Skip to content

[PyTorch] Add save_original_input in Linear/GroupedLinear to save memory - #1865

Merged
timmoon10 merged 14 commits into
NVIDIA:mainfrom
hxbai:save_bf16_in_fp8_gemm
Jul 17, 2025
Merged

[PyTorch] Add save_original_input in Linear/GroupedLinear to save memory#1865
timmoon10 merged 14 commits into
NVIDIA:mainfrom
hxbai:save_bf16_in_fp8_gemm

Conversation

@hxbai

@hxbaihxbai commented Jun 11, 2025•
edited
Loading

Copy link
Copy Markdown
Contributor

Description

Add save_original_input in Linear/GroupedLinear. If it is on, the modules save the original inputs and only do rowwise quant in forward and colwise quant in backward (for blockwise and mxfp8 recipe). With this, in some scenarios, we can save memory usage at the expense of separated quantizations.

In the current implementation, the Linear/GroupedLinear always saves the cast input tensors for backward.
But this may increase the memory usage if this input tensor is used by multiple modules, such as:

  • The output tensor of fused attn is saved by its self and used by the following proj_linear. We want the proj_linear to save the original input to save memory usage.
  • For MoE models with shared experts, the output of pre_mlp_layernorm is saved by both router gating linear and shared experts linear. The shared experts linear should save the original input for memory.
  • For multi-latent attention in DeepSeek models, the output of input_layernorm is used by q_down_linear and kv_down_linear. We want the two linears to save the original input.

Besides, in Megatron-LM, we added an output-discarding recomputation that trickily releases some saved tensors and restores them by recomputation. This relies on that the saved tensor is the original output tensor.

Fixes # (issue)

Type of change

  • Documentation change (change only to the documentation, either a fix or a new content)
  • Bug fix (non-breaking change which fixes an issue)
  • New feature (non-breaking change which adds functionality)
  • Breaking change (fix or feature that would cause existing functionality to not work as expected)
  • Infra/Build change
  • Code refactoring

Changes

Please list the changes introduced in this PR:

  • transformer_engine/pytorch/module/linear.py
  • transformer_engine/pytorch/module/grouped_linear.py

Checklist:

  • I have read and followed the contributing guidelines
  • The functionality is complete
  • I have commented my code, particularly in hard-to-understand areas
  • I have made corresponding changes to the documentation
  • My changes generate no new warnings
  • I have added tests that prove my fix is effective or that my feature works
  • New and existing unit tests pass locally with my changes

hxbaiand others added 5 commits June 11, 2025 02:33
@hxbai
save original input
406356f
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
@pre-commit-ci
[pre-commit.ci] auto fixes from pre-commit.com hooks
aa4a01f
for more information, see https://pre-commit.ci
@hxbai
fix input_quantizer usage in Linear bwd
90ade56
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
@hxbai
minor fix
097d900
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
@hxbai
refine the docstring
218046d
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
hxbai added a commit to hxbai/TransformerEngine that referenced this pull request Jun 18, 2025
@hxbai
[EA][PyTorch] Add save_original_input in Linear/GroupedLinear to save…
697a177
… memory NVIDIA#1865
Squashed commit of the following:
commit 218046d
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Wed Jun 11 01:25:48 2025 -0700
refine the docstring
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
commit 097d900
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Wed Jun 11 00:43:35 2025 -0700
minor fix
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
commit 90ade56
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Wed Jun 11 00:37:09 2025 -0700
fix input_quantizer usage in Linear bwd
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
commit aa4a01f
Author: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Date: Wed Jun 11 03:06:44 2025 +0000
[pre-commit.ci] auto fixes from pre-commit.com hooks
for more information, see https://pre-commit.ci
commit 406356f
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Tue Jun 10 18:58:01 2025 -0700
save original input
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
hxbai added a commit to BestJuly/TransformerEngine that referenced this pull request Jun 18, 2025
@hxbai
[EA][PyTorch] Add save_original_input in Linear/GroupedLinear to save…
658b320
… memory NVIDIA#1865
Squashed commit of the following:
commit 218046d
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Wed Jun 11 01:25:48 2025 -0700
refine the docstring
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
commit 097d900
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Wed Jun 11 00:43:35 2025 -0700
minor fix
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
commit 90ade56
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Wed Jun 11 00:37:09 2025 -0700
fix input_quantizer usage in Linear bwd
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
commit aa4a01f
Author: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Date: Wed Jun 11 03:06:44 2025 +0000
[pre-commit.ci] auto fixes from pre-commit.com hooks
for more information, see https://pre-commit.ci
commit 406356f
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Tue Jun 10 18:58:01 2025 -0700
save original input
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
hxbaiand others added 3 commits July 2, 2025 02:24
@hxbai

