update

app.py

@@ -9,7 +9,7 @@ from src.about import (
     CITATION_BUTTON_TEXT,
     EVALUATION_QUEUE_TEXT,
     INTRODUCTION_TEXT,
-    LLM_BENCHMARKS_TEXT,
+    DETECTOR_BENCHMARKS_TEXT,
     TITLE,
 )
 from src.display.css_html_js import custom_css
@@ -63,28 +63,28 @@ def init_leaderboard(dataframe):
     return Leaderboard(
         value=dataframe,
         datatype=[c.type for c in fields(AutoEvalColumn)],
-        select_columns=SelectColumns(
-            default_selection=[c.name for c in fields(AutoEvalColumn) if c.displayed_by_default],
-            cant_deselect=[c.name for c in fields(AutoEvalColumn) if c.never_hidden],
-            label="Select Columns to Display:",
-        ),
-        search_columns=[AutoEvalColumn.model.name, AutoEvalColumn.license.name],
-        hide_columns=[c.name for c in fields(AutoEvalColumn) if c.hidden],
-        filter_columns=[
-            ColumnFilter(AutoEvalColumn.model_type.name, type="checkboxgroup", label="Model types"),
-            ColumnFilter(AutoEvalColumn.precision.name, type="checkboxgroup", label="Precision"),
-            ColumnFilter(
-                AutoEvalColumn.params.name,
-                type="slider",
-                min=0.01,
-                max=150,
-                label="Select the number of parameters (B)",
-            ),
-            ColumnFilter(
-                AutoEvalColumn.still_on_hub.name, type="boolean", label="Deleted/incomplete", default=True
-            ),
-        ],
-        bool_checkboxgroup_label="Hide models",
+        # select_columns=SelectColumns(
+        #     default_selection=[c.name for c in fields(AutoEvalColumn) if c.displayed_by_default],
+        #     cant_deselect=[c.name for c in fields(AutoEvalColumn) if c.never_hidden],
+        #     label="Select Columns to Display:",
+        # ),
+        search_columns=[AutoEvalColumn.model.name],
+        # hide_columns=[c.name for c in fields(AutoEvalColumn) if c.hidden],
+        # filter_columns=[
+        #     ColumnFilter(AutoEvalColumn.model_type.name, type="checkboxgroup", label="Model types"),
+        #     ColumnFilter(AutoEvalColumn.precision.name, type="checkboxgroup", label="Precision"),
+        #     ColumnFilter(
+        #         AutoEvalColumn.params.name,
+        #         type="slider",
+        #         min=0.01,
+        #         max=150,
+        #         label="Select the number of parameters (B)",
+        #     ),
+        #     ColumnFilter(
+        #         AutoEvalColumn.still_on_hub.name, type="boolean", label="Deleted/incomplete", default=True
+        #     ),
+        # ],
+        # bool_checkboxgroup_label="Hide models",
         interactive=False,
     )
 
@@ -95,98 +95,106 @@ with demo:
     gr.Markdown(INTRODUCTION_TEXT, elem_classes="markdown-text")
 
     with gr.Tabs(elem_classes="tab-buttons") as tabs:
-        with gr.TabItem("🏅 LLM Benchmark", elem_id="llm-benchmark-tab-table", id=0):
+        with gr.TabItem("🏅 Detector Leaderboard", elem_id="detector-benchmark-tab-table", id=0):
             leaderboard = init_leaderboard(LEADERBOARD_DF)
 
