From 3fc6fce04c63faa2b18b52e67e6ecaffc6d8d789 Mon Sep 17 00:00:00 2001
From: dohe0342 <kimdohe1070@gmail.com>
Date: Tue, 11 Apr 2023 15:46:18 +0900
Subject: [PATCH] from local

---
 .../.optim.py.swp                             | Bin 4096 -> 16384 bytes
 .../.prompt_tuning.py.swp                     | Bin 86016 -> 86016 bytes
 .../prompt_tuning.py                          |   5 +----
 3 files changed, 1 insertion(+), 4 deletions(-)

diff --git a/egs/librispeech/ASR/pruned_transducer_stateless_d2v_v2/.optim.py.swp b/egs/librispeech/ASR/pruned_transducer_stateless_d2v_v2/.optim.py.swp
index ed4f747e0d1a08e16c200bf8ac2e9dd9f32c1860..af20f364edde5dbddab8ddf7860235af4e35536c 100644
GIT binary patch
literal 16384
zcmeHNTZ|-C87@E+5EL(ws1Xjm!R_AkR?jp`gxMX!>;;&(%&t4Lz-BVkobEc^eVpp5
zqRy%5X(kJ=Mqkw6gYgkG(L^+I(S-P-qHiXw#0PJW#7BLQ74P4F>Qq&C&#VNDFQh8@
zrl!y3KmYmuOVv4bI&%*%Kg8~F=Nw!&JI?9bp6-6{+SjfbeavzEXdve9JkqAB|8|{3
zJCVx65Ap8tDB3!kL~amuc`)1O$y{J-6eU|K=3UW}UD4w~&=P&s3S=jdDi)$UXf3Ut
zYsE<#3g6QS50#&GMdB&VwFrb#p5NZ~w%e^J)^g~^<JqCA-xUKD1A7>_$+@#VZ$kY*
z?RNH&51-%DqAI8us2Hdis2Hdis2Hdis2Hdis2F(nGN9udou@FptL+p!_I}mO{onSw
z&z5&*>i=cy`?ma<nfia)`ZZgAgB>6r=N~imkImHo-PS*5`~P63{%^K^)7IZ?CzQ*7
z)7IyD$20YRwe|V@e>GEoc}D(S*JK0b^52;0|FxNV`_pMz{Vvbc+v4h0F;Fp3F;Fp3
zF;Fp3F;Fp3F;Fp3F;Fp3G4Ow3fcuW~AxM6a7P9#MpRa#^^g+jY3D^O)fg`}5Kj1jO
z0A2x}2EGcM2Cf5se3RphfQNzezzN_8@M+*`;L?qb^D6K&;2VGj)`5F~yMfOD9|dj%
zo`1jNYy#H<KfeLv0WSmJ1-=P93p@eb1KbUK9Qe)kj`K_4%fRP>TYyW~InI}WCE$(s
zInG~z=YSN@z(c^jz$bvq?{%ElfnNhZ0iFjGa1v+%w*j{THv>Pt)^WZLTmbF@8o&*}
zo9}^cz*9g7)PUQ8*Rj_B7O)FE3N(N^@F(ms&_2T#fG2^2!0%v(-vTcIWRn+wy%)0=
z&O>y!a6}^J2_NE6gzRdz$mmvYI8NTG!*zEc6l&J(zxmw@`A{ftON`aJ^mjHkST9N#
z{$ny^DoQkDQ~9i@tlxydtl%AsZ4?8L&N%PU7xh4db*pYBz0qhgXu25iVaMl8F0iR;
zJBw1+Qhz9#{iqk??Cpw1fmC{WOqZysHzvp=1*bZ%qsfNL<5+}#o%n<~rsrduPc@ur
z0vQ;;u5?n5U0;u5fyo|<T{YnNx6qob6E~S1g^cusjO)hwUO$Oa)R|lE5XQWXg`z)K
zN+g;QwG*8hY!P!u**<+sFKRG$fNk2jY_cJb73(F@km)oI1gjgZJ+5#_Cutta6lXdz
zq_U{TP+aH%4o!JcFvTdC@Np_A4Ycv(qS61VK^VGcDNuwKZ>`Fel&IBeg+iExxkdxo
z9WbfbX*ke2Cc5D|whFhstS%S}nphJYJ&i3$E3#;2$BF)#q$m`IMG^Ge3EQbXBP}pK
zAuyB13JrN-k77`Dl&L|K20rVEV%VjmuNF$<A8U0ZU))<%XFB1p>rN_w0_&pYq*jx~
zHdA0;*2JYVGB%QWz!c3WtIQKw=7{SxqoS&9G#FQV`yZQDX&j}j%R}Zz<q-R-y+>uE
zE9`AFF+bMhGL6_N6{acGniywVm9>FD5*5f%s)>F|3+|Kg{78oV306mm)S@&N&9S>h
zF{1}UCajhS9@H3}rPDGBCrrIIfV1_|uuJ9Sc)sXH3C_m(#1v=UDAZy{BS=Huhg&HU
