MCPcopy Index your code
hub / github.com/openai/guided-diffusion / _update_ema

Method _update_ema

guided_diffusion/train_util.py:216–218  ·  view source on GitHub ↗
(self)

Source from the content-addressed store, hash-verified

214 self.mp_trainer.backward(loss)
215
216 def _update_ema(self):
217 for rate, params in zip(self.ema_rate, self.ema_params):
218 update_ema(params, self.mp_trainer.master_params, rate=rate)
219
220 def _anneal_lr(self):
221 if not self.lr_anneal_steps:

Callers 1

run_stepMethod · 0.95

Calls 1

update_emaFunction · 0.85

Tested by

no test coverage detected