Think Strategies

{ "strategyStages": { "linear": [ "problem_reception", "initial_thought_planning", "thought_generation", "thought_evaluation", "thought_revision", "continuation_decision", "thought_adjustment", "branch_creation", "hypothesis_generation", "hypothesis_verification", "solution_finalization", "final_response" ], "chain_of_thought": [ "problem_reception", "step_decomposition", "sequential_reasoning", "solution_formulation", "answer_verification", "final_response" ], "react": [ "problem_reception", "initial_reasoning", "action_planning", "action_execution", "observation_reception", "reasoning_update", "evaluation_checkpoint", "solution_formulation", "final_response" ], "rewoo": [ "problem_reception", "planning_phase", "tool_call_specification", "working_phase", "evidence_collection", "solving_phase", "final_response" ], "scratchpad": [ "problem_reception", "scratchpad_initialization", "iterative_calculation", "state_tracking", "continuation_decision", "result_extraction", "final_response" ], "self_ask": [ "problem_reception", "problem_decomposition", "sub_question_formulation", "sub_question_answering", "answer_integration", "completion_check", "solution_formulation", "final_response" ], "self_consistency": [ "problem_reception", "multiple_path_sampling", "reasoning_path_execution", "answer_collection", "consistency_analysis", "majority_selection", "final_response" ], "step_back": [ "problem_reception", "abstraction", "principle_identification", "approach_selection", "specific_application", "step_by_step_solution", "solution_verification", "final_response" ], "tree_of_thoughts": [ "problem_reception", "approach_exploration", "branch_creation", "branch_development", "branch_evaluation", "branch_selection", "continuation_decision", "solution_formulation", "path_justification", "final_response" ], "trilemma": [ "problem_reception", "trilemma_identification", "objective_initialization", "trade_off_evaluation", "satisficing_iteration", "equilibrium_check", "propagation_decision", "solution_refinement", "final_balance", "final_response" ], "system_architect": [ "problem_reception", "requirements_analysis", "architecture_design", "design_evaluation", "final_response" ], "cyclic_reasoning": [ "problem_reception", "approach_selection", "element_initialization", "thought_processing", "question_processing", "solution_processing", "cycle_evaluation", "approach_adjustment", "final_response" ] }, "stageDescriptions": { "problem_reception": "Receive and understand the problem statement", "initial_thought_planning": "Estimate the number of thinking steps needed and plan the initial approach", "thought_generation": "Generate a thought step that can include analytical steps, revisions, questions, or realizations", "thought_evaluation": "Evaluate if the current thought is satisfactory or needs revision", "thought_revision": "Revise a previous thought based on new insights or realizations", "continuation_decision": "Decide if more thoughts are needed, even if at what seemed like the end", "thought_adjustment": "Adjust the total number of thoughts needed based on problem complexity", "branch_creation": "Create a new reasoning branch from a specific previous thought", "hypothesis_generation": "Generate a solution hypothesis based on the chain of thought", "hypothesis_verification": "Verify the hypothesis against the chain of thought steps", "solution_finalization": "Finalize the solution based on the verified hypothesis", "final_response": "Provide a single, correct answer as the final output", "step_decomposition": "Break down the problem into logical reasoning steps", "sequential_reasoning": "Process each step in order, building toward the solution", "solution_formulation": "Derive the final answer based on the previous reasoning steps", "answer_verification": "Verify the answer against the original problem and reasoning", "initial_reasoning": "Initial analysis of the problem to determine needed information", "action_planning": "Formulate a specific action to gather required information", "action_execution": "Execute the planned action to interact with external tools/environment", "observation_reception": "Receive and process observations from the executed action", "reasoning_update": "Update reasoning based on new observations", "evaluation_checkpoint": "Determine if more actions are needed or if sufficient information is available", "planning_phase": "Generate a comprehensive plan outlining all steps and required tool calls", "tool_call_specification": "Specify all required