OSError: Unable to open file (unable to open file)

Question

I am trying to load a pre_trained model named "tr_model.h5" for my assignment but I get the following error:

Traceback (most recent call last):
 File "Trigger_Project.py", line 84, in <module>
model = load_model(filename)
 File "Trigger_Project.py", line 84, in <module>
model = load_model(filename)
 File "/home/neeraj/anaconda3/lib/python3.6/site-packages/h5py/_hl/files.py", line 99, in make_fid
fid = h5f.open(name, flags, fapl=fapl)
File "h5py/_objects.pyx", line 54, in h5py._objects.with_phil.wrapper

File "h5py/_objects.pyx", line 55, in h5py._objects.with_phil.wrapper

File "h5py/h5f.pyx", line 78, in h5py.h5f.open
OSError: Unable to open file (unable to open file: name = 'tr_model.h5', errno = 2, error message = 'No such file or directory', flags = 0, o_flags = 0)

I have made sure that the file is present. I don't why it is showing os error. I am using linux 18.04 and all the required libraries are upgraded. Any help is much appreciated.

Is there any issue where it can not be opened through terminal ? — Neeraj, Jul 30 '18 at 12:12
Can you add a `print(os.getcwd())` and check that it is the correct directory you're assuming? — Edward Minnix, Jul 30 '18 at 12:27
Yes, I did check that and it is reporting the same directory I am assuming. — Neeraj, Jul 30 '18 at 12:49
All these solutions does not work for me. Interestingly, I managed to solve by uninstall the h5py package (3.15) with an older version, say, 2.15 by `pip3 install h5py==2.15` — YinchaoOnline, Aug 19 '22 at 07:36

score 4 · Answer 1 · answered Aug 13 '19 at 04:34

I encounter the same issue as I posted in my question:

h5py.File(path) doesn't recognize folder path

My initial reasoning is that h5py.File(path) doesn't handle the standard subfolder path as its argument, e.g. load_model("neunet.h5") has no issue, but load_model("subfolder/neunet.h5") will give the same error.

In a nutshell, my solution is to simply put whatever .h5 files into the working home folder for in my jupyter notebook, which is the place you create the .ipynb file. You could use print(os.getcwd()) in jupyter notebook to see where is your current working directory.

score 3 · Answer 2 · answered Jul 23 '20 at 13:51

I had the same issue when setting the path for training checkpoints using tensorflow.keras.callbacks.ModelCheckpoint I had set my path to:

path = os.path.join(subdir,filename)

using an f string solved the problem:

path = f'{subdir}/{filename}'

So I would also check how you are importing the load_model, try tensorflow.keras.models instead of only keras.models

score 1 · Answer 3 · answered Sep 15 '19 at 15:52

1

If you are working on colab and google drive, mount the drive to the colab using,

# Load the Drive helper and mount
from google.colab import drive

# This will prompt for authorization.
drive.mount('/content/drive')

answered Sep 15 '19 at 15:52

Shirantha Madusanka

1,461
1
11
16

score 1 · Answer 4 · answered Dec 31 '19 at 03:49

I solved this problem by specifying an absolute path instead. Getting the absolute path of the working folder and then appending the path to the file from the working directory. in my case, the file is in a directory named datasets, so I tried the following code

file_name = os.path.dirname(__file__) +'\\datasets\\test_catvnoncat.h5'
test_dataset = h5py.File(file_name, "r")

score 0 · Answer 5 · answered Apr 24 '19 at 17:15

I encountered with this error when I was working with docker image. Since customized docker has different path than root path, I got the same error.

OSError: Unable to open file (unable to open file: name = '', errno = 2, error message = 'No such file or directory', flags = 0, o_flags = 0)

Basically mismatch of the paths raise this error

Check relative, absolute path
Check which path your program expects

score 0 · Answer 6 · answered Mar 31 '20 at 01:54

0

If you run a.py got the error, but you load the model.h5 in b.py
Please cp model.h5 to the same level directory with a.py
Edit b.py's load('xx/model.h5') to load('model.h5')

It worked for me.

answered Mar 31 '20 at 01:54

Ivan Yang

1

score 0 · Answer 7 · answered Jul 13 '20 at 06:43

I encountered the same issue in colab after mounting the drive by setting "%cd /gdrive" and when I wrote "model.save('/gdrive/mnist.h5')". Then I realised files can be created and saved in '/gdrive/My Drive/' and not just '/gdrive'. So, it is important to check if you can create files in the path specified.

So, model.save('/gdrive/My Drive/mnist.h5') worked for me.

score 0 · Answer 8 · answered Nov 03 '21 at 22:41

I had loss always equal to inf, therefore the model file had never been written to disk on callbacks as it had only been written for best cases for loss.

When I eliminated all NaN values from the dataset, all went smoothly and the model appeared on the file system.

Enabling verbosity in the callback helped to diagnose this.

For details, see the official Keras docs on callbacks.

OSError: Unable to open file (unable to open file)

8 Answers8

Linked