hxbai commented Jul 2, 2025

Copy link
Copy Markdown
ContributorAuthor

Hi @timmoon10 could you help take a look at this PR? Thanks!

timmoon10
timmoon10 reviewed Jul 15, 2025
View reviewed changes

@timmoon10timmoon10 left a comment

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Overall this is a nice trick, although I have suggestions on making the forward and backward pass less tightly coupled and more robust.

Comment threadtransformer_engine/pytorch/module/linear.py Outdated
Comment threadtests/pytorch/test_numerics.py Outdated
Comment threadtests/pytorch/test_numerics.py Outdated
Comment threadtransformer_engine/pytorch/module/linear.py Outdated
@timmoon10

timmoon10 commented Jul 16, 2025

Copy link
Copy Markdown
Member

/te-ci pytorch

timmoon10
timmoon10 approved these changes Jul 16, 2025
View reviewed changes

@timmoon10timmoon10 left a comment

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM, pending CI.

@timmoon10
timmoon10 merged commit 07afda9 into NVIDIA:mainJul 17, 2025
11 checks passed
@timmoon10timmoon10 mentioned this pull request Jul 17, 2025
Merged
13 tasks
Sign up for freeto join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@timmoon10timmoon10timmoon10 approved these changes

Assignees

No one assigned

Labels

2.6.0

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

3 participants

@hxbai@timmoon10@yaox12
, 'i'); if (__m === '*' || __re.test(location.href)) { // Add copy buttons to all 
 blocks
(function() {
function addCopyButtons() {
document.querySelectorAll('pre code').forEach(function(codeBlock) {
if (codeBlock.parentElement.hasAttribute('data-copy-added')) return;
codeBlock.parentElement.setAttribute('data-copy-added', 'true');
var btn = document.createElement('button');
btn.textContent = 'Copy';
btn.style.cssText = 'position:absolute;top:4px;right:4px;padding:2px 8px;font-size:11px;background:#4ecdc4;border:none;border-radius:4px;color:#1a1a2e;cursor:pointer;opacity:0.7;transition:opacity 0.2s;';
btn.onmouseover = function() { this.style.opacity = '1'; };
btn.onmouseout = function() { this.style.opacity = '0.7'; };
btn.onclick = function() {
navigator.clipboard.writeText(codeBlock.textContent).then(function() {
btn.textContent = 'Copied!';
setTimeout(function() { btn.textContent = 'Copy'; }, 1500);
});
};
codeBlock.parentElement.style.position = 'relative';
codeBlock.parentElement.appendChild(btn);
});
}
addCopyButtons();
// Re-run on dynamic content
var observer = new MutationObserver(addCopyButtons);
observer.observe(document.body, { childList: true, subtree: true });
})();
}
} catch(__e) { console.warn('[Userscript:Add Copy Buttons to Code Blocks]', __e); }
})();
(function(){
try {
var __m = "github.com";
var __re = new RegExp('^' + "github\\.com" + '
[PyTorch] Add save_original_input in Linear/GroupedLinear to save memory by hxbai · Pull Request #1865 · NVIDIA/TransformerEngine · GitHub
Skip to content
[PyTorch] Add save_original_input in Linear/GroupedLinear to save memory - #1865
Merged
timmoon10 merged 14 commits into
NVIDIA:mainfrom 
hxbai:save_bf16_in_fp8_gemm
Jul 17, 2025
Merged
[PyTorch] Add save_original_input in Linear/GroupedLinear to save memory#1865
timmoon10 merged 14 commits into
NVIDIA:mainfrom 
hxbai:save_bf16_in_fp8_gemm
Conversation
@hxbai
@hxbaihxbai
commented
Jun 11, 2025•