-        with gr.TabItem("📝 About", elem_id="llm-benchmark-tab-table", id=2):
-            gr.Markdown(LLM_BENCHMARKS_TEXT, elem_classes="markdown-text")
-
-        with gr.TabItem("🚀 Submit here! ", elem_id="llm-benchmark-tab-table", id=3):
-            with gr.Column():
-                with gr.Row():
-                    gr.Markdown(EVALUATION_QUEUE_TEXT, elem_classes="markdown-text")
-
-                with gr.Column():
-                    with gr.Accordion(
-                        f"✅ Finished Evaluations ({len(finished_eval_queue_df)})",
-                        open=False,
-                    ):
-                        with gr.Row():
-                            finished_eval_table = gr.components.Dataframe(
-                                value=finished_eval_queue_df,
-                                headers=EVAL_COLS,
-                                datatype=EVAL_TYPES,
-                                row_count=5,
-                            )
-                    with gr.Accordion(
-                        f"🔄 Running Evaluation Queue ({len(running_eval_queue_df)})",
-                        open=False,
-                    ):
-                        with gr.Row():
-                            running_eval_table = gr.components.Dataframe(
-                                value=running_eval_queue_df,
-                                headers=EVAL_COLS,
-                                datatype=EVAL_TYPES,
-                                row_count=5,
-                            )
-
-                    with gr.Accordion(
-                        f"⏳ Pending Evaluation Queue ({len(pending_eval_queue_df)})",
-                        open=False,
-                    ):
-                        with gr.Row():
-                            pending_eval_table = gr.components.Dataframe(
-                                value=pending_eval_queue_df,
-                                headers=EVAL_COLS,
-                                datatype=EVAL_TYPES,
-                                row_count=5,
-                            )
-            with gr.Row():
-                gr.Markdown("# ✉️✨ Submit your model here!", elem_classes="markdown-text")
-
+        with gr.TabItem("🔍 Detector Playground ", elem_id="detector-playground-tab-table", id=1):
             with gr.Row():
                 with gr.Column():
-                    model_name_textbox = gr.Textbox(label="Model name")
-                    revision_name_textbox = gr.Textbox(label="Revision commit", placeholder="main")
-                    model_type = gr.Dropdown(
-                        choices=[t.to_str(" : ") for t in ModelType if t != ModelType.Unknown],
-                        label="Model type",
-                        multiselect=False,
-                        value=None,
-                        interactive=True,
-                    )
-
-                with gr.Column():
-                    precision = gr.Dropdown(
-                        choices=[i.value.name for i in Precision if i != Precision.Unknown],
-                        label="Precision",
-                        multiselect=False,
-                        value="float16",
-                        interactive=True,
-                    )
-                    weight_type = gr.Dropdown(
-                        choices=[i.value.name for i in WeightType],
-                        label="Weights type",
-                        multiselect=False,
-                        value="Original",
-                        interactive=True,
-                    )
-                    base_model_name_textbox = gr.Textbox(label="Base model (for delta or adapter weights)")
-
-            submit_button = gr.Button("Submit Eval")
-            submission_result = gr.Markdown()
-            submit_button.click(
-                add_new_eval,
-                [
-                    model_name_textbox,
-                    base_model_name_textbox,
-                    revision_name_textbox,
-                    precision,
-                    weight_type,
-                    model_type,
-                ],
-                submission_result,
-            )
+                    # print(LEADERBOARD_DF.keys())
+                    gr.Dropdown(LEADERBOARD_DF['Model'].tolist())
+                    gr.Image()
+                    gr.Button("Submit")
+
+        with gr.TabItem("📝 About", elem_id="detector-benchmark-tab-table", id=2):
+            gr.Markdown(DETECTOR_BENCHMARKS_TEXT, elem_classes="markdown-text")
+
+        # with gr.TabItem("🚀 Submit here! ", elem_id="detector-benchmark-tab-table", id=3):
+        #     with gr.Column():
+        #         with gr.Row():
+        #             gr.Markdown(EVALUATION_QUEUE_TEXT, elem_classes="markdown-text")
+
+        #         with gr.Column():
+        #             with gr.Accordion(
+        #                 f"✅ Finished Evaluations ({len(finished_eval_queue_df)})",
+        #                 open=False,
+        #             ):
+        #                 with gr.Row():
+        #                     finished_eval_table = gr.components.Dataframe(
+        #                         value=finished_eval_queue_df,
+        #                         headers=EVAL_COLS,
+        #                         datatype=EVAL_TYPES,
+        #                         row_count=5,
+        #                     )
+        #             with gr.Accordion(
+        #                 f"🔄 Running Evaluation Queue ({len(running_eval_queue_df)})",
+        #                 open=False,
+        #             ):
+        #                 with gr.Row():
+        #                     running_eval_table = gr.components.Dataframe(
+        #                         value=running_eval_queue_df,
+        #                         headers=EVAL_COLS,
+        #                         datatype=EVAL_TYPES,
+        #                         row_count=5,
+        #                     )
+
+        #             with gr.Accordion(
+        #                 f"⏳ Pending Evaluation Queue ({len(pending_eval_queue_df)})",
+        #                 open=False,
+        #             ):
+        #                 with gr.Row():
+        #                     pending_eval_table = gr.components.Dataframe(
+        #                         value=pending_eval_queue_df,
+        #                         headers=EVAL_COLS,
+        #                         datatype=EVAL_TYPES,
+        #                         row_count=5,
+        #                     )
+        #     with gr.Row():
+        #         gr.Markdown("# ✉️✨ Submit your model here!", elem_classes="markdown-text")
+
+        #     with gr.Row():
+        #         with gr.Column():
+        #             model_name_textbox = gr.Textbox(label="Model name")
+        #             revision_name_textbox = gr.Textbox(label="Revision commit", placeholder="main")
+        #             model_type = gr.Dropdown(
+        #                 choices=[t.to_str(" : ") for t in ModelType if t != ModelType.Unknown],
+        #                 label="Model type",
+        #                 multiselect=False,
+        #                 value=None,
+        #                 interactive=True,
+        #             )
+
+        #         with gr.Column():
+        #             precision = gr.Dropdown(
+        #                 choices=[i.value.name for i in Precision if i != Precision.Unknown],
+        #                 label="Precision",
+        #                 multiselect=False,
+        #                 value="float16",
+        #                 interactive=True,
+        #             )
+        #             weight_type = gr.Dropdown(
+        #                 choices=[i.value.name for i in WeightType],
+        #                 label="Weights type",
+        #                 multiselect=False,
+        #                 value="Original",
+        #                 interactive=True,
+        #             )
+        #             base_model_name_textbox = gr.Textbox(label="Base model (for delta or adapter weights)")
+
+        #     submit_button = gr.Button("Submit Eval")
+        #     submission_result = gr.Markdown()
+        #     submit_button.click(
+        #         add_new_eval,
+        #         [
+        #             model_name_textbox,
+        #             base_model_name_textbox,
+        #             revision_name_textbox,
+        #             precision,
+        #             weight_type,
+        #             model_type,
+        #         ],
+        #         submission_result,
+        #     )
 
