Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Make algo tests much faster to run #93

Merged
merged 11 commits into from
Dec 4, 2024
Merged

Make algo tests much faster to run #93

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
11 commits
Select commit Hold shift + click to select a range
fdbb343
Make algo tests much faster to run
lebrice Dec 2, 2024
aa9c7b3
Fix issues with test signature change
lebrice Dec 3, 2024
ff46362
Update outdated regression files
lebrice Dec 3, 2024
884f9ab
Update jax regression test files
lebrice Dec 3, 2024
9304585
Reduce some redundancy in llm_finetuning_test.py
lebrice Dec 3, 2024
19af480
Remove outdated code in `project/conftest.py`
lebrice Dec 3, 2024
c69eae0
Add xfail on jax tests on self-hosted runner
lebrice Dec 3, 2024
fe98f25
Add broad xfail for jax_image_classifier tests :(
lebrice Dec 4, 2024
3406068
try fix for `IN_SELF_HOSTED_GITHUB_CI`
lebrice Dec 4, 2024
2a32003
Try to solve issues by updating regression files
lebrice Dec 4, 2024
eec9b4b
Add back the xfail on jax image classifier tests
lebrice Dec 4, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
68 changes: 34 additions & 34 deletions ...ge_classifier_test/test_backward_pass_is_reproducible/fcnet_cifar10_image_classifier.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,14 +1,14 @@
batch.0:
device: cuda:0
max: '2.126e+00'
mean: '-6.179e-03'
mean: '6.869e-03'
min: '-1.989e+00'
shape:
- 128
- 3
- 32
- 32
sum: '-2.43e+03'
sum: '2.701e+03'
batch.1:
device: cuda:0
max: 9
Expand All @@ -19,71 +19,71 @@ batch.1:
sum: 583
grads.network.0.1.bias:
device: cuda:0
max: '6.107e-03'
mean: '1.775e-04'
min: '-5.292e-03'
max: '5.928e-03'
mean: '3.020e-04'
min: '-3.916e-03'
shape:
- 128
sum: '2.272e-02'
sum: '3.866e-02'
grads.network.0.1.weight:
device: cuda:0
max: '1.307e-02'
mean: '4.693e-05'
min: '-1.141e-02'
max: '1.229e-02'
mean: '1.095e-04'
min: '-1.115e-02'
shape:
- 128
- 3072
sum: '1.845e+01'
sum: '4.306e+01'
grads.network.1.0.bias:
device: cuda:0
max: '1.041e-02'
mean: '6.975e-04'
min: '-8.782e-03'
max: '1.187e-02'
mean: '6.403e-04'
min: '-9.623e-03'
shape:
- 128
sum: '8.928e-02'
sum: '8.196e-02'
grads.network.1.0.weight:
device: cuda:0
max: '1.584e-02'
mean: '1.481e-04'
min: '-1.507e-02'
max: '1.566e-02'
mean: '1.344e-04'
min: '-1.467e-02'
shape:
- 128
- 128
sum: '2.426e+00'
sum: '2.202e+00'
grads.network.2.0.bias:
device: cuda:0
max: '3.282e-02'
mean: '-1.956e-09'
min: '-2.134e-02'
max: '3.269e-02'
mean: '-2.887e-09'
min: '-2.157e-02'
shape:
- 10
sum: '-1.956e-08'
sum: '-2.887e-08'
grads.network.2.0.weight:
device: cuda:0
max: '2.200e-02'
mean: '-2.561e-10'
min: '-5.831e-02'
max: '2.914e-02'
mean: '-2.98e-10'
min: '-3.501e-02'
shape:
- 10
- 128
sum: '-3.278e-07'
sum: '-3.814e-07'
outputs.logits:
device: cuda:0
max: '7.036e-01'
mean: '-8.651e-03'
min: '-8.180e-01'
max: '8.135e-01'
mean: '-8.627e-03'
min: '-7.944e-01'
shape:
- 128
- 10
sum: '-1.107e+01'
sum: '-1.104e+01'
outputs.loss:
device: cuda:0
max: '2.316e+00'
mean: '2.316e+00'
min: '2.316e+00'
max: '2.319e+00'
mean: '2.319e+00'
min: '2.319e+00'
shape: []
sum: '2.316e+00'
sum: '2.319e+00'
outputs.y:
device: cuda:0
max: 9
Expand Down
68 changes: 34 additions & 34 deletions ...ssifier_test/test_backward_pass_is_reproducible/fcnet_fashion_mnist_image_classifier.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,14 +1,14 @@
batch.0:
device: cuda:0
max: '2.821e+00'
mean: '4.822e-01'
mean: '4.772e-01'
min: '-4.242e-01'
shape:
- 128
- 1
- 28
- 28
sum: '4.839e+04'
sum: '4.789e+04'
batch.1:
device: cuda:0
max: 9
Expand All @@ -19,71 +19,71 @@ batch.1:
sum: 583
grads.network.0.1.bias:
