Created by: Melonzhou

paddle-cpu-1.6.0 组网部分： `def create_net(slots, is_inference=False): """create net""" text_a = slots[0] text_a_mask = slots[1] text_a_lens = slots[2] label_a = slots[3] label_a_mask = slots[4] label_a_lens = slots[5]

unpad_words_emb  = fluid.layers.sequence_unpad(text_a, length=text_a_lens)
unpad_labels = fluid.layers.sequence_unpad(label_a, length=label_a_lens)


word_emb_dim = 768
grnn_hidden_dim = 768
emb_lr = 5
crf_lr = 0.2
bigru_num = 2
init_bound = 0.1
vocab_size = 17964
num_labels = 7

def _bigru_layer(input_feature):
    """
    define the bidirectional gru layer
    """
    pre_gru = fluid.layers.fc(
        input=input_feature,
        size=grnn_hidden_dim * 3,
        param_attr=fluid.ParamAttr(
            initializer=fluid.initializer.Uniform(
                low=-init_bound, high=init_bound),
            regularizer=fluid.regularizer.L2DecayRegularizer(
                regularization_coeff=1e-4)))
    gru = fluid.layers.dynamic_gru(
        input=pre_gru,
        size=grnn_hidden_dim,
        param_attr=fluid.ParamAttr(
            initializer=fluid.initializer.Uniform(
                low=-init_bound, high=init_bound),
            regularizer=fluid.regularizer.L2DecayRegularizer(
                regularization_coeff=1e-4)))

    pre_gru_r = fluid.layers.fc(
        input=input_feature,
        size=grnn_hidden_dim * 3,
        param_attr=fluid.ParamAttr(
            initializer=fluid.initializer.Uniform(
                low=-init_bound, high=init_bound),
            regularizer=fluid.regularizer.L2DecayRegularizer(
                regularization_coeff=1e-4)))
    gru_r = fluid.layers.dynamic_gru(
        input=pre_gru_r,
        size=grnn_hidden_dim,
        is_reverse=True,
        param_attr=fluid.ParamAttr(
            initializer=fluid.initializer.Uniform(
                low=-init_bound, high=init_bound),
            regularizer=fluid.regularizer.L2DecayRegularizer(
                regularization_coeff=1e-4)))

    bi_merge = fluid.layers.concat(input=[gru, gru_r], axis=1)
    return bi_merge


word_embedding = fluid.layers.embedding(
    input=unpad_words_emb,
    size=[vocab_size, word_emb_dim],
    dtype='float32',
    param_attr=fluid.ParamAttr(
        learning_rate=emb_lr,
        name="word_emb",
        initializer=fluid.initializer.Uniform(
            low=-init_bound, high=init_bound)))

input_feature = word_embedding
for i in range(bigru_num):
    bigru_output = _bigru_layer(input_feature)
    input_feature = bigru_output

emission = fluid.layers.fc(
    size=num_labels,
    input=bigru_output,
    param_attr=fluid.ParamAttr(
        initializer=fluid.initializer.Uniform(
            low=-init_bound, high=init_bound),
        regularizer=fluid.regularizer.L2DecayRegularizer(
            regularization_coeff=1e-4)))
crf_cost = fluid.layers.linear_chain_crf(
    input=emission,
    label=unpad_labels,
    param_attr=fluid.ParamAttr(
        name='crfw',
        learning_rate=crf_lr))
crf_decode = fluid.layers.crf_decoding(
    input=emission, param_attr=fluid.ParamAttr(name='crfw'))

if is_inference:
    feed_targets_name = [text_a.name, text_a_lens.name]
    output_targets_name = [crf_decode]
    return feed_targets_name, output_targets_name

avg_cost = fluid.layers.mean(x=crf_cost)
graph_vars = collections.OrderedDict()
graph_vars["loss"] = avg_cost
graph_vars["sequence_label_infer"] = crf_decode
graph_vars["label"] = unpad_labels

(precision, recall, f1_score, num_infer_chunks, num_label_chunks, num_correct_chunks) = fluid.layers.chunk_eval( input=graph_vars["sequence_label_infer"], label=graph_vars["label"], chunk_scheme="plain", num_chunk_types=51)

        graph_vars["precision"] = precision
        graph_vars["recall"] = recall
        graph_vars["f1_score"] = f1_score
        graph_vars["num_infer_chunks"] = num_infer_chunks
        graph_vars["num_label_chunks"] = num_label_chunks
        graph_vars["num_correct_chunks"] = num_correct_chunks

`

数据及label_map.json 报错信息

Error Message Summary:

PaddleCheckError: Expected static_cast<size_t>(*std::max_element(lbl, lbl + seq_length)) < tag_num, but received static_cast<size_t>(*std::max_element(lbl, lbl + seq_length)):22 >= tag_num:7. An invalid tag label that execesses the largest tag number. at [/home/teamcity/work/ef54dc8a5b211854/paddle/fluid/operators/linear_chain_crf_op.h:207] [operator < linear_chain_crf > error]

没理解这个22和7是从哪里来的