     with gr.Row():
         with gr.Accordion("📙 Citation", open=False):

src/about.py

@@ -12,8 +12,11 @@ class Task:
 # ---------------------------------------------------
 class Tasks(Enum):
     # task_key in the json file, metric_key in the json file, name to display in the leaderboard 
-    task0 = Task("anli_r1", "acc", "ANLI")
-    task1 = Task("logiqa", "acc_norm", "LogiQA")
+    task0 = Task("miragenews", "acc", "MiRAGeNews")
+    task1 = Task("genimage", "acc", "GenImage")
+    # task2 = Task("cnn_det", "acc_norm", "CNN Detection")
+    # task3 = Task("forgery_net", "acc_norm", "Forgery Net")
+    # task4 = Task("deepfake_det", "acc_norm", "Deepfake Detection")
 
 NUM_FEWSHOT = 0 # Change with your few shot
 # ---------------------------------------------------
@@ -21,15 +24,15 @@ NUM_FEWSHOT = 0 # Change with your few shot
 
 
 # Your leaderboard name
-TITLE = """<h1 align="center" id="space-title">Demo leaderboard</h1>"""
+TITLE = """<h1 align="center" id="space-title"> 🕵️ AI-Generated Image Detector Benchmark</h1>"""
 
 # What does your leaderboard evaluate?
 INTRODUCTION_TEXT = """
-Intro text
+AI-Generated Image Detector Benchmark is a platform for evaluating the performance of existing detectors on various data sources and tasks. We collected X images from Y generators with 2 different tasks: Full Image Generation Detection and Partial Image Manipulation Detection.
 """
 
 # Which evaluations are you running? how can people reproduce what you have?
-LLM_BENCHMARKS_TEXT = f"""
+DETECTOR_BENCHMARKS_TEXT = f"""
 ## How it works
 
