Merge branch 'hn-tokenizer-imports' into 'main'

Fix some more imports from the refactor

See merge request ADLR/megatron-lm!1325

Author: jaredcasper

megatron/training/__init__.py

@@ -14,7 +14,7 @@
 from .global_vars import get_adlr_autoresume
 from .global_vars import get_timers
 from .initialize  import initialize_megatron
-from .training import pretrain, get_model
+from .training import pretrain, get_model, get_train_valid_test_num_samples
 
 from .utils import (print_rank_0,
                     is_last_rank,

tools/retro/preprocess_data.py

@@ -36,7 +36,7 @@
     get_config_path,
     get_gpt_data_dir,
 )
-from megatron.tokenizer.tokenizer import (
+from megatron.training.tokenizer.tokenizer import (
     _BertWordPieceTokenizer,
     _GPT2BPETokenizer,
     _GPTSentencePieceTokenizer,