docs: some changes in kaggle doc (#461)

* add kaggle scenario description for webapp

* catch download button error

* change kaggle auto_submit to False default

* Added guidance for custom templates in kaggle doc

* CI

* CI

* add kaggle data set instruction(NOTE)

* delete repeated description for kaggle scenario

---------

Co-authored-by: TPLin22 <tplin2@163.com>

docs/scens/kaggle_agent.rst

@@ -100,6 +100,10 @@ You can try our demo by running the following command:
 
     The `competition name` parameter must match the name used with the API on the Kaggle platform.
 
+(NOTE: The code for crawling Kaggle competition information may not be applicable to your environment. 
+
+If you cannot execute it normally, you can refer to the following **Example Guide: Running a Specific Experiment**.)
+
 
 📋 Competition List Available
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
@@ -149,6 +153,8 @@ You can try our demo by running the following command:
 
   - Alternatively, you can manually place the dataset in the specified location in advance.
 
+  - you can download the kaggle_data.zip in the release and unzip it to the directory configured by the `KG_LOCAL_DATA_PATH` environment variable. 
+
 
 - 🚀 **Run the Application**
 
@@ -166,6 +172,26 @@ You can try our demo by running the following command:
 
 For more information about Kaggle API Settings, refer to the `Kaggle API <https://github.com/Kaggle/kaggle-api>`_.
 
+
+🎨 Customize one template for a new competition
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+In order to facilitate RD-Agent to generate competition codes, we have specified a competition code structure:
+
+.. image:: kaggle_template.png
+   :alt: Design of Kaggle Code Template
+   :align: center
+
+- **feature directory** contains the feature engineering code. Generally no modification is required.
+- **model directory** contains the model codes.
+  select_xx.py is used to select different features according to different models.
+  model_xx.py is the basic code of different models. Generally, only some initial parameters need to be adjusted.
+- **fea_share_preprocess.py** is some basic preprocessing code shared by different models. The degree of customization here is high, but the preprocess_script() function needs to be retained, which will be called by train.py
+- **train.py** is the main code, which connects all the codes and is also the code called during the final execution.
+
+**We will soon provide a tool for automatic/semi-automatic template generation.**
+If you want to try a different competition now, you can refer to our current template structure and content to write a new template.
+
+
 🎯 Roadmap
 ~~~~~~~~~~~
 

rdagent/app/kaggle/conf.py

@@ -69,7 +69,7 @@ class KaggleBasePropSetting(BasePropSetting):
     knowledge_base_path: str = "kg_graph.pkl"
     """Advanced version of graph-based RAG"""
 
-    auto_submit: bool = True
+    auto_submit: bool = False
     """Automatically upload and submit each experiment result to Kaggle platform"""
 
     mini_case: bool = False

rdagent/log/ui/app.py

@@ -560,12 +560,16 @@ def feedback_window():
                     st.markdown(
                         f":green[**Exp Workspace**]: {str(fbe[0].content.experiment_workspace.workspace_path.absolute())}"
                     )
-                    st.download_button(
-                        label="**Download** submission.csv",
-                        data=submission_path.read_bytes(),
-                        file_name="submission.csv",
-                        mime="text/csv",
-                    )
+                    try:
+                        data = submission_path.read_bytes()
+                        st.download_button(
+                            label="**Download** submission.csv",
+                            data=data,
+                            file_name="submission.csv",
+                            mime="text/csv",
+                        )
+                    except Exception as e:
+                        st.markdown(f":red[**Download Button Error**]: {e}")
 
 
 def evolving_window():

rdagent/scenarios/kaggle/experiment/scenario.py

@@ -221,7 +221,29 @@ The model code should follow the simulator:
     @property
     def rich_style_description(self) -> str:
         return f"""
-This is the Kaggle scenario for the competition: {self.competition}
+### Kaggle Agent: Automated Feature Engineering & Model Tuning Evolution
+
+#### [Overview](#_summary)
+
+In this scenario, our automated system proposes hypothesis, choose action, implements code, conducts validation, and utilizes feedback in a continuous, iterative process.
+
+#### Kaggle Competition info
+
+Current Competition: [{self.competition}](https://www.kaggle.com/competitions/{self.competition})
+
+#### [Automated R&D](#_rdloops)
+
+- **[R (Research)](#_research)**
+- Iteration of ideas and hypotheses.
+- Continuous learning and knowledge construction.
+
+- **[D (Development)](#_development)**
+- Evolving code generation, model refinement, and features generation.
+- Automated implementation and testing of models/features.
+
+#### [Objective](#_summary)
+
+To automatically optimize performance metrics within the validation set or Kaggle Leaderboard, ultimately discovering the most efficient features and models through autonomous research and development.
 """
 
     def get_scenario_all_desc(self, task: Task | None = None) -> str:

rdagent/scenarios/kaggle/kaggle_crawler.py

@@ -96,7 +96,7 @@ def crawl_descriptions(competition: str, wait: float = 3.0, force: bool = False)
     return descriptions
 
 
-def download_data(competition: str, local_path: str = "/data/userdata/share/kaggle") -> None:
+def download_data(competition: str, local_path: str = KAGGLE_IMPLEMENT_SETTING.local_data_path) -> None:
     data_path = f"{local_path}/{competition}"
     if not Path(data_path).exists():
         subprocess.run(["kaggle", "competitions", "download", "-c", competition, "-p", data_path])
@@ -136,7 +136,7 @@ def score_rank(competition: str, score: float) -> tuple[int, float]:
 
 
 def download_notebooks(
-    competition: str, local_path: str = "/data/userdata/share/kaggle/notebooks", num: int = 15
+    competition: str, local_path: str = f"{KAGGLE_IMPLEMENT_SETTING.local_data_path}/notebooks", num: int = 15
 ) -> None:
     data_path = Path(f"{local_path}/{competition}")
     from kaggle.api.kaggle_api_extended import KaggleApi
@@ -183,7 +183,9 @@ def notebook_to_knowledge(notebook_text: str) -> str:
     return response
 
 
-def convert_notebooks_to_text(competition: str, local_path: str = "/data/userdata/share/kaggle/notebooks") -> None:
+def convert_notebooks_to_text(
+    competition: str, local_path: str = f"{KAGGLE_IMPLEMENT_SETTING.local_data_path}/notebooks"
+) -> None:
     data_path = Path(f"{local_path}/{competition}")
     converted_num = 0
 
@@ -219,7 +221,7 @@ def convert_notebooks_to_text(competition: str, local_path: str = "/data/userdat
     print(f"Converted {converted_num} notebooks to text files.")
 
 
-def collect_knowledge_texts(local_path: str = "/data/userdata/share/kaggle") -> dict[str, list[str]]:
+def collect_knowledge_texts(local_path: str = KAGGLE_IMPLEMENT_SETTING.local_data_path) -> dict[str, list[str]]:
     """
     {
         "competition1": [

rdagent/utils/env.py

@@ -186,8 +186,6 @@ class KGDockerConf(DockerConf):
     #     Path("git_ignore_folder/data").resolve(): "/root/.data/"
     # }
 
-    # local_data_path: str = "/data/userdata/share/kaggle"
-
 
 # physionet.org/files/mimic-eicu-fiddle-feature/1.0.0/FIDDLE_mimic3
 class DockerEnv(Env[DockerConf]):