mirror of
https://github.com/browsermt/bergamot-translator.git
synced 2024-09-11 05:35:33 +03:00
Updated config for min inference time
- This combination gives min inference time (~ 200 WPS) on local machine
This commit is contained in:
parent
c5c5339489
commit
921c2eedf8
@ -75,17 +75,25 @@ vocabs:
|
||||
beam-size: 1
|
||||
normalize: 1.0
|
||||
word-penalty: 0
|
||||
mini-batch: 32
|
||||
maxi-batch: 100
|
||||
maxi-batch-sort: src
|
||||
max-input-sentence-tokens: 128
|
||||
max-input-tokens: 1024
|
||||
workspace: 128
|
||||
max-length-factor: 2.0
|
||||
skip-cost: true
|
||||
cpu-threads: 1
|
||||
quiet: true
|
||||
quiet-translation: true
|
||||
shortlist:
|
||||
- /lex.${lang}.s2t
|
||||
- 50
|
||||
- 50
|
||||
`;
|
||||
/*
|
||||
This config is not valid anymore in new APIs
|
||||
mini-batch: 32
|
||||
maxi-batch: 100
|
||||
maxi-batch-sort: src
|
||||
*/
|
||||
// TODO: Use in model config when wormhole is enabled:
|
||||
// gemm-precision: int8shift
|
||||
// TODO: Use in model config when loading of binary models is supported and we use model.intgemm.alphas.bin:
|
||||
|
Loading…
Reference in New Issue
Block a user