github/tinygrad
1
1
Fork 0
mirror of https://github.com/tinygrad/tinygrad.git synced 2026-01-10 07:28:15 -05:00
Code Issues Packages Projects Releases Wiki Activity

add RESET_STEP in bert to control reset (#6818)

Browse Source 
same as resnet
This commit is contained in:
chenyu
2024-09-30 09:39:04 -04:00
committed by GitHub
parent 0c24fec9f4
commit f59517754e
1 changed files with 2 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
examples/mlperf/model_train.py
View File

@@ -802,7 +802,7 @@ def train_bert():
if i % eval_step_freq == 0 or (BENCHMARK and i == BENCHMARK):
if MLLOGGER and RUNMLPERF:
MLLOGGER.start(key=mllog_constants.EVAL_START, value=None, metadata={"epoch_num": 1, "epoch_count": 1, "step_num": i})
train_step_bert.reset()
if getenv("RESET_STEP", 1): train_step_bert.reset()
eval_lm_losses = []
eval_clsf_losses = []
eval_lm_accs = []
@@ -840,7 +840,7 @@ def train_bert():
MLLOGGER.event(key=mllog_constants.INIT_STOP, value=None)
return
eval_step_bert.reset()
if getenv("RESET_STEP", 1): eval_step_bert.reset()
del eval_data, eval_result
avg_lm_loss = sum(eval_lm_losses) / len(eval_lm_losses)
avg_clsf_loss = sum(eval_clsf_losses) / len(eval_clsf_losses)