device: cuda:0
max: '6.875e-03'
mean: '2.096e-04'
min: '-8.370e-03'
max: '7.419e-03'
mean: '4.543e-04'
min: '-4.832e-03'
shape:
- 128
sum: '2.683e-02'
sum: '5.816e-02'
grads.network.0.1.weight:
device: cuda:0
max: '1.948e-02'
mean: '2.916e-04'
min: '-2.213e-02'
max: '1.735e-02'
mean: '2.23e-04'
min: '-1.552e-02'
shape:
- 128
- 784
sum: '2.926e+01'
sum: '2.238e+01'
grads.network.1.0.bias:
device: cuda:0
max: '1.109e-02'
mean: '2.213e-04'
min: '-1.267e-02'
max: '1.157e-02'
mean: '2.873e-04'
min: '-1.017e-02'
shape:
- 128
sum: '2.832e-02'
sum: '3.678e-02'
grads.network.1.0.weight:
device: cuda:0
max: '2.374e-02'
mean: '9.326e-05'
min: '-2.32e-02'
max: '2.752e-02'
mean: '1.217e-04'
min: '-3.079e-02'
shape:
- 128
- 128
sum: '1.528e+00'
sum: '1.994e+00'
grads.network.2.0.bias:
device: cuda:0
max: '3.847e-02'
mean: '-3.353e-09'
min: '-4.706e-02'
max: '3.865e-02'
mean: '-9.313e-10'
min: '-4.547e-02'
shape:
- 10
sum: '-3.353e-08'
sum: '-9.313e-09'
grads.network.2.0.weight:
device: cuda:0
max: '5.741e-02'
mean: '-3.929e-10'
min: '-6.431e-02'
max: '4.74e-02'
mean: '-2.085e-10'
min: '-6.661e-02'
shape:
- 10
- 128
sum: '-5.029e-07'
sum: '-2.668e-07'
outputs.logits:
device: cuda:0
max: '9.872e-01'
mean: '-1.288e-02'
min: '-7.225e-01'
max: '8.907e-01'
mean: '-1.669e-02'
min: '-6.486e-01'
shape:
- 128
- 10
sum: '-1.648e+01'
sum: '-2.136e+01'
outputs.loss:
device: cuda:0
max: '2.311e+00'
mean: '2.311e+00'
min: '2.311e+00'
max: '2.309e+00'
mean: '2.309e+00'
min: '2.309e+00'
shape: []
sum: '2.311e+00'
sum: '2.309e+00'
outputs.y:
device: cuda:0
max: 9
Expand Down
66 changes: 33 additions & 33 deletions ...mage_classifier_test/test_backward_pass_is_reproducible/fcnet_mnist_image_classifier.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,14 +1,14 @@
batch.0:
device: cuda:0
max: '2.821e+00'
mean: '1.432e-02'
mean: '1.477e-02'
min: '-4.242e-01'
shape:
- 128
- 1
- 28
- 28
sum: '1.437e+03'
sum: '1.482e+03'
batch.1:
device: cuda:0
max: 9
Expand All @@ -19,71 +19,71 @@ batch.1:
sum: 543
grads.network.0.1.bias:
device: cuda:0
max: '1.075e-02'
mean: '2.421e-04'
min: '-7.844e-03'
max: '8.396e-03'
mean: '1.867e-04'
min: '-6.027e-03'
shape:
- 128
sum: '3.099e-02'
sum: '2.389e-02'
grads.network.0.1.weight:
device: cuda:0
max: '2.006e-02'
mean: '5.258e-05'
min: '-1.844e-02'
max: '1.893e-02'
mean: '4.891e-05'
min: '-1.587e-02'
shape:
- 128
- 784
sum: '5.277e+00'
sum: '4.909e+00'
grads.network.1.0.bias:
device: cuda:0
max: '1.169e-02'
mean: '4.285e-04'
max: '1.069e-02'
mean: '7.139e-05'
min: '-1.152e-02'
shape:
- 128
sum: '5.485e-02'
sum: '9.138e-03'
grads.network.1.0.weight:
device: cuda:0
max: '1.753e-02'
mean: '1.016e-04'
min: '-2.219e-02'
max: '1.619e-02'
mean: '3.114e-05'
min: '-1.955e-02'
shape:
- 128
- 128
sum: '1.665e+00'
sum: '5.102e-01'
grads.network.2.0.bias:
device: cuda:0
max: '3.969e-02'
mean: '-1.490e-09'
min: '-7.979e-02'
max: '3.893e-02'
mean: '-7.451e-10'
min: '-7.559e-02'
shape:
- 10
sum: '-1.490e-08'
sum: '-7.451e-09'
grads.network.2.0.weight:
device: cuda:0
max: '3.221e-02'
mean: '-1.928e-10'
min: '-6.755e-02'
max: '3.259e-02'
mean: '-9.604e-11'
min: '-4.695e-02'
shape:
- 10
- 128
sum: '-2.468e-07'
sum: '-1.229e-07'
outputs.logits:
device: cuda:0
max: '7.029e-01'
mean: '-3.564e-02'
min: '-7.781e-01'
max: '6.222e-01'
mean: '-3.729e-02'
min: '-6.079e-01'
shape:
- 128
- 10
sum: '-4.562e+01'
sum: '-4.773e+01'
outputs.loss:
device: cuda:0
max: '2.304e+00'
mean: '2.304e+00'
min: '2.304e+00'
max: '2.308e+00'
mean: '2.308e+00'
min: '2.308e+00'
shape: []
sum: '2.304e+00'
sum: '2.308e+00'
outputs.y:
device: cuda:0
max: 9
Expand Down
Loading
Loading