
+            "model_size": "Unknown",
+            "is_multimodal": true,
+            "results": {
+                "Signal": {
+                    "accuracy": 68.15,
+                    "subcategories": {
+                        "Spectrum Type Classification": {
+                            "accuracy": 96.36
+                        },
+                        "Spectrum Quality Assessment": {
+                            "accuracy": 28.33
+                        },
+                        "Basic Feature Extraction": {
+                            "accuracy": 76.47
+                        },
+                        "Impurity Peak Detection": {
+                            "accuracy": 71.43
+                        }
+                    }
+                },
+                "Perception": {
+                    "accuracy": 74.85,
+                    "subcategories": {
+                        "Functional Group Recognition": {
+                            "accuracy": 60.0
+                        },
+                        "Elemental Compositional Prediction": {
+                            "accuracy": 77.78
+                        },
+                        "Peak Assignment": {
+                            "accuracy": 76.32
+                        },
+                        "Basic Property Prediction": {
+                            "accuracy": 85.29
+                        }
+                    }
+                },
+                "Semantic": {
+                    "accuracy": 82.11,
+                    "subcategories": {
+                        "Molecular Structure Elucidation": {
+                            "accuracy": 82.5
+                        },
+                        "Fusing Spectroscopic Modalities": {
+                            "accuracy": 69.23
+                        },
+                        "Multimodal Molecular Reasoning": {
+                            "accuracy": 94.59
+                        }
+                    }
+                },
+                "Generation": {
+                    "accuracy": 6.67,
+                    "subcategories": {
+                        "Forward Problems": {
+                            "accuracy": 20.0
+                        },
+                        "Inverse Problems": {
+                            "accuracy": 0
+                        },
+                        "De Novo Generation": {
+                            "accuracy": 0
+                        }
+                    }
+                },
+                "overall_accuracy": 57.95
+            },
+            "model_info": {
+                "homepage": "https://www.anthropic.com/claude",
+                "paper": "",
+                "code": "",
+                "description": "Claude 3.5 Sonnet with advanced reasoning and multimodal capabilities"
+            }
+        },
+        {
+            "name": "Claude-4-Opus",
+            "name_link": "https://www.anthropic.com/claude",
+            "submitter": "Anthropic Team",
+            "submitter_link": "mailto:support@anthropic.com",
+            "submission_time": "2025-08-01T17:09:29.919719Z",
+            "model_type": "proprietary",
+            "model_size": "Unknown",
+            "is_multimodal": true,
+            "results": {
+                "Signal": {
+                    "accuracy": 77.2,
+                    "subcategories": {
+                        "Spectrum Type Classification": {
+                            "accuracy": 96.36
+                        },
+                        "Spectrum Quality Assessment": {
+                            "accuracy": 33.33
+                        },
+                        "Basic Feature Extraction": {
+                            "accuracy": 86.27
+                        },
+                        "Impurity Peak Detection": {
+                            "accuracy": 92.86
+                        }
+                    }
+                },
+                "Perception": {
+                    "accuracy": 78.25,
+                    "subcategories": {
+                        "Functional Group Recognition": {
+                            "accuracy": 73.33
+                        },
+                        "Elemental Compositional Prediction": {
+                            "accuracy": 83.33
+                        },
+                        "Peak Assignment": {
+                            "accuracy": 71.05
+                        },
+                        "Basic Property Prediction": {
+                            "accuracy": 85.29
+                        }
+                    }
+                },
+                "Semantic": {
+                    "accuracy": 65.3,
+                    "subcategories": {
+                        "Molecular Structure Elucidation": {
+                            "accuracy": 32.5
+                        },
+                        "Fusing Spectroscopic Modalities": {
+                            "accuracy": 76.92
+                        },
+                        "Multimodal Molecular Reasoning": {
+                            "accuracy": 86.49
+                        }
+                    }
+                },
+                "Generation": {
+                    "accuracy": 7.31,
+                    "subcategories": {
+                        "Forward Problems": {
+                            "accuracy": 16.67
+                        },
