#!/bin/bash -v

DATA_DIR=/raid/data/nmt_data
python -m nmt.nmt \
    --src=en --tgt=de \
    --vocab_prefix=${DATA_DIR}/vocab.bpe.32000  \
    --train_prefix=${DATA_DIR}/train.tok.clean.bpe.32000 \
    --dev_prefix=${DATA_DIR}/newstest2014.tok.clean.bpe.32000  \
    --test_prefix=${DATA_DIR}/newstest2015.tok.clean.bpe.32000 \
    --out_dir=/raid/skyw/train_dir/nmt_model/test \
    --num_train_steps=12000 \
    --steps_per_stats=100 \
    --num_layers=2 \
    --num_units=128 \
    --dropout=0.2 \
    --metrics=bleu \
    --bpe_delimiter=@@ \
    --hparams_path nmt/standard_hparams/wmt16_en_de.json
# Job id 0
# Loading hparams from /raid/skyw/train_dir/nmt_model/test/hparams
# Loading standard hparams from nmt/standard_hparams/wmt16_en_de.json
# hparams:
  src=en
  tgt=de
  train_prefix=/raid/data/nmt_data/train.tok.clean.bpe.32000
  dev_prefix=/raid/data/nmt_data/newstest2014.tok.clean.bpe.32000
  test_prefix=/raid/data/nmt_data/newstest2015.tok.clean.bpe.32000
  out_dir=/raid/skyw/train_dir/nmt_model/test
# Vocab file /raid/data/nmt_data/vocab.bpe.32000.en exists
The first 3 vocab words [-e <unk>, <s>, </s>] are not [<unk>, <s>, </s>]
# Vocab file /raid/data/nmt_data/vocab.bpe.32000.de exists
The first 3 vocab words [-e <unk>, <s>, </s>] are not [<unk>, <s>, </s>]
  saving hparams to /raid/skyw/train_dir/nmt_model/test/hparams
  saving hparams to /raid/skyw/train_dir/nmt_model/test/best_bleu/hparams
  attention=normed_bahdanau
  attention_architecture=standard
  batch_size=128
  beam_width=10
  best_bleu=0
  best_bleu_dir=/raid/skyw/train_dir/nmt_model/test/best_bleu
  bpe_delimiter=@@
  colocate_gradients_with_ops=True
  decay_factor=0.5
  decay_steps=17000
  dev_prefix=/raid/data/nmt_data/newstest2014.tok.clean.bpe.32000
  dropout=0.2
  encoder_type=bi
  eos=</s>
  epoch_step=0
  forget_bias=1.0
  infer_batch_size=32
  init_weight=0.1
  learning_rate=1.0
  length_penalty_weight=0.0
  log_device_placement=False
  max_gradient_norm=5.0
  max_train=0
  metrics=[u'bleu']
