MCPcopy Create free account
hub / github.com/deepspeedai/DeepSpeed / steps_per_print

Method steps_per_print

deepspeed/runtime/engine.py:1373–1374  ·  view source on GitHub ↗
(self)

Source from the content-addressed store, hash-verified

1371 return self._config.gradient_predivide_factor
1372
1373 def steps_per_print(self):
1374 return self._config.steps_per_print
1375
1376 def zero_allgather_partitions(self):
1377 return self._config.zero_config.allgather_partitions

Callers 6

__init__Method · 0.95
_take_model_stepMethod · 0.95
__init__Method · 0.80
train_batchMethod · 0.80
tput_logMethod · 0.80
_exec_optimizer_stepMethod · 0.80

Calls

no outgoing calls

Tested by

no test coverage detected