+                        "Inverse Problems": {
+                            "accuracy": 0
+                        },
+                        "De Novo Generation": {
+                            "accuracy": 5.26
+                        }
+                    }
+                },
+                "overall_accuracy": 57.02
+            },
+            "model_info": {
+                "homepage": "https://www.anthropic.com/claude",
+                "paper": "",
+                "code": "",
+                "description": "Claude 4 Opus - most capable model in the Claude family"
+            }
+        },
+        {
+            "name": "Doubao-1.5-Vision-Pro",
+            "name_link": "https://www.volcengine.com/product/doubao",
+            "submitter": "ByteDance Team",
+            "submitter_link": "https://www.volcengine.com/",
+            "submission_time": "2025-08-01T17:09:29.932862Z",
+            "model_type": "proprietary",
+            "model_size": "Unknown",
+            "is_multimodal": true,
+            "results": {
+                "Signal": {
+                    "accuracy": 75.7,
+                    "subcategories": {
+                        "Spectrum Type Classification": {
+                            "accuracy": 98.18
+                        },
+                        "Spectrum Quality Assessment": {
+                            "accuracy": 33.33
+                        },
+                        "Basic Feature Extraction": {
+                            "accuracy": 78.43
+                        },
+                        "Impurity Peak Detection": {
+                            "accuracy": 92.86
+                        }
+                    }
+                },
+                "Perception": {
+                    "accuracy": 76.66,
+                    "subcategories": {
+                        "Functional Group Recognition": {
+                            "accuracy": 66.67
+                        },
+                        "Elemental Compositional Prediction": {
+                            "accuracy": 83.33
+                        },
+                        "Peak Assignment": {
+                            "accuracy": 68.42
+                        },
+                        "Basic Property Prediction": {
+                            "accuracy": 88.24
+                        }
+                    }
+                },
+                "Semantic": {
+                    "accuracy": 71.03,
+                    "subcategories": {
+                        "Molecular Structure Elucidation": {
+                            "accuracy": 67.5
+                        },
+                        "Fusing Spectroscopic Modalities": {
+                            "accuracy": 56.41
+                        },
+                        "Multimodal Molecular Reasoning": {
+                            "accuracy": 89.19
+                        }
+                    }
+                },
+                "Generation": {
+                    "accuracy": 2.22,
+                    "subcategories": {
+                        "Forward Problems": {
+                            "accuracy": 6.67
+                        },
+                        "Inverse Problems": {
+                            "accuracy": 0
+                        },
+                        "De Novo Generation": {
+                            "accuracy": 0
+                        }
+                    }
+                },
+                "overall_accuracy": 56.4
+            },
+            "model_info": {
+                "homepage": "https://www.volcengine.com/product/doubao",
+                "paper": "",
+                "code": "",
+                "description": "Doubao 1.5 Vision Pro multimodal model"
+            }
+        },
+        {
+            "name": "InternVL3-78B",
+            "name_link": "https://internvl.github.io/",
+            "submitter": "Shanghai AI Laboratory",
+            "submitter_link": "https://www.shlab.org.cn/",
+            "submission_time": "2025-08-01T17:09:29.926853Z",
+            "model_type": "open_source",
+            "model_size": "78B",
+            "is_multimodal": true,
+            "results": {
+                "Signal": {
+                    "accuracy": 69.18,
+                    "subcategories": {
+                        "Spectrum Type Classification": {
+                            "accuracy": 96.36
+                        },
+                        "Spectrum Quality Assessment": {
+                            "accuracy": 38.33
+                        },
+                        "Basic Feature Extraction": {
+                            "accuracy": 70.59
+                        },
+                        "Impurity Peak Detection": {
+                            "accuracy": 71.43
+                        }
+                    }
+                },
+                "Perception": {
+                    "accuracy": 73.33,
+                    "subcategories": {
+                        "Functional Group Recognition": {
+                            "accuracy": 48.49
+                        },
+                        "Elemental Compositional Prediction": {
+                            "accuracy": 75.0
+                        },
+                        "Peak Assignment": {
+                            "accuracy": 81.58
+                        },