 ## Reproducibility

src/display/utils.py

@@ -23,22 +23,22 @@ class ColumnContent:
 ## Leaderboard columns
 auto_eval_column_dict = []
 # Init
-auto_eval_column_dict.append(["model_type_symbol", ColumnContent, ColumnContent("T", "str", True, never_hidden=True)])
+# auto_eval_column_dict.append(["model_type_symbol", ColumnContent, ColumnContent("T", "str", True, never_hidden=True)])
 auto_eval_column_dict.append(["model", ColumnContent, ColumnContent("Model", "markdown", True, never_hidden=True)])
 #Scores
 auto_eval_column_dict.append(["average", ColumnContent, ColumnContent("Average ⬆️", "number", True)])
 for task in Tasks:
     auto_eval_column_dict.append([task.name, ColumnContent, ColumnContent(task.value.col_name, "number", True)])
 # Model information
-auto_eval_column_dict.append(["model_type", ColumnContent, ColumnContent("Type", "str", False)])
-auto_eval_column_dict.append(["architecture", ColumnContent, ColumnContent("Architecture", "str", False)])
-auto_eval_column_dict.append(["weight_type", ColumnContent, ColumnContent("Weight type", "str", False, True)])
-auto_eval_column_dict.append(["precision", ColumnContent, ColumnContent("Precision", "str", False)])
-auto_eval_column_dict.append(["license", ColumnContent, ColumnContent("Hub License", "str", False)])
-auto_eval_column_dict.append(["params", ColumnContent, ColumnContent("#Params (B)", "number", False)])
-auto_eval_column_dict.append(["likes", ColumnContent, ColumnContent("Hub ❤️", "number", False)])
-auto_eval_column_dict.append(["still_on_hub", ColumnContent, ColumnContent("Available on the hub", "bool", False)])
-auto_eval_column_dict.append(["revision", ColumnContent, ColumnContent("Model sha", "str", False, False)])
+# auto_eval_column_dict.append(["model_type", ColumnContent, ColumnContent("Type", "str", False)])
+# auto_eval_column_dict.append(["architecture", ColumnContent, ColumnContent("Architecture", "str", False)])
+# auto_eval_column_dict.append(["weight_type", ColumnContent, ColumnContent("Weight type", "str", False, True)])
+# auto_eval_column_dict.append(["precision", ColumnContent, ColumnContent("Precision", "str", False)])
+# auto_eval_column_dict.append(["license", ColumnContent, ColumnContent("Hub License", "str", False)])
+# auto_eval_column_dict.append(["params", ColumnContent, ColumnContent("#Params (B)", "number", False)])
+# auto_eval_column_dict.append(["likes", ColumnContent, ColumnContent("Hub ❤️", "number", False)])
+# auto_eval_column_dict.append(["still_on_hub", ColumnContent, ColumnContent("Available on the hub", "bool", False)])
+# auto_eval_column_dict.append(["revision", ColumnContent, ColumnContent("Model sha", "str", False, False)])
 
 # We use make dataclass to dynamically fill the scores from Tasks
 AutoEvalColumn = make_dataclass("AutoEvalColumn", auto_eval_column_dict, frozen=True)

src/leaderboard/read_evals.py

@@ -20,17 +20,17 @@ class EvalResult:
     full_model: str # org/model (path on hub)
     org: str 
     model: str
-    revision: str # commit hash, "" if main
+    # revision: str # commit hash, "" if main
     results: dict
-    precision: Precision = Precision.Unknown
-    model_type: ModelType = ModelType.Unknown # Pretrained, fine tuned, ...
-    weight_type: WeightType = WeightType.Original # Original or Adapter
-    architecture: str = "Unknown" 
-    license: str = "?"
-    likes: int = 0
-    num_params: int = 0
-    date: str = "" # submission date of request file
-    still_on_hub: bool = False
+    # precision: Precision = Precision.Unknown
+    # model_type: ModelType = ModelType.Unknown # Pretrained, fine tuned, ...
+    # weight_type: WeightType = WeightType.Original # Original or Adapter
+    # architecture: str = "Unknown" 
+    # license: str = "?"
+    # likes: int = 0
+    # num_params: int = 0
+    # date: str = "" # submission date of request file
+    # still_on_hub: bool = False
 
     @classmethod
     def init_from_json_file(self, json_filepath):
@@ -41,7 +41,7 @@ class EvalResult:
         config = data.get("config")
 
         # Precision
-        precision = Precision.from_str(config.get("model_dtype"))
+        # precision = Precision.from_str(config.get("model_dtype"))
 
