Fix loading sampler state dict.

k2-fsa · Jun 14, 2022 · fd7646d · fd7646d
1 parent 53f38c0
commit fd7646d
Show file tree

Hide file tree

Showing 4 changed files with 0 additions and 44 deletions.
diff --git a/egs/librispeech/ASR/pruned_transducer_stateless/train.py b/egs/librispeech/ASR/pruned_transducer_stateless/train.py
@@ -402,9 +402,6 @@ def load_checkpoint_if_available(
         if "cur_epoch" in saved_params:
             params["start_epoch"] = saved_params["cur_epoch"]
 
-        if "cur_batch_idx" in saved_params:
-            params["cur_batch_idx"] = saved_params["cur_batch_idx"]
-
     return saved_params
 
 
@@ -610,13 +607,7 @@ def maybe_log_weights(tag: str):
                 global_step=params.batch_idx_train,
             )
 
-    cur_batch_idx = params.get("cur_batch_idx", 0)
-
     for batch_idx, batch in enumerate(train_dl):
-        if batch_idx < cur_batch_idx:
-            continue
-        cur_batch_idx = batch_idx
-
         params.batch_idx_train += 1
         batch_size = len(batch["supervisions"]["text"])
 
@@ -664,7 +655,6 @@ def maybe_log_weights(tag: str):
             params.batch_idx_train > 0
             and params.batch_idx_train % params.save_every_n == 0
         ):
-            params.cur_batch_idx = batch_idx
             save_checkpoint_with_global_batch_idx(
                 out_dir=params.exp_dir,
                 global_batch_idx=params.batch_idx_train,
@@ -674,7 +664,6 @@ def maybe_log_weights(tag: str):
                 sampler=train_dl.sampler,
                 rank=rank,
             )
-            del params.cur_batch_idx
             remove_checkpoints(
                 out_dir=params.exp_dir,
                 topk=params.keep_last_k,

diff --git a/egs/librispeech/ASR/pruned_transducer_stateless2/train.py b/egs/librispeech/ASR/pruned_transducer_stateless2/train.py
@@ -449,9 +449,6 @@ def load_checkpoint_if_available(
         if "cur_epoch" in saved_params:
             params["start_epoch"] = saved_params["cur_epoch"]
 
-        if "cur_batch_idx" in saved_params:
-            params["cur_batch_idx"] = saved_params["cur_batch_idx"]
-
     return saved_params
 
 
@@ -661,13 +658,7 @@ def train_one_epoch(
 
     tot_loss = MetricsTracker()
 
-    cur_batch_idx = params.get("cur_batch_idx", 0)
-
     for batch_idx, batch in enumerate(train_dl):
-        if batch_idx < cur_batch_idx:
-            continue
-        cur_batch_idx = batch_idx
-
         params.batch_idx_train += 1
         batch_size = len(batch["supervisions"]["text"])
 
@@ -702,7 +693,6 @@ def train_one_epoch(
             params.batch_idx_train > 0
             and params.batch_idx_train % params.save_every_n == 0
         ):
-            params.cur_batch_idx = batch_idx
             save_checkpoint_with_global_batch_idx(
                 out_dir=params.exp_dir,
                 global_batch_idx=params.batch_idx_train,
@@ -714,7 +704,6 @@ def train_one_epoch(
                 scaler=scaler,
                 rank=rank,
             )
-            del params.cur_batch_idx
             remove_checkpoints(
                 out_dir=params.exp_dir,
                 topk=params.keep_last_k,

diff --git a/egs/librispeech/ASR/pruned_transducer_stateless4/train.py b/egs/librispeech/ASR/pruned_transducer_stateless4/train.py
@@ -471,9 +471,6 @@ def load_checkpoint_if_available(
         if "cur_epoch" in saved_params:
             params["start_epoch"] = saved_params["cur_epoch"]
 
-        if "cur_batch_idx" in saved_params:
-            params["cur_batch_idx"] = saved_params["cur_batch_idx"]
-
     return saved_params
 
 
@@ -694,13 +691,7 @@ def train_one_epoch(
 
     tot_loss = MetricsTracker()
 
-    cur_batch_idx = params.get("cur_batch_idx", 0)
-
     for batch_idx, batch in enumerate(train_dl):
-        if batch_idx < cur_batch_idx:
-            continue
-        cur_batch_idx = batch_idx
-
         params.batch_idx_train += 1
         batch_size = len(batch["supervisions"]["text"])
 
@@ -742,7 +733,6 @@ def train_one_epoch(
             params.batch_idx_train > 0
             and params.batch_idx_train % params.save_every_n == 0
         ):
-            params.cur_batch_idx = batch_idx
             save_checkpoint_with_global_batch_idx(
                 out_dir=params.exp_dir,
                 global_batch_idx=params.batch_idx_train,
@@ -755,7 +745,6 @@ def train_one_epoch(
                 scaler=scaler,
                 rank=rank,
             )
-            del params.cur_batch_idx
             remove_checkpoints(
                 out_dir=params.exp_dir,
                 topk=params.keep_last_k,

diff --git a/egs/librispeech/ASR/pruned_transducer_stateless5/train.py b/egs/librispeech/ASR/pruned_transducer_stateless5/train.py
@@ -512,9 +512,6 @@ def load_checkpoint_if_available(
         if "cur_epoch" in saved_params:
             params["start_epoch"] = saved_params["cur_epoch"]
 
-        if "cur_batch_idx" in saved_params:
-            params["cur_batch_idx"] = saved_params["cur_batch_idx"]
-
     return saved_params
 
 
@@ -735,13 +732,7 @@ def train_one_epoch(
 
     tot_loss = MetricsTracker()
 
-    cur_batch_idx = params.get("cur_batch_idx", 0)
-
     for batch_idx, batch in enumerate(train_dl):
-        if batch_idx < cur_batch_idx:
-            continue
-        cur_batch_idx = batch_idx
-
         params.batch_idx_train += 1
         batch_size = len(batch["supervisions"]["text"])
 
@@ -787,7 +778,6 @@ def train_one_epoch(
             params.batch_idx_train > 0
             and params.batch_idx_train % params.save_every_n == 0
         ):
-            params.cur_batch_idx = batch_idx
             save_checkpoint_with_global_batch_idx(
                 out_dir=params.exp_dir,
                 global_batch_idx=params.batch_idx_train,
@@ -800,7 +790,6 @@ def train_one_epoch(
                 scaler=scaler,
                 rank=rank,
             )
-            del params.cur_batch_idx
             remove_checkpoints(
                 out_dir=params.exp_dir,
                 topk=params.keep_last_k,