+                        "Basic Property Prediction": {
+                            "accuracy": 88.24
+                        }
+                    }
+                },
+                "Semantic": {
+                    "accuracy": 71.84,
+                    "subcategories": {
+                        "Molecular Structure Elucidation": {
+                            "accuracy": 62.5
+                        },
+                        "Fusing Spectroscopic Modalities": {
+                            "accuracy": 69.23
+                        },
+                        "Multimodal Molecular Reasoning": {
+                            "accuracy": 83.78
+                        }
+                    }
+                },
+                "Generation": {
+                    "accuracy": 9.53,
+                    "subcategories": {
+                        "Forward Problems": {
+                            "accuracy": 23.33
+                        },
+                        "Inverse Problems": {
+                            "accuracy": 0
+                        },
+                        "De Novo Generation": {
+                            "accuracy": 5.26
+                        }
+                    }
+                },
+                "overall_accuracy": 55.97
+            },
+            "model_info": {
+                "homepage": "https://internvl.github.io/",
+                "paper": "https://arxiv.org/abs/2312.14238",
+                "code": "https://github.com/OpenGVLab/InternVL",
+                "description": "InternVL3-78B large-scale multimodal foundation model"
+            }
+        },
+        {
+            "name": "GPT-4o",
+            "name_link": "https://openai.com/gpt-4",
+            "submitter": "OpenAI Team",
+            "submitter_link": "mailto:research@openai.com",
+            "submission_time": "2025-08-01T17:09:29.920289Z",
+            "model_type": "proprietary",
+            "model_size": "Unknown",
+            "is_multimodal": true,
+            "results": {
+                "Signal": {
+                    "accuracy": 72.8,
+                    "subcategories": {
+                        "Spectrum Type Classification": {
+                            "accuracy": 96.36
+                        },
+                        "Spectrum Quality Assessment": {
+                            "accuracy": 33.33
+                        },
+                        "Basic Feature Extraction": {
+                            "accuracy": 68.63
+                        },
+                        "Impurity Peak Detection": {
+                            "accuracy": 92.86
+                        }
+                    }
+                },
+                "Perception": {
+                    "accuracy": 69.53,
+                    "subcategories": {
+                        "Functional Group Recognition": {
+                            "accuracy": 57.78
+                        },
+                        "Elemental Compositional Prediction": {
+                            "accuracy": 77.78
+                        },
+                        "Peak Assignment": {
+                            "accuracy": 63.16
+                        },
+                        "Basic Property Prediction": {
+                            "accuracy": 79.41
+                        }
+                    }
+                },
+                "Semantic": {
+                    "accuracy": 77.44,
+                    "subcategories": {
+                        "Molecular Structure Elucidation": {
+                            "accuracy": 78.75
+                        },
+                        "Fusing Spectroscopic Modalities": {
+                            "accuracy": 58.97
+                        },
+                        "Multimodal Molecular Reasoning": {
+                            "accuracy": 94.59
+                        }
+                    }
+                },
+                "Generation": {
+                    "accuracy": 3.33,
+                    "subcategories": {
+                        "Forward Problems": {
+                            "accuracy": 10.0
+                        },
+                        "Inverse Problems": {
+                            "accuracy": 0
+                        },
+                        "De Novo Generation": {
+                            "accuracy": 0
+                        }
+                    }
+                },
+                "overall_accuracy": 55.77
+            },
+            "model_info": {
+                "homepage": "https://openai.com/gpt-4",
+                "paper": "https://arxiv.org/abs/2303.08774",
+                "code": "",
+                "description": "GPT-4o with enhanced multimodal capabilities"
+            }
+        },
+        {
+            "name": "Grok-2-Vision",
+            "name_link": "https://grok.x.ai/",
+            "submitter": "xAI Team",
+            "submitter_link": "mailto:team@x.ai",
+            "submission_time": "2025-08-01T17:09:29.922697Z",
+            "model_type": "proprietary",
+            "model_size": "Unknown",
+            "is_multimodal": true,
+            "results": {
+                "Signal": {
+                    "accuracy": 72.5,
+                    "subcategories": {
+                        "Spectrum Type Classification": {
+                            "accuracy": 94.55
+                        },