  num_buckets=5
  num_gpus=1
  num_layers=4
  num_residual_layers=0
  num_train_steps=340000
  num_units=1024
  optimizer=sgd
  out_dir=/raid/skyw/train_dir/nmt_model/test
  pass_hidden_state=True
  random_seed=None
  residual=False
  share_vocab=False
  sos=<s>
  source_reverse=False
  src=en
  src_max_len=50
  src_max_len_infer=None
  src_vocab_file=/raid/skyw/train_dir/nmt_model/test/vocab.bpe.32000.en
  src_vocab_size=36551
  start_decay_step=170000
  steps_per_external_eval=None
  steps_per_stats=100
  test_prefix=/raid/data/nmt_data/newstest2015.tok.clean.bpe.32000
  tgt=de
  tgt_max_len=50
  tgt_max_len_infer=None
  tgt_vocab_file=/raid/skyw/train_dir/nmt_model/test/vocab.bpe.32000.de
  tgt_vocab_size=36551
  time_major=True
  train_prefix=/raid/data/nmt_data/train.tok.clean.bpe.32000
  unit_type=lstm
  vocab_prefix=/raid/data/nmt_data/vocab.bpe.32000
# creating train graph ...
  num_bi_layers = 2, num_bi_residual_layers=0
  cell 0  LSTM, forget_bias=1  DropoutWrapper, dropout=0.2   DeviceWrapper, device=/gpu:0
  cell 1  LSTM, forget_bias=1  DropoutWrapper, dropout=0.2   DeviceWrapper, device=/gpu:0
  cell 0  LSTM, forget_bias=1  DropoutWrapper, dropout=0.2   DeviceWrapper, device=/gpu:0
  cell 1  LSTM, forget_bias=1  DropoutWrapper, dropout=0.2   DeviceWrapper, device=/gpu:0
  cell 0  LSTM, forget_bias=1  DropoutWrapper, dropout=0.2   DeviceWrapper, device=/gpu:0
  cell 1  LSTM, forget_bias=1  DropoutWrapper, dropout=0.2   DeviceWrapper, device=/gpu:0
  cell 2  LSTM, forget_bias=1  DropoutWrapper, dropout=0.2   DeviceWrapper, device=/gpu:0
  cell 3  LSTM, forget_bias=1  DropoutWrapper, dropout=0.2   DeviceWrapper, device=/gpu:0
  start_decay_step=170000, learning_rate=1, decay_steps 17000,decay_factor 0.5
# Trainable variables
  embeddings/encoder/embedding_encoder:0, (36551, 1024), 
  embeddings/decoder/embedding_decoder:0, (36551, 1024), 
  dynamic_seq2seq/encoder/bidirectional_rnn/fw/multi_rnn_cell/cell_0/basic_lstm_cell/kernel:0, (2048, 4096), /device:GPU:0
  dynamic_seq2seq/encoder/bidirectional_rnn/fw/multi_rnn_cell/cell_0/basic_lstm_cell/bias:0, (4096,), /device:GPU:0
  dynamic_seq2seq/encoder/bidirectional_rnn/fw/multi_rnn_cell/cell_1/basic_lstm_cell/kernel:0, (2048, 4096), /device:GPU:0
  dynamic_seq2seq/encoder/bidirectional_rnn/fw/multi_rnn_cell/cell_1/basic_lstm_cell/bias:0, (4096,), /device:GPU:0
  dynamic_seq2seq/encoder/bidirectional_rnn/bw/multi_rnn_cell/cell_0/basic_lstm_cell/kernel:0, (2048, 4096), /device:GPU:0
  dynamic_seq2seq/encoder/bidirectional_rnn/bw/multi_rnn_cell/cell_0/basic_lstm_cell/bias:0, (4096,), /device:GPU:0
  dynamic_seq2seq/encoder/bidirectional_rnn/bw/multi_rnn_cell/cell_1/basic_lstm_cell/kernel:0, (2048, 4096), /device:GPU:0
  dynamic_seq2seq/encoder/bidirectional_rnn/bw/multi_rnn_cell/cell_1/basic_lstm_cell/bias:0, (4096,), /device:GPU:0
  dynamic_seq2seq/decoder/memory_layer/kernel:0, (2048, 1024), 
