From 177bce2db41d6289db7117af8d763eb49cf0fb90 Mon Sep 17 00:00:00 2001
From: dohe0342 <kimdohe1070@gmail.com>
Date: Wed, 24 May 2023 13:37:08 +0900
Subject: [PATCH] from local

---
 .../.train_lora.py.swp                        | Bin 90112 -> 90112 bytes
 .../train_lora.py                             |  13 +++++++------
 2 files changed, 7 insertions(+), 6 deletions(-)

diff --git a/egs/librispeech/ASR/pruned_transducer_stateless_d2v_v2/.train_lora.py.swp b/egs/librispeech/ASR/pruned_transducer_stateless_d2v_v2/.train_lora.py.swp
index 914514cac6bdf6ef1e8279e9af706c553bf711da..3b7f923d95df056458cb198f893361111d218b57 100644
GIT binary patch
delta 292
zcmZoTz}j$tRV>LM%+puFQqO<^2m}}y{!Y$KkuumQ_DhhFyID|ZyPy!`XI|D|prrS9
zP6fv6jDjl6Kn@VEWMW{b0b=jz9~BvcCx6fr=gG-0N{r9VPbtkwEuNmJ#Hc0!W-Dao
zDZoT0DlulWLphUs;w1%<^(fdXM8|I5&CXcIw0*lOqc0b0IX?r#;^~|QjM9vYx2qa3
z-sk5H;bUN!1H>TLa!(g@U=)_y2voHai0gnj7l`G7_%~3=b+FRyjt-1be3SQ0<e0w4
lkMZL4nf{Cx^6Kj93<^MymS3a*astSWNKV;)#-Gtd004uOP`Cg9

delta 285
zcmZoTz}j$tRV>LM%+puFQqO<^2m}}yK1|L{`KZ59?3W-TSF@ndc0nP=&%CUDKuNdl
zoC=KB83k3D85q7WF)+*p;xZt1oBmOeF_<YQf4a94qx5urB}UfiIZBL*Tnd?a3c2|y
zr8%k7=PEH~+h*!Oxgd$uywcp%qQsI^4WN?L96gA_Vhv4AD+UE1$jL8CjE4v+*eXQF
zZr{z$SjV(|yDFnE7i%#;1H=63oCb{2jPtjv8Zh4H=Plr4VAul0AU|+T7j$40mRtl>
zH4%tgf!G{~g@O1zNZEEr2gYc=i4SBZ@0-XmeTg6A#p%_dj6Bmz{TVH`U-W0x7XSb!
CuuXmd

diff --git a/egs/librispeech/ASR/pruned_transducer_stateless_d2v_v2/train_lora.py b/egs/librispeech/ASR/pruned_transducer_stateless_d2v_v2/train_lora.py
index d83874c08..2260aff8a 100755
--- a/egs/librispeech/ASR/pruned_transducer_stateless_d2v_v2/train_lora.py
+++ b/egs/librispeech/ASR/pruned_transducer_stateless_d2v_v2/train_lora.py
@@ -1591,17 +1591,18 @@ def run_adapter(rank, world_size, args, wb=None):
         logging.info("Using DDP")
         model = DDP(model, device_ids=[rank], find_unused_parameters=True)
     
-    lora_module = []
-    for i, module in enumerate(model.modules()):
+    lora_modules = []
+    for modules in model.modules():
         if isinstance(module, fairseq.modules.multihead_attention.MultiheadAttention):
-            for m in module.modules():
-                lora_module.append(LoRAHook(m))
+            for module in modules.modules():
+                lora_modules.append(LoRAHook(m))
     
     adapter_names = []
     adapter_param = []
+    for lora in lora_module
+    '''
     for n, p  in model.named_parameters():
         print(n)
-        '''
         if 'adapters' in n:# or 'joiner' in n or 'simple' in n or 'ctc' in n:
             adapter_names.append(n)
             adapter_param.append(p)
@@ -1609,7 +1610,7 @@ def run_adapter(rank, world_size, args, wb=None):
             p.requires_grad = True
         else:
             p.requires_grad = False
-        '''
+    '''
     optimizer_adapter = ScaledAdam(
             adapter_param,
             lr=params.adapter_lr,