+                        "Spectrum Quality Assessment": {
+                            "accuracy": 31.67
+                        },
+                        "Basic Feature Extraction": {
+                            "accuracy": 74.51
+                        },
+                        "Impurity Peak Detection": {
+                            "accuracy": 89.29
+                        }
+                    }
+                },
+                "Perception": {
+                    "accuracy": 75.26,
+                    "subcategories": {
+                        "Functional Group Recognition": {
+                            "accuracy": 64.44
+                        },
+                        "Elemental Compositional Prediction": {
+                            "accuracy": 80.56
+                        },
+                        "Peak Assignment": {
+                            "accuracy": 73.68
+                        },
+                        "Basic Property Prediction": {
+                            "accuracy": 82.35
+                        }
+                    }
+                },
+                "Semantic": {
+                    "accuracy": 61.75,
+                    "subcategories": {
+                        "Molecular Structure Elucidation": {
+                            "accuracy": 37.5
+                        },
+                        "Fusing Spectroscopic Modalities": {
+                            "accuracy": 66.67
+                        },
+                        "Multimodal Molecular Reasoning": {
+                            "accuracy": 81.08
+                        }
+                    }
+                },
+                "Generation": {
+                    "accuracy": 7.78,
+                    "subcategories": {
+                        "Forward Problems": {
+                            "accuracy": 23.33
+                        },
+                        "Inverse Problems": {
+                            "accuracy": 0
+                        },
+                        "De Novo Generation": {
+                            "accuracy": 0
+                        }
+                    }
+                },
+                "overall_accuracy": 54.32
+            },
+            "model_info": {
+                "homepage": "https://grok.x.ai/",
+                "paper": "",
+                "code": "",
+                "description": "Grok-2 with vision capabilities for multimodal reasoning"
+            }
+        },
+        {
+            "name": "Claude-3.5-Haiku",
+            "name_link": "https://www.anthropic.com/claude",
+            "submitter": "Anthropic Team",
+            "submitter_link": "mailto:support@anthropic.com",
+            "submission_time": "2025-08-01T17:09:29.919136Z",
+            "model_type": "proprietary",
+            "model_size": "Unknown",
+            "is_multimodal": true,
+            "results": {
+                "Signal": {
+                    "accuracy": 67.52,
+                    "subcategories": {
+                        "Spectrum Type Classification": {
+                            "accuracy": 94.55
+                        },
+                        "Spectrum Quality Assessment": {
+                            "accuracy": 31.67
+                        },
+                        "Basic Feature Extraction": {
+                            "accuracy": 50.98
+                        },
+                        "Impurity Peak Detection": {
+                            "accuracy": 92.86
+                        }
+                    }
+                },
+                "Perception": {
+                    "accuracy": 73.61,
+                    "subcategories": {
+                        "Functional Group Recognition": {
+                            "accuracy": 66.67
+                        },
+                        "Elemental Compositional Prediction": {
+                            "accuracy": 75.0
+                        },
+                        "Peak Assignment": {
+                            "accuracy": 76.32
+                        },
+                        "Basic Property Prediction": {
+                            "accuracy": 76.47
+                        }
+                    }
+                },
+                "Semantic": {
+                    "accuracy": 70.89,
+                    "subcategories": {
+                        "Molecular Structure Elucidation": {
+                            "accuracy": 67.5
+                        },
+                        "Fusing Spectroscopic Modalities": {
+                            "accuracy": 64.1
+                        },
+                        "Multimodal Molecular Reasoning": {
+                            "accuracy": 81.08
+                        }
+                    }
+                },
+                "Generation": {
+                    "accuracy": 3.33,
+                    "subcategories": {
+                        "Forward Problems": {
+                            "accuracy": 10.0
+                        },
+                        "Inverse Problems": {
+                            "accuracy": 0
+                        },
+                        "De Novo Generation": {
+                            "accuracy": 0
+                        }
+                    }
+                },
+                "overall_accuracy": 53.84
+            },
+            "model_info": {
