[contrib] Improve FusedAdamSWA interface and add unit tests #1759
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
Maybe @crcrpar, would you please review? Thanks! |
crcrpar
reviewed
Dec 14, 2023
| kBF16 = 1 | ||
| kFP32 = 2 | ||
| kFP64 = 3 | ||
| @unique |
There was a problem hiding this comment.
Didn't know there exists this decorator
There was a problem hiding this comment.
Yeah, this is a neat decorator ensuring unique enum values ;)
Comment on lines
+212
to
+225
| params: List[nn.Parameter], | ||
| compute_params: List[nn.Parameter], | ||
| swa_params: List[nn.Parameter], | ||
| swa_decay_rate: float, | ||
| lr: float = 1e-3, | ||
| bias_correction: bool = True, | ||
| betas: Tuple[float, float] = (0.9, 0.999), | ||
| eps: float = 1e-8, | ||
| adam_math_mode: AdamMathType = AdamMathType.PyTorchAdam, | ||
| weight_decay: float = 0.0, | ||
| amsgrad: bool = False, | ||
| set_grad_none: bool = True, | ||
| capturable: bool = False, | ||
| master_weights: bool = False, |
| chain(moments_gt, velocities_gt), | ||
| ) | ||
| ): | ||
| assert torch.allclose(m, m_gt, rtol=rtol, atol=atol) |
There was a problem hiding this comment.
Suggested change
| assert torch.allclose(m, m_gt, rtol=rtol, atol=atol) | |
| torch.testing.assert_close(m, m_gt, rtol=rtol, atol=atol) |
There was a problem hiding this comment.
Good catch! Fixed.
| chain(state_params_gt, compute_params_gt, swa_params_gt), | ||
| ) | ||
| ): | ||
| assert torch.allclose(p_test, p_gt, rtol=rtol, atol=atol) |
There was a problem hiding this comment.
Suggested change
| assert torch.allclose(p_test, p_gt, rtol=rtol, atol=atol) | |
| torch.testing.assert_close(p_test, p_gt, rtol=rtol, atol=atol) |
lirundong
force-pushed
the
lirundong/adam-swa-remove-sync
branch
from
d7101b4
to
e7afced
Compare
crcrpar
approved these changes
Dec 14, 2023
There was a problem hiding this comment.
idieally I want this to be compatible with python standard unittest module.
In this repository some file uses pytorch's TestCase class and runner as in https://github.com/NVIDIA/apex/blob/37d83fce4dcbb59897dfd951906493a6fe7fae37/tests/L0/run_fused_layer_norm/test_fused_layer_norm.py
There was a problem hiding this comment.
Thanks for the pointer. Have refactored to unittest.TestCase.
Why? - FusedAdamSWA interface was loosely typed and error-prone - The training critical path of FusedAdamSWA (i.e., its step function) could contain unnecessary GPU-host sync when grad_clip_scale is set to a non-CUDA-tensor variable - FusedAdamSWA didn't have any unit test What? - Encapsulated FusedAdamSWA math types and internal numerical type into Python enumerations to improve type robustness and readability - Accept grad_clip_scale as either a tensor or a number, for the latter case we move it to GPU in a non-blocking manner to eliminate a GPU-host sync - Add unit test to guarentee numerical correctness and demostrate usage
lirundong
force-pushed
the
lirundong/adam-swa-remove-sync
branch
from
e7afced
to
499bfdc
Compare
crcrpar
approved these changes
Dec 15, 2023
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Why?
What?