Could not load dynamic library 'libnvinfer.so. Allocation of 376320000 exceeds 10% of system memory. Tests failed on 1 cell(s)! These tests could be hidden6'

Course Q&A TensorFlow: Advanced Techniques Specialization Custom Models, Layers and Loss Functions with TF
1

My Assignment 3 Submission - when i run on my laptop seems to go thru! But when submitted _ i get this error.

[ValidateApp | INFO] Validating ‘/home/jovyan/work/submitted/courseraLearner/W1_Assignment/C1W1_Assignment.ipynb’

[ValidateApp | INFO] Executing notebook with kernel: python3

2021-08-22 23:45:29.700911: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library ‘libnvinfer.so.6’; dlerror: libnvinfer.so.6: cannot open shared object file: No such file or directory

2021-08-22 23:45:29.701009: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library ‘libnvinfer_plugin.so.6’; dlerror: libnvinfer_plugin.so.6: cannot open shared object file: No such file or directory

2021-08-22 23:45:29.701021: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:30] 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.

2021-08-22 23:45:31.630063: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library ‘libcuda.so.1’; dlerror: libcuda.so.1: cannot open shared object file: No such file or directory

2021-08-22 23:45:31.630105: E tensorflow/stream_executor/cuda/cuda_driver.cc:351] failed call to cuInit: UNKNOWN ERROR (303)

2021-08-22 23:45:31.630132: I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:156] kernel driver does not appear to be running on this host (somehost): /proc/driver/nvidia/version does not exist

2021-08-22 23:45:31.630303: I tensorflow/core/platform/cpu_feature_guard.cc:142] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 AVX512F FMA

2021-08-22 23:45:31.637440: I tensorflow/core/platform/profile_utils/cpu_utils.cc:94] CPU Frequency: 3000000000 Hz

[ValidateApp | INFO] Validating ‘/home/jovyan/work/submitted/courseraLearner/W3_Assignment/C1W3_Assignment.ipynb’ [ValidateApp | INFO] Executing notebook with kernel: python3 2021-08-23 03:51:22.973540: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library ‘libnvinfer.so.6’; dlerror: libnvinfer.so.6: cannot open shared object file: No such file or directory 2021-08-23 03:51:22.973622: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library ‘libnvinfer_plugin.so.6’; dlerror: libnvinfer_plugin.so.6: cannot open shared object file: No such file or directory 2021-08-23 03:51:22.973633: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:30] 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. 2021-08-23 03:51:23.618880: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library ‘libcuda.so.1’; dlerror: libcuda.so.1: cannot open shared object file: No such file or directory 2021-08-23 03:51:23.618915: E tensorflow/stream_executor/cuda/cuda_driver.cc:351] failed call to cuInit: UNKNOWN ERROR (303) 2021-08-23 03:51:23.618940: I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:156] kernel driver does not appear to be running on this host (somehost): /proc/driver/nvidia/version does not exist 2021-08-23 03:51:23.619210: I tensorflow/core/platform/cpu_feature_guard.cc:142] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 AVX512F FMA 2021-08-23 03:51:23.624868: I tensorflow/core/platform/profile_utils/cpu_utils.cc:94] CPU Frequency: 2999995000 Hz 2021-08-23 03:51:23.625595: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x7f01ab2ec910 initialized for platform Host (this does not guarantee that XLA will be used). Devices: 2021-08-23 03:51:23.625626: I tensorflow/compiler/xla/service/service.cc:176] StreamExecutor device (0): Host, Default Version 2021-08-23 03:51:24.586326: W tensorflow/core/framework/cpu_allocator_impl.cc:81] Allocation of 376320000 exceeds 10% of system memory. Tests failed on 1 cell(s)! These tests could be hidden. Please check your submission.

2021-08-23 04:05:41.603950: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library ‘libnvinfer.so.6’; dlerror: libnvinfer.so.6: cannot open shared object file: No such file or directory 2021-08-23 04:05:41.604034: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library ‘libnvinfer_plugin.so.6’; dlerror: libnvinfer_plugin.so.6: cannot open shared object file: No such file or directory 2021-08-23 04:05:41.604045: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:30] 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. 2021-08-23 04:05:42.387451: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library ‘libcuda.so.1’; dlerror: libcuda.so.1: cannot open shared object file: No such file or directory 2021-08-23 04:05:42.387487: E tensorflow/stream_executor/cuda/cuda_driver.cc:351] failed call to cuInit: UNKNOWN ERROR (303) 2021-08-23 04:05:42.387511: I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:156] kernel driver does not appear to be running on this host (somehost): /proc/driver/nvidia/version does not exist

2

Hello there,

This is week 1 assignment not week 3, I would say there should be something wrong with your nvidia card (driver). Try to submit from another pc.

3

Thank you for your time and response,

This is week 3 assessment of the Custom Functional Layers Advanced tensor Flow course specialization.

Will try from another laptop. But since I am submitting via Colab or the Coursera notebook environment, is there a setting issue?

When I run the notebook without submitting, there is no error.

Thank you again. Much appreciated.
VK

4

I said week 1 because its writen C1W1 on the error response. Implement a Quadratic Layer is the W3 assignment and this is run on the coursera platform.

Not any that I can think of! Try from another PC thats my sugestion.

5

Good morning again Gent

I tried from another laptop; and get the same/similar issue; pasting below the ‘grader output’ when i u submit.