+                "homepage": "https://www.anthropic.com/claude",
+                "paper": "",
+                "code": "",
+                "description": "Claude 3.5 Haiku - fast and efficient multimodal model"
+            }
+        },
+        {
+            "name": "Claude-4-Sonnet",
+            "name_link": "https://www.anthropic.com/claude",
+            "submitter": "Anthropic Team",
+            "submitter_link": "mailto:support@anthropic.com",
+            "submission_time": "2025-08-01T17:09:29.918518Z",
+            "model_type": "proprietary",
+            "model_size": "Unknown",
+            "is_multimodal": true,
+            "results": {
+                "Signal": {
+                    "accuracy": 78.11,
+                    "subcategories": {
+                        "Spectrum Type Classification": {
+                            "accuracy": 96.36
+                        },
+                        "Spectrum Quality Assessment": {
+                            "accuracy": 35.0
+                        },
+                        "Basic Feature Extraction": {
+                            "accuracy": 88.24
+                        },
+                        "Impurity Peak Detection": {
+                            "accuracy": 92.86
+                        }
+                    }
+                },
+                "Perception": {
+                    "accuracy": 65.78,
+                    "subcategories": {
+                        "Functional Group Recognition": {
+                            "accuracy": 62.22
+                        },
+                        "Elemental Compositional Prediction": {
+                            "accuracy": 63.89
+                        },
+                        "Peak Assignment": {
+                            "accuracy": 60.53
+                        },
+                        "Basic Property Prediction": {
+                            "accuracy": 76.47
+                        }
+                    }
+                },
+                "Semantic": {
+                    "accuracy": 41.57,
+                    "subcategories": {
+                        "Molecular Structure Elucidation": {
+                            "accuracy": 16.25
+                        },
+                        "Fusing Spectroscopic Modalities": {
+                            "accuracy": 43.59
+                        },
+                        "Multimodal Molecular Reasoning": {
+                            "accuracy": 64.86
+                        }
+                    }
+                },
+                "Generation": {
+                    "accuracy": 8.13,
+                    "subcategories": {
+                        "Forward Problems": {
+                            "accuracy": 3.33
+                        },
+                        "Inverse Problems": {
+                            "accuracy": 0
+                        },
+                        "De Novo Generation": {
+                            "accuracy": 21.05
+                        }
+                    }
+                },
+                "overall_accuracy": 48.4
+            },
+            "model_info": {
+                "homepage": "https://www.anthropic.com/claude",
+                "paper": "",
+                "code": "",
+                "description": "Claude 4 Sonnet with advanced reasoning capabilities"
+            }
+        },
+        {
+            "name": "Qwen2.5-VL-32B-Instruct",
+            "name_link": "https://qwenlm.github.io/",
+            "submitter": "Alibaba DAMO Academy",
+            "submitter_link": "https://damo.alibaba.com/",
+            "submission_time": "2025-08-01T17:09:29.924718Z",
+            "model_type": "open_source",
+            "model_size": "32B",
+            "is_multimodal": true,
+            "results": {
+                "Signal": {
+                    "accuracy": 57.02,
+                    "subcategories": {
+                        "Spectrum Type Classification": {
+                            "accuracy": 92.73
+                        },
+                        "Spectrum Quality Assessment": {
+                            "accuracy": 26.67
+                        },
+                        "Basic Feature Extraction": {
+                            "accuracy": 37.25
+                        },
+                        "Impurity Peak Detection": {
+                            "accuracy": 71.43
+                        }
+                    }
+                },
+                "Perception": {
+                    "accuracy": 48.89,
+                    "subcategories": {
+                        "Functional Group Recognition": {
+                            "accuracy": 57.78
+                        },
+                        "Elemental Compositional Prediction": {
+                            "accuracy": 44.44
+                        },
+                        "Peak Assignment": {
+                            "accuracy": 31.58
+                        },
+                        "Basic Property Prediction": {
+                            "accuracy": 61.76
+                        }
+                    }
+                },
+                "Semantic": {
+                    "accuracy": 17.03,
+                    "subcategories": {
+                        "Molecular Structure Elucidation": {
+                            "accuracy": 0.0