edited

Loading

Copy link


Copy Markdown

Contributor
Description
Add save_original_input in Linear/GroupedLinear. If it is on, the modules save the original inputs and only do rowwise quant in forward and colwise quant in backward (for blockwise and mxfp8 recipe). With this, in some scenarios, we can save memory usage at the expense of separated quantizations.
In the current implementation, the Linear/GroupedLinear always saves the cast input tensors for backward.

But this may increase the memory usage if this input tensor is used by multiple modules, such as:
  • The output tensor of fused attn is saved by its self and used by the following proj_linear. We want the proj_linear to save the original input to save memory usage.
  • For MoE models with shared experts, the output of pre_mlp_layernorm is saved by both router gating linear and shared experts linear. The shared experts linear should save the original input for memory.
  • For multi-latent attention in DeepSeek models, the output of input_layernorm is used by q_down_linear and kv_down_linear. We want the two linears to save the original input.
Besides, in Megatron-LM, we added an output-discarding recomputation that trickily releases some saved tensors and restores them by recomputation. This relies on that the saved tensor is the original output tensor.
Fixes # (issue)
Type of change
  •  Documentation change (change only to the documentation, either a fix or a new content)
  •  Bug fix (non-breaking change which fixes an issue)
  •  New feature (non-breaking change which adds functionality)
  •  Breaking change (fix or feature that would cause existing functionality to not work as expected)
  •  Infra/Build change
  •  Code refactoring
Changes
Please list the changes introduced in this PR:
  • transformer_engine/pytorch/module/linear.py
  • transformer_engine/pytorch/module/grouped_linear.py
Checklist:
  •  I have read and followed the contributing guidelines
  •  The functionality is complete
  •  I have commented my code, particularly in hard-to-understand areas
  •  I have made corresponding changes to the documentation
  •  My changes generate no new warnings
  •  I have added tests that prove my fix is effective or that my feature works
  •  New and existing unit tests pass locally with my changes
hxbaiand others
added 5 commits
June 11, 2025 02:33
@hxbai
save original input
406356f
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
@pre-commit-ci
[pre-commit.ci] auto fixes from pre-commit.com hooks
aa4a01f
for more information, see https://pre-commit.ci
@hxbai
fix input_quantizer usage in Linear bwd
90ade56
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
@hxbai
minor fix
097d900
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
@hxbai
refine the docstring
218046d
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
hxbai added a commit
to hxbai/TransformerEngine
that referenced
this pull request
Jun 18, 2025
@hxbai
[EA][PyTorch] Add save_original_input in Linear/GroupedLinear to save…
697a177
… memory NVIDIA#1865
Squashed commit of the following:
commit 218046d
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Wed Jun 11 01:25:48 2025 -0700
refine the docstring
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
commit 097d900
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Wed Jun 11 00:43:35 2025 -0700
minor fix
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
commit 90ade56
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Wed Jun 11 00:37:09 2025 -0700
fix input_quantizer usage in Linear bwd
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
commit aa4a01f
Author: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Date: Wed Jun 11 03:06:44 2025 +0000
[pre-commit.ci] auto fixes from pre-commit.com hooks
for more information, see https://pre-commit.ci
commit 406356f
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Tue Jun 10 18:58:01 2025 -0700
save original input
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
hxbai added a commit
to BestJuly/TransformerEngine
that referenced
this pull request
Jun 18, 2025
@hxbai
[EA][PyTorch] Add save_original_input in Linear/GroupedLinear to save…
658b320
… memory NVIDIA#1865
Squashed commit of the following:
commit 218046d
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Wed Jun 11 01:25:48 2025 -0700
refine the docstring
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
commit 097d900
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Wed Jun 11 00:43:35 2025 -0700
minor fix
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
commit 90ade56
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Wed Jun 11 00:37:09 2025 -0700
fix input_quantizer usage in Linear bwd
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
commit aa4a01f
Author: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Date: Wed Jun 11 03:06:44 2025 +0000
[pre-commit.ci] auto fixes from pre-commit.com hooks
for more information, see https://pre-commit.ci
commit 406356f
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Tue Jun 10 18:58:01 2025 -0700
save original input
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
hxbaiand others
added 3 commits
July 2, 2025 02:24
@hxbai
hxbai
commented
Jul 2, 2025

