Iterative Prompt Trainer for Accurate LLM Search Results
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional large language models (LLMs) are unable to automatically refine search queries to return accurate and relevant results, failing to adjust search scope when prompts are too broad or narrow.
Innovation Solution
An iterative prompt trainer that refines search prompts and adjusts scope through iterative adjustments to improve the search results of LLMs, ensuring the results meet predetermined thresholds before transmission.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional LLMs use fixed search prompts, then the system complexity is low, but the search result accuracy deteriorates
Solution Approach 1:
The patent implements dynamic prompt adjustment where the search prompt is not fixed but automatically modified based on analysis of search results. The system iteratively refines prompts by adjusting scope and parameters according to whether errors are detected in results, transforming a static system into a dynamic adaptive one that improves accuracy without requiring complex manual intervention
Solution Approach 2:
The system performs self-service through automatic prompt refinement. The LLM analyzes its own search results and automatically generates improved prompts without external human intervention. This self-correcting mechanism allows the system to improve its own accuracy while maintaining relatively simple overall architecture
2Reliability
If conventional LLMs perform single-pass search, then the processing time is short, but the search result relevance deteriorates
Solution Approach 1:
The patent implements periodic action through iterative search cycles. Instead of a single-pass search, the system performs multiple sequential search iterations, each with refined prompts. The process periodically evaluates results and decides whether to continue refining or terminate, achieving high relevance while controlling time loss through automatic termination conditions
Solution Approach 2:
The system uses feedback mechanisms where search results are analyzed to determine whether errors exist, and this feedback drives automatic prompt refinement for the next iteration. The feedback loop continues until termination conditions are met, ensuring high result relevance while managing processing time through automated decision-making
3Measurement precision
If conventional LLMs do not adjust search scope, then the operation is simple, but the search result quality deteriorates
Solution Approach 1:
The patent makes the search scope dynamic rather than fixed. The system automatically adjusts scope parameters based on analysis of search results, expanding or contracting the search as needed. This dynamic adjustment improves result quality while the automation maintains operational simplicity by eliminating manual scope adjustment
4Measurement precision
If iterative prompt refinement is implemented, then the search result accuracy is improved, but the computational resources consumed increase
Solution Approach 1:
The system uses self-service through automatic termination conditions that monitor computational resource usage. The iterative refinement process automatically stops when results meet quality thresholds or when resource constraints are approached, preventing excessive resource consumption while maintaining high accuracy through intelligent self-regulation
Data Source
AI summary
An iterative prompt training apparatus receives a search request for searching for information and automatically generates a first prompt instructing a large language model to search for the information requested in the search request. The iterative prompt training apparatus provides the first prompt to the large language model and analyzes a first search result, output by the large language model based on the first prompt, to determine whether an error exists in the first search result. In response to determining that the error exists in the first search result, the iterative prompt training apparatus automatically executes prompt adjustment to generate a second prompt that is different from the first prompt and provides the second prompt to the large language model.