  dynamic_seq2seq/decoder/attention/multi_rnn_cell/cell_0/basic_lstm_cell/kernel:0, (3072, 4096), /device:GPU:0
  dynamic_seq2seq/decoder/attention/multi_rnn_cell/cell_0/basic_lstm_cell/bias:0, (4096,), /device:GPU:0
  dynamic_seq2seq/decoder/attention/multi_rnn_cell/cell_1/basic_lstm_cell/kernel:0, (2048, 4096), /device:GPU:0
  dynamic_seq2seq/decoder/attention/multi_rnn_cell/cell_1/basic_lstm_cell/bias:0, (4096,), /device:GPU:0
  dynamic_seq2seq/decoder/attention/multi_rnn_cell/cell_2/basic_lstm_cell/kernel:0, (2048, 4096), /device:GPU:0
  dynamic_seq2seq/decoder/attention/multi_rnn_cell/cell_2/basic_lstm_cell/bias:0, (4096,), /device:GPU:0
  dynamic_seq2seq/decoder/attention/multi_rnn_cell/cell_3/basic_lstm_cell/kernel:0, (2048, 4096), /device:GPU:0
  dynamic_seq2seq/decoder/attention/multi_rnn_cell/cell_3/basic_lstm_cell/bias:0, (4096,), /device:GPU:0
  dynamic_seq2seq/decoder/attention/bahdanau_attention/query_layer/kernel:0, (1024, 1024), /device:GPU:0
  dynamic_seq2seq/decoder/attention/bahdanau_attention/attention_v:0, (1024,), /device:GPU:0
  dynamic_seq2seq/decoder/attention/bahdanau_attention/attention_g:0, (), /device:GPU:0
  dynamic_seq2seq/decoder/attention/bahdanau_attention/attention_b:0, (1024,), /device:GPU:0
  dynamic_seq2seq/decoder/attention/attention_layer/kernel:0, (3072, 1024), /device:GPU:0
  dynamic_seq2seq/decoder/output_projection/kernel:0, (1024, 36551), /device:GPU:0
# creating eval graph ...
  num_bi_layers = 2, num_bi_residual_layers=0
  cell 0  LSTM, forget_bias=1  DeviceWrapper, device=/gpu:0
  cell 1  LSTM, forget_bias=1  DeviceWrapper, device=/gpu:0
  cell 0  LSTM, forget_bias=1  DeviceWrapper, device=/gpu:0
  cell 1  LSTM, forget_bias=1  DeviceWrapper, device=/gpu:0
  cell 0  LSTM, forget_bias=1  DeviceWrapper, device=/gpu:0
  cell 1  LSTM, forget_bias=1  DeviceWrapper, device=/gpu:0
  cell 2  LSTM, forget_bias=1  DeviceWrapper, device=/gpu:0
  cell 3  LSTM, forget_bias=1  DeviceWrapper, device=/gpu:0
  start_decay_step=170000, learning_rate=1, decay_steps 17000,decay_factor 0.5
# Trainable variables
  embeddings/encoder/embedding_encoder:0, (36551, 1024), 
  embeddings/decoder/embedding_decoder:0, (36551, 1024), 
  dynamic_seq2seq/encoder/bidirectional_rnn/fw/multi_rnn_cell/cell_0/basic_lstm_cell/kernel:0, (2048, 4096), /device:GPU:0
  dynamic_seq2seq/encoder/bidirectional_rnn/fw/multi_rnn_cell/cell_0/basic_lstm_cell/bias:0, (4096,), /device:GPU:0
  dynamic_seq2seq/encoder/bidirectional_rnn/fw/multi_rnn_cell/cell_1/basic_lstm_cell/kernel:0, (2048, 4096), /device:GPU:0
  dynamic_seq2seq/encoder/bidirectional_rnn/fw/multi_rnn_cell/cell_1/basic_lstm_cell/bias:0, (4096,), /device:GPU:0
  dynamic_seq2seq/encoder/bidirectional_rnn/bw/multi_rnn_cell/cell_0/basic_lstm_cell/kernel:0, (2048, 4096), /device:GPU:0
  dynamic_seq2seq/encoder/bidirectional_rnn/bw/multi_rnn_cell/cell_0/basic_lstm_cell/bias:0, (4096,), /device:GPU:0
