diff --git a/fairseq/data/transforms/__init__.py b/fairseq/data/encoders/__init__.py
similarity index 83%
rename from fairseq/data/transforms/__init__.py
rename to fairseq/data/encoders/__init__.py
index a64954edd6..1e7e69fbea 100644
--- a/fairseq/data/transforms/__init__.py
+++ b/fairseq/data/encoders/__init__.py
@@ -24,8 +24,8 @@
 )
 
 
-# automatically import any Python files in the transforms/ directory
+# automatically import any Python files in the encoders/ directory
 for file in os.listdir(os.path.dirname(__file__)):
     if file.endswith('.py') and not file.startswith('_'):
         module = file[:file.find('.py')]
-        importlib.import_module('fairseq.data.transforms.' + module)
+        importlib.import_module('fairseq.data.encoders.' + module)
diff --git a/fairseq/data/transforms/gpt2_bpe.py b/fairseq/data/encoders/gpt2_bpe.py
similarity index 99%
rename from fairseq/data/transforms/gpt2_bpe.py
rename to fairseq/data/encoders/gpt2_bpe.py
index 38361a49ac..b411a5aa39 100644
--- a/fairseq/data/transforms/gpt2_bpe.py
+++ b/fairseq/data/encoders/gpt2_bpe.py
@@ -6,7 +6,7 @@
 # can be found in the PATENTS file in the same directory.
 
 from fairseq import file_utils
-from fairseq.data.transforms import register_bpe
+from fairseq.data.encoders import register_bpe
 
 
 @register_bpe('gpt2')
diff --git a/fairseq/data/transforms/moses_tokenizer.py b/fairseq/data/encoders/moses_tokenizer.py
similarity index 96%
rename from fairseq/data/transforms/moses_tokenizer.py
rename to fairseq/data/encoders/moses_tokenizer.py
index dc6016d914..4964a822c2 100644
--- a/fairseq/data/transforms/moses_tokenizer.py
+++ b/fairseq/data/encoders/moses_tokenizer.py
@@ -5,7 +5,7 @@
 # the root directory of this source tree. An additional grant of patent rights
 # can be found in the PATENTS file in the same directory.
 
-from fairseq.data.transforms import register_tokenizer
+from fairseq.data.encoders import register_tokenizer
 
 
 @register_tokenizer('moses')
diff --git a/fairseq/data/transforms/nltk_tokenizer.py b/fairseq/data/encoders/nltk_tokenizer.py
similarity index 93%
rename from fairseq/data/transforms/nltk_tokenizer.py
rename to fairseq/data/encoders/nltk_tokenizer.py
index 206243c7fb..61325efc42 100644
--- a/fairseq/data/transforms/nltk_tokenizer.py
+++ b/fairseq/data/encoders/nltk_tokenizer.py
@@ -5,7 +5,7 @@
 # the root directory of this source tree. An additional grant of patent rights
 # can be found in the PATENTS file in the same directory.
 
-from fairseq.data.transforms import register_tokenizer
+from fairseq.data.encoders import register_tokenizer
 
 
 @register_tokenizer('nltk')
diff --git a/fairseq/data/transforms/sentencepiece_bpe.py b/fairseq/data/encoders/sentencepiece_bpe.py
similarity index 95%
rename from fairseq/data/transforms/sentencepiece_bpe.py
rename to fairseq/data/encoders/sentencepiece_bpe.py
index 11b8dfe20c..9b27460194 100644
--- a/fairseq/data/transforms/sentencepiece_bpe.py
+++ b/fairseq/data/encoders/sentencepiece_bpe.py
@@ -6,7 +6,7 @@
 # can be found in the PATENTS file in the same directory.
 
 from fairseq import file_utils
-from fairseq.data.transforms import register_bpe
+from fairseq.data.encoders import register_bpe
 
 
 @register_bpe('sentencepiece')
diff --git a/fairseq/data/transforms/space_tokenizer.py b/fairseq/data/encoders/space_tokenizer.py
similarity index 91%
rename from fairseq/data/transforms/space_tokenizer.py
rename to fairseq/data/encoders/space_tokenizer.py
index 95d68a45d6..b804b969d8 100644
--- a/fairseq/data/transforms/space_tokenizer.py
+++ b/fairseq/data/encoders/space_tokenizer.py
@@ -7,7 +7,7 @@
 
 import re
 
-from fairseq.data.transforms import register_tokenizer
+from fairseq.data.encoders import register_tokenizer
 
 
 @register_tokenizer('space')
diff --git a/fairseq/data/transforms/subword_nmt_bpe.py b/fairseq/data/encoders/subword_nmt_bpe.py
similarity index 97%
rename from fairseq/data/transforms/subword_nmt_bpe.py
rename to fairseq/data/encoders/subword_nmt_bpe.py
index c5f2722340..b2c1fa33b9 100644
--- a/fairseq/data/transforms/subword_nmt_bpe.py
+++ b/fairseq/data/encoders/subword_nmt_bpe.py
@@ -6,7 +6,7 @@
 # can be found in the PATENTS file in the same directory.
 
 from fairseq import file_utils
-from fairseq.data.transforms import register_bpe
+from fairseq.data.encoders import register_bpe
 
 
 @register_bpe('subword_nmt')
diff --git a/fairseq/hub_utils.py b/fairseq/hub_utils.py
index 1218b9da7c..02c3291fda 100644
--- a/fairseq/hub_utils.py
+++ b/fairseq/hub_utils.py
@@ -9,7 +9,7 @@
 import torch
 
 from fairseq import utils
-from fairseq.data import transforms
+from fairseq.data import encoders
 
 
 class Generator(object):
@@ -44,8 +44,8 @@ def __init__(self, args, task, models):
         # (None if no unknown word replacement, empty if no path to align dictionary)
         self.align_dict = utils.load_align_dict(getattr(args, 'replace_unk', None))
 
-        self.tokenizer = transforms.build_tokenizer(args)
-        self.bpe = transforms.build_bpe(args)
+        self.tokenizer = encoders.build_tokenizer(args)
+        self.bpe = encoders.build_bpe(args)
 
     def generate(self, src_str, verbose=False):
 
diff --git a/interactive.py b/interactive.py
index eea7fcbda4..632a16f3ed 100644
--- a/interactive.py
+++ b/interactive.py
@@ -15,7 +15,7 @@
 import torch
 
 from fairseq import checkpoint_utils, options, tasks, utils
-from fairseq.data import transforms
+from fairseq.data import encoders
 
 
 Batch = namedtuple('Batch', 'ids src_tokens src_lengths')
@@ -103,8 +103,8 @@ def main(args):
     generator = task.build_generator(args)
 
     # Handle tokenization and BPE
-    tokenizer = transforms.build_tokenizer(args)
-    bpe = transforms.build_bpe(args)
+    tokenizer = encoders.build_tokenizer(args)
+    bpe = encoders.build_bpe(args)
 
     def encode_fn(x):
         if tokenizer is not None: