Merge pull request google#732 from google:mattdavidow-support-simple

PiperOrigin-RevId: 647752765

MaxText/layers/models.py

@@ -203,7 +203,7 @@ def get_decoder_layer(self):
  raise ValueError(f"Incorrect decoder_block name {self.config.decoder_block=}")
 
  def get_norm_layer(self):
- if self.config.decoder_block in ("default", "llama2", "mistral", "gemma"):
+ if self.config.decoder_block in ("default", "llama2", "mistral", "gemma", "simple"):
  return RMSNorm
  elif self.config.decoder_block == "gpt3":
  from layers import gpt3

MaxText/tests/simple_decoder_layer_test.py

@@ -0,0 +1,34 @@
+"""
+Copyright 2024 Google LLC
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+ https://www.apache.org/licenses/LICENSE-2.0
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+"""
+
+import unittest
+import pytest
+from train import main as train_main
+
+
+class SimpleDecoderLayerTest(unittest.TestCase):
+ @pytest.mark.tpu
+ def test_simple_decoder_layer(self):
+ train_main([
+ None,
+ "configs/base.yml",
+ r"base_output_directory=gs:https://runner-maxtext-logs",
+ "run_name=runner_simple_decoder_layer_test",
+ r"dataset_path=gs:https://maxtext-dataset",
+ "decoder_block=simple",
+ "enable_checkpointing=False",
+ "tokenizer_path=../assets/tokenizer.llama2",
+ ])
+
+if __name__ == "__main__":
+ unittest.main()