I've put a complex codebase into a single 120K-token prompt, and asked 7 questions GPT-4 and Gemini 1.5. Here are the results!
I'm the author of HVM1, which is currently being updated to HVM2. These are 2 complex codebases that implement a parallel inet runtime; basically, hard compiler stuff. User @SullyOmarr on X, who gained Gemini 1.5 access, kindly offered me a prompt. So, I've concatenated both HVM codebases into a single 120K-token
file,
and asked 7 questions to both Gemini and GPT-4. Here are the complete
results.
Breakdown:
1. Which was based in a term-like calculus, and which was based on raw interaction combin...
Read more at reddit.com