Amazon web services AWS SageMaker+；Tensorflow+；GPU

我试图提取ELMo嵌入，并在AWS SageMaker和TensorFlow上运行代码。代码在SageMaker的cpu实例上运行良好，但我想在GPU上运行它。以下是我执行的步骤和错误列表：

已启动AWS SageMaker实例：ml.p3.8xlarge-它有8个GPU

启动JuperterLab并为笔记本选择conda_tensorflow_p36

运行以下代码：

!！pip3安装tensorflow gpu==1.15

---

!pip3安装“tensorflow Hub”你有没有想过？你有没有想过？

url = "https://tfhub.dev/google/elmo/2"
embed = hub.Module(url)

def defineEmbeddings(start, end, extractions):    
    embeddings = embed(extractions[start:end],signature="default",as_dict=True)["default"]
    return embeddings
def scoreExtractions (embeddings):

    config = tf.compat.v1.ConfigProto(log_device_placement=True)
    config.gpu_options.allow_growth = True
    config.gpu_options.per_process_gpu_memory_fraction = 0.9
    with tf.compat.v1.Session(config=config) as sess:
        sess.run(tf.global_variables_initializer())
        sess.run(tf.tables_initializer())
        x = sess.run(embeddings)
    return x

with tf.device('/device:GPU:3'):
    a = tf.constant([1.0, 2.0, 3.0, 4.0, 5.0, 6.0], shape=[2, 3], name='a')
    b = tf.constant([1.0, 2.0, 3.0, 4.0, 5.0, 6.0], shape=[3, 2], name='b')
    c = tf.matmul(a, b)
with tf.Session() as sess:
    print(sess.run(c))