z$u*OjC@Rfb`aqM$WC2&2k*Ar~1CkYvNkJ{UX}}?f@(~JsmL;ju_;tu0$sm9ev3M|6
z5(fZ-ai%=ff&<V1HA8RBc!~9BDyE4RzO`xEY|^x-F0&4%mk1R{p$`f(icoLdhL|lU
zQiw5AfW8UM24WERnfgtmztjq$#(gP5@Pdg3APHm9^YmyTBsNuSv&Xl@W;sg64|>@S
ztR8KHX1;@>xki>2X-;JiRfuGyugRH;)se8x{E+QO1Cf;RH*MiW=rjq@UgK<Jm{Hqg
zV<`e(Avp4wF2=}{6w;`Tn<BKx$q|>72jB9Mai8@g9L#j>zt*El7gA{~lKOG;is7H3
z<CyXZkziUFr{Q{@45jwG>~Vu=u!&t|&ot7nXfmPMWJ4P-`IK0uJCYN_HIb&^tZ>0Z
z&`HtEbHa4kfmJ~FmPkSoDEK#GW>C7LIX+7IW`2xV(jvsGg33}8vI(Ln0lx@f@;hG+
zQID8&88a4wy(K~|^dXZnADv$Hv(68kJ+GQZ$viTQvlwSA-PQOa{1f3$-iWA%6iQ_#
zWRdc7Mon&`bF)UI8N7>dh2#3@g`AQ7DlA~e>GGI&WFYmJTI8}OyprNF%e|IRqB)Z^
zYtnil8ZQ)%lRf&Qq&skpTqcJxUKm-?YNAD6w`R!ADirbqQEIy_A(d^LG<Tc_Jj7w&
ztXVLKv=VNrCCq2r_T-L-cSJuLdJ1$qwpf!gxnkpZLO4B^n?|!J<Uy0I(Hl#XoeyEj
zY?y8o1j1y~Vw8zjkbHz}(EL{r_4^nk|5?V0;~YRl4kZpNQ`%-J+#@meIE~}ZA*O={
zt0K!ww#eAs<f&M6WlwhDt6@Jy_GW#`Eg?+@kv5cy{Erf|4=Il<uRVC?{2F^?>D;-c
zm9^zltL)4<cH+#+$>p`>Gb{LYKU-S4z&^XYa<a(;p106M?8L~WiqWh7Fb<^fNg*F3
zC9;#!Lu8M}<{2~50Fi>=s4cCs<<%NHzO=f$+JubrAsx;}5I&wtL?Y{q8D5M7B&-fz
z5d%J=F?m0sAb~@|`REadsUcrPJv~CQb`C%eT?4J-g;r}c8sX&{Z%M+9l0MFe8wOSD
z^zw;QE32mtw{dixx?Dia4EdNv9nB$^o`ErbA=M)J=OlBDmK0soGK@4!6<GmM+>=mH
z>~uwpr$LC$L=5?17R(buh$|~^lBgx4V_wr_4~v8h#h@S3Hgu@jPPN8SqcAyu>*n(?
z_DA3)WP9T`V6o2=lXix96_)gp%~L%<b}BexCosuG_6OQf#@ci3He-)U9t|bxk{Ed^
zyXv;<0z1h=JU7@`c-6S!(E9%p*1zAu8kyGrExXu!0qgxf@JZkj*7siomVh_0mVX6!
z1~>-1hV}Y$z*m3>co291xE1&ma2ad(mw*?69pG-@4XoFH2y6o>pn(g(G2j;9kF<6N
zUIo4nYykrJ9B>?10uBOK0hgioAAsKjF9S4>Zv)j;F;Fp3F;Fp3F;Fq^ZeW1kV>~UC
z_JohU%4`9d70_A&+pgT_!}>t_sK!?Kuu(2j%R+4PwCCm=5{3n#D7SE?nqW;=S7YV!
zq`$pBzmbh4duEd~FK@7W*}Pqx+7$?@jAXkqaAnww>UODHZsOMZSP;^xKUM5-8b-`L
z%BTpv*x~u?e_!@`?7>q@Cz)HryYWB-!QxtyX8QxQldsZFoJ3gZDi^QD-7POlb)4#Y
zHs@TQqS2IEeW6cd?Sf@{!&UfSY>S|t>%6>j|Cvlzv!0>`9v0GA$76poiakBR#)hxz
z^ZPZLs03S{+5Gc7+Fq`C-VpEQo>!yABiJ`rKByt8SIg#rhJ9=rq`vuNvAd`6Vy)zI
zMt}^#!A$?zvdQjPr)7({%v7~(C2MwM@{cK?-;}y#DSZ#H6JWCvOe^yR21X*&(RM=`
zV25G<{!`=kYGwm2NYHn;><N2$RN)1%7^z>3KGS;{CR*IZo?>zli(KqgFpYhyZ~?+}
z6tlzRb%~xNz?3gC?Al;$=-MT(M?W!CIS#E3VU$Q=AJB~x_yXw33-EoUY?<>?YLDH|
z*fV<!HfHlCbkC^8226y_DDGde1qs=wGKk>+AOVg{Tc=n*?k9o<GT#pH<s?dmYH^ot
z^G)-&V{Yy2ZwK7Kq$Z!BnYqt5vs)*-x9zQS(IL6}UA-C^g5`z~afChoU2}t73>9oQ
z#J_yI@6i<4ZvNkbd!&zw0A>sl{16xSer=#3rws4xr-X5$(h5^Pv_3bvy<PUUGjDg`
ze+Tby+u|U*gVl%Y?G3jh>B|n~Sr^*};2T~xoD76g3VkOsE!%s=j07!K->#tfS%v1_
znL_h33PH@<R*0PDN(3D`)NZ&#!SQr*)#C|1!dja4wa@|MjaMY;od~-UQCF7rAC_^0
As{jB1

