LLM Prompting : Are you asking the Right Questions ?
Do you do something for the sake of doing ? Or do you happen to go through the understanding ? If so, here’s your chance to do what can help you through the process in a smarter way. One might use Chatgpt / other tools and then again the important part is :
“Asking the right questions / prompting ?”
Let’s get down to the business then. Look into our prompt :
Prompt : “Top brainstorming questions to ask when learning programming concepts”
Now let’s see how Bard , GPT-4 and lmsys.org where I chose the mistral next model
Gemini giving some quite good output where it organizes the information in sections with an added “Bonus” category.
If we consider the output of GPT-4 we can see that it gives you a section wise understandable parts. It also says about engaging with community and seeking resources, help as an extra option. Now if we go through again the outputs of two different Opensource models : mistral-next and Anthropic claude-2.1 , one can see there are no probing sections in the output.
From the above discussion we can come to a conclusion that both GPT-4 and Gemini has quite an organized understanding and outputs the pattern of questions in sectionwise. If we consider the opensource model like mistral-next and claude-2.1 then we can see it gives a direct output. This provokes another understanding of large language models as mistral-next and claude-2.1 are finetuned for systemic output.
Have you used the above large language models ? If so, how and how did that help you brainstorm things , get better at your coding skills ?
If the above post helped you somehow feel free to connect to me in linkedin or support me in patreon.
#LanguageModels, #ProgrammingConcepts, #CodingSkills, #BrainstormingQuestions, and #AIForCoders #42 #42born2code
