jiahuiyu / neuralgym Goto Github PK

When I run the code with tensorflow 2.0, it shows

AttributeError: module tensorflow has no attribute contrib

and I find the bug is about tensorflow 2.0, so I wonder if the repo will be updated to follow the version of tensorflow.

Hi,

As per my understanding, neuralgym.data.DataFromFNames is able to read only image files. Is there a way to read matrices or csv files?
if not then could you also suggest a way to do it in tensorflow?

TIA

File "/usr/local/lib/python3.6/dist-packages/neuralgym/ops/layers.py", line 115, in
func=tf.image.resize_bilinear, name='resize'):
AttributeError: module 'tensorflow._api.v2.image' has no attribute 'resize_bilinear'

It concerns are placed here:

114    def resize(x, scale=2, to_shape=None, align_corners=True, dynamic=False,
115              func=tf.image.resize_bilinear, name='resize'):

I tried Tensorflow 2.0 , 2.1 and 2.4.1

Is the problem only mine or will an update according to the link below help?

https://www.tensorflow.org/api_docs/python/tf/compat/v1/image/resize_bilinear

Hi,

Thanks for providing the package. I have a question about the requirements. I am trying to train the generative inpainting code [https://github.com/JiahuiYu/generative_inpainting] which is using neuralgym package. In the start of the training procedure [after assigning the GPUs and even generating the graphs], I am getting a few errors related to the version of cuDNN and CUDA library that I am using. However, previously I didn't have any issue in running the test codes of the generative inpainting model.

Ok, long story short, previously I had:
cuDNN v7.1.4
CUDA 9.0
Nvidia Driver 384.130

After the training error, it suggests me to upgrade to cuDNN v.7.2.x, now, with this new driver I am getting:
CUDNN_STATUS_NOT_INITIALIZED

and again, it suggests I need to upgrade my driver. So, I upgrade to Nvidia driver 390.87, but now, it has issues to catch all of the GPUs and saying not enough gpus, well, which I have enough gpu.

My question is that, is there any safe, and tested combination of Tensorflow, Nvidia driver version, CUDA, and cuDNN library version that works well with neuralgym. I really apperitiate even if you can share your versions.

-Many Thanks, Shek

hello,

i want to input a image with 16bit.

In data_from_fnames.py , i changed cv2.imread(..) commnet to cv2.imread(..,-1) to read 16 bit.

but it cause index error

please tell me the solution

I am getting this error for tensorflow. I am unable to figure out the reason. I have passed the variables to var_list which are not empty either.

Hi Jiahui,

Opencv's expected shape is (width,height).
The fix it's really easy, so, I'm submitting PR.

In data_from_fnames.py line 166, it appears that random is always true. I believe it's supposed to be self.random

if random: filenames = random.choice(self.fnamelists_)

should add NVSMI_SHOW_ALL_DEVICES=1

Hi Jiahui,

Thanks for your toolkit, it's really helpful !

In my world, an epoch means running through the training dataset once.
So I'm not sure why you set an epoch as TRAIN_SPE steps ignoring the size of training dataset.

see progress_logger

The requirements file appears to be an empty file. As a result, none of the dependencies get installed. Since most package managers will by default select dependencies that are incompatible (i.e. Tensorflow2), this makes getting this repo (as well as all downstream repos that need it) working a bunch of trial-and-error guesswork.

2023-06-18 07:23:12.649587: W tensorflow/compiler/xla/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'libnvinfer.so.7'; dlerror: libnvinfer.so.7: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: /usr/local/nvidia/lib:/usr/local/nvidia/lib64
2023-06-18 07:23:12.652543: W tensorflow/compiler/xla/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'libnvinfer_plugin.so.7'; dlerror: libnvinfer_plugin.so.7: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: /usr/local/nvidia/lib:/usr/local/nvidia/lib64
2023-06-18 07:23:12.652586: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Cannot dlopen some TensorRT libraries. If you would like to use Nvidia GPU with TensorRT, please make sure the missing libraries mentioned above are installed properly.
Traceback (most recent call last):
File "/content/drive/MyDrive/geninpaint_code1/train.py", line 5, in
import neuralgym as ng
File "/usr/local/lib/python3.10/dist-packages/neuralgym/init.py", line 27, in
from . import callbacks
File "/usr/local/lib/python3.10/dist-packages/neuralgym/callbacks/init.py", line 3, in
from .hyper_param_scheduler import HyperParamScheduler
File "/usr/local/lib/python3.10/dist-packages/neuralgym/callbacks/hyper_param_scheduler.py", line 6, in
from ..ops.summary_ops import scalar_summary
File "/usr/local/lib/python3.10/dist-packages/neuralgym/ops/init.py", line 4, in
from . import layers
File "/usr/local/lib/python3.10/dist-packages/neuralgym/ops/layers.py", line 93, in
activation_fn=None, w_init=tf.contrib.layers.xavier_initializer_conv2d(),
AttributeError: module 'tensorflow' has no attribute 'contrib'

hi jiahui,i'am using your inpainting test model. when the code comes to this np.get_gpus(1) line, an weird error happen.

i was wandering whether my environment was wrong: i'm using anaconda2 with py3.5 +cuda9.0+cudnn7.1
my GPU is 1050Ti.