mirrors/icefall
1
0
Fork 0
mirror of https://github.com/k2-fsa/icefall.git synced 2025-09-19 05:54:20 +00:00
Code Issues Packages Projects Releases Wiki Activity

Some code reworking and fixes, rationalizing how speedup is done and fix an issue affecting learning rate.

Browse Source
This commit is contained in:
Daniel Povey 2022-06-23 15:40:51 +08:00
parent c34344e98f
commit 683b8e1504
1 changed files with 21 additions and 14 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

35
egs/librispeech/ASR/pruned_transducer_stateless7/optim.py
View File

@ -244,6 +244,10 @@ class NeutralGradient(Optimizer):
state["exp_avg_sq"] = torch.zeros_like(p) state["exp_avg_sq"] = torch.zeros_like(p)
kwargs = {'device':p.device, 'dtype':p.dtype} kwargs = {'device':p.device, 'dtype':p.dtype}
# grad_times_step is an estimate of the importance of each parameter, for purposes
# of determining step_period period (by which we takes steps only periodically, to save
# time)
state["grad_times_step"] = torch.zeros((), **kwargs)
if p.numel() > 1: if p.numel() > 1:
is_one_axis = (p.numel() in list(p.shape)) # e.g. a bias is_one_axis = (p.numel() in list(p.shape)) # e.g. a bias
@ -376,6 +380,10 @@ class NeutralGradient(Optimizer):
this_delta = grad / grad_rms this_delta = grad / grad_rms
alpha = (-(1-beta1) * lr * (bias_correction2 ** 0.5)) * scale alpha = (-(1-beta1) * lr * (bias_correction2 ** 0.5)) * scale
delta.add_(this_delta, alpha=alpha) delta.add_(this_delta, alpha=alpha)
if step % 10 == 0: # this periodicity is just to save time
# divide by 1-beta1 because we want the actual step...
state["grad_times_step"].mul_(beta1).add_( (this_delta * grad).sum(), alpha=-alpha/n_cached_grads)
else: else:
# The full update. # The full update.
step_within_period = state["step_within_period"] step_within_period = state["step_within_period"]
@ -409,7 +417,7 @@ class NeutralGradient(Optimizer):
cur_grad = self._change_coordinates(cur_grad, state, forward=False) cur_grad = self._change_coordinates(cur_grad, state, forward=False)
if random.random() < 0.00005: if random.random() < 0.00005:
# This is only for debug. The logic below would not be valid for n_cache_grads>0, # This is only for debug. The logic below would not be valid for n_cached_grads>0,
# anyway we will delete this code at some point. # anyway we will delete this code at some point.
# in principle, the cur_grad is supposed to have the same rms as params, on average. # in principle, the cur_grad is supposed to have the same rms as params, on average.
cur_grad_rms = (cur_grad**2).mean().sqrt() cur_grad_rms = (cur_grad**2).mean().sqrt()
@ -432,11 +440,16 @@ class NeutralGradient(Optimizer):
if param_pow != 1.0 or grad_pow != 1.0: if param_pow != 1.0 or grad_pow != 1.0:
# Renormalize scale of cur_grad # Renormalize scale of cur_grad
scalar_exp_avg_sq = state["scalar_exp_avg_sq"] scalar_exp_avg_sq = state["scalar_exp_avg_sq"]
scalar_exp_avg_sq.mul_(beta2).add_((cur_grad**2).mean(), alpha=(1-beta2)) scalar_exp_avg_sq.mul_(beta2).add_((cur_grad**2).mean()/n_cached_grads, alpha=(1-beta2))
alpha = alpha * scale / ((scalar_exp_avg_sq / bias_correction2).sqrt() + grad_eps) alpha = alpha * scale / ((scalar_exp_avg_sq / bias_correction2).sqrt() + grad_eps)
delta.add_(cur_grad, alpha=alpha) delta.add_(cur_grad, alpha=alpha)
if step % 10 == 0: # this periodicity is just to save time
# divide by 1-beta1 because we want the actual step...
state["grad_times_step"].mul_(beta1).add_( (cur_grad * grad).sum(),
alpha=-alpha/n_cached_grads)
state["step_within_period"] += 1 state["step_within_period"] += 1
else: else:
# p.numel() == 1. # p.numel() == 1.
@ -449,6 +462,8 @@ class NeutralGradient(Optimizer):
alpha = -lr*(1-beta1)*(bias_correction2 ** 0.5) alpha = -lr*(1-beta1)*(bias_correction2 ** 0.5)
delta.add_(this_delta, alpha=alpha) delta.add_(this_delta, alpha=alpha)
state["grad_times_step"].mul_(beta2).add_(this_delta * grad, alpha=-alpha*(1-beta2)/((1-beta1)*n_cached_grads))
if random.random() < 0.0001: if random.random() < 0.0001:
logging.info(f"Delta rms = {(delta**2).mean().item()}, shape = {delta.shape}") logging.info(f"Delta rms = {(delta**2).mean().item()}, shape = {delta.shape}")
p.add_(delta) p.add_(delta)
@ -894,17 +909,9 @@ class NeutralGradient(Optimizer):
state = self.state[p] state = self.state[p]
step = state["step"] step = state["step"]
prod = state["exp_avg_sq"].sqrt().sum() # the sqrt() is just a heuristic, it seems to give periods that work better.
is_one_axis = (p.numel() in list(p.shape)) # e.g. a bias param_prods.append(state["grad_times_step"].sqrt())
if is_one_axis or p.numel() == 1:
if is_one_axis:
prod *= state["scale"]
bias_correction2 = 1 - beta2 ** (step + 1)
else:
step_within_period = state["step_within_period"]
bias_correction2 = 1 - beta2 ** (min(step, step_within_period) + 1)
prod /= bias_correction2 ** 0.5
param_prods.append(prod)
param_prods = torch.stack(param_prods).to('cpu') param_prods = torch.stack(param_prods).to('cpu')
# TEMP # TEMP
@ -1598,7 +1605,7 @@ def _test_eve_cain():
estimate_period=500, stats_steps=100) estimate_period=500, stats_steps=100)
elif iter == 3: optim = NeutralGradient(m.parameters(), lr=0.04, max_fullcov_size=1000, elif iter == 3: optim = NeutralGradient(m.parameters(), lr=0.04, max_fullcov_size=1000,
estimate_period=500, stats_steps=100) estimate_period=500, stats_steps=100)
scheduler = Eden(optim, lr_batches=200, lr_epochs=10, verbose=False) scheduler = Eden(optim, lr_batches=200, lr_epochs=5, verbose=False)
start = timeit.default_timer() start = timeit.default_timer()
avg_loss = 0.0 avg_loss = 0.0