From 1e29fe07196ea325db60b3e3a92c1ab2a34f2b28 Mon Sep 17 00:00:00 2001
From: dohe0342 <kimdohe1070@gmail.com>
Date: Mon, 9 Jan 2023 20:26:14 +0900
Subject: [PATCH] from local

---
 .../incremental_transf/.identity_train.py.swp | Bin 65536 -> 73728 bytes
 .../ASR/incremental_transf/identity_train.py  |  33 +++++-------------
 2 files changed, 8 insertions(+), 25 deletions(-)

diff --git a/egs/librispeech/ASR/incremental_transf/.identity_train.py.swp b/egs/librispeech/ASR/incremental_transf/.identity_train.py.swp
index fd33ee62242c219f9054f10024e2aa2af9dd40ad..65b825032275da595035268d43164405afd2ad5b 100644
GIT binary patch
delta 1720
zcmZA1ZERCj7{Kv!yRwb7U0ZOBqTB9ZVPzZRiplCCi*GX-Bsdb{1jWqZ+F%@PP?TZR
z7EI7=0oNT-BJ#2cbi|O*-8dbSuu0<lK+Fcyfy{)&*Gy_KY6R5H|0#E0Jn8S=d!EzY
zw$F2(8;#ndgZ7+_M%~KQdS$sG7#<-)w+5aIByy5&%_Gx1Nv%fy=E+^cLTcEp^M|%O
z7n(J9SZAB`dUL&AYt~|&`%RxWp53e4qpr{_A-dCvbH^I#JvtuIW(zT%Da0U-U>`PO
z1?J)J3?aV7ezc<&bCH20ce;pE_yFq8fAI|;<7HG~1^(f8_aTgXa666n;Po_7Dz<Xb
zf*R=fhW~Z|b(oLa{K?}usF{bmoU`mYL!KVXm21x1eXaFP!7bscP_W^}?c0K>NtsB!
z<3dkC-sUi&=5ScA(l<9Z2Alk;CAC~!Tx{mYf7dc*CeD*L(<3jqg?I}pJA89$#I4vN
z&aK#irxAb~7qW%ejY{}1>=NR0G~rP=@U>HjW7vfN?uCIrsoOBxQR@&9@i-UvBY+C{
zFqI|51U|uDY)3sR;YTqtkdCh?<|%YyCqh__aunef1s%l*`tdd-Ho(Y?2=Omf9mNPf
z#Q@IWINGoYk6<CDsP-lFpaVfvAP-rXrt(*C8Sh~O0;oV4@^ORR@Ey*h_ZbfR@G7dX
z6aqKc6+<u)!4?=Oz$A6Lj85#q16YiFIN`uA?4C)@yp~9~3RS(;nAsNA4iTv!hehV;
zb88-oG&J6+>`f}Fc7ysQYMi0sY8O;Mt&N}uhp`f+ScpQnah2*Hz)L8=1XX_rb$9~v
zag`z;#9q{6nL}{o^&=G>$}%<k)NC2=s+5VbdGc(RNA`~8q#5Og+&Y}g+$XmUtM~$I
z-KW;2`$RUe5^K$$ik<E9Tl1N?VST3DT7TG@JO3BY=W`2e`VuRgX`LS=JV%yw>$0fZ
zZx(fDYN^C+B2m@kT_vtXWsNaS!Y*PIA0dY4u^1C1N+JxEI322m4j~8wQz}u2p&f0g
zMInA5S0~Y<TBwtY7&;KfI;d8fgP%y~m*|JYZtOw>p2ae#R=Q5BT*4sYcoVN-ElN>>
zx%ic~Num#Jcv7`cDHkPB56>Ljg=yO5dwhoX@e0;qDQuXeUCu$Z(GE0XJ?_JD{6VYq
zA%+g@K`T~4J!JFX#Lw&5sN;ASZ{R7c!rgG;3N7;m+OQMVSdQzI>l*eWidKXW#AB$o
M#fF;AWMaVeA6}h7-2eap