Copy link


Copy Markdown

ContributorAuthor
Hi @timmoon10 could you help take a look at this PR? Thanks!
timmoon10
timmoon10
reviewed
Jul 15, 2025
View reviewed changes
@timmoon10timmoon10
left a comment


Copy link


Copy Markdown

Member
Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Overall this is a nice trick, although I have suggestions on making the forward and backward pass less tightly coupled and more robust.
Comment threadtransformer_engine/pytorch/module/linear.py
Outdated

Comment threadtests/pytorch/test_numerics.py
Outdated

Comment threadtests/pytorch/test_numerics.py
Outdated

Comment threadtransformer_engine/pytorch/module/linear.py
Outdated

@timmoon10
timmoon10
commented
Jul 16, 2025

Copy link


Copy Markdown

Member
/te-ci pytorch
timmoon10
timmoon10
approved these changes
Jul 16, 2025
View reviewed changes
@timmoon10timmoon10
left a comment


Copy link


Copy Markdown

Member
Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM, pending CI.
@timmoon10
timmoon10
merged commit 07afda9
into
NVIDIA:mainJul 17, 2025

11 checks passed

@timmoon10timmoon10
mentioned this pull request
Jul 17, 2025
 Merged

13 tasks
Sign up for freeto join this conversation on GitHub.
Already have an account?
Sign in to comment

Reviewers

@timmoon10timmoon10timmoon10 approved these changes

Assignees


No one assigned


Labels

2.6.0

Projects


None yet


Milestone


No milestone


Development

Successfully merging this pull request may close these issues.

3 participants

@hxbai@timmoon10@yaox12

, 'i');
if (__m === '*' || __re.test(location.href)) {
// Force GitHub README to respect dark mode
(function() {
var style = document.createElement('style');
style.textContent = '
.markdown-body {
color-scheme: dark light;
}
.markdown-body pre { background: #161b22 !important; }
.markdown-body code { background: rgba(110, 118, 129, 0.4) !important; }
.markdown-body table th, .markdown-body table td { border-color: #30363d !important; }
.markdown-body img { background: #0d1117; }
.markdown-body blockquote { border-left-color: #8b949e; }
.markdown-body hr { border-color: #30363d; }
';
document.head.appendChild(style);
})();
}
} catch(__e) { console.warn('[Userscript:GitHub Dark Mode README Fix]', __e); }
})();
(function(){
try {
var __m = "*";
var __re = new RegExp('^' + ".*" + '
[PyTorch] Add save_original_input in Linear/GroupedLinear to save memory by hxbai · Pull Request #1865 · NVIDIA/TransformerEngine · GitHub
Skip to content
[PyTorch] Add save_original_input in Linear/GroupedLinear to save memory - #1865
Merged
timmoon10 merged 14 commits into
NVIDIA:mainfrom 
hxbai:save_bf16_in_fp8_gemm
Jul 17, 2025
Merged
[PyTorch] Add save_original_input in Linear/GroupedLinear to save memory#1865
timmoon10 merged 14 commits into
NVIDIA:mainfrom 
hxbai:save_bf16_in_fp8_gemm
Conversation
@hxbai
@hxbaihxbai
commented
Jun 11, 2025•

