Spaces:
Running
Coding test runs into a loop
Tested this model with the following prompt:
"Assume the role of an experienced software developer as you create a detailed plan for a simple Python 3 command line tool that efficiently identifies and lists all duplicate files in a folder, regardless of their path or filename, while also considering any sub-directories of varying depths. Ensure that two files are defined as duplicates only if they share identical file size and content. The tool's output should present the duplicate files alongside their respective duplicates count, with a summary at the end providing the total number of duplicate files and the amount of disk space that could be conserved through deduplication (retaining only a single version of each duplicate file). Emphasize that the tool should not delete any files. Begin by using a tree of thoughts decision-making process i.e. brainstorm, evaluate, and consider multiple approaches to achieve the most promising results, detailing the step by step implementation for the chosen approach in Python 3."
Once it gave a reasonable plan I asked it to implement the plan in Python 3.
It started with promising Python code looking legit, but then went into an infinite loop generating the same code fragment again and again.
Is something wrong with the default repetition penalty or so?