 Kai Wu
|
bb990be967
not working, need create dataloader function
|
преди 7 месеца |
|
Kai Wu
|
ee204ccb98
working now
|
преди 8 месеца |
|
Kai Wu
|
b566582a86
finetune not working with fsdp
|
преди 8 месеца |
 Matthias Reso
|
7a8c52cb38
Remove pkg_resources.packaging
|
преди 11 месеца |
 simwiki
|
66e1867120
Fix save metric FileNotFoundError when finetuning
|
преди 1 година |
|
Kai Wu
|
26e877fd42
changed readme, unified the context interface and added get_flops_per_sec()
|
преди 1 година |
|
Kai Wu
|
d9558c11ca
changed context name and add more docs
|
преди 1 година |
|
Kai Wu
|
03f1ca7817
fixed some typo to pass spellcheck
|
преди 1 година |
|
Kai Wu
|
7b1a9413d2
fixed a typo
|
преди 1 година |
|
Kai Wu
|
41434dc825
formatted and removed duplicated or unused function get_total_flops() and byte2mb()
|
преди 1 година |
|
Kai Wu
|
f2e80bae22
created a FlopMeasure class on top of FlopCounterMode instead of keep of copy of our own tflop_counter.py
|
преди 1 година |
|
Kai Wu
|
69e46887b4
handling incorrect profiling early stop caused by max_train_steps and add profiler.step() for each train step
|
преди 1 година |
|
Kai Wu
|
34e0bf4c6e
second draft of this feature, seems to be working now
|
преди 1 година |
|
Kai Wu
|
a35519ee90
fixed typo and handling unexpected exit
|
преди 1 година |
|
Kai Wu
|
2a5de9b448
first draft of flop counter feature
|
преди 1 година |
|
Kai Wu
|
e6f69f84ad
add max_steps_reached to reduce redundancy
|
преди 1 година |
|
Kai Wu
|
fa0a389f74
add max_step feature for training and eval
|
преди 1 година |
 jpgard
|
6954b16b3b
only save training params on rank 0
|
преди 1 година |
 Hamid Shojanazeri
|
761b7e6e51
adding wandb_run ro eval
|
преди 1 година |
|
Hamid Shojanazeri
|
ffdc93f00a
Merge branch 'main' into wandb_logging
|
преди 1 година |
|
Matthias Reso
|
c5a382e509
Make tests run on cpu only machines
|
преди 1 година |
|
Hamid Shojanazeri
|
162be4c045
Revert "Flop counter, profiling and GC (#357)"
|
преди 1 година |
|
Hamid Shojanazeri
|
1a09fb5d27
add logging for setting profiler
|
преди 1 година |
|
Hamid Shojanazeri
|
71d137c722
Merge branch 'main' into flop_counter_gc
|
преди 1 година |
|
Hamid Shojanazeri
|
8bf474b455
clean up
|
преди 1 година |
|
Hamid Shojanazeri
|
19089269d3
add gc
|
преди 1 година |
|
Hamid Shojanazeri
|
dbfea484c6
Feature : Enable Intel GPU/XPU finetuning and inference (#116)
|
преди 1 година |
 Beto
|
1f5b202c18
Adding tests for the save_metrics param in the train function
|
преди 1 година |
|
Beto
|
7474514fe0
Merging with main
|
преди 1 година |
 gaopengzhi
|
c7d410725b
Merge branch 'main' into grad_clip
|
преди 1 година |