+                        },
+                        "Fusing Spectroscopic Modalities": {
+                            "accuracy": 5.13
+                        },
+                        "Multimodal Molecular Reasoning": {
+                            "accuracy": 45.95
+                        }
+                    }
+                },
+                "Generation": {
+                    "accuracy": 6.67,
+                    "subcategories": {
+                        "Forward Problems": {
+                            "accuracy": 20.0
+                        },
+                        "Inverse Problems": {
+                            "accuracy": 0
+                        },
+                        "De Novo Generation": {
+                            "accuracy": 0
+                        }
+                    }
+                },
+                "overall_accuracy": 32.4
+            },
+            "model_info": {
+                "homepage": "https://qwenlm.github.io/",
+                "paper": "https://arxiv.org/abs/2407.10671",
+                "code": "https://github.com/QwenLM/Qwen2.5",
+                "description": "Qwen2.5-VL-32B open-source vision-language model"
+            }
+        },
+        {
+            "name": "DeepSeek-VL2",
+            "name_link": "https://www.deepseek.com/",
+            "submitter": "DeepSeek Team",
+            "submitter_link": "https://github.com/deepseek-ai",
+            "submission_time": "2025-08-01T17:09:29.931327Z",
+            "model_type": "open_source",
+            "model_size": "Unknown",
+            "is_multimodal": true,
+            "results": {
+                "Signal": {
+                    "accuracy": 33.51,
+                    "subcategories": {
+                        "Spectrum Type Classification": {
+                            "accuracy": 52.73
+                        },
+                        "Spectrum Quality Assessment": {
+                            "accuracy": 23.33
+                        },
+                        "Basic Feature Extraction": {
+                            "accuracy": 29.41
+                        },
+                        "Impurity Peak Detection": {
+                            "accuracy": 28.57
+                        }
+                    }
+                },
+                "Perception": {
+                    "accuracy": 28.91,
+                    "subcategories": {
+                        "Functional Group Recognition": {
+                            "accuracy": 8.89
+                        },
+                        "Elemental Compositional Prediction": {
+                            "accuracy": 27.78
+                        },
+                        "Peak Assignment": {
+                            "accuracy": 28.95
+                        },
+                        "Basic Property Prediction": {
+                            "accuracy": 50.0
+                        }
+                    }
+                },
+                "Semantic": {
+                    "accuracy": 20.94,
+                    "subcategories": {
+                        "Molecular Structure Elucidation": {
+                            "accuracy": 15.0
+                        },
+                        "Fusing Spectroscopic Modalities": {
+                            "accuracy": 15.38
+                        },
+                        "Multimodal Molecular Reasoning": {
+                            "accuracy": 32.43
+                        }
+                    }
+                },
+                "Generation": {
+                    "accuracy": 6.75,
+                    "subcategories": {
+                        "Forward Problems": {
+                            "accuracy": 10.0
+                        },
+                        "Inverse Problems": {
+                            "accuracy": 5.0
+                        },
+                        "De Novo Generation": {
+                            "accuracy": 5.26
+                        }
+                    }
+                },
+                "overall_accuracy": 22.53
+            },
+            "model_info": {
+                "homepage": "https://www.deepseek.com/",
+                "paper": "https://arxiv.org/abs/2412.10302",
+                "code": "https://github.com/deepseek-ai/DeepSeek-VL2",
+                "description": "DeepSeek-VL2 advanced vision-language model"
+            }
+        },
+        {
+            "name": "Llama-3.2-90B-Vision-Instruct",
+            "name_link": "https://llama.meta.com/",
+            "submitter": "Meta AI",
+            "submitter_link": "https://ai.meta.com/",
+            "submission_time": "2025-08-01T17:09:29.929751Z",
+            "model_type": "open_source",
+            "model_size": "90B",
+            "is_multimodal": true,
+            "results": {
+                "Signal": {
+                    "accuracy": 27.12,
+                    "subcategories": {
+                        "Spectrum Type Classification": {
+                            "accuracy": 38.18
+                        },
+                        "Spectrum Quality Assessment": {
+                            "accuracy": 10.0
+                        },
+                        "Basic Feature Extraction": {
+                            "accuracy": 35.29
+                        },
+                        "Impurity Peak Detection": {