  dynamic_seq2seq/encoder/bidirectional_rnn/bw/multi_rnn_cell/cell_1/basic_lstm_cell/kernel:0, (2048, 4096), /device:GPU:0
  dynamic_seq2seq/encoder/bidirectional_rnn/bw/multi_rnn_cell/cell_1/basic_lstm_cell/bias:0, (4096,), /device:GPU:0
  dynamic_seq2seq/decoder/memory_layer/kernel:0, (2048, 1024), 
  dynamic_seq2seq/decoder/attention/multi_rnn_cell/cell_0/basic_lstm_cell/kernel:0, (3072, 4096), /device:GPU:0
  dynamic_seq2seq/decoder/attention/multi_rnn_cell/cell_0/basic_lstm_cell/bias:0, (4096,), /device:GPU:0
  dynamic_seq2seq/decoder/attention/multi_rnn_cell/cell_1/basic_lstm_cell/kernel:0, (2048, 4096), /device:GPU:0
  dynamic_seq2seq/decoder/attention/multi_rnn_cell/cell_1/basic_lstm_cell/bias:0, (4096,), /device:GPU:0
  dynamic_seq2seq/decoder/attention/multi_rnn_cell/cell_2/basic_lstm_cell/kernel:0, (2048, 4096), /device:GPU:0
  dynamic_seq2seq/decoder/attention/multi_rnn_cell/cell_2/basic_lstm_cell/bias:0, (4096,), /device:GPU:0
  dynamic_seq2seq/decoder/attention/multi_rnn_cell/cell_3/basic_lstm_cell/kernel:0, (2048, 4096), /device:GPU:0
  dynamic_seq2seq/decoder/attention/multi_rnn_cell/cell_3/basic_lstm_cell/bias:0, (4096,), /device:GPU:0
  dynamic_seq2seq/decoder/attention/bahdanau_attention/query_layer/kernel:0, (1024, 1024), /device:GPU:0
  dynamic_seq2seq/decoder/attention/bahdanau_attention/attention_v:0, (1024,), /device:GPU:0
  dynamic_seq2seq/decoder/attention/bahdanau_attention/attention_g:0, (), /device:GPU:0
  dynamic_seq2seq/decoder/attention/bahdanau_attention/attention_b:0, (1024,), /device:GPU:0
  dynamic_seq2seq/decoder/attention/attention_layer/kernel:0, (3072, 1024), /device:GPU:0
  dynamic_seq2seq/decoder/output_projection/kernel:0, (1024, 36551), /device:GPU:0
# creating infer graph ...
  num_bi_layers = 2, num_bi_residual_layers=0
  cell 0  LSTM, forget_bias=1  DeviceWrapper, device=/gpu:0
  cell 1  LSTM, forget_bias=1  DeviceWrapper, device=/gpu:0
  cell 0  LSTM, forget_bias=1  DeviceWrapper, device=/gpu:0
  cell 1  LSTM, forget_bias=1  DeviceWrapper, device=/gpu:0
  cell 0  LSTM, forget_bias=1  DeviceWrapper, device=/gpu:0
  cell 1  LSTM, forget_bias=1  DeviceWrapper, device=/gpu:0
  cell 2  LSTM, forget_bias=1  DeviceWrapper, device=/gpu:0
  cell 3  LSTM, forget_bias=1  DeviceWrapper, device=/gpu:0
  start_decay_step=170000, learning_rate=1, decay_steps 17000,decay_factor 0.5
# Trainable variables
  embeddings/encoder/embedding_encoder:0, (36551, 1024), 
  embeddings/decoder/embedding_decoder:0, (36551, 1024), 
  dynamic_seq2seq/encoder/bidirectional_rnn/fw/multi_rnn_cell/cell_0/basic_lstm_cell/kernel:0, (2048, 4096), /device:GPU:0
  dynamic_seq2seq/encoder/bidirectional_rnn/fw/multi_rnn_cell/cell_0/basic_lstm_cell/bias:0, (4096,), /device:GPU:0
  dynamic_seq2seq/encoder/bidirectional_rnn/fw/multi_rnn_cell/cell_1/basic_lstm_cell/kernel:0, (2048, 4096), /device:GPU:0
  dynamic_seq2seq/encoder/bidirectional_rnn/fw/multi_rnn_cell/cell_1/basic_lstm_cell/bias:0, (4096,), /device:GPU:0
