Could you provide access to your MiBench-derived IR dataset?
#6
by
CarrotLoveProgramming
- opened
Hello, I’d like to evaluate the exact match rate of your llm-compiler-7b model on the compiler emulation task using your dataset samples, and compare its performance with my own model. Would it be possible to gain access to your dataset? Otherwise, I would have to test your model only on my dataset, which I believe may not provide a fair comparison for your model.