gavindiaz/NexQuant
1
0
Fork 0
mirror of https://github.com/NicolasBohn/NexQuant.git synced 2026-08-01 09:27:43 +00:00
Code Issues Packages Projects Releases Wiki Activity

costeer max loop as hyperp & skip workflow in eval (#737)

Browse Source
This commit is contained in:
Xu Yang
2025-04-02 00:18:42 -06:00
committed by GitHub
parent 1474478768
commit bc8e9dc23a
11 changed files with 69 additions and 13 deletions
Download Patch File Download Diff File Expand all files Collapse all files
rdagent/app/data_science/conf.py
+3
View File
@@ -26,5 +26,8 @@ class DataScienceBasePropSetting(KaggleBasePropSetting):
proposal_version: str = "v1"
coder_max_loop: int = 10
runner_max_loop: int = 3
DS_RD_SETTING = DataScienceBasePropSetting()
rdagent/components/coder/data_science/ensemble/__init__.py
+10 -1
View File
@@ -154,4 +154,13 @@ class EnsembleCoSTEER(CoSTEER):
eva = CoSTEERMultiEvaluator(EnsembleCoSTEEREvaluator(scen=scen), scen=scen)
es = EnsembleMultiProcessEvolvingStrategy(scen=scen, settings=settings)
super().__init__(*args, settings=settings, eva=eva, es=es, evolving_version=2, scen=scen, **kwargs)
super().__init__(
*args,
settings=settings,
eva=eva,
es=es,
evolving_version=2,
scen=scen,
max_loop=DS_RD_SETTING.coder_max_loop,
**kwargs,
)
rdagent/components/coder/data_science/ensemble/eval.py
+1 -1
View File
@@ -67,7 +67,7 @@ class EnsembleCoSTEEREvaluator(CoSTEEREvaluator):
stdout += f"\nNOTE: the above scripts run with return code {ret_code}"
if "main.py" in implementation.file_dict:
if "main.py" in implementation.file_dict and ret_code == 0:
workflow_stdout = implementation.execute(env=env, entry="python main.py")
workflow_stdout = re.sub(r"=== Start of EDA part ===(.*)=== End of EDA part ===", "", workflow_stdout)
else:
rdagent/components/coder/data_science/feature/__init__.py
+10 -1
View File
@@ -130,4 +130,13 @@ class FeatureCoSTEER(CoSTEER):
) # Please specify whether you agree running your eva in parallel or not
es = FeatureMultiProcessEvolvingStrategy(scen=scen, settings=settings)
super().__init__(*args, settings=settings, eva=eva, es=es, evolving_version=2, scen=scen, **kwargs)
super().__init__(
*args,
settings=settings,
eva=eva,
es=es,
evolving_version=2,
scen=scen,
max_loop=DS_RD_SETTING.coder_max_loop,
**kwargs,
)
rdagent/components/coder/data_science/feature/eval.py
+2 -2
View File
@@ -52,9 +52,9 @@ class FeatureCoSTEEREvaluator(CoSTEEREvaluator):
test_code = (DIRNAME / "eval_tests" / "feature_test.txt").read_text()
implementation.inject_files(**{fname: test_code})
stdout = implementation.execute(env=env, entry=f"python {fname}")
stdout, ret_code = implementation.execute_ret_code(env=env, entry=f"python {fname}")
if "main.py" in implementation.file_dict:
if "main.py" in implementation.file_dict and ret_code == 0:
workflow_stdout = implementation.execute(env=env, entry="python main.py")
workflow_stdout = re.sub(r"=== Start of EDA part ===(.*)=== End of EDA part ===", "", workflow_stdout)
else:
rdagent/components/coder/data_science/model/__init__.py
+10 -1
View File
@@ -162,4 +162,13 @@ class ModelCoSTEER(CoSTEER):
# eva = ModelGeneralCaseSpecEvaluator(scen=scen)
es = ModelMultiProcessEvolvingStrategy(scen=scen, settings=settings)
super().__init__(*args, settings=settings, eva=eva, es=es, evolving_version=2, scen=scen, **kwargs)
super().__init__(
*args,
settings=settings,
eva=eva,
es=es,
evolving_version=2,
scen=scen,
max_loop=DS_RD_SETTING.coder_max_loop,
**kwargs,
)
rdagent/components/coder/data_science/model/eval.py
+3 -2
View File
@@ -67,17 +67,18 @@ class ModelGeneralCaseSpecEvaluator(CoSTEEREvaluator):
(DIRNAME / "eval_tests" / "model_test.txt").read_text().replace("model01", target_task.name)
) # only check the model changed this time
implementation.inject_files(**{fname: test_code})
stdout = implementation.execute(env=env, entry=f"python {fname}")
stdout, ret_code = implementation.execute_ret_code(env=env, entry=f"python {fname}")
if stdout is None:
raise CoderError(
"The execution output contains too many progress bars and results in the LLM's token size exceeding the limit."