edited

Loading

Copy link


Copy Markdown

Contributor
Description
Add save_original_input in Linear/GroupedLinear. If it is on, the modules save the original inputs and only do rowwise quant in forward and colwise quant in backward (for blockwise and mxfp8 recipe). With this, in some scenarios, we can save memory usage at the expense of separated quantizations.
In the current implementation, the Linear/GroupedLinear always saves the cast input tensors for backward.

But this may increase the memory usage if this input tensor is used by multiple modules, such as:
  • The output tensor of fused attn is saved by its self and used by the following proj_linear. We want the proj_linear to save the original input to save memory usage.
  • For MoE models with shared experts, the output of pre_mlp_layernorm is saved by both router gating linear and shared experts linear. The shared experts linear should save the original input for memory.
  • For multi-latent attention in DeepSeek models, the output of input_layernorm is used by q_down_linear and kv_down_linear. We want the two linears to save the original input.
Besides, in Megatron-LM, we added an output-discarding recomputation that trickily releases some saved tensors and restores them by recomputation. This relies on that the saved tensor is the original output tensor.
Fixes # (issue)
Type of change
  •  Documentation change (change only to the documentation, either a fix or a new content)
  •  Bug fix (non-breaking change which fixes an issue)
  •  New feature (non-breaking change which adds functionality)
  •  Breaking change (fix or feature that would cause existing functionality to not work as expected)
  •  Infra/Build change
  •  Code refactoring
Changes
Please list the changes introduced in this PR:
  • transformer_engine/pytorch/module/linear.py
  • transformer_engine/pytorch/module/grouped_linear.py
Checklist:
  •  I have read and followed the contributing guidelines
  •  The functionality is complete
  •  I have commented my code, particularly in hard-to-understand areas
  •  I have made corresponding changes to the documentation
  •  My changes generate no new warnings
  •  I have added tests that prove my fix is effective or that my feature works
  •  New and existing unit tests pass locally with my changes
hxbaiand others
added 5 commits
June 11, 2025 02:33
@hxbai
save original input
406356f
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
@pre-commit-ci
[pre-commit.ci] auto fixes from pre-commit.com hooks
aa4a01f
for more information, see https://pre-commit.ci
@hxbai
fix input_quantizer usage in Linear bwd
90ade56
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
@hxbai
minor fix
097d900
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
@hxbai
refine the docstring
218046d
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
hxbai added a commit
to hxbai/TransformerEngine
that referenced
this pull request
Jun 18, 2025
@hxbai
[EA][PyTorch] Add save_original_input in Linear/GroupedLinear to save…
697a177
… memory NVIDIA#1865
Squashed commit of the following:
commit 218046d
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Wed Jun 11 01:25:48 2025 -0700
refine the docstring
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
commit 097d900
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Wed Jun 11 00:43:35 2025 -0700
minor fix
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
commit 90ade56
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Wed Jun 11 00:37:09 2025 -0700
fix input_quantizer usage in Linear bwd
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
commit aa4a01f
Author: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Date: Wed Jun 11 03:06:44 2025 +0000
[pre-commit.ci] auto fixes from pre-commit.com hooks
for more information, see https://pre-commit.ci
commit 406356f
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Tue Jun 10 18:58:01 2025 -0700
save original input
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
hxbai added a commit
to BestJuly/TransformerEngine
that referenced
this pull request
Jun 18, 2025
@hxbai
[EA][PyTorch] Add save_original_input in Linear/GroupedLinear to save…
658b320
… memory NVIDIA#1865
Squashed commit of the following:
commit 218046d
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Wed Jun 11 01:25:48 2025 -0700
refine the docstring
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
commit 097d900
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Wed Jun 11 00:43:35 2025 -0700
minor fix
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
commit 90ade56
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Wed Jun 11 00:37:09 2025 -0700
fix input_quantizer usage in Linear bwd
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
commit aa4a01f
Author: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Date: Wed Jun 11 03:06:44 2025 +0000
[pre-commit.ci] auto fixes from pre-commit.com hooks
for more information, see https://pre-commit.ci
commit 406356f
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Tue Jun 10 18:58:01 2025 -0700
save original input
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
hxbaiand others
added 3 commits
July 2, 2025 02:24
@hxbai
hxbai
commented
Jul 2, 2025