delta 890
zcmX}qT}YEr9LMqh<JMl(y&+4)i#>%5typv2I7Kul(2K|*x`{T_v@OQw%L;m9iz2)z
zmOa)YA{5f~0@YeUWbaZWM1>KCkzGVV<iPAABB;It^ADfH;hg6@9M1oje56k%_0A6O
zw$@Fm-{Td^F4FdEywjd_=52~q)7kS%K>KblH<uK4sa~ztQJxDq&RPdwH{VqCNlO{4
z8HItf0nITyo%m!Dd4a1qhbUUG5=MMAii{wQ?Qr3PLF5iDAciJ5pu@y6kz;5@3ru*T
z7fGNR22AKgQaGfO)pC%@R;<C7B9V*eKotzga^1p}$CSax&h$XB;!nRSRV?Y0Rk7vf
zN3Q9$KhKuehP%~hEUw05eVrjsdn8gF?hkf{qv3d0O}(g#wX~)-s@m|VvqE*NJ$-wk
zT|0x3Sbu+TX<xMUqh&dF=A)8xT>fE5l=5P8i_g}}yik{H+zb1$4?Ez8123&2N6~;v
zOj<;qBZ`f%;2l35!x8wg4nF*(dy_bUEodu|gw!+fqYjl2%odB>MGAY-js~nk4a_iM
z+AMM%LpXpCHp2@ye$(YQn8G;D;4}j8;jb~l2ef?(4{#s%kil{6#s*YD@Rk@{!71!Q
z9b71dg7-vX2ImogA9e6R#R3uejHkGTlR<_;ewwfrMOYv%g&f_+Vf4U<6__J7GZ?}E
n8c_`uR#=dyorRWuLvA*kwWOmJiZMO!QtYPxdCA>)*w*|H7>=D0

diff --git a/egs/librispeech/ASR/incremental_transf/identity_train.py b/egs/librispeech/ASR/incremental_transf/identity_train.py
index 66e59465c..45a5d8e08 100755
--- a/egs/librispeech/ASR/incremental_transf/identity_train.py
+++ b/egs/librispeech/ASR/incremental_transf/identity_train.py
@@ -665,6 +665,7 @@ def compute_loss(
     y = k2.RaggedTensor(y).to(device)
 
     with torch.set_grad_enabled(is_training):
+        '''
         simple_loss, pruned_loss = model(
             x=feature,
             x_lens=feature_lens,
@@ -678,36 +679,18 @@ def compute_loss(
         )
         simple_loss_is_finite = torch.isfinite(simple_loss)
         pruned_loss_is_finite = torch.isfinite(pruned_loss)
-        is_finite = simple_loss_is_finite & pruned_loss_is_finite
-        if not torch.all(is_finite):
-            logging.info(
-                "Not all losses are finite!\n"
-                f"simple_loss: {simple_loss}\n"
-                f"pruned_loss: {pruned_loss}"
-            )
-            display_and_save_batch(batch, params=params, sp=sp)
-            simple_loss = simple_loss[simple_loss_is_finite]
-            pruned_loss = pruned_loss[pruned_loss_is_finite]
+        '''
+        mse_loss = model(
+                x=feature,
+                x_lens=feautre_lens,
+        )
 
-            # If the batch contains more than 10 utterances AND
-            # if either all simple_loss or pruned_loss is inf or nan,
-            # we stop the training process by raising an exception
-            if torch.all(~simple_loss_is_finite) or torch.all(~pruned_loss_is_finite):
-                raise ValueError(
-                    "There are too many utterances in this batch "
-                    "leading to inf or nan losses."
-                )
-
-        simple_loss = simple_loss.sum()
-        pruned_loss = pruned_loss.sum()
         # after the main warmup step, we keep pruned_loss_scale small
         # for the same amount of time (model_warm_step), to avoid
         # overwhelming the simple_loss and causing it to diverge,
         # in case it had not fully learned the alignment yet.
-        pruned_loss_scale = (
-            0.0 if warmup < 1.0 else (0.1 if warmup > 1.0 and warmup < 2.0 else 1.0)
-        )
-        loss = params.simple_loss_scale * simple_loss + pruned_loss_scale * pruned_loss
+        #loss = params.simple_loss_scale * simple_loss + pruned_loss_scale * pruned_loss
+        loss = mse_loss
 
     assert loss.requires_grad == is_training