List your company

Transformer Inference Latency Estimation on LLaMA 3 (8B)

283.157Inference Latency (s)Bifrost
182.2878863.15491,544.0222,224.8891Jun 16, 2026

Evaluation results

MethodLinks
BifrostSystem Type=Bifrost, Evaluation Protocol=Euston-style estimator projection2026.06283.1579.906
EustonSystem Type=Pure FHE, Evaluation Protocol=Euston-style estimator projection2026.062,804.887—