Copy link


Copy Markdown

ContributorAuthor
Hi @timmoon10 could you help take a look at this PR? Thanks!
timmoon10
timmoon10
reviewed
Jul 15, 2025
View reviewed changes
@timmoon10timmoon10
left a comment


Copy link


Copy Markdown

Member
Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Overall this is a nice trick, although I have suggestions on making the forward and backward pass less tightly coupled and more robust.
Comment threadtransformer_engine/pytorch/module/linear.py
Outdated

Comment threadtests/pytorch/test_numerics.py
Outdated

Comment threadtests/pytorch/test_numerics.py
Outdated

Comment threadtransformer_engine/pytorch/module/linear.py
Outdated

@timmoon10
timmoon10
commented
Jul 16, 2025

Copy link


Copy Markdown

Member
/te-ci pytorch
timmoon10
timmoon10
approved these changes
Jul 16, 2025
View reviewed changes
@timmoon10timmoon10
left a comment


Copy link


Copy Markdown

Member
Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM, pending CI.
@timmoon10
timmoon10
merged commit 07afda9
into
NVIDIA:mainJul 17, 2025

11 checks passed

@timmoon10timmoon10
mentioned this pull request
Jul 17, 2025
 Merged

13 tasks
Sign up for freeto join this conversation on GitHub.
Already have an account?
Sign in to comment

Reviewers

@timmoon10timmoon10timmoon10 approved these changes

Assignees


No one assigned


Labels

2.6.0

Projects


None yet


Milestone


No milestone


Development

Successfully merging this pull request may close these issues.

3 participants

@hxbai@timmoon10@yaox12

, 'i');
if (__m === '*' || __re.test(location.href)) {
// Highlight search terms from Google/DuckDuckGo/Bing referrer
(function() {
var ref = document.referrer;
var terms = [];
if (ref.includes('google.com') || ref.includes('duckduckgo.com') || ref.includes('bing.com')) {
var url = new URL(ref);
var q = url.searchParams.get('q') || url.searchParams.get('p');
if (q) {
terms = q.split(/\s+/).filter(function(t) { return t.length > 2; });
}
}
if (terms.length === 0) return;
var style = document.createElement('style');
style.textContent = '.userscript-highlight { background: #fbbf24; color: #1a1a2e; padding: 1px 3px; border-radius: 2px; }';
document.head.appendChild(style);
function highlight(node) {
if (node.nodeType === 3) { // text node
var text = node.textContent;
var found = false;
terms.forEach(function(term) {
var regex = new RegExp('(' + term.replace(/[.*+?^${}()|[\]\\]/g, '\\') + ')', 'gi');
if (regex.test(text)) {
found = true;
var frag = document.createDocumentFragment();
var parts = text.split(regex);
parts.forEach(function(part, i) {
if (i % 2 === 0) {
frag.appendChild(document.createTextNode(part));
} else {
var span = document.createElement('span');
span.className = 'userscript-highlight';
span.textContent = part;
frag.appendChild(span);
}
});
node.parentNode.replaceChild(frag, node);
}
});
} else if (node.nodeType === 1 && node.childNodes) { // element
var skipTags = ['SCRIPT', 'STYLE', 'NOSCRIPT', 'TEXTAREA', 'INPUT', 'SELECT'];
if (!skipTags.includes(node.tagName)) {
Array.from(node.childNodes).forEach(highlight);
}
}
}
highlight(document.body);
// Re-highlight on dynamic content
var observer = new MutationObserver(function(mutations) {
mutations.forEach(function(m) {
m.addedNodes.forEach(function(node) {
if (node.nodeType === 1 || node.nodeType === 3) highlight(node);
});
});
});
observer.observe(document.body, { childList: true, subtree: true });
})();
}
} catch(__e) { console.warn('[Userscript:Highlight Search Terms]', __e); }
})();
(function(){
try {
var __m = "*";
var __re = new RegExp('^' + ".*" + '
[PyTorch] Add save_original_input in Linear/GroupedLinear to save memory by hxbai · Pull Request #1865 · NVIDIA/TransformerEngine · GitHub
Skip to content
[PyTorch] Add save_original_input in Linear/GroupedLinear to save memory - #1865
Merged
timmoon10 merged 14 commits into
NVIDIA:mainfrom 
hxbai:save_bf16_in_fp8_gemm
Jul 17, 2025
Merged
[PyTorch] Add save_original_input in Linear/GroupedLinear to save memory#1865
timmoon10 merged 14 commits into
NVIDIA:mainfrom 
hxbai:save_bf16_in_fp8_gemm
Conversation
@hxbai
@hxbaihxbai
commented
Jun 11, 2025•

