Stuck on an issue?

Lightrun Answers was designed to reduce the constant googling that comes with debugging 3rd party libraries. It collects links to all the places you might be looking at while hunting down a tough bug.

And, if you’re still stuck at the end, we’re happy to hop on a call to see how we can help out.

New update in PyTorch break forces

See original GitHub issue

@farhadrgh @IgnacioJPickering Just tested python tests/test_forces.py with a newest freshly compiled PyTorch, get the following error:

FFF.F
======================================================================
FAIL: testBenzeneMD (__main__.TestForce)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "tests/test_forces.py", line 89, in testBenzeneMD
    self.assertLess(max_diff, tolerance)
AssertionError: 30.85953140258789 not less than 1e-05

======================================================================
FAIL: testIsomers (__main__.TestForce)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "tests/test_forces.py", line 42, in testIsomers
    self.assertLess(max_diff, self.tolerance)
AssertionError: 0.5897072553634644 not less than 1e-05

======================================================================
FAIL: testNIST (__main__.TestForce)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "tests/test_forces.py", line 125, in testNIST
    self.assertLess(max_diff, self.tolerance)
AssertionError: 4.202903747558594 not less than 1e-05

======================================================================
FAIL: testTripeptideMD (__main__.TestForce)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "tests/test_forces.py", line 108, in testTripeptideMD
    self.assertLess(max_diff, tolerance)
AssertionError: 22.5817928314209 not less than 2e-06

----------------------------------------------------------------------
Ran 5 tests in 64.475s

FAILED (failures=4)

Issue Analytics

State:
Created 4 years ago
Comments:5 (5 by maintainers)

Top GitHub Comments

1reaction

zasdfgbnmcommented, Aug 6, 2019

torch_nightly==1.2.0.dev20190801 works well torch_nightly==1.2.0.dev20190802 is broken

0reactions

zasdfgbnmcommented, Aug 6, 2019

Tests in test_aevs and test_energies are also failing.

Top Results From Across the Web

PyTorch 2.0

Introducing PyTorch 2.0, our first steps toward the next generation 2-series release of PyTorch. ... Let us break down the compiler into three...

Releases · pytorch/pytorch - GitHub

PyTorch 1.10 updates are focused on improving training and performance of PyTorch, and developer usability. Highlights include: CUDA Graphs APIs are integrated ...

Understanding PyTorch with an example: a step-by-step tutorial

Next, let's split our synthetic data into train and validation sets, ... In the final step, we use the gradients to update the...

Debugging and Visualisation in PyTorch using Hooks

We go over PyTorch hooks and using them to debug our backpass, visualise activations and modify ... First, because they force us to...

PyTorch 1.9.0 Now Available - Exxact Corporation

PyTorch just released version 1.9 with support scientific computing, support for large scale distributed training with GPU support, ...