1
2
3
4
5
Drag to adjust the number of frozen columns
Beam.cloud Performance Metrics
GPT- Neo (1.3B)
GPT- Neo (125M)
Roberta Large
With Cold Start
99 secs
75.37 secs
93.71 secs
Inference time
2.1 secs
2.5 secs
1.6 secs
Latency
101.1 secs
77.87 secs
95.31 secs
Variability
Cold start - Variable
Cold start - Variable
Cold start - Variable
Autoscaling (c=5)
Does not hold latency
Does not hold latency
Does not hold latency
5 records

Alert

Lorem ipsum
Okay
View larger version