clarify discovery

README.md

@@ -105,7 +105,7 @@ The following are tested in my workflow, but may not generalize well to other wo
 
 - In general, if there's an error in the code, copy pasting the last few rows of stacktrace to the LLM seems to work.
 - Adding "Now, reflect." after a failed attempt at code generation sometimes allows Claude-3.5-Sonnet to generate the correct version.
-- If GPT-4o reasons correctly in its first response and the conversation is then sent to GPT-4-mini, the mini model can maintain comparable level of reasoning/accuracy as GPT-4o.
+- If GPT-4o reasons correctly in its first response and the conversation is then continued with GPT-4-mini, the mini model can maintain comparable level of reasoning/accuracy as GPT-4o.
 
 <br>