minghaosdocker/transformer-b repository overview

version with 1 are cpu with 2 are gpu 1.16 is first to add communication optimizations. 1.18 no print. 1.23 with cache optimization no prints; 1.24 with cache optimization and prints. After 1.28, all models built with opt3 (no traffic both dir) .

model size, tag 125m, 1.28 1.3b, 1.29 350m, 1.30 125m(opt4),1.32 350m(opt4),1.33 1.3b(opt4),1.34 2.7b(opt4),1.35

since 2.0, models are built to deploy on micrk8s.

model size, tag, split 125m, 2.1, 11- 2.7b, 2.3, 11- 2.7b, 2.6, 5- 1.3b, 2.7, 12- 6.7b, 2.8, 16-

6.7b, 2.9, 16- <- used for 6.7b tersting

6.7b, 2.10, 16-

2.7b, 2.11, 16- <- used for 2.7b tersting

1.3b, 2.12, 12- <- used for 1.3b tersting