tool calls with potential variables and dependencies", "working_phase": "Execute all tool calls in the plan, potentially in parallel", "evidence_collection": "Gather and organize all results from tool calls", "solving_phase": "Use the results of all tool calls to generate the final answer", "scratchpad_initialization": "Set up the scratchpad environment with relevant variables and initial state", "iterative_calculation": "Work through calculations or logical steps, updating the program state at each step", "state_tracking": "Document and track changes to variables or logical state after each step", "result_extraction": "Extract the final result from the program state", "problem_decomposition": "Break down the main question into smaller, answerable sub-questions", "sub_question_formulation": "Formulate a specific sub-question that needs to be answered", "sub_question_answering": "Answer the current sub-question based on available knowledge", "answer_integration": "Integrate the sub-question answer into the overall reasoning process", "completion_check": "Determine if all necessary sub-questions have been answered", "multiple_path_sampling": "Sample multiple diverse reasoning paths with different temperatures", "reasoning_path_execution": "Execute each reasoning path in parallel, collecting all reasoning steps", "answer_collection": "Collect all answers generated from the different reasoning paths", "consistency_analysis": "Analyze the frequency and consistency of answers across reasoning paths", "majority_selection": "Select the most frequently occurring answer as the final result", "abstraction": "Step back to identify the general type or category of the problem", "principle_identification": "Identify the general principles, methods, or approaches that apply to this type of problem", "approach_selection": "Select the most appropriate general approach for the specific problem", "specific_application": "Apply the selected approach to the specific problem instance", "step_by_step_solution": "Work through the solution steps using the selected approach", "solution_verification": "Verify the solution against both specific problem requirements and general principles", "approach_exploration": "Generate multiple diverse approaches to solving the problem", "branch_development": "Develop each branch by exploring its reasoning path step by step", "branch_evaluation": "Evaluate the promise or feasibility of each reasoning branch", "branch_selection": "Select the most promising branch(es) to continue pursuing", "path_justification": "Justify why the selected path was most effective compared to alternatives", "trilemma_identification": "Identify the three competing objectives that cannot all be maximized simultaneously", "objective_initialization": "Initialize scoring metrics and thresholds for each of the three objectives", "trade_off_evaluation": "Evaluate current trade-offs between the three objectives and score each (0-1)", "satisficing_iteration": "Iterate on solutions that satisfice across objectives, adjusting weights based on priorities", "equilibrium_check": "Check if current balance meets satisficing thresholds for all three objectives", "propagation_decision": "Decide whether to propagate current solution forward or explore new trade-offs", "solution_refinement": "Refine the solution based on accumulated trade-off insights across iterations", "final_balance": "Present the final balanced solution with trade-off scores and justification", "approach_selection": "Select the optimal reasoning approach (thought-first, question-first, or solution-first) based on problem domain", "element_initialization": "Initialize the cyclic reasoning with the first element in the selected approach order", "thought_processing": "Generate analytical thoughts, understand principles, and build conceptual understanding", "question_processing": "Formulate key questions, identify uncertainties, and explore problem boundaries", "solution_processing": "Generate potential solutions, test approaches, and refine implementation strategies", "cycle_evaluation": "Evaluate if the current cycle has provided sufficient insight or if more cycles are needed", "approach_adjustment": "Optionally adjust the reasoning approach based on insights gained during