deeprnn / image_captioning Goto Github PK
View Code? Open in Web Editor NEWTensorflow implementation of "Show, Attend and Tell: Neural Image Caption Generation with Visual Attention"
License: MIT License
Tensorflow implementation of "Show, Attend and Tell: Neural Image Caption Generation with Visual Attention"
License: MIT License
Hi all,
I have adapted this repo into python3-compatible version. Please refer to here for the code. I did not pull request as the new version seems not compatible with old python 2 (mainly as different data encoding). Hope it helps!
If I just put some of the training data into the /images fold, there is always an error shown below.
Training the model...
epoch: 0%| | 0/100 [00:00<?, ?it/s]
Traceback (most recent call last): | 0/11290 [00:00<?, ?it/s]
File "main.py", line 72, in
tf.app.run()
File "/anaconda3/lib/python3.7/site-packages/tensorflow/python/platform/app.py", line 125, in run
_sys.exit(main(argv))
File "main.py", line 53, in main
model.train(sess, data)
File "/Users/brian/Downloads/show-attend-and-tell-master/base_model.py", line 50, in train
images = self.image_loader.load_images(image_files)
File "/Users/brian/Downloads/show-attend-and-tell-master/utils/misc.py", line 34, in load_images
images.append(self.load_image(image_file))
File "/Users/brian/Downloads/show-attend-and-tell-master/utils/misc.py", line 18, in load_image
temp = image.swapaxes(0, 2)
AttributeError: 'NoneType' object has no attribute 'swapaxes'
I found that in the paper, the formula of MLP attention is usually desribed as below:
where vi is i-th feature map,ht is the output of lstm.
But in the code, the implementation goes like this:
def attend(self, contexts, output):
""" Attention Mechanism. """
config = self.config
reshaped_contexts = tf.reshape(contexts, [-1, self.dim_ctx])
reshaped_contexts = self.nn.dropout(reshaped_contexts)
output = self.nn.dropout(output)
if config.num_attend_layers == 1:
# use 1 fc layer to attend
logits1 = self.nn.dense(reshaped_contexts,
units = 1,
activation = None,
use_bias = False,
name = 'fc_a')
logits1 = tf.reshape(logits1, [-1, self.num_ctx])
logits2 = self.nn.dense(output,
units = self.num_ctx,
activation = None,
use_bias = False,
name = 'fc_b')
logits = logits1 + logits2
else:
# use 2 fc layers to attend
temp1 = self.nn.dense(reshaped_contexts,
units = config.dim_attend_layer,
activation = tf.tanh,
name = 'fc_1a')
temp2 = self.nn.dense(output,
units = config.dim_attend_layer,
activation = tf.tanh,
name = 'fc_1b')
temp2 = tf.tile(tf.expand_dims(temp2, 1), [1, self.num_ctx, 1])
temp2 = tf.reshape(temp2, [-1, config.dim_attend_layer])
temp = temp1 + temp2
temp = self.nn.dropout(temp)
logits = self.nn.dense(temp,
units = 1,
activation = None,
use_bias = False,
name = 'fc_2')
logits = tf.reshape(logits, [-1, self.num_ctx])
alpha = tf.nn.softmax(logits)
return alpha
Here I only consider the 2-fc branch.
I think the fomula of the code is : wa(tanh(Wva vi) + tanh(Wha ht)), which is slightly different with the paper. But tanh(A) + tanh(B) != tanh(A+B)
So I wonder if there could be some problems that this difference may cause. Anyone can help?
Can your code achieve the same or similar BLEU scores published in the original paper?
i'm using mac os x el capitan
python 3.6.5
the last line error which given to me is 'You may need to pass the encoding= option to numpy.load'
log of process:
loading annotations into memory...
Done (t=4.42s)
creating index...
index created!
Filtering the captions by length...
creating index...
index created!
Building the vocabulary...
Vocabulary built.
Number of words = 5000
Filtering the captions by words...
creating index...
index created!
Processing the captions...
Captions processed.
Number of captions = 515671
Building the dataset...
Dataset built.
Building the CNN...
CNN built.
Building the RNN...
RNN built.
Loading the CNN from vgg16_no_fc.npy...
loading annotations into memory...
Done (t=3.82s)
creating index...
index created!
Filtering the captions by length...
creating index...
index created!
Building the vocabulary...
Vocabulary built.
Number of words = 5000
Filtering the captions by words...
creating index...
index created!