+                            "accuracy": 25.0
+                        }
+                    }
+                },
+                "Perception": {
+                    "accuracy": 23.77,
+                    "subcategories": {
+                        "Functional Group Recognition": {
+                            "accuracy": 17.78
+                        },
+                        "Elemental Compositional Prediction": {
+                            "accuracy": 27.78
+                        },
+                        "Peak Assignment": {
+                            "accuracy": 28.95
+                        },
+                        "Basic Property Prediction": {
+                            "accuracy": 20.59
+                        }
+                    }
+                },
+                "Semantic": {
+                    "accuracy": 23.21,
+                    "subcategories": {
+                        "Molecular Structure Elucidation": {
+                            "accuracy": 21.25
+                        },
+                        "Fusing Spectroscopic Modalities": {
+                            "accuracy": 5.13
+                        },
+                        "Multimodal Molecular Reasoning": {
+                            "accuracy": 43.24
+                        }
+                    }
+                },
+                "Generation": {
+                    "accuracy": 0,
+                    "subcategories": {
+                        "Forward Problems": {
+                            "accuracy": 0
+                        },
+                        "Inverse Problems": {
+                            "accuracy": 0
+                        },
+                        "De Novo Generation": {
+                            "accuracy": 0
+                        }
+                    }
+                },
+                "overall_accuracy": 18.52
+            },
+            "model_info": {
+                "homepage": "https://llama.meta.com/",
+                "paper": "https://arxiv.org/abs/2407.21783",
+                "code": "https://github.com/meta-llama/llama3",
+                "description": "Llama 3.2 90B with vision capabilities"
+            }
+        },
+        {
+            "name": "Llama-3.2-11B-Vision-Instruct",
+            "name_link": "https://llama.meta.com/",
+            "submitter": "Meta AI",
+            "submitter_link": "https://ai.meta.com/",
+            "submission_time": "2025-08-01T17:09:29.928044Z",
+            "model_type": "open_source",
+            "model_size": "11B",
+            "is_multimodal": true,
+            "results": {
+                "Signal": {
+                    "accuracy": 21.24,
+                    "subcategories": {
+                        "Spectrum Type Classification": {
+                            "accuracy": 34.55
+                        },
+                        "Spectrum Quality Assessment": {
+                            "accuracy": 11.67
+                        },
+                        "Basic Feature Extraction": {
+                            "accuracy": 13.73
+                        },
+                        "Impurity Peak Detection": {
+                            "accuracy": 25.0
+                        }
+                    }
+                },
+                "Perception": {
+                    "accuracy": 26.72,
+                    "subcategories": {
+                        "Functional Group Recognition": {
+                            "accuracy": 20.0
+                        },
+                        "Elemental Compositional Prediction": {
+                            "accuracy": 41.67
+                        },
+                        "Peak Assignment": {
+                            "accuracy": 15.79
+                        },
+                        "Basic Property Prediction": {
+                            "accuracy": 29.41
+                        }
+                    }
+                },
+                "Semantic": {
+                    "accuracy": 11.42,
+                    "subcategories": {
+                        "Molecular Structure Elucidation": {
+                            "accuracy": 7.5
+                        },
+                        "Fusing Spectroscopic Modalities": {
+                            "accuracy": 5.13
+                        },
+                        "Multimodal Molecular Reasoning": {
+                            "accuracy": 21.62
+                        }
+                    }
+                },
+                "Generation": {
+                    "accuracy": 0,
+                    "subcategories": {
+                        "Forward Problems": {
+                            "accuracy": 0
+                        },
+                        "Inverse Problems": {
+                            "accuracy": 0
+                        },
+                        "De Novo Generation": {
+                            "accuracy": 0
+                        }
+                    }
+                },
+                "overall_accuracy": 14.84
+            },
+            "model_info": {
+                "homepage": "https://llama.meta.com/",
+                "paper": "https://arxiv.org/abs/2407.21783",
+                "code": "https://github.com/meta-llama/llama3",
+                "description": "Llama 3.2 11B with vision capabilities"
+            }
+        }
+    ]
+}