From 274e9de876897b1665543f76be53bee68789089c Mon Sep 17 00:00:00 2001
From: dohe0342 <kimdohe1070@gmail.com>
Date: Tue, 14 Feb 2023 00:51:47 +0900
Subject: [PATCH] from local

---
 egs/tedlium2/ASR/conformer_ctc2/.train.py.swp | Bin 49152 -> 53248 bytes
 egs/tedlium2/ASR/conformer_ctc2/train.py      |  27 ++++++++++++------
 2 files changed, 18 insertions(+), 9 deletions(-)

diff --git a/egs/tedlium2/ASR/conformer_ctc2/.train.py.swp b/egs/tedlium2/ASR/conformer_ctc2/.train.py.swp
index 7e1995bf3d8c37f9e3363e27221b86001c8959c4..d054c650fb2f883eb680071626206bb2227b5585 100644
GIT binary patch
delta 898
zcmY+?OGp%P9KiA4v=29RS2t)i4VoGxD_N~Wp<rPMsi_g96dq)B*IF7M&F&fsp%G$r
z2y3cqg+iB##yP}d52B!#MP0f`)S*)ey+{greaFY4KYaeL`9EfUzZo9(3}5y{Ix1T0
zo7Bo^g;4TEthNuGT@7|*Hc!cMJo!OI*=&Ww1>s!xIi+Tyk?YQ?@T1VA>KzNt21KTp
zGtYJw-+VCig^@e+5RY*dqu7IHXn5`J!YEp>8y-w$iHu_q1K5iKJoAejLklXf2ETmV
zflIi6^Jw+SVJW6lgcn|s6F7hp6yv8y<OiN&45x7lZ79RHWg=hk3RiIky=aDp0RAW<
zzcH;i&uxETi%LnwW5!aONO!B%>b`iaCTZ5{DZQyTmNw$0x_L0ULEJ@}GndKs59u+H
zm%Qg^KE@5-Qgr|3iM+=(T)`piL=7T%5)!$GLDXS8=7J(Oa1q@wP>XEbC4GIUgN7CO
zlp{lOD@WuUPGS`G2qB33*&_XDL?J?$C37Ef8^<w-0W_cp@5tZ`Y#hZA^q~<2h~T@M
zO-yE~{BLwj>_QpVVJ+N<-0?e#SNB;n_D0K&7FuuZO<p&6&WycP$#UaOoUPSPG*~?!
zeMZq(yx$sg_LnR^jiyXJ(bFCy2S%?Ri<`+(agUg<G|TT)Dw?&!DW~pn$^+VBcC2xy
zO3|D*&WLaQ!U>(}RFahbf2=!My0m2?qgf-FmAP8nNXDxcFPO+|E7THQYM*ZE-N|TI
O+A!LUM2DHE68Q(-tht>4

delta 350
zcmXBPze@sf7{~GFd#87odF4eY6v0hFH;71!HMH8$>L3_V2L1RS^2UOO>_BRW1f3cp
z{X*0h1qCe)4UrKD(Go2+H201^@PY@P&oew>(+op1y__q}FZhLQPDo0mQ2tnXD|XVK
z!z!W8l&jQw<arrg^e%?#SAN%z?YZMrU(|n}?b%K3dqCt1PiUfnMO2W-yTb(<sG^J!
z+}k2u>|h&9m_*MK*+ms|$YKzWagkHB(ZVW%xNOQ4!)r`rA8VLK00%anO_5_9;s6_%
zg#m@WA##o*Y@vcYGH{W^-^)E!?@UCs+dz-qC6zYxwH5UroTzj%YR`!HME_hStY6^k
BKcoNv

diff --git a/egs/tedlium2/ASR/conformer_ctc2/train.py b/egs/tedlium2/ASR/conformer_ctc2/train.py
index 6c6149955..a3cf82b07 100755
--- a/egs/tedlium2/ASR/conformer_ctc2/train.py
+++ b/egs/tedlium2/ASR/conformer_ctc2/train.py
@@ -935,12 +935,14 @@ def run(rank, world_size, args):
     if params.print_diagnostics:
         diagnostic = diagnostics.attach_diagnostics(model)
 
-    librispeech = LibriSpeechAsrDataModule(args)
+    #librispeech = LibriSpeechAsrDataModule(args)
+    tedlium = TedAsrDataModule(args)
 
-    if params.full_libri:
-        train_cuts = librispeech.train_all_shuf_cuts()
-    else:
-        train_cuts = librispeech.train_clean_100_cuts()
+    #if params.full_libri:
+    #    train_cuts = librispeech.train_all_shuf_cuts()
+    #else:
+    #    train_cuts = librispeech.train_clean_100_cuts()
+    train_cuts = tedlium.train_cuts()
 
     def remove_short_and_long_utt(c: Cut):
         # Keep only utterances with duration between 1 second and 20 seconds
@@ -975,13 +977,20 @@ def run(rank, world_size, args):
     else:
         sampler_state_dict = None
 
-    train_dl = librispeech.train_dataloaders(
+    #train_dl = librispeech.train_dataloaders(
+    #    train_cuts, sampler_state_dict=sampler_state_dict
+    #)
+
+    #valid_cuts = librispeech.dev_clean_cuts()
+    #valid_cuts += librispeech.dev_other_cuts()
+    #valid_dl = librispeech.valid_dataloaders(valid_cuts)
+    
+    train_dl = tedlium.train_dataloaders(
         train_cuts, sampler_state_dict=sampler_state_dict
     )
 
-    valid_cuts = librispeech.dev_clean_cuts()
-    valid_cuts += librispeech.dev_other_cuts()
-    valid_dl = librispeech.valid_dataloaders(valid_cuts)
+    valid_cuts = tedlium.dev_cuts()
+    valid_dl = tedlium.valid_dataloaders(valid_cuts)
 
     if params.print_diagnostics:
         scan_pessimistic_batches_for_oom(