Processing the captions...
Captions processed.
Number of captions = 515671
Building the dataset...
Dataset built.
Building the CNN...
CNN built.
Building the RNN...
RNN built.
Loading the CNN from vgg16_no_fc.npy...
What command-line arguments should be used to get the best performance (BLEU, etc.)? I assume --load_cnn_model
should be set to True so that a pretrained CNN model can be used. What other settings should be used? For example, should --num_lstm
be increased from 1 to 2? Should --init_lstm_with_fc_feats
be set to True?
Dear all
there is something error at this line, so I make some revise
145 plt.savefig(os.path.join("test/results/",image_name.split("/")[-1]+'_result.jpg'))
Loading the model from ./models/289999.npy...
100%|██████████████████████████████████████████████████████████████████████████████████████████████| 47/47 [00:01<00:00, 23.69it/s]
47 tensors loaded.
Evaluating the model ...
batch: 100%|█████████████████████████████████████████████████████████████████████████████████| 1266/1266 [2:02:42<00:00, 5.83s/it]
Loading and preparing results...
DONE (t=0.13s)
creating index...
index created!
tokenization...
Traceback (most recent call last):
File "main.py", line 69, in
tf.app.run()
File "/home/viktor/anaconda2/envs/captureimage4/lib/python2.7/site-packages/tensorflow/python/platform/app.py", line 48, in run
_sys.exit(main(_sys.argv[:1] + flags_passthrough))
File "main.py", line 58, in main
model.eval(sess, coco, data, vocabulary)
File "/home/viktor/anaconda2/envs/captureimage4/scr/image_captioning-master/base_model.py", line 108, in eval
scorer.evaluate()
File "/home/viktor/anaconda2/envs/captureimage4/scr/image_captioning-master/utils/coco/pycocoevalcap/eval.py", line 31, in evaluate
gts = tokenizer.tokenize(gts)
File "/home/viktor/anaconda2/envs/captureimage4/scr/image_captioning-master/utils/coco/pycocoevalcap/tokenizer/ptbtokenizer.py", line 52, in tokenize
stdout=subprocess.PIPE)
File "/home/viktor/anaconda2/envs/captureimage4/lib/python2.7/subprocess.py", line 390, in init
errread, errwrite)
File "/home/viktor/anaconda2/envs/captureimage4/lib/python2.7/subprocess.py", line 1024, in _execute_child
raise child_exception
OSError: [Errno 2] No such file or directory
(captureimage4) viktor@viktor-System-Product-Name:
Hello
I'm trying to run it with weigths provided, however I didn't found our environment's setting (python version? dependencies versions?) So I'm running it on python 2.7.
I got the following error:
(captioning) rola93@rola93-Latitude-E5520:~/no_version/image_captioning$ python main.py --phase=test --model_file='./models/289999/289999.npy' --beam_size=3
2018-08-08 14:50:20.015861: 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.
2018-08-08 14:50:20.015898: 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.
2018-08-08 14:50:20.015922: 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.
Building the vocabulary...
Vocabulary built.
Number of words = 5000
Building the dataset...
Dataset built.
Building the CNN...
CNN built.
Building the RNN...
Traceback (most recent call last):
File "main.py", line 69, in <module>
tf.app.run()
File "/home/rola93/.pyenv/versions/captioning_2/lib/python2.7/site-packages/tensorflow/python/platform/app.py", line 48, in run
_sys.exit(main(_sys.argv[:1] + flags_passthrough))
File "main.py", line 63, in main
model = CaptionGenerator(config)
File "/home/rola93/no_version/image_captioning/base_model.py", line 27, in __init__
self.build()
File "/home/rola93/no_version/image_captioning/model.py", line 10, in build
self.build_rnn()
File "/home/rola93/no_version/image_captioning/model.py", line 228, in build_rnn
lstm = tf.nn.rnn_cell.LSTMCell(
AttributeError: 'module' object has no attribute 'rnn_cell'
According to this tf.nn.rnn_cell.LSTMCell was moved to contrib, so instead of lstm = tf.nn.rnn_cell.LSTMCell
it should be lstm = tf.contrib.rnn.LSTMCell
I tried it, and worked (actually it breaks anyway, but i'm sure it's another problem).
Does it make sense?
thank you!
