Is it possible to release the speculators trained for different block sizes (Figure 4) for more rigorous benchmarking by researchers?
Is it possible to release the speculators trained for different block sizes (Figure 4) for more rigorous benchmarking by researchers?