我在Google的colab上写了一本jupyter笔记本,以微调(用于文本分类)我只接受过阿拉伯语培训的BERT版本。培训开始时,我无法解决这个错误。
我关注了Google在github上给出的笔记本
建筑模型代码:
model_fn = model_fn_builder(
bert_config=modeling.BertConfig.from_json_file(CONFIG_FILE),
num_labels=len(label_list),
init_checkpoint=INIT_CHECKPOINT,
learning_rate=LEARNING_RATE,
num_train_steps=num_train_steps,
num_warmup_steps=num_warmup_steps,
use_tpu=True,
use_one_hot_embeddings=True
)
tpu_cluster_resolver = tf.contrib.cluster_resolver.TPUClusterResolver(TPU_ADDRESS)
run_config = tf.contrib.tpu.RunConfig(
cluster=tpu_cluster_resolver,
model_dir=OUTPUT_DIR,
save_checkpoints_steps=SAVE_CHECKPOINTS_STEPS,
tpu_config=tf.contrib.tpu.TPUConfig(
iterations_per_loop=ITERATIONS_PER_LOOP,
num_shards=NUM_TPU_CORES,
per_host_input_for_training=tf.contrib.tpu.InputPipelineConfig.PER_HOST_V2))
estimator = tf.contrib.tpu.TPUEstimator(
use_tpu=USE_TPU,
model_fn=model_fn,
config=run_config,
train_batch_size=TRAIN_BATCH_SIZE,
eval_batch_size=EVAL_BATCH_SIZE,
predict_batch_size=PREDICT_BATCH_SIZE,)
train_input_fn = input_fn_builder(
features=train_features,
seq_length=MAX_SEQ_LENGTH,
is_training=True,
drop_remainder=False)
#tf.reset_default_graph()
print(f'Beginning Training!')
current_time = datetime.now()
estimator.train(input_fn=train_input_fn, max_steps=TRAIN_STEPS)
print("Training took time ", datetime.now() - current_time)
错误代码:
/usr/local/lib/python3.6/dist-packages/tensorflow/python/tpu/tpu_sharding.py in _unshard_shape(self, shape)
214 (shape.as_list(), self._shard_dimension))
215 dims = shape.as_list()
--> 216 dims[self._shard_dimension] *= self._number_of_shards
217 return tensor_shape.as_shape(dims)
218
TypeError: unsupported operand type(s) for *=: 'NoneType' and 'int'
参数和其余代码在以下colab笔记本的共享副本中:colab_link
答案 0 :(得分:0)
在本部分中提及答案(即使在“评论”部分中也已回答),以使社区受益。
在函数drop_remainder
中将参数True
设置为input_fn_builder
可以解决此问题。
各个代码段如下所示:
train_input_fn = input_fn_builder(
features=train_features,
seq_length=MAX_SEQ_LENGTH,
is_training=True,
drop_remainder=False)