)
else:
ret_code = 0
if_model_removed = True
stdout = f"Model {target_task.name} removal succeeded."
if "main.py" in implementation.file_dict:
if "main.py" in implementation.file_dict and ret_code == 0:
workflow_stdout = implementation.execute(env=env, entry="python main.py")
workflow_stdout = re.sub(r"=== Start of EDA part ===(.*)=== End of EDA part ===", "", workflow_stdout)
else:
rdagent/components/coder/data_science/raw_data_loader/__init__.py
+10 -1
View File
@@ -211,7 +211,16 @@ class DataLoaderCoSTEER(CoSTEER):
) # Please specify whether you agree running your eva in parallel or not
es = DataLoaderMultiProcessEvolvingStrategy(scen=scen, settings=settings)
super().__init__(*args, settings=settings, eva=eva, es=es, evolving_version=2, scen=scen, **kwargs)
super().__init__(
*args,
settings=settings,
eva=eva,
es=es,
evolving_version=2,
scen=scen,
max_loop=DS_RD_SETTING.coder_max_loop,
**kwargs,
)
def develop(self, exp):
new_exp = super().develop(exp)
rdagent/components/coder/data_science/raw_data_loader/eval.py
+2 -2
View File
@@ -54,14 +54,14 @@ class DataLoaderCoSTEEREvaluator(CoSTEEREvaluator):
fname = "test/data_loader_test.py"
test_code = (DIRNAME / "eval_tests" / "data_loader_test.txt").read_text()
implementation.inject_files(**{fname: test_code})
stdout = implementation.execute(env=env, entry=f"python {fname}")
stdout, ret_code = implementation.execute_ret_code(env=env, entry=f"python {fname}")
match = re.search(r"(.*?)=== Start of EDA part ===(.*)=== End of EDA part ===(.*)", stdout, re.DOTALL)
stdout_part_1, eda_output, stdout_part_2 = match.groups() if match else (stdout, None, "")
stdout = stdout_part_1 + stdout_part_2
if eda_output is not None and len(eda_output.split(" ")) > 10000:
eda_output += "Length of EDA output is too long, truncated. Please reject this implementation and motivate it to reduce the length of EDA output."
if "main.py" in implementation.file_dict:
if "main.py" in implementation.file_dict and ret_code == 0:
workflow_stdout = implementation.execute(env=env, entry="python main.py")
workflow_stdout = re.sub(r"=== Start of EDA part ===(.*)=== End of EDA part ===", "", workflow_stdout)
else:
rdagent/components/coder/data_science/workflow/__init__.py
+10 -1
View File
@@ -123,4 +123,13 @@ class WorkflowCoSTEER(CoSTEER):
WorkflowGeneralCaseSpecEvaluator(scen=scen), scen=scen
) # Please specify whether you agree running your eva in parallel or not
es = WorkflowMultiProcessEvolvingStrategy(scen=scen, settings=settings)
super().__init__(*args, settings=settings, eva=eva, es=es, evolving_version=2, scen=scen, **kwargs)
super().__init__(
*args,
settings=settings,
eva=eva,
es=es,
evolving_version=2,
scen=scen,
max_loop=DS_RD_SETTING.coder_max_loop,
**kwargs,
)
rdagent/scenarios/data_science/dev/runner/__init__.py
+8 -1
View File
@@ -93,7 +93,14 @@ class DSCoSTEERRunner(CoSTEER):
# In runner, we don't need very big loops, so we set max_loop to 3
super().__init__(
*args, settings=CoSTEER_SETTINGS, eva=eva, es=es, evolving_version=2, scen=scen, max_loop=3, **kwargs
*args,
settings=CoSTEER_SETTINGS,
eva=eva,
es=es,
evolving_version=2,
scen=scen,
max_loop=DS_RD_SETTING.runner_max_loop,
**kwargs,
)
def develop(self, exp):