Megatron-DeepSpeed
Adding language specific validation sets for Multilingual model training
#97
Merged

Adding language specific validation sets for Multilingual model training #97

hadyelsahar
hadyelsahar adding multiple datasets to args
073c98ef
hadyelsahar hadyelsahar assigned hadyelsahar hadyelsahar 4 years ago
hadyelsahar hadyelsahar added enhancement
hadyelsahar hadyelsahar added multilinguality
hadyelsahar hadyelsahar requested a review from TevenLeScao TevenLeScao 4 years ago
hadyelsahar hadyelsahar marked this pull request as draft 4 years ago
sbmaruf
hadyelsahar Extra_valid_dataset arg & display on Tensorboard
287f1247
hadyelsahar hadyelsahar marked this pull request as ready for review 4 years ago
hadyelsahar
hadyelsahar code cleaning
5a64b937
hadyelsahar code cleaning
20c48380
hadyelsahar add run script example for multilingual validation
64476947
hadyelsahar
stas00
ibeltagy
TevenLeScao
hadyelsahar
stas00
hadyelsahar
stas00
hadyelsahar
hadyelsahar rename extra-validation to periodic-eval
f7714076
hadyelsahar hadyelsahar changed the title WIP: Adding language specific validation steps WIP: Adding language specific validation steps (periodic evaluation) 4 years ago
hadyelsahar
TevenLeScao
sbmaruf
TevenLeScao
TevenLeScao
TevenLeScao
TevenLeScao
TevenLeScao requested changes on 2021-10-26
hadyelsahar hadyelsahar changed the title WIP: Adding language specific validation steps (periodic evaluation) Adding language specific validation sets (periodic evaluation) 4 years ago
stas00
TevenLeScao
stas00
TevenLeScao
stas00
stas00
TevenLeScao
TevenLeScao
stas00
TevenLeScao
hadyelsahar
TevenLeScao bugfix + logging amount of training data per language
1fdf4c36
TevenLeScao Merge branch 'main' into main
51ac6022
TevenLeScao
hadyelsahar change "periodic-eval" to "extra-eval"
da7ab8ba
hadyelsahar elaborate tree sturcture in Multiling run script
0f7a89bc
hadyelsahar initialize all_x_datasets to None if empty
bca5606e
hadyelsahar clean kwargs.get
7fe7206d
sbmaruf
TevenLeScao
sbmaruf
stas00
TevenLeScao
sbmaruf
stas00
hadyelsahar
stas00
hadyelsahar add mode two of data loading
38e1fc1b
hadyelsahar
sbmaruf
hadyelsahar
TevenLeScao
stas00
hadyelsahar adding option2 for data loading
5d37e043
hadyelsahar fix missing argument and range of split
64e97765
hadyelsahar add multilingual run script
83260327
hadyelsahar
hadyelsahar hadyelsahar changed the title Adding language specific validation sets (periodic evaluation) Adding language specific validation sets for Multilingual model training 4 years ago
TevenLeScao small cleanups + fixed bugs with test iterator
4c6fa821
TevenLeScao
TevenLeScao
TevenLeScao Fixed prefixlm
ec007069
TevenLeScao
TevenLeScao adding variable cutoff for last epoch rather than 80%
389936ef
TevenLeScao
stas00
TevenLeScao
TevenLeScao TevenLeScao merged 846c0879 into main 4 years ago

Login to write a write a comment.

Login via GitHub

Reviewers
Assignees
Labels
Milestone