delta 7
OcmZo@U~EuWAOHXgJOWYx

diff --git a/egs/librispeech/ASR/pruned_transducer_stateless_d2v_v2/.prompt_tuning.py.swp b/egs/librispeech/ASR/pruned_transducer_stateless_d2v_v2/.prompt_tuning.py.swp
index 9586782e54015429271bbbb05a6bf99bc4fb8500..14bf7df31b6cc6e64ac2b2dd177215ac99156c3d 100644
GIT binary patch
delta 296
zcmZozz}m2Yb;AWg#>&YT1?3qzCVvEy{F?=ZKI`%dFfcF(Fo6gLhRV%?4(AzVm6;hB
zc!2mJP<S#B=L4}f5QCI)OlC|Ho?gYp$hBF>NriFy0ddAkrp?=3eYryQ`4|}5ff!^G
zKac`}-$2H7Al?JSvw*k?h<$+A35czLSPY0kithpS?E~VeK->?+Aam-0xO{VDfE=HM
z0uZRHt23a})9XbTRW`4V^I;TXNK9d1xCPV#HWKIrj>&=v!h*|yhSvab3=oF{vDjwQ
egok{S`SNE@ewZ&cIWM1Ux`;NT*yfk{;tv2;_&J*Z

delta 278
zcmZozz}m2Yb;AWg#;VB|1?3qzCw~Nz0-FVeKI`(bFfcH1FaZex28OE5f)3{yWsR5_
z7^H!i1&D6~rMiJQ9f+NQm}4?ylJI5;CpAW9b#?XayTuvnm^N>B_2mlH;$vWF17eUt
zKm-QA`572K1Mztv-U7svfVc>VU4YmMh>d`l4~TyN^<4wvZ9v=$#O*)~GD~T5WPlvs
z^m=<n?#UM`g*VTQOJbaC7tbNYkeI^2a1N*fq#I-%=VZYIVZp^f-aH^K0pbuK_TOxp
e@Q`ovkv#FqrukLVm)bH4On#X!wE1Md`2zsW5IZ#h

diff --git a/egs/librispeech/ASR/pruned_transducer_stateless_d2v_v2/prompt_tuning.py b/egs/librispeech/ASR/pruned_transducer_stateless_d2v_v2/prompt_tuning.py
index 67f6abf81..4ad3dc93a 100755
--- a/egs/librispeech/ASR/pruned_transducer_stateless_d2v_v2/prompt_tuning.py
+++ b/egs/librispeech/ASR/pruned_transducer_stateless_d2v_v2/prompt_tuning.py
@@ -1579,7 +1579,6 @@ def run_adapter(rank, world_size, args, wb=None):
         logging.info("Using DDP")
         model = DDP(model, device_ids=[rank], find_unused_parameters=True)
     
-    '''
     adapter_names = []
     adapter_param = []
     for n, p  in model.named_parameters():
@@ -1590,14 +1589,12 @@ def run_adapter(rank, world_size, args, wb=None):
             p.requires_grad = True
         else:
             p.requires_grad = False
-    '''
     
     for n, p in model.named_parameters():
         p.requires_grad = False
     
     prompt = torch.nn.Parameter(torch.randn(50, 512)).to(device)
     
-    '''
     optimizer_adapter = ScaledAdam(
             adapter_param,
             lr=params.adapter_lr,
@@ -1611,7 +1608,7 @@ def run_adapter(rank, world_size, args, wb=None):
             clipping_scale=5.0,
             parameters_names=['prompt'],
     )
-
+    '''
     scheduler_adapter = Eden(optimizer_adapter, 10000, 7) #params.lr_batche, params.lr_epochs)
 
     optimizer, scheduler = optimizer_adapter, scheduler_adapter