I am unable to train the model on multiple GPUs. Am I missing something? Where do I need to configure the script for multi-gpu training? Thanks
Hi @DeepRNN Is there a way I can access the attention point from the image. I know I can do that using alpha, but I want to use those attention points and utilize that for further data processing. Can you please guide how can I get alpha from build_rnn() to the main file and access the positions of each annotation while testing it with an image? something like - this
Hello @DeepRNN !
I took a look at attentions that model generates in test mode.
I did the following: in base_model.py:200
i changed the code as following
memory, output, scores, attentions = sess.run(
[self.memory, self.output, self.probs, self.attentions],
feed_dict = {self.contexts: contexts,
self.last_word: last_word,
self.last_memory: last_memory,
self.last_output: last_output})
So after that, every attentions
array has the shape batch_size, 196, beam_size
and for simplicity I set beam_size=1
when testing. Next, I simply stack all attentions
in one numpy array and visualize its content.
I found two concerns:
1.jpg
the maximum difference between 1st and 2nd tokens is ~e^-9
).However, the caption of the image is both grammatically and semantically correct.
I would like to discuss these results.
File "/home/cao/semantic-IQA/image-caption-tensorflow/image_captioning-master/model.py", line 449, in build_rnn
opt_op = solver.apply_gradients(zip(gs, tvars), global_step=self.global_step)
ValueError: Variable emb_w/Adam/ does not exist, or was not created with tf.get_variable(). Did you mean to set reuse=None in VarScope?
Hello, Thank you for your efforts for writing and putting up your code on github. I really appreciate it.
I have an issue with it when I use anaconda environment while executing it, it gives me the following error:
/image_captioning-master/utils/coco/_mask.so: undefined symbol: PyFPE_jbuf
I search for it online and also tried removing the numpy library and then again executing it, but I still face this error. Can you please help me out ?
Thank You
In lines 404 to 415 of the model.py file, why do you want to add the logit of the image and the logit of the hiddien state as the final logit? Why not directly multiply the image features and the hidden state as the final logit?
Why don't you convert the weighted image features into the state c of the cell, and concatenate it with word embedding as input?
I was trying to run the model with ResNet as CNN. I've supplied model weights file in parameters, and also changed it in configuration. But in Tensorboard it still seems like the network still uses VGG.
Is there a way to actually use ResNet? If so, how?
Hi, How to visualization visual attention
with tf.Session() as sess:
if FLAGS.phase == 'train':
# training phase
data = prepare_train_data(config)
tf.get_default_graph().finalize()
model = CaptionGenerator(config)
sess.run(tf.global_variables_initializer())
if FLAGS.load:
model.load(sess, FLAGS.model_file)
if FLAGS.load_cnn:
model.load_cnn(sess, FLAGS.cnn_model_file)
tf.get_default_graph().finalize()
model.train(sess, data)
Vocabulary built.
Number of words = 5000
Filtering the captions by words...
100%|██████████| 409884/409884 [00:48<00:00, 8534.57it/s]
creating index...
index created!
Processing the captions...
Captions processed.
Number of captions = 361254
Building the dataset...
Dataset built.
Traceback (most recent call last):
File "", line 83, in
tf.app.run()
File "C:\ProgramData\Anaconda3\lib\site-packages\tensorflow\python\platform\app.py", line 125, in run
_sys.exit(main(argv))
File "", line 54, in main
model = CaptionGenerator(config)
File "H:\First Neural Network\image_captioning-master\base_model.py", line 26, in init
trainable = False)
File "C:\ProgramData\Anaconda3\lib\site-packages\tensorflow\python\ops\variables.py", line 259, in init
constraint=constraint)
File "C:\ProgramData\Anaconda3\lib\site-packages\tensorflow\python\ops\variables.py", line 380, in _init_from_args
initial_value, name="initial_value", dtype=dtype)
File "C:\ProgramData\Anaconda3\lib\site-packages\tensorflow\python\framework\ops.py", line 1011, in convert_to_tensor
as_ref=False)
File "C:\ProgramData\Anaconda3\lib\site-packages\tensorflow\python\framework\ops.py", line 1107, in internal_convert_to_tensor
ret = conversion_func(value, dtype=dtype, name=name, as_ref=as_ref)
File "C:\ProgramData\Anaconda3\lib\site-packages\tensorflow\python\framework\constant_op.py", line 217, in _constant_tensor_conversion_function
return constant(v, dtype=dtype, name=name)
File "C:\ProgramData\Anaconda3\lib\site-packages\tensorflow\python\framework\constant_op.py", line 202, in constant
name=name).outputs[0]
File "C:\ProgramData\Anaconda3\lib\site-packages\tensorflow\python\framework\ops.py", line 3386, in create_op
self._check_not_finalized()
File "C:\ProgramData\Anaconda3\lib\site-packages\tensorflow\python\framework\ops.py", line 3024, in _check_not_finalized
raise RuntimeError("Graph is finalized and cannot be modified.")