===

[ValidateApp | INFO] Validating ‘/home/jovyan/work/submitted/courseraLearner/W3_Assignment/C1W3_Assignment.ipynb’
[ValidateApp | INFO] Executing notebook with kernel: python3
2021-08-24 11:29:15.312913: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library ‘libnvinfer.so.6’; dlerror: libnvinfer.so.6: cannot open shared object file: No such file or directory
2021-08-24 11:29:15.312999: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library ‘libnvinfer_plugin.so.6’; dlerror: libnvinfer_plugin.so.6: cannot open shared object file: No such file or directory
2021-08-24 11:29:15.313010: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:30] 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.
2021-08-24 11:29:15.985948: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library ‘libcuda.so.1’; dlerror: libcuda.so.1: cannot open shared object file: No such file or directory
2021-08-24 11:29:15.985988: E tensorflow/stream_executor/cuda/cuda_driver.cc:351] failed call to cuInit: UNKNOWN ERROR (303)
2021-08-24 11:29:15.986013: I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:156] kernel driver does not appear to be running on this host (somehost): /proc/driver/nvidia/version does not exist
2021-08-24 11:29:15.986307: I tensorflow/core/platform/cpu_feature_guard.cc:142] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 AVX512F FMA
2021-08-24 11:29:15.992093: I tensorflow/core/platform/profile_utils/cpu_utils.cc:94] CPU Frequency: 2999995000 Hz
2021-08-24 11:29:15.992935: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x7f0b79d84990 initialized for platform Host (this does not guarantee that XLA will be used). Devices:
2021-08-24 11:29:15.992967: I tensorflow/compiler/xla/service/service.cc:176] StreamExecutor device (0): Host, Default Version
2021-08-24 11:29:17.179993: W tensorflow/core/framework/cpu_allocator_impl.cc:81] Allocation of 376320000 exceeds 10% of system memory.
Tests failed on 1 cell(s)! These tests could be hidden. Please check your submission.

6

I ran the lab myself and it gives me this warning but in the end passes all the tests, now I believe your code might not be implemented right, you need to check your code and see either there is a mistake or something is expected differently. Here is my run:

Grader output

[ValidateApp | INFO] Validating ‘/home/jovyan/work/submitted/courseraLearner/W3_Assignment/C1W3_Assignment.ipynb’ [ValidateApp | INFO] Executing notebook with kernel: python3 2021-08-24 11:50:24.495049: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library ‘libnvinfer.so.6’; dlerror: libnvinfer.so.6: cannot open shared object file: No such file or directory 2021-08-24 11:50:24.495142: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library ‘libnvinfer_plugin.so.6’; dlerror: libnvinfer_plugin.so.6: cannot open shared object file: No such file or directory 2021-08-24 11:50:24.495153: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:30] 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. 2021-08-24 11:50:25.157401: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library ‘libcuda.so.1’; dlerror: libcuda.so.1: cannot open shared object file: No such file or directory 2021-08-24 11:50:25.157438: E tensorflow/stream_executor/cuda/cuda_driver.cc:351] failed call to cuInit: UNKNOWN ERROR (303) 2021-08-24 11:50:25.157462: I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:156] kernel driver does not appear to be running on this host (somehost): /proc/driver/nvidia/version does not exist 2021-08-24 11:50:25.157675: I tensorflow/core/platform/cpu_feature_guard.cc:142] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 AVX512F FMA 2021-08-24 11:50:25.164670: I tensorflow/core/platform/profile_utils/cpu_utils.cc:94] CPU Frequency: 3000000000 Hz 2021-08-24 11:50:25.166558: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x7f86a95c5b60 initialized for platform Host (this does not guarantee that XLA will be used). Devices: 2021-08-24 11:50:25.166592: I tensorflow/compiler/xla/service/service.cc:176] StreamExecutor device (0): Host, Default Version 2021-08-24 11:50:25.920146: W tensorflow/core/framework/cpu_allocator_impl.cc:81] Allocation of 376320000 exceeds 10% of system memory. Success! Your notebook passes all the tests.

7

Thank you Gent; i will surelly take a look - maybe slighter later today or tomorrow. Somehow when i just run the code ( without submitting) - the error is not thrown. But will surely take a look.

Not sure what the ‘hidden test’ could be meant; i looked over the utils.py script - did not see any such error explicitly referenced in that code. Thanks Much for having tried out - maybe there’s an error elsewhere , or earlier in the code somewhere - that triggers this failure. Will keep you posted - after I have resolved the issue.

==
Tests failed on 1 cell(s)! These tests could be hidden. Please check your submission.

8

I think now i see the issue; the ‘hidden cell’ refers to the ‘train/evaluate’ cell ; I ran it again ( without submitting) AND noticed that i got an accuracy around .924… Reviewing the assignment - the expected accuracy is 97%

Now, i will revisit my code and try to improve ; atleast i know functionally the code is OK. Thanks again for your clarifications and help. Will let you know when I have resolved and get a ‘pass’

== 60000/60000 [==============================] - 12s 195us/sample - loss: 0.2951 - accuracy: 0.9178 10000/10000 [==============================] - 1s 74us/sample - loss: 0.2809 - accuracy: 0.9245

Out[5]:

[0.28086821455657485, 0.9245]

1 Like
9

Indeed Krishnam, i think the problem is there too, specifically how the model is built!

10

Thank you for your good wishes, inspirational motivation by consistent , persitent follow-up responses!

Guess what - i ‘passed’ the submission just now! It was a simple omission :slight_smile:

I was simply using ‘return’ ; had to include the return self.activation… when i re-read the assignment notebook - there was a comment to include the 'return self.activation…

1 Like
11

Great, keep up the good work!