Learn Before
Backtracking in Step-Level Search
Step-by-step verification enables dynamic adjustments to the reasoning process through backtracking. If a candidate step, denoted as , is deemed incorrect or unpromising by the verifier , the search algorithm can backtrack to explore alternative steps from the immediately preceding partial path , or even from an earlier node (where ). Conversely, computational resources are focused on extending paths that receive high ratings.
0
1
Tags
Foundations of Large Language Models
Ch.5 Inference - Foundations of Large Language Models
Foundations of Large Language Models Course
Computing Sciences
Related
A system is generating a multi-step solution to a complex problem. After the first step, it has produced four potential partial solutions. A separate evaluation component has assigned a confidence score to each partial solution, indicating its likelihood of being on the correct path:
- Partial Solution A: Score 0.9
- Partial Solution B: Score 0.8
- Partial Solution C: Score 0.4
- Partial Solution D: Score 0.1
Which of the following best describes the fundamental trade-off the system's search component must navigate when deciding how to proceed?
Analysis of a Flawed Search Strategy
A reasoning system is at a step where it has generated several possible next steps, each with a score from a verifier. Match each search strategy with the description of how it would proceed.
Backtracking in Step-Level Search