RuntimeError: Graph is finalized and cannot be modified.
As far as I know, image caption can be generated without the mechanism of attention.
How can i remove the mechanism of attention effectively for training?
can u give me advice pls?
Can the author provide more details for training the model?
It's useful to describe where to get the dataset, how should we prepare the input files and how should we start the training process.
Thanks!
I want to try testing serval images with this amazing method, would you pls release a pretrained caption model?
In keras we can use:
keras.utils.vis_utils import plot_model
plot_model(model, to_file='model_plot.png', show_shapes=True, show_layer_names=True)
how to do it inTF
I am implementing a network similar to this one, but want to use the pre-trained CNN with max accuracy over 2012ILSVRC dataset, i.e., NASNet-large. Usually, people go by extracting the last convolution layer features. But NASNet's architecture is relatively complex and I couldn't find a direct Conv layer. Below is the Tensorboard visualization of the "final_layer" cell of NASNet:
And below is the second last cell:
To me, the relu node I've selected in first image([1,11,11,4032]) seems close to what's needed for attention, but I am not sure. Any help will be highly appreciated.
hi,guys! when i tried to run train.py on Ubuntu16.04 with python3.5, i got this error, could anyone tell me how to fix this problem? i don't have the root right& i don't want to reinstall python from source. thank you, please!
In the code,you use fully_connect layer to calculate attention,why don't use formula in <>?
I have been training for a few days, I do not know how long I need to train
OMP: Error #15: Initializing libiomp5.dylib, but found libiomp5.dylib already initialized.
I have run that:
python main.py --phase=test \ --model_file='./models/xxxxxx.npy' \ --beam_size=3
now, I want to test another image , but don't want to reload the model, how can I do?
IOError: [Errno 2] No such file or directory: './train/captions_train2014.json'
What are the pre-requisites needed to run this code?
How do I get around this error:
Traceback (most recent call last):
File "main.py", line 69, in
tf.app.run()
File "C:\Python36\lib\site-packages\tensorflow\python\platform\app.py", line 126, in run
_sys.exit(main(argv))
File "main.py", line 50, in main
model.train(sess, data)
File "C:\Users\pranj\Desktop\img_cap\base_model.py", line 50, in train
images = self.image_loader.load_images(image_files)
File "C:\Users\pranj\Desktop\img_cap\utils\misc.py", line 35, in load_images
images.append(self.load_image(image_file))
File "C:\Users\pranj\Desktop\img_cap\utils\misc.py", line 19, in load_image
temp = image.swapaxes(0, 2)
AttributeError: 'NoneType' object has no attribute 'swapaxes'
Thanks for you sharing.
I met an AttributeError: 'module' object has no attribute 'AUTO_SIZE'. I wonder that my tensorflow version(1.2.0 ) is low. Maybe higher than 1.4.0 is needed?
How can I solve the problem?
Anyone with experience Serving the TF model for Docker? In particular with producing the signature_def_map.
It take a long time for training, who can tell optimal hyperparameters
it shows:
Traceback (most recent call last):
File "test.py", line 12, in
for param_name, data in data_dict[op_name].iteritems():
AttributeError: 'list' object has no attribute 'iteritems'
and i print the data_dict[op_name],it is list,what should i do.
def train(self, sess, train_data):
""" Train the model using the COCO train2014 data. """
print("Training the model...")
config = self.config
if not os.path.exists(config.summary_dir):
os.mkdir(config.summary_dir)
train_writer = tf.summary.FileWriter(config.summary_dir,
sess.graph)
for _ in tqdm(list(range(config.num_epochs)), desc='epoch'):
for _ in tqdm(list(range(train_data.num_batches)), desc='batch'):
batch = train_data.__next__batch()
image_files, sentences, masks = batch
images = self.image_loader.load_images('./train/images')
feed_dict = {self.images: images,
self.sentences: sentences,
self.masks: masks}
_, summary, global_step = sess.run([self.opt_op,
self.summary,
self.global_step],
feed_dict=feed_dict)
if (global_step + 1) % config.save_period == 0:
self.save()
train_writer.add_summary(summary, global_step)
train_data.reset()
self.save()
train_writer.close()
print("Training complete.")