processing" }, "stageTransitions": { "base_sequential": { "problem_reception": ["initial_thought_planning"], "initial_thought_planning": ["thought_generation"], "thought_generation": ["thought_evaluation"], "thought_evaluation": ["thought_revision", "continuation_decision"], "thought_revision": ["continuation_decision"], "continuation_decision": ["thought_adjustment", "branch_creation", "hypothesis_generation"], "thought_adjustment": ["thought_generation"], "branch_creation": ["thought_generation"], "hypothesis_generation": ["hypothesis_verification"], "hypothesis_verification": ["solution_finalization", "continuation_decision"], "solution_finalization": ["final_response"], "final_response": [] }, "chain_of_thought": { "problem_reception": ["step_decomposition"], "step_decomposition": ["sequential_reasoning"], "sequential_reasoning": ["solution_formulation"], "solution_formulation": ["answer_verification"], "answer_verification": ["final_response"], "final_response": [] }, "react": { "problem_reception": ["initial_reasoning"], "initial_reasoning": ["action_planning"], "action_planning": ["action_execution"], "action_execution": ["observation_reception"], "observation_reception": ["reasoning_update"], "reasoning_update": ["evaluation_checkpoint"], "evaluation_checkpoint": ["action_planning", "solution_formulation"], "solution_formulation": ["final_response"], "final_response": [] }, "rewoo": { "problem_reception": ["planning_phase"], "planning_phase": ["tool_call_specification"], "tool_call_specification": ["working_phase"], "working_phase": ["evidence_collection"], "evidence_collection": ["solving_phase"], "solving_phase": ["final_response"], "final_response": [] }, "scratchpad": { "problem_reception": ["scratchpad_initialization"], "scratchpad_initialization": ["iterative_calculation"], "iterative_calculation": ["state_tracking"], "state_tracking": ["continuation_decision"], "continuation_decision": ["iterative_calculation", "result_extraction"], "result_extraction": ["final_response"], "final_response": [] }, "self_ask": { "problem_reception": ["problem_decomposition"], "problem_decomposition": ["sub_question_formulation"], "sub_question_formulation": ["sub_question_answering"], "sub_question_answering": ["answer_integration"], "answer_integration": ["completion_check"], "completion_check": ["sub_question_formulation", "solution_formulation"], "solution_formulation": ["final_response"], "final_response": [] }, "self_consistency": { "problem_reception": ["multiple_path_sampling"], "multiple_path_sampling": ["reasoning_path_execution"], "reasoning_path_execution": ["answer_collection"], "answer_collection": ["consistency_analysis"], "consistency_analysis": ["majority_selection"], "majority_selection": ["final_response"], "final_response": [] }, "step_back": { "problem_reception": ["abstraction"], "abstraction": ["principle_identification"], "principle_identification": ["approach_selection"], "approach_selection": ["specific_application"], "specific_application": ["step_by_step_solution"], "step_by_step_solution": ["solution_verification"], "solution_verification": ["final_response"], "final_response": [] }, "tree_of_thoughts": { "problem_reception": ["approach_exploration"], "approach_exploration": ["branch_creation"], "branch_creation": ["branch_development"], "branch_development": ["branch_evaluation"], "branch_evaluation": ["branch_selection"], "branch_selection": ["continuation_decision"], "continuation_decision": ["branch_development", "branch_creation", "solution_formulation"], "solution_formulation": ["path_justification"], "path_justification": ["final_response"], "final_response": [] }, "trilemma": { "problem_reception": ["trilemma_identification"], "trilemma_identification": ["objective_initialization"], "objective_initialization": ["trade_off_evaluation"], "trade_off_evaluation": ["satisficing_iteration"], "satisficing_iteration": ["equilibrium_check"], "equilibrium_check": ["propagation_decision", "final_balance"], "propagation_decision": ["trade_off_evaluation", "solution_refinement"], "solution_refinement": ["equilibrium_check"], "final_balance": ["final_response"], "final_response": [] }, "system_architect": { "problem_reception": ["requirements_analysis"], "requirements_analysis": ["architecture_design"], "architecture_design": ["design_evaluation"], "design_evaluation": ["architecture_design", "final_response"], "final_response": [] }, "cyclic_reasoning": { "problem_reception": ["approach_selection"], "approach_selection": ["element_initialization"], "element_initialization": ["thought_processing", "question_processing", "solution_processing"], "thought_processing": ["question_processing", "solution_processing", "cycle_evaluation"], "question_processing": ["thought_processing", "solution_processing", "cycle_evaluation"], "solution_processing": ["thought_processing", "question_processing", "cycle_evaluation"], "cycle_evaluation": ["thought_processing", "question_processing", "solution_processing", "approach_adjustment", "final_response"], "approach_adjustment": ["element_initialization"], "final_response": [] } }, "strategyParameters": { "base_sequential": { "thought_visibility": "explicit", "requires_external_tools": false, "path_linearity": "adaptive", "backtracking_allowed": true, "verification_step": true, "dynamic_thought_count": true, "revision_allowed": true, "branching_allowed": true, "multiple_paths": false, "path_evaluation": false, "state_tracking": false, "question_decomposition": false, "parallel_processing": false, "consensus_required": false, "abstraction_level": "medium" }, "chain_of_thought": { "thought_visibility": "explicit", "requires_external_tools": false, "path_linearity": "linear", "backtracking_allowed": false, "verification_step": true, "dynamic_thought_count": false, "revision_allowed": false, "branching_allowed": false, "multiple_paths": false, "path_evaluation": false, "state_tracking": false, "question_decomposition": false, "parallel_processing": false, "consensus_required": false, "abstraction_level": "low" }, "react": { "thought_visibility": "explicit", "requires_external_tools": true, "path_linearity": "cyclic", "backtracking_allowed": true, "verification_step": true, "dynamic_thought_count": true, "revision_allowed": true, "branching_allowed": false, "multiple_paths": false, "path_evaluation": false, "state_tracking": true, "question_decomposition": false, "parallel_processing": false, "consensus_required": false, "abstraction_level": "medium" }, "rewoo": { "thought_visibility": "explicit", "requires_external_tools": true, "path_linearity": "modular", "backtracking_allowed": false, "verification_step": false, "dynamic_thought_count": false, "revision_allowed": false, "branching_allowed": false, "multiple_paths": false, "path_evaluation": false, "state_tracking": true, "question_decomposition": false, "parallel_processing": true, "consensus_required": false, "abstraction_level": "medium" }, "scratchpad": { "thought_visibility": "explicit", "requires_external_tools": false, "path_linearity": "iterative", "backtracking_allowed": true, "verification_step": false, "dynamic_thought_count": true, "revision_allowed": true, "branching_allowed": false, "multiple_paths": false, "path_evaluation": false, "state_tracking": true, "question_decomposition": false, "parallel_processing": false, "consensus_required": false, "abstraction_level": "low" }, "self_ask": { "thought_visibility": "explicit", "requires_external_tools": false, "path_linearity": "hierarchical", "backtracking_allowed": true, "verification_step": false, "dynamic_thought_count": true, "revision_allowed": false, "branching_allowed": false, "multiple_paths": false, "path_evaluation": false, "state_tracking": false, "question_decomposition": true, "parallel_processing": false, "consensus_required": false, "abstraction_level": "medium" }, "self_consistency": { "thought_visibility": "explicit", "requires_external_tools": false, "path_linearity": "parallel", "backtracking_allowed": false, "verification_step": true, "dynamic_thought_count": false, "revision_allowed": false, "branching_allowed": false, "multiple_paths": true, "path_evaluation": true, "state_tracking": false, "question_decomposition": false, "parallel_processing": true, "consensus_required": true, "abstraction_level": "low" }, "step_back": { "thought_visibility": "explicit", "requires_external_tools": false, "path_linearity": "hierarchical", "backtracking_allowed": true, "verification_step": true, "dynamic_thought_count": false, "revision_allowed": true, "branching_allowed": false, "multiple_paths": false, "path_evaluation": false, "state_tracking": false, "question_decomposition": false, "parallel_processing": false, "consensus_required": false, "abstraction_level": "high" }, "tree_of_thoughts": { "thought_visibility": "explicit", "requires_external_tools": false, "path_linearity": "branching", "backtracking_allowed": true, "verification_step": true, "dynamic_thought_count": true, "revision_allowed": true, "branching_allowed": true, "multiple_paths": true, "path_evaluation": true, "state_tracking": false, "question_decomposition": false, "parallel_processing": false, "consensus_required": false, "abstraction_level": "medium" }, "trilemma": { "thought_visibility": "explicit", "requires_external_tools": false, "path_linearity": "iterative", "backtracking_allowed": true, "verification_step": true, "dynamic_thought_count": true, "revision_allowed": true, "branching_allowed": false, "multiple_paths": false, "path_evaluation": true, "state_tracking": true, "question_decomposition": false, "parallel_processing": false, "consensus_required": false, "abstraction_level": "medium", "objective_tracking": true, "trade_off_scoring": true, "satisficing_mode": true }, "system_architect": { "thought_visibility": "explicit", "requires_external_tools": false, "path_linearity": "iterative", "backtracking_allowed": true, "verification_step": true, "dynamic_thought_count": true, "revision_allowed": true, "branching_allowed": false, "multiple_paths": false, "path_evaluation": true, "state_tracking": false, "question_decomposition": false, "parallel_processing": false, "consensus_required": false, "abstraction_level": "high" }, "cyclic_reasoning": { "thought_visibility": "explicit", "requires_external_tools": false, "path_linearity": "cyclic", "backtracking_allowed": true, "verification_step": true, "dynamic_thought_count": true, "revision_allowed": true, "branching_allowed": false, "multiple_paths": false, "path_evaluation": true, "state_tracking": true, "question_decomposition": true, "parallel_processing": false, "consensus_required": false, "abstraction_level": "medium", "approach_adaptation": true, "domain_awareness": true, "element_cycling": true } }, "promptTemplates": { "base_sequential": "Let me solve this problem step by step, adjusting my approach as needed.\n\nProblem: {problem}\n\nInitial plan: I'll start with approximately {count} thoughts, but may adjust as needed.\n\n{{thought_sequence}}", "chain_of_thought": "Let me think through this step by step.\n\nProblem: {problem}\n\n{{thought_sequence}}\n\nTherefore, the answer is: {{final_answer}}", "react": "Problem: {problem}\n\nI'll solve this by thinking and taking actions as needed.\n\nThought: {{initial_reasoning}}\nAction: {{action}}\nObservation: {{observation}}\nThought: {{reasoning_update}}\n...\nFinal Answer: {{final_answer}}", "rewoo": "Problem: {problem}\n\nLet me plan how to approach this:\n{{planning_phase}}\n\nNow I'll solve the problem using the evidence gathered:\n{{solving_phase}}\n\nFinal Answer: {{final_answer}}", "scratchpad": "Problem: {problem}\n\n<scratchpad>\n{{iterative_calculation}}\n{{state_tracking}}\n</scratchpad>\n\nTherefore, the answer is: {{final_answer}}", "self_ask": "Problem: {problem}\n\nTo answer this question, I need to break it down into smaller questions:\n1. {{sub_question_1}}?\n Answer: {{sub_answer_1}}\n2. {{sub_question_2}}?\n Answer: {{sub_answer_2}}\n...\nTherefore, the answer to the original question is: {{final_answer}}", "self_consistency": "Problem: {problem}\n\nLet me explore multiple reasoning paths:\n\nPath 1: {{reasoning_path_1}}\nAnswer 1: {{answer_1}}\n\nPath 2: {{reasoning_path_2}}\nAnswer 2: {{answer_2}}\n...\n\nBased on consistency across paths, the most reliable answer is: {{final_answer}}", "step_back": "Problem: {problem}\n\nBefore solving this specific problem, let's step back and consider the general principles or methods that apply to this type of problem.\n\nGeneral approach: {{general_principle}}\n\nNow let's apply this to our specific problem:\n{{step_by_step_solution}}\n\nTherefore, the answer is: {{final_answer}}", "tree_of_thoughts": "Problem: {problem}\n\nLet's solve this step by step, exploring multiple approaches:\n\n1. First approach: {{approach_1}}\n2. Second approach: {{approach_2}}\n3. Third approach: {{approach_3}}\n\nEvaluating these approaches, the most promising solution is:\n{{solution_formulation}}\n\nTherefore, the answer is: {{final_answer}}", "trilemma": "Problem: {problem}\n\nI'll solve this by balancing three competing objectives through iterative satisficing.\n\nTrilemma Identification:\n1. Objective A: {{objective_1}}\n2. Objective B: {{objective_2}}\n3. Objective C: {{objective_3}}\n\nIterative Trade-off Analysis:\n{{trade_off_iterations}}\n\nFinal Balance:\n- Objective A Score: {{score_1}}/1.0 ({{threshold_1}} threshold)\n- Objective B Score: {{score_2}}/1.0 ({{threshold_2}} threshold)\n- Objective C Score: {{score_3}}/1.0 ({{threshold_3}} threshold)\n\nSolution: {{final_balance}}", "cyclic_reasoning": "Problem: {problem}\n\nI'll solve this using cyclic reasoning, cycling through thought, question, and solution elements.\n\nApproach: {{reasoning_approach}} ({{domain_rationale}})\nCycle Order: {{cycle_order}}\n\n{{reasoning_cycles}}\n\nFinal Answer: {{final_answer}}" } }