You can not select more than 25 topics Topics must start with a chinese character,a letter or number, can include dashes ('-') and can be up to 35 characters long.
 
 
 
taoht e465a81dd4 Update 2 years ago
..
__pycache__ Hybrid parallel implementation 2 years ago
__init__.py Hybrid parallel implementation 2 years ago
beam_search.py Hybrid parallel implementation 2 years ago
config.py Update 2 years ago
dataset.py Update 2 years ago
eval_config.py Hybrid parallel implementation 2 years ago
lr_schedule.py Update 2 years ago
process_output.py Hybrid parallel implementation 2 years ago
tokenization.py Hybrid parallel implementation 2 years ago
transformer_for_train.py Update 2 years ago
transformer_model.py Update 2 years ago
weight_init.py Hybrid parallel implementation 2 years ago

让模型的训练更有效率(10B以内),支持训练更大规模的模型(>10B、50B、100B),构建支持分布式混合并行的典型模型案例,是该项目的初衷。

Python Shell Perl

Apache-2.0

Contributors (2)