File "H:\First Neural Network\image_captioning-master\base_model.py", line 44, in train
batch = train_data.__next__batch()
AttributeError: 'DataSet' object has no attribute '_BaseModel__next__batch'
Hello, thank you for your work first.
There are some problem when I run main.py, I am confused why there is a Syntax Error in bleu_scorer.py .Error message as follow:
Traceback (most recent call last):
File "main.py", line 5, in
from model import CaptionGenerator
File "D:\AI_Prj\image_captioning-master\model.py", line 4, in
from base_model import BaseModel
File "D:\AI_Prj\image_captioning-master\base_model.py", line 13, in
from utils.coco.pycocoevalcap.eval import COCOEvalCap
File "D:\AI_Prj\image_captioning-master\utils\coco\pycocoevalcap\eval.py", line 3, in
from utils.coco.pycocoevalcap.bleu.bleu import Bleu
File "D:\AI_Prj\image_captioning-master\utils\coco\pycocoevalcap\bleu\bleu.py", line 11, in
from utils.coco.pycocoevalcap.bleu.bleu_scorer import BleuScorer
File "D:\AI_Prj\image_captioning-master\utils\coco\pycocoevalcap\bleu\bleu_scorer.py", line 60
def cook_test(test, (reflen, refmaxcounts), eff=None, n=4):
^
SyntaxError: invalid syntax
Anyone have the file:vgg16_no_fc.npy. The address is invaild. Thank you !
COCO have no methods like filter_by_words() in line 75 and all_captions() in line 69 in dataset.py.
Do you have a QQ? I run your code, why are all the results of the test picture output the same? I checked the code, and there was no error. Why the description of the image is the same
Hi, I' am trying to reproduce your work.
May I ask, how much are the totoal_loss and accuracy after training?
I train the model for 60 epochs on 1/10 of the train data, and get a total_loss of about 1.6, an accuracy of about 65%, but when generating captions for the test images, the model just repeats all the same word, quite strange!!!
Any ideas? Thanks very much.
Hello
I'm trying to run it with provided weigths on some images to get its captions.
I'm running it on python 2.7, and tensorflow 1.1.0., however wher I run it I get this:
$ python main.py --phase=test --model_file='./models/289999/289999.npy' --beam_size=3
2018-08-08 15:12:08.881870: 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.
2018-08-08 15:12:08.881907: 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.
2018-08-08 15:12:08.881922: 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.
Building the vocabulary...
Vocabulary built.
Number of words = 5000
Building the dataset...
Dataset built.
Building the CNN...
CNN built.
Building the RNN...
RNN built.
Loading the model from ./models/289999/289999.npy...
100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████| 47/47 [00:01<00:00, 45.43it/s]
45 tensors loaded.
Testing the model ...
path: 0%| | 0/1 [00:00<?, ?it/s]Traceback (most recent call last):
File "main.py", line 69, in <module>
tf.app.run()
File "/home/rola93/.pyenv/versions/captioning_2/lib/python2.7/site-packages/tensorflow/python/platform/app.py", line 48, in run
_sys.exit(main(_sys.argv[:1] + flags_passthrough))
File "main.py", line 66, in main
model.test(sess, data, vocabulary)
File "/home/rola93/no_version/image_captioning/base_model.py", line 124, in test
caption_data = self.beam_search(sess, batch, vocabulary)
File "/home/rola93/no_version/image_captioning/base_model.py", line 202, in beam_search
self.last_output: last_output})
File "/home/rola93/.pyenv/versions/captioning_2/lib/python2.7/site-packages/tensorflow/python/client/session.py", line 778, in run
run_metadata_ptr)
File "/home/rola93/.pyenv/versions/captioning_2/lib/python2.7/site-packages/tensorflow/python/client/session.py", line 982, in _run
feed_dict_string, options, run_metadata)
File "/home/rola93/.pyenv/versions/captioning_2/lib/python2.7/site-packages/tensorflow/python/client/session.py", line 1032, in _do_run
target_list, options, run_metadata)
File "/home/rola93/.pyenv/versions/captioning_2/lib/python2.7/site-packages/tensorflow/python/client/session.py", line 1052, in _do_call
raise type(e)(node_def, op, message)
