File size: 4,794 Bytes
4c30587 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 | """
Interactive Designer
"""
from typing import List, Any
from src.designer import Designer, DesignerCase
from prompts.designer_prompts_interactive import (
INTERACTIVE_DESIGNER_ANALYSIS_PROMPT,
INTERACTIVE_DESIGNER_REFLECTION_PROMPT,
INTERACTIVE_DESIGNER_REFINEMENT_PROMPT,
)
class InteractiveDesigner(Designer):
"""Designer specialized for interactive objective-driven failures."""
def _format_failure_cases_details(self, cases: List[DesignerCase]) -> str:
case_details = []
for i, case in enumerate(cases):
case_str = f"### Case {i + 1}\n"
case_str += f"**Objective:** {case.question}\n"
case_str += f"**Outcome:** {case.prediction}\n"
if case.evidence:
case_str += "**Trajectory:**\n"
case_str += f"{case.evidence}\n"
if case.retrieved_memories:
case_str += f"**Retrieved Memories ({len(case.retrieved_memories)}):**\n"
for j, mem in enumerate(case.retrieved_memories[:20]):
case_str += f" {j + 1}. {mem}\n"
else:
case_str += "**Retrieved Memories:** None\n"
case_details.append(case_str)
return "\n".join(case_details)
def build_analysis_prompt(self, cases: List[DesignerCase], operation_bank,
evolution_feedback: str = "") -> str:
cases = self._normalize_cases_for_prompt(cases)
operation_bank_description = self._format_operation_bank_description(operation_bank)
failure_cases_details = self._format_failure_cases_details(cases)
new_skill_hint = ""
if getattr(self.args, "designer_new_skill_hint", False):
new_skill_hint = ("Note: If failures indicate a capability gap, it is encouraged to recommend "
"adding a new skill.")
return INTERACTIVE_DESIGNER_ANALYSIS_PROMPT.format(
operation_bank_description=operation_bank_description,
evolution_feedback=evolution_feedback,
num_failure_cases=len(cases),
failure_cases_details=failure_cases_details,
new_skill_hint=new_skill_hint,
max_changes=self._get_max_changes()
)
def build_reflection_prompt(self, analysis_feedback: str, cases: List[Any], operation_bank,
evolution_feedback: str = "", reflection_round: int = 2,
reflection_round_total: int = 2) -> str:
cases = self._normalize_cases_for_prompt(cases)
operation_bank_description = self._format_operation_bank_description(operation_bank)
failure_cases_details = self._format_failure_cases_details(cases)
new_skill_hint = ""
if getattr(self.args, "designer_new_skill_hint", False):
new_skill_hint = ("Note: If failures indicate a capability gap, it is encouraged to recommend "
"adding a new skill.")
return INTERACTIVE_DESIGNER_REFLECTION_PROMPT.format(
analysis_feedback=analysis_feedback,
operation_bank_description=operation_bank_description,
evolution_feedback=evolution_feedback,
num_failure_cases=len(cases),
failure_cases_details=failure_cases_details,
reflection_round=reflection_round,
reflection_round_total=reflection_round_total,
new_skill_hint=new_skill_hint,
max_changes=self._get_max_changes()
)
def build_refinement_prompt(self, analysis_feedback: str, operation_bank,
evolution_feedback: str = "") -> str:
ops = operation_bank.get_all_operations()
op_full_details = []
for op in ops:
op_detail = f"### {op.name}\n"
op_detail += f"- **Type:** {op.update_type}\n"
op_detail += f"- **Description:** {op.description}\n"
op_detail += f"- **Instruction Template:**\n```\n{op.instruction_template}\n```\n"
op_full_details.append(op_detail)
operation_bank_full = "\n".join(op_full_details) if op_full_details else "(No evolvable operations available)"
new_skill_hint = ""
if getattr(self.args, "designer_new_skill_hint", False):
new_skill_hint = ("Note: If failures indicate a capability gap, it is encouraged to recommend "
"adding a new skill.")
return INTERACTIVE_DESIGNER_REFINEMENT_PROMPT.format(
analysis_feedback=analysis_feedback,
operation_bank_full=operation_bank_full,
evolution_feedback=evolution_feedback,
new_skill_hint=new_skill_hint,
max_changes=self._get_max_changes()
)
|