         # Get model and org
         org_and_model = config.get("model_name", config.get("model_args", None))
@@ -50,11 +50,11 @@ class EvalResult:
         if len(org_and_model) == 1:
             org = None
             model = org_and_model[0]
-            result_key = f"{model}_{precision.value.name}"
+            result_key = f"{model}"
         else:
             org = org_and_model[0]
             model = org_and_model[1]
-            result_key = f"{org}_{model}_{precision.value.name}"
+            result_key = f"{org}_{model}"
         full_model = "/".join(org_and_model)
 
         still_on_hub, _, model_config = is_model_on_hub(
@@ -85,15 +85,15 @@ class EvalResult:
             org=org,
             model=model,
             results=results,
-            precision=precision,  
-            revision= config.get("model_sha", ""),
-            still_on_hub=still_on_hub,
-            architecture=architecture
+            # precision=precision,  
+            # revision= config.get("model_sha", ""),
+            # still_on_hub=still_on_hub,
+            # architecture=architecture
         )
 
     def update_with_request_file(self, requests_path):
         """Finds the relevant request file for the current model and updates info with it"""
-        request_file = get_request_file_for_model(requests_path, self.full_model, self.precision.value.name)
+        request_file = get_request_file_for_model(requests_path, self.full_model)
 
         try:
             with open(request_file, "r") as f:
@@ -112,27 +112,25 @@ class EvalResult:
         average = sum([v for v in self.results.values() if v is not None]) / len(Tasks)
         data_dict = {
             "eval_name": self.eval_name,  # not a column, just a save name,
-            AutoEvalColumn.precision.name: self.precision.value.name,
-            AutoEvalColumn.model_type.name: self.model_type.value.name,
-            AutoEvalColumn.model_type_symbol.name: self.model_type.value.symbol,
-            AutoEvalColumn.weight_type.name: self.weight_type.value.name,
-            AutoEvalColumn.architecture.name: self.architecture,
+            # AutoEvalColumn.precision.name: self.precision.value.name,
+            # AutoEvalColumn.model_type.name: self.model_type.value.name,
+            # AutoEvalColumn.model_type_symbol.name: self.model_type.value.symbol,
+            # AutoEvalColumn.weight_type.name: self.weight_type.value.name,
+            # AutoEvalColumn.architecture.name: self.architecture,
             AutoEvalColumn.model.name: make_clickable_model(self.full_model),
-            AutoEvalColumn.revision.name: self.revision,
+            # AutoEvalColumn.revision.name: self.revision,
             AutoEvalColumn.average.name: average,
-            AutoEvalColumn.license.name: self.license,
-            AutoEvalColumn.likes.name: self.likes,
-            AutoEvalColumn.params.name: self.num_params,
-            AutoEvalColumn.still_on_hub.name: self.still_on_hub,
+            # AutoEvalColumn.license.name: self.license,
+            # AutoEvalColumn.likes.name: self.likes,
+            # AutoEvalColumn.params.name: self.num_params,
+            # AutoEvalColumn.still_on_hub.name: self.still_on_hub,
         }
-
         for task in Tasks:
             data_dict[task.value.col_name] = self.results[task.value.benchmark]
-
         return data_dict
 
 
-def get_request_file_for_model(requests_path, model_name, precision):
+def get_request_file_for_model(requests_path, model_name):
     """Selects the correct request file for a given model. Only keeps runs tagged as FINISHED"""
     request_files = os.path.join(
         requests_path,
@@ -148,7 +146,7 @@ def get_request_file_for_model(requests_path, model_name, precision):
             req_content = json.load(f)
             if (
                 req_content["status"] in ["FINISHED"]
-                and req_content["precision"] == precision.split(".")[-1]
+                # and req_content["precision"] == precision.split(".")[-1]
             ):
                 request_file = tmp_request_file
     return request_file
@@ -176,7 +174,7 @@ def get_raw_eval_results(results_path: str, requests_path: str) -> list[EvalResu
     for model_result_filepath in model_result_filepaths:
         # Creation of result
         eval_result = EvalResult.init_from_json_file(model_result_filepath)
-        eval_result.update_with_request_file(requests_path)
+        # eval_result.update_with_request_file(requests_path)
 
         # Store results of same eval together
         eval_name = eval_result.eval_name
@@ -187,10 +185,12 @@ def get_raw_eval_results(results_path: str, requests_path: str) -> list[EvalResu
 
     results = []
     for v in eval_results.values():
+        # print(v)
         try:
             v.to_dict() # we test if the dict version is complete
             results.append(v)
         except KeyError:  # not all eval values present
+            # print(v)
             continue
-
+    # print("RES", results)
     return results