  dynamic_seq2seq/encoder/bidirectional_rnn/bw/multi_rnn_cell/cell_0/basic_lstm_cell/kernel:0, (2048, 4096), /device:GPU:0
  dynamic_seq2seq/encoder/bidirectional_rnn/bw/multi_rnn_cell/cell_0/basic_lstm_cell/bias:0, (4096,), /device:GPU:0
  dynamic_seq2seq/encoder/bidirectional_rnn/bw/multi_rnn_cell/cell_1/basic_lstm_cell/kernel:0, (2048, 4096), /device:GPU:0
  dynamic_seq2seq/encoder/bidirectional_rnn/bw/multi_rnn_cell/cell_1/basic_lstm_cell/bias:0, (4096,), /device:GPU:0
  dynamic_seq2seq/decoder/memory_layer/kernel:0, (2048, 1024), 
  dynamic_seq2seq/decoder/attention/multi_rnn_cell/cell_0/basic_lstm_cell/kernel:0, (3072, 4096), /device:GPU:0
  dynamic_seq2seq/decoder/attention/multi_rnn_cell/cell_0/basic_lstm_cell/bias:0, (4096,), /device:GPU:0
  dynamic_seq2seq/decoder/attention/multi_rnn_cell/cell_1/basic_lstm_cell/kernel:0, (2048, 4096), /device:GPU:0
  dynamic_seq2seq/decoder/attention/multi_rnn_cell/cell_1/basic_lstm_cell/bias:0, (4096,), /device:GPU:0
  dynamic_seq2seq/decoder/attention/multi_rnn_cell/cell_2/basic_lstm_cell/kernel:0, (2048, 4096), /device:GPU:0
  dynamic_seq2seq/decoder/attention/multi_rnn_cell/cell_2/basic_lstm_cell/bias:0, (4096,), /device:GPU:0
  dynamic_seq2seq/decoder/attention/multi_rnn_cell/cell_3/basic_lstm_cell/kernel:0, (2048, 4096), /device:GPU:0
  dynamic_seq2seq/decoder/attention/multi_rnn_cell/cell_3/basic_lstm_cell/bias:0, (4096,), /device:GPU:0
  dynamic_seq2seq/decoder/attention/bahdanau_attention/query_layer/kernel:0, (1024, 1024), /device:GPU:0
  dynamic_seq2seq/decoder/attention/bahdanau_attention/attention_v:0, (1024,), /device:GPU:0
  dynamic_seq2seq/decoder/attention/bahdanau_attention/attention_g:0, (), /device:GPU:0
  dynamic_seq2seq/decoder/attention/bahdanau_attention/attention_b:0, (1024,), /device:GPU:0
  dynamic_seq2seq/decoder/attention/attention_layer/kernel:0, (3072, 1024), /device:GPU:0
  dynamic_seq2seq/decoder/output_projection/kernel:0, (1024, 36551), 
# log_file=/raid/skyw/train_dir/nmt_model/test/log_1500309031
2017-07-17 09:30:31.459101: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use SSE4.1 instructions, but these are available on your machine and could speed up CPU computations.
2017-07-17 09:30:31.459116: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use SSE4.2 instructions, but these are available on your machine and could speed up CPU computations.
2017-07-17 09:30:31.459119: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use AVX instructions, but these are available on your machine and could speed up CPU computations.
2017-07-17 09:30:31.459122: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use AVX2 instructions, but these are available on your machine and could speed up CPU computations.
2017-07-17 09:30:31.459124: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use FMA instructions, but these are available on your machine and could speed up CPU computations.
