The essay compares five different workloads for each model and each method. However, figure11.py only runs two of them per model-method pair.
And I wonder how alpha and budget have been chosen in https://github.com/snu-comparch/InfiniGen/blob/main/accuracy/lm_eval/figure11-config.json
The essay compares five different workloads for each model and each method. However, figure11.py only runs two of them per model-method pair.
And I wonder how alpha and budget have been chosen in https://github.com/snu-comparch/InfiniGen/blob/main/accuracy/lm_eval/figure11-config.json