From 7134a51afa71ab146b52987c194adace62f8b034 Mon Sep 17 00:00:00 2001
From: Yuante Li <104308117+WinstonLiyt@users.noreply.github.com>
Date: Fri, 1 Aug 2025 13:20:09 +0800
Subject: [PATCH] fix: fix some bugs in RD-Agent(Q) (#1143)

* fix some bugs in RD-Agent(Q)

* fix factor from report

* fix ci
---
 .../app/qlib_rd_loop/factor_from_report.py    | 21 +++++++++++++++++--
 .../scenarios/qlib/developer/factor_runner.py |  2 +-
 .../scenarios/qlib/developer/model_runner.py  |  2 +-
 .../scenarios/qlib/proposal/quant_proposal.py |  2 +-
 4 files changed, 22 insertions(+), 5 deletions(-)

diff --git a/rdagent/app/qlib_rd_loop/factor_from_report.py b/rdagent/app/qlib_rd_loop/factor_from_report.py
index 984c7519..defb8d1f 100644
--- a/rdagent/app/qlib_rd_loop/factor_from_report.py
+++ b/rdagent/app/qlib_rd_loop/factor_from_report.py
@@ -12,7 +12,7 @@ from rdagent.components.document_reader.document_reader import (
     load_and_process_pdfs_by_langchain,
 )
 from rdagent.core.conf import RD_AGENT_SETTINGS
-from rdagent.core.proposal import Hypothesis
+from rdagent.core.proposal import Hypothesis, HypothesisFeedback
 from rdagent.log import rdagent_logger as logger
 from rdagent.oai.llm_utils import APIBackend
 from rdagent.scenarios.qlib.experiment.factor_experiment import QlibFactorExperiment
@@ -136,8 +136,25 @@ class FactorReportLoop(FactorRDLoop, metaclass=LoopMeta):
         logger.log_object(exp.sub_workspace_list, tag="coder result")
         return exp
 
+    def feedback(self, prev_out: dict[str, Any]):
+        e = prev_out.get(self.EXCEPTION_KEY, None)
+        if e is not None:
+            feedback = HypothesisFeedback(
+                observations=str(e),
+                hypothesis_evaluation="",
+                new_hypothesis="",
+                reason="",
+                decision=False,
+            )
+            logger.log_object(feedback, tag="feedback")
+            self.trace.hist.append((prev_out["direct_exp_gen"]["exp_gen"], feedback))
+        else:
+            feedback = self.summarizer.generate_feedback(prev_out["running"], self.trace)
+            logger.log_object(feedback, tag="feedback")
+            self.trace.hist.append((prev_out["running"], feedback))
 
-def main(report_folder=None, path=None, all_duration: str | None = None, checkout: bool = True):
+
+def main(report_folder=None, path=None, all_duration=None, checkout=True):
     """
     Auto R&D Evolving loop for fintech factors (the factors are extracted from finance reports).
 
diff --git a/rdagent/scenarios/qlib/developer/factor_runner.py b/rdagent/scenarios/qlib/developer/factor_runner.py
index ea69a9e5..667c334a 100644
--- a/rdagent/scenarios/qlib/developer/factor_runner.py
+++ b/rdagent/scenarios/qlib/developer/factor_runner.py
@@ -145,7 +145,7 @@ class QlibFactorRunner(CachedRunner[QlibFactorExperiment]):
                             "lr": str(sota_training_hyperparameters.get("lr", "2e-4")),
                             "early_stop": str(sota_training_hyperparameters.get("early_stop", 10)),
                             "batch_size": str(sota_training_hyperparameters.get("batch_size", 256)),
-                            "weight_decay": str(sota_training_hyperparameters.get("weight_decay", 0.0)),
+                            "weight_decay": str(sota_training_hyperparameters.get("weight_decay", 0.0001)),
                         }
                     )
                 sota_model_type = sota_model_exp.sub_tasks[0].model_type
diff --git a/rdagent/scenarios/qlib/developer/model_runner.py b/rdagent/scenarios/qlib/developer/model_runner.py
index 3d3dea71..47c934e6 100644
--- a/rdagent/scenarios/qlib/developer/model_runner.py
+++ b/rdagent/scenarios/qlib/developer/model_runner.py
@@ -65,7 +65,7 @@ class QlibModelRunner(CachedRunner[QlibModelExperiment]):
             env_to_use.update(
                 {
                     "n_epochs": str(training_hyperparameters.get("n_epochs", "100")),
-                    "lr": str(training_hyperparameters.get("lr", "1e-3")),
+                    "lr": str(training_hyperparameters.get("lr", "2e-4")),
                     "early_stop": str(training_hyperparameters.get("early_stop", 10)),
                     "batch_size": str(training_hyperparameters.get("batch_size", 256)),
                     "weight_decay": str(training_hyperparameters.get("weight_decay", 0.0001)),
diff --git a/rdagent/scenarios/qlib/proposal/quant_proposal.py b/rdagent/scenarios/qlib/proposal/quant_proposal.py
index a31bc051..b61ce4c0 100644
--- a/rdagent/scenarios/qlib/proposal/quant_proposal.py
+++ b/rdagent/scenarios/qlib/proposal/quant_proposal.py
@@ -61,7 +61,7 @@ class QlibQuantHypothesisGen(FactorAndModelHypothesisGen):
         # ========= LLM ==========
         elif QUANT_PROP_SETTING.action_selection == "llm":
             hypothesis_and_feedback = (
-                T("scenarios.qlib.prompts:hypothesis_and_feedback").render(trace=trace)
+                T("scenarios.qlib.prompts:hypothesis_and_feedback").r(trace=trace)
                 if len(trace.hist) > 0
                 else "No previous hypothesis and feedback available since it's the first round."
             )