2017-07-17 09:30:31.923512: I tensorflow/core/common_runtime/gpu/gpu_device.cc:940] Found device 0 with properties: 
name: Tesla P100-DGXS-16GB
major: 6 minor: 0 memoryClockRate (GHz) 1.4805
pciBusID 0000:07:00.0
Total memory: 15.89GiB
Free memory: 15.60GiB
2017-07-17 09:30:31.923534: I tensorflow/core/common_runtime/gpu/gpu_device.cc:961] DMA: 0 
2017-07-17 09:30:31.923538: I tensorflow/core/common_runtime/gpu/gpu_device.cc:971] 0:   Y 
2017-07-17 09:30:31.923544: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1030] Creating TensorFlow device (/gpu:0) -> (device: 0, name: Tesla P100-DGXS-16GB, pci bus id: 0000:07:00.0)
2017-07-17 09:30:31.925513: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1030] Creating TensorFlow device (/gpu:0) -> (device: 0, name: Tesla P100-DGXS-16GB, pci bus id: 0000:07:00.0)
2017-07-17 09:30:31.926086: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1030] Creating TensorFlow device (/gpu:0) -> (device: 0, name: Tesla P100-DGXS-16GB, pci bus id: 0000:07:00.0)
  created train model with fresh parameters, time 0.00s.
2017-07-17 09:30:32.616890: W tensorflow/core/framework/op_kernel.cc:1158] Failed precondition: HashTable has different value for same key. Key <s> has 1 and trying to add value 4
2017-07-17 09:30:32.619153: W tensorflow/core/framework/op_kernel.cc:1158] Failed precondition: HashTable has different value for same key. Key <s> has 1 and trying to add value 4
Traceback (most recent call last):
  File "/usr/lib/python2.7/runpy.py", line 162, in _run_module_as_main
    "__main__", fname, loader, pkg_name)
  File "/usr/lib/python2.7/runpy.py", line 72, in _run_code
    exec code in run_globals
  File "/raid/skyw/google-git/nmt/nmt/nmt.py", line 479, in <module>
    tf.app.run(main=main, argv=[sys.argv[0]] + unparsed)
  File "/raid/skyw/venv/tensorflow-pip-py27/local/lib/python2.7/site-packages/tensorflow/python/platform/app.py", line 48, in run
    _sys.exit(main(_sys.argv[:1] + flags_passthrough))
  File "/raid/skyw/google-git/nmt/nmt/nmt.py", line 292, in main
    train.train(hparams)
  File "nmt/train.py", line 212, in train
    train_model, model_dir, train_sess, hparams.out_dir, "train")
  File "nmt/model_helper.py", line 218, in create_or_load_model
    session.run(tf.tables_initializer())
  File "/raid/skyw/venv/tensorflow-pip-py27/local/lib/python2.7/site-packages/tensorflow/python/client/session.py", line 789, in run
    run_metadata_ptr)
  File "/raid/skyw/venv/tensorflow-pip-py27/local/lib/python2.7/site-packages/tensorflow/python/client/session.py", line 997, in _run
    feed_dict_string, options, run_metadata)
  File "/raid/skyw/venv/tensorflow-pip-py27/local/lib/python2.7/site-packages/tensorflow/python/client/session.py", line 1132, in _do_run
    target_list, options, run_metadata)
  File "/raid/skyw/venv/tensorflow-pip-py27/local/lib/python2.7/site-packages/tensorflow/python/client/session.py", line 1152, in _do_call
    raise type(e)(node_def, op, message)
tensorflow.python.framework.errors_impl.FailedPreconditionError: HashTable has different value for same key. Key <s> has 1 and trying to add value 4
	 [[Node: string_to_index_1/hash_table/table_init = InitializeTableFromTextFileV2[delimiter="\t", key_index=-2, value_index=-1, vocab_size=-1, _device="/job:localhost/replica:0/task:0/cpu:0"](string_to_index_1/hash_table, string_to_index_1/hash_table/table_init/asset_filepath)]]
