mirrors/icefall
1
0
Fork 0
mirror of https://github.com/k2-fsa/icefall.git synced 2025-09-18 21:44:18 +00:00
Code Issues Packages Projects Releases Wiki Activity

Using a more flexible test. Moved to simpler update , tuned diffrently.

Browse Source
This commit is contained in:
Daniel Povey 2022-07-24 09:20:53 +08:00
parent b8a9485011
commit 962e95f119
1 changed files with 27 additions and 9 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

36
egs/librispeech/ASR/pruned_transducer_stateless7/optim.py
View File

@ -159,8 +159,8 @@ param_rms_smooth1: Smoothing proportion for parameter matrix, if assumed rank of
lr=3e-02,
betas=(0.9, 0.98),
size_lr_scale=0.1,
param_cov_min=(0.05, 0.01, 0.01),
param_cov_max=(10.0, 40.0, 10.0),
param_cov_min=(0.05, 0.01, 0.04),
param_cov_max=(10.0, 40.0, 5.0),
param_pow=(1.0, 1.0, 1.0),
param_rms_smooth0=0.4,
param_rms_smooth1=0.2,
@ -418,7 +418,8 @@ param_rms_smooth1: Smoothing proportion for parameter matrix, if assumed rank of
# Only update the parameter-dependent part of the learning
# rate matrices at most every other time we reach here, and
# less frequently than that later in training.
self._update_param_scales(group, p, state, P_proj)
#self._update_param_scales(group, p, state, P_proj)
self._update_param_scales_simple(group, p, state, P_proj)
# We won't be doing this any more.
#self._diagonalize_grad_cov(group, p, state)
@ -599,6 +600,13 @@ param_rms_smooth1: Smoothing proportion for parameter matrix, if assumed rank of
# individual tensor dims
this_P_proj /= _mean(this_P_proj, exclude_dims=[0], keepdim=True)
if True:
# debug info.
scale = this_P_proj.sqrt()
step = state["step"]
scale_min, scale_max, scale_mean = scale.min().item(), scale.max().item(), scale.mean().item()
logging.info(f"step={step}, dim={dim}, size={size}, scale min,max,mean={scale_min,scale_max,scale_mean}")
Q *= this_P_proj.sqrt()
def _update_param_scales(self,
@ -775,7 +783,9 @@ param_rms_smooth1: Smoothing proportion for parameter matrix, if assumed rank of
scale = cur_scales[dim].reshape(batch_size, num_blocks, block_size, 1)
# Geometrically interpolate scale with P_proj[dim].sqrt()
scale = (scale * P_proj[dim].reshape(batch_size, num_blocks, block_size, 1).sqrt()).sqrt()
P_proj_weight = 0.5
scale = ((scale ** (1-P_proj_weight)) *
(P_proj[dim].reshape(batch_size, num_blocks, block_size, 1) ** (P_proj_weight * 0.5)))
# The following normalization step will ensure the Frobenius
# norm is unchanged, from applying this scale: at least,
@ -787,9 +797,15 @@ param_rms_smooth1: Smoothing proportion for parameter matrix, if assumed rank of
# individual tensor dims
scale /= _mean(scale**2, exclude_dims=[0], keepdim=True).sqrt()
if True:
# debug info.
step = state["step"]
scale_min, scale_max, scale_mean = scale.min().item(), scale.max().item(), scale.mean().item()
logging.info(f"step={step}, dim={dim}, size={size}, scale min,max,mean={scale_min,scale_max,scale_mean}")
# Q is indexed [batch_index, block_index, diagonalized_coordinate, canonical_coordinate],
# want to multiply on the diagonalized co-ordinate.
# else: Q is indexed [batch_index, canonical_coordinate].
state[f"Q_{dim}"] *= scale
state["last_param_scale_update"] = state["step"]
@ -2163,11 +2179,13 @@ def _test_eve_cain():
fix_random_seed(42)
Linear = torch.nn.Linear if iter == 0 else ScaledLinear
# TODO: find out why this is not converging...
m = torch.nn.Sequential(Linear(E, 200),
hidden_dim = 512
m = torch.nn.Sequential(Linear(E, hidden_dim),
torch.nn.PReLU(),
Linear(200, 200),
Linear(hidden_dim, hidden_dim),
torch.nn.PReLU(),
Linear(200, E),
Linear(hidden_dim, E),
).to(device)
train_pairs = [ (100.0 * torch.randn(B, T, E, device=device, dtype=dtype) * input_magnitudes,
@ -2176,7 +2194,7 @@ def _test_eve_cain():
if iter == 0: optim = Eve(m.parameters(), lr=0.003)
elif iter == 1: optim = Cain(m.parameters(), lr=0.03)
elif iter == 2: optim = PrAdam(m.parameters(), lr=0.03)
elif iter == 3: optim = PrAdam(m.parameters(), lr=0.03, max_block_size=100)
elif iter == 3: optim = PrAdam(m.parameters(), lr=0.03, max_block_size=256)
scheduler = Eden(optim, lr_batches=200, lr_epochs=5, verbose=False)
#TEMP