Created by: Akeepers

paddle version: 1.6.3 python 2.7 错误信息：

  "The following exception is not an EOF exception.")
Traceback (most recent call last):
  File "run_seq2seq.py", line 236, in <module>
    main(args)
  File "run_seq2seq.py", line 207, in main
    train(args)
  File "run_seq2seq.py", line 106, in train
    preds_res, loss_val = exe.run(program=prog, feed=data, fetch_list=[preds_res, loss])
  File "/home/yangpan/anaconda3/envs/paddle-py2.7/lib/python2.7/site-packages/paddle/fluid/executor.py", line 780, in run
    six.reraise(*sys.exc_info())
  File "/home/yangpan/anaconda3/envs/paddle-py2.7/lib/python2.7/site-packages/paddle/fluid/executor.py", line 775, in run
    use_program_cache=use_program_cache)
  File "/home/yangpan/anaconda3/envs/paddle-py2.7/lib/python2.7/site-packages/paddle/fluid/executor.py", line 834, in _run_impl
    return_numpy=return_numpy)  File "/home/yangpan/anaconda3/envs/paddle-py2.7/lib/python2.7/site-packages/paddle/fluid/executor.py", line 674, in _run_parallel
    tensors = exe.run(fetch_var_names)._move_to_list()
paddle.fluid.core_avx.EnforceNotMet: 

--------------------------------------------
C++ Call Stacks (More useful to developers):
--------------------------------------------
0   std::string paddle::platform::GetTraceBackString<std::string const&>(std::string const&, char const*, int)
1   paddle::platform::EnforceNotMet::EnforceNotMet(std::string const&, char const*, int)
2   paddle::framework::LoDTensor::MergeLoDTensor(std::vector<paddle::framework::LoDTensor const*, std::allocator<paddle::framework::LoDTensor const*> > const&, paddle::platform::Place)
3   paddle::framework::details::FetchOpHandle::WaitAndMergeCPUTensors() const
4   paddle::framework::details::FetchOpHandle::RunImpl()
5   paddle::framework::details::FastThreadedSSAGraphExecutor::RunOpSync(paddle::framework::details::OpHandleBase*)
6   paddle::framework::details::FastThreadedSSAGraphExecutor::RunOp(paddle::framework::details::OpHandleBase*, std::shared_ptr<paddle::framework::BlockingQueue<unsigned long> > const&, unsigned long*)
7   std::_Function_handler<std::unique_ptr<std::__future_base::_Result_base, std::__future_base::_Result_base::_Deleter> (), std::__future_base::_Task_setter<std::unique_ptr<std::__future_base::_Result<void>, std::__future_base::_Result_base::_Deleter>, void> >::_M_invoke(std::_Any_data const&)
8   std::__future_base::_State_base::_M_do_set(std::function<std::unique_ptr<std::__future_base::_Result_base, std::__future_base::_Result_base::_Deleter> ()>&, bool&)
9   ThreadPool::ThreadPool(unsigned long)::{lambda()#1}::operator()() const

----------------------
Error Message Summary:
----------------------
Error: Paddle internal Check failed. (Please help us create a new issue, here we need to find the developer to add a user friendly error message)
  [Hint: Expected framework::product(new_dim) / new_dim[0] == framework::product(t->dims()) / t->dims()[0], but received framework::product(new_dim) / new_dim[0]:50 != framework::product(t->dims()) / t->dims()[0]:63.] at (/paddle/paddle/fluid/framework/lod_tensor.cc:363)

报错部分的paddle代码： 报错部分： preds_res, loss_val = exe.run(program=prog, feed=data, fetch_list=[preds_res, loss]) Note: 如果将fetch_list中的preds_res去除，代码运行正常

 for current_epoch in range(args.epoch):
        step = 0
        for data in loader():
            step += 1
            preds_res, loss_val = exe.run(program=prog, feed=data, fetch_list=[preds_res, loss])
            loss_val = np.mean(loss_val)
            np_preds_res = np.array(preds_res)

            current_example, current_epoch = reader.get_train_progress()
            print('epoch: %d, batch_id: %d/%d, loss: %f' %
                  (current_epoch, step, num_examples, loss_val))

            if step % args.save_steps == 0:
                save_path = os.path.join(args.checkpoints,
                                         "step_" + str(step))
                fluid.io.save_persistables(exe, save_path, train_prog)
                print("step: %d, save model in %s" % (step, save_path))
        # except fluid.core.EOFException:
        # save_path = os.path.join(args.checkpoints,
        #                          "step_" + str(step))
        # fluid.io.save_persistables(exe, save_path, train_prog)
        # print("step: %d, save model in %s" % (step, save_path))
        # loader.reset()

program定义：

    train_prog = fluid.Program()
    startup_prog = fluid.Program()
    with fluid.program_guard(train_prog, startup_prog):
        with fluid.unique_name.guard():
            # For training:
            inputs, loader = data_func(args, is_train=True)
            logits = model_func(args, inputs, ernie_config, is_train=True)
            loss, preds_res = loss_func(logits, inputs[-1], inputs[-2])
            optimizer = optimizer_func()
            optimizer.minimize(loss)

loss func定义：

def loss_func(logits, label, trg_sequence_length):
    probs = layers.softmax(logits)
    res = layers.argmax(probs, axis=-1)
    loss = layers.cross_entropy(input=probs, label=label)
    trg_mask = layers.sequence_mask(
        trg_sequence_length, maxlen=layers.shape(logits)[1], dtype="float32")
    avg_cost = layers.reduce_sum(loss * trg_mask) / layers.reduce_sum(trg_mask)
    return avg_cost, res