averages the eval score over multiple seeds

harvard-edge / dataperf-speech-example

Example workflow for our data-centric speech benchmark

17 stars 11 forks source link

Closed colbybanbury closed 1 year ago

colbybanbury commented 1 year ago

Tested it and got the following:

en size 25 Score: 0.37435393268857786

en size 60 Score: 0.4132873524553394

pt size 25 Score: 0.4226180970526988

pt size 60 Score: 0.5262337789479115

id size 25 Score: 0.3620309417727216

id size 60 Score: 0.43527109648343154