New Framework for idea proposal and implementation on RD-Agent (#34)

* Commit init framework * Co-authored-by: Yuante Li (FESCO Adecco Human Resources) <v-yuanteli@microsoft.com> Co-authored-by: XianBW <XianBW@users.noreply.github.com> * add an import * refine the whole framework * benchmark related framework * fix black and isort errors * move requirements to folder * fix black again --------- Co-authored-by: Young <afe.young@gmail.com> Co-authored-by: xuyang1 <xuyang1@microsoft.com>
2026-07-29 00:17:44 +00:00 · 2024-06-28 11:45:23 +08:00
parent bc8d96e96c
commit 6b626eb56d
62 changed files with 654 additions and 1120 deletions
@@ -0,0 +1,48 @@
+from pathlib import Path
+from typing import Literal, Union
+
+from pydantic_settings import BaseSettings
+
+SELECT_METHOD = Literal["random", "scheduler"]
+
+
+class FactorImplementSettings(BaseSettings):
+    file_based_execution_data_folder: str = str(
+        (Path().cwd() / "git_ignore_folder" / "factor_implementation_source_data").absolute(),
+    )
+    file_based_execution_workspace: str = str(
+        (Path().cwd() / "git_ignore_folder" / "factor_implementation_workspace").absolute(),
+    )
+    implementation_execution_cache_location: str = str(
+        (Path().cwd() / "git_ignore_folder" / "factor_implementation_execution_cache").absolute(),
+    )
+    enable_execution_cache: bool = True  # whether to enable the execution cache
+
+    # TODO: the factor implement specific settings should not appear in this settings
+    # Evolving should have a method specific settings
+    # evolving related config
+    fail_task_trial_limit: int = 20
+
+    v1_query_former_trace_limit: int = 5
+    v1_query_similar_success_limit: int = 5
+
+    v2_query_component_limit: int = 1
+    v2_query_error_limit: int = 1
+    v2_query_former_trace_limit: int = 1
+    v2_error_summary: bool = False
+    v2_knowledge_sampler: float = 1.0
+
+    evo_multi_proc_n: int = 16  # how many processes to use for evolving (including eval & generation)
+
+    file_based_execution_timeout: int = 120  # seconds for each factor implementation execution
+
+    select_method: SELECT_METHOD = "random"
+    select_ratio: float = 0.5
+
+    max_loop: int = 10
+
+    knowledge_base_path: Union[str, None] = None
+    new_knowledge_base_path: Union[str, None] = None
+
+
+FACTOR_IMPLEMENT_SETTINGS = FactorImplementSettings()
@@ -0,0 +1,56 @@
+from pathlib import Path
+
+import pandas as pd
+
+# render it with jinja
+from jinja2 import Template
+
+from rdagent.components.task_implementation.factor_implementation.evolving.factor import (
+    FactorImplementTask,
+)
+from rdagent.components.task_implementation.factor_implementation.share_modules.factor_implementation_config import (
+    FACTOR_IMPLEMENT_SETTINGS,
+)
+
+TPL = """
+{{file_name}}
+```{{type_desc}}
+{{content}}
+````
+"""
+# Create a Jinja template from the string
+JJ_TPL = Template(TPL)
+
+
+def get_data_folder_intro():
+    """Direclty get the info of the data folder.
+    It is for preparing prompting message.
+    """
+    content_l = []
+    for p in Path(FACTOR_IMPLEMENT_SETTINGS.file_based_execution_data_folder).iterdir():
+        if p.name.endswith(".h5"):
+            df = pd.read_hdf(p)
+            # get  df.head() as string with full width
+            pd.set_option("display.max_columns", None)  # or 1000
+            pd.set_option("display.max_rows", None)  # or 1000
+            pd.set_option("display.max_colwidth", None)  # or 199
+            rendered = JJ_TPL.render(
+                file_name=p.name,
+                type_desc="generated by `pd.read_hdf(filename).head()`",
+                content=df.head().to_string(),
+            )
+            content_l.append(rendered)
+        elif p.name.endswith(".md"):
+            with open(p) as f:
+                content = f.read()
+                rendered = JJ_TPL.render(
+                    file_name=p.name,
+                    type_desc="markdown",
+                    content=content,
+                )
+                content_l.append(rendered)
+        else:
+            raise NotImplementedError(
+                f"file type {p.name} is not supported. Please implement its description function.",
+            )
+    return "\n ----------------- file spliter -------------\n".join(content_l)