	 [[Node: string_to_index_1/hash_table/table_init/_14 = _Recv[client_terminated=false, recv_device="/job:localhost/replica:0/task:0/gpu:0", send_device="/job:localhost/replica:0/task:0/cpu:0", send_device_incarnation=1, tensor_name="edge_8_string_to_index_1/hash_table/table_init", tensor_type=DT_FLOAT, _device="/job:localhost/replica:0/task:0/gpu:0"]()]]

Caused by op u'string_to_index_1/hash_table/table_init', defined at:
  File "/usr/lib/python2.7/runpy.py", line 162, in _run_module_as_main
    "__main__", fname, loader, pkg_name)
  File "/usr/lib/python2.7/runpy.py", line 72, in _run_code
    exec code in run_globals
  File "/raid/skyw/google-git/nmt/nmt/nmt.py", line 479, in <module>
    tf.app.run(main=main, argv=[sys.argv[0]] + unparsed)
  File "/raid/skyw/venv/tensorflow-pip-py27/local/lib/python2.7/site-packages/tensorflow/python/platform/app.py", line 48, in run
    _sys.exit(main(_sys.argv[:1] + flags_passthrough))
  File "/raid/skyw/google-git/nmt/nmt/nmt.py", line 292, in main
    train.train(hparams)
  File "nmt/train.py", line 159, in train
    create_train_model(model_creator, hparams, scope))
  File "nmt/train.py", line 58, in create_train_model
    tgt_vocab_file, default_value=vocab_utils.UNK_ID)
  File "/raid/skyw/venv/tensorflow-pip-py27/local/lib/python2.7/site-packages/tensorflow/python/ops/lookup_ops.py", line 947, in index_table_from_file
    init, default_value, shared_name=shared_name, name=hash_table_scope)
  File "/raid/skyw/venv/tensorflow-pip-py27/local/lib/python2.7/site-packages/tensorflow/python/ops/lookup_ops.py", line 275, in __init__
    super(HashTable, self).__init__(table_ref, default_value, initializer)
  File "/raid/skyw/venv/tensorflow-pip-py27/local/lib/python2.7/site-packages/tensorflow/python/ops/lookup_ops.py", line 161, in __init__
    self._init = initializer.initialize(self)
  File "/raid/skyw/venv/tensorflow-pip-py27/local/lib/python2.7/site-packages/tensorflow/python/ops/lookup_ops.py", line 519, in initialize
    name=scope)
  File "/raid/skyw/venv/tensorflow-pip-py27/local/lib/python2.7/site-packages/tensorflow/python/ops/gen_lookup_ops.py", line 187, in _initialize_table_from_text_file_v2
    name=name)
  File "/raid/skyw/venv/tensorflow-pip-py27/local/lib/python2.7/site-packages/tensorflow/python/framework/op_def_library.py", line 767, in apply_op
    op_def=op_def)
  File "/raid/skyw/venv/tensorflow-pip-py27/local/lib/python2.7/site-packages/tensorflow/python/framework/ops.py", line 2506, in create_op
    original_op=self._default_original_op, op_def=op_def)
  File "/raid/skyw/venv/tensorflow-pip-py27/local/lib/python2.7/site-packages/tensorflow/python/framework/ops.py", line 1269, in __init__
    self._traceback = _extract_stack()

FailedPreconditionError (see above for traceback): HashTable has different value for same key. Key <s> has 1 and trying to add value 4
	 [[Node: string_to_index_1/hash_table/table_init = InitializeTableFromTextFileV2[delimiter="\t", key_index=-2, value_index=-1, vocab_size=-1, _device="/job:localhost/replica:0/task:0/cpu:0"](string_to_index_1/hash_table, string_to_index_1/hash_table/table_init/asset_filepath)]]
	 [[Node: string_to_index_1/hash_table/table_init/_14 = _Recv[client_terminated=false, recv_device="/job:localhost/replica:0/task:0/gpu:0", send_device="/job:localhost/replica:0/task:0/cpu:0", send_device_incarnation=1, tensor_name="edge_8_string_to_index_1/hash_table/table_init", tensor_type=DT_FLOAT, _device="/job:localhost/replica:0/task:0/gpu:0"]()]]