 Hamid Shojanazeri
|
df03fd4b12
Recipe to add a new language to Llama2 (#429)
|
1 rok pred |
 Rahul A R
|
2fa8e69b62
add new argument: tokenizer_name
|
1 rok pred |
 Kai Wu
|
fa0a389f74
add max_step feature for training and eval
|
1 rok pred |
|
Hamid Shojanazeri
|
ffdc93f00a
Merge branch 'main' into wandb_logging
|
1 rok pred |
|
Hamid Shojanazeri
|
162be4c045
Revert "Flop counter, profiling and GC (#357)"
|
1 rok pred |
|
Hamid Shojanazeri
|
71d137c722
Merge branch 'main' into flop_counter_gc
|
1 rok pred |
 Beto
|
7474514fe0
Merging with main
|
1 rok pred |
 kldarek
|
989b6ee812
wandb logging feedback
|
1 rok pred |
 gaopengzhi
|
e2797abe9b
Add gradient_clipping and gradient_clipping_threshold parameters
|
1 rok pred |
|
kldarek
|
cf373529f7
basic wandb logging instrumentation
|
1 rok pred |
|
Beto
|
17d02c3b44
Adding config to conditionally save stats
|
2 rokov pred |
|
gaopengzhi
|
04befdef69
Add gradient clipping feature
|
2 rokov pred |
 Matthias Reso
|
a647955fc8
Make packing/padding a training setting
|
2 rokov pred |
|
Hamid Shojanazeri
|
35b394e49f
adding profiler and flop_counter
|
2 rokov pred |
|
Hamid Shojanazeri
|
d56d5c469d
adding flop counter
|
2 rokov pred |
|
Matthias Reso
|
72a9832571
Merge branch 'main' into feature/package_distribution
|
2 rokov pred |
|
Matthias Reso
|
cf678b9bf0
Adjust imports to package structure + cleaned up imports
|
2 rokov pred |
|
Matthias Reso
|
4c9cc7d223
Move modules into separate src folder
|
2 rokov pred |
