Support --iter in export.py (#360)

egs/librispeech/ASR/pruned_transducer_stateless2/export.py

@@ -51,7 +51,11 @@ import sentencepiece as spm
 import torch
 from train import get_params, get_transducer_model
 
-from icefall.checkpoint import average_checkpoints, load_checkpoint
+from icefall.checkpoint import (
+    average_checkpoints,
+    find_checkpoints,
+    load_checkpoint,
+)
 from icefall.utils import str2bool
 
 
@@ -64,8 +68,19 @@ def get_parser():
         "--epoch",
         type=int,
         default=28,
-        help="It specifies the checkpoint to use for decoding."
+        help="""It specifies the checkpoint to use for averaging.
-        "Note: Epoch counts from 0.",
+        Note: Epoch counts from 0.
+        You can specify --avg to use more checkpoints for model averaging.""",
+    )
+
+    parser.add_argument(
+        "--iter",
+        type=int,
+        default=0,
+        help="""If positive, --epoch is ignored and it
+        will use the checkpoint exp_dir/checkpoint-iter.pt.
+        You can specify --avg to use more checkpoints for model averaging.
+        """,
     )
 
     parser.add_argument(
@@ -74,7 +89,7 @@ def get_parser():
         default=15,
         help="Number of checkpoints to average. Automatically select "
         "consecutive checkpoints before the checkpoint specified by "
-        "'--epoch'. ",
+        "'--epoch' and '--iter'",
     )
 
     parser.add_argument(
@@ -141,7 +156,24 @@ def main():
 
     model.to(device)
 
-    if params.avg == 1:
+    if params.iter > 0:
+        filenames = find_checkpoints(params.exp_dir, iteration=-params.iter)[
+            : params.avg
+        ]
+        if len(filenames) == 0:
+            raise ValueError(
+                f"No checkpoints found for"
+                f" --iter {params.iter}, --avg {params.avg}"
+            )
+        elif len(filenames) < params.avg:
+            raise ValueError(
+                f"Not enough checkpoints ({len(filenames)}) found for"
+                f" --iter {params.iter}, --avg {params.avg}"
+            )
+        logging.info(f"averaging {filenames}")
+        model.to(device)
+        model.load_state_dict(average_checkpoints(filenames, device=device))
+    elif params.avg == 1:
         load_checkpoint(f"{params.exp_dir}/epoch-{params.epoch}.pt", model)
     else:
         start = params.epoch - params.avg + 1

egs/librispeech/ASR/pruned_transducer_stateless3/export.py

@@ -52,7 +52,11 @@ import sentencepiece as spm
 import torch
 from train import get_params, get_transducer_model
 
-from icefall.checkpoint import average_checkpoints, load_checkpoint
+from icefall.checkpoint import (
+    average_checkpoints,
+    find_checkpoints,
+    load_checkpoint,
+)
 from icefall.utils import str2bool
 
 
@@ -65,8 +69,19 @@ def get_parser():
         "--epoch",
         type=int,
         default=28,
-        help="It specifies the checkpoint to use for decoding."
+        help="""It specifies the checkpoint to use for averaging.
-        "Note: Epoch counts from 0.",
+        Note: Epoch counts from 0.
+        You can specify --avg to use more checkpoints for model averaging.""",
+    )
+
+    parser.add_argument(
+        "--iter",
+        type=int,
+        default=0,
+        help="""If positive, --epoch is ignored and it
+        will use the checkpoint exp_dir/checkpoint-iter.pt.
+        You can specify --avg to use more checkpoints for model averaging.
+        """,
     )
 
     parser.add_argument(
@@ -75,7 +90,7 @@ def get_parser():
         default=15,
         help="Number of checkpoints to average. Automatically select "
         "consecutive checkpoints before the checkpoint specified by "
-        "'--epoch'. ",
+        "'--epoch' and '--iter'",
     )
 
     parser.add_argument(
@@ -142,7 +157,24 @@ def main():
 
     model.to(device)
 
-    if params.avg == 1:
+    if params.iter > 0:
+        filenames = find_checkpoints(params.exp_dir, iteration=-params.iter)[
+            : params.avg
+        ]
+        if len(filenames) == 0:
+            raise ValueError(
+                f"No checkpoints found for"
+                f" --iter {params.iter}, --avg {params.avg}"
+            )
+        elif len(filenames) < params.avg:
+            raise ValueError(
+                f"Not enough checkpoints ({len(filenames)}) found for"
+                f" --iter {params.iter}, --avg {params.avg}"
+            )
+        logging.info(f"averaging {filenames}")
+        model.to(device)
+        model.load_state_dict(average_checkpoints(filenames, device=device))
+    elif params.avg == 1:
         load_checkpoint(f"{params.exp_dir}/epoch-{params.epoch}.pt", model)
     else:
         start = params.epoch - params.avg + 1