On 18/4/25 13:35, jeremy ardley wrote:
Another strategy is to use two models and compare the outputs or use
the same model in two sessions.
I tried the two model approach and compared perplexity with Claude
Sonnet and asked Claude Sonnet to check the results.
*Query*
I have checked your output with the same query to Perplexity. Here are
the markdown versions with perplexity the second one. Please check all
figures and text for differences other than formatting
<snip markdown>
*Claude Response*
I'll compare the two markdown versions and identify any differences in
the content beyond just formatting.
Comparing the two markdown tables, I found the following differences:
1. Value discrepancy: In the "Dark-green vegetables" row:
* My version: Cost of $1.06
* Perplexity's version: Cost of $1.86
2. Formatting/labeling differences:
* Perplexity added bold formatting to main category headers
* Perplexity slightly abbreviated some of the longer category
descriptions
* Perplexity rounded 14.377% to 14.38% for Pre-prepared entrees
3. Title repetition:
* Perplexity included the table title twice, once as a main
heading and once as a secondary heading
All other numerical values appear to match between the two versions.
Only the dark-green vegetables cost figure ($1.06 vs $1.86) is a true
data discrepancy. The original image shows $1.06, so my value is correct
for that entry.