tensorflow.python.framework.errors_impl.FailedPreconditionError: Attempting to use uninitialized value lstm/lstm_cell/biases
[[Node: lstm/lstm_cell/biases/read = Identity[T=DT_FLOAT, _class=["loc:@lstm/lstm_cell/biases"], _device="/job:localhost/replica:0/task:0/cpu:0"](lstm/lstm_cell/biases)]]
Caused by op u'lstm/lstm_cell/biases/read', defined at:
File "main.py", line 69, in <module>
tf.app.run()
File "/home/rola93/.pyenv/versions/captioning_2/lib/python2.7/site-packages/tensorflow/python/platform/app.py", line 48, in run
_sys.exit(main(_sys.argv[:1] + flags_passthrough))
File "main.py", line 63, in main
model = CaptionGenerator(config)
File "/home/rola93/no_version/image_captioning/base_model.py", line 27, in __init__
self.build()
File "/home/rola93/no_version/image_captioning/model.py", line 10, in build
self.build_rnn()
File "/home/rola93/no_version/image_captioning/model.py", line 279, in build_rnn
output, state = lstm(current_input, last_state)
File "/home/rola93/.pyenv/versions/captioning_2/lib/python2.7/site-packages/tensorflow/contrib/rnn/python/ops/core_rnn_cell_impl.py", line 404, in __call__
lstm_matrix = _linear([inputs, m_prev], 4 * self._num_units, bias=True)
File "/home/rola93/.pyenv/versions/captioning_2/lib/python2.7/site-packages/tensorflow/contrib/rnn/python/ops/core_rnn_cell_impl.py", line 1056, in _linear
initializer=init_ops.constant_initializer(bias_start, dtype=dtype))
File "/home/rola93/.pyenv/versions/captioning_2/lib/python2.7/site-packages/tensorflow/python/ops/variable_scope.py", line 1049, in get_variable
use_resource=use_resource, custom_getter=custom_getter)
File "/home/rola93/.pyenv/versions/captioning_2/lib/python2.7/site-packages/tensorflow/python/ops/variable_scope.py", line 948, in get_variable
use_resource=use_resource, custom_getter=custom_getter)
File "/home/rola93/.pyenv/versions/captioning_2/lib/python2.7/site-packages/tensorflow/python/ops/variable_scope.py", line 356, in get_variable
validate_shape=validate_shape, use_resource=use_resource)
File "/home/rola93/.pyenv/versions/captioning_2/lib/python2.7/site-packages/tensorflow/python/ops/variable_scope.py", line 341, in _true_getter
use_resource=use_resource)
File "/home/rola93/.pyenv/versions/captioning_2/lib/python2.7/site-packages/tensorflow/python/ops/variable_scope.py", line 714, in _get_single_variable
validate_shape=validate_shape)
File "/home/rola93/.pyenv/versions/captioning_2/lib/python2.7/site-packages/tensorflow/python/ops/variables.py", line 197, in __init__
expected_shape=expected_shape)
File "/home/rola93/.pyenv/versions/captioning_2/lib/python2.7/site-packages/tensorflow/python/ops/variables.py", line 316, in _init_from_args
self._snapshot = array_ops.identity(self._variable, name="read")
File "/home/rola93/.pyenv/versions/captioning_2/lib/python2.7/site-packages/tensorflow/python/ops/gen_array_ops.py", line 1338, in identity
result = _op_def_lib.apply_op("Identity", input=input, name=name)
File "/home/rola93/.pyenv/versions/captioning_2/lib/python2.7/site-packages/tensorflow/python/framework/op_def_library.py", line 768, in apply_op
op_def=op_def)
File "/home/rola93/.pyenv/versions/captioning_2/lib/python2.7/site-packages/tensorflow/python/framework/ops.py", line 2336, in create_op
original_op=self._default_original_op, op_def=op_def)
File "/home/rola93/.pyenv/versions/captioning_2/lib/python2.7/site-packages/tensorflow/python/framework/ops.py", line 1228, in __init__
self._traceback = _extract_stack()
FailedPreconditionError (see above for traceback): Attempting to use uninitialized value lstm/lstm_cell/biases
[[Node: lstm/lstm_cell/biases/read = Identity[T=DT_FLOAT, _class=["loc:@lstm/lstm_cell/biases"], _device="/job:localhost/replica:0/task:0/cpu:0"](lstm/lstm_cell/biases)]]
can anybody share the pretrained model from another way?
I was told "No such file or directory: './words/word_table.pickle'" when running main.py
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google ❤️ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.