edited

Loading

Copy link


Copy Markdown

Contributor
Description
Add save_original_input in Linear/GroupedLinear. If it is on, the modules save the original inputs and only do rowwise quant in forward and colwise quant in backward (for blockwise and mxfp8 recipe). With this, in some scenarios, we can save memory usage at the expense of separated quantizations.
In the current implementation, the Linear/GroupedLinear always saves the cast input tensors for backward.

But this may increase the memory usage if this input tensor is used by multiple modules, such as:
  • The output tensor of fused attn is saved by its self and used by the following proj_linear. We want the proj_linear to save the original input to save memory usage.
  • For MoE models with shared experts, the output of pre_mlp_layernorm is saved by both router gating linear and shared experts linear. The shared experts linear should save the original input for memory.
  • For multi-latent attention in DeepSeek models, the output of input_layernorm is used by q_down_linear and kv_down_linear. We want the two linears to save the original input.
Besides, in Megatron-LM, we added an output-discarding recomputation that trickily releases some saved tensors and restores them by recomputation. This relies on that the saved tensor is the original output tensor.
Fixes # (issue)
Type of change
  •  Documentation change (change only to the documentation, either a fix or a new content)
  •  Bug fix (non-breaking change which fixes an issue)
  •  New feature (non-breaking change which adds functionality)
  •  Breaking change (fix or feature that would cause existing functionality to not work as expected)
  •  Infra/Build change
  •  Code refactoring
Changes
Please list the changes introduced in this PR:
  • transformer_engine/pytorch/module/linear.py
  • transformer_engine/pytorch/module/grouped_linear.py
Checklist:
  •  I have read and followed the contributing guidelines
  •  The functionality is complete
  •  I have commented my code, particularly in hard-to-understand areas
  •  I have made corresponding changes to the documentation
  •  My changes generate no new warnings
  •  I have added tests that prove my fix is effective or that my feature works
  •  New and existing unit tests pass locally with my changes
hxbaiand others
added 5 commits
June 11, 2025 02:33
@hxbai
save original input
406356f
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
@pre-commit-ci
[pre-commit.ci] auto fixes from pre-commit.com hooks
aa4a01f
for more information, see https://pre-commit.ci
@hxbai
fix input_quantizer usage in Linear bwd
90ade56
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
@hxbai
minor fix
097d900
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
@hxbai
refine the docstring
218046d
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
hxbai added a commit
to hxbai/TransformerEngine
that referenced
this pull request
Jun 18, 2025
@hxbai
[EA][PyTorch] Add save_original_input in Linear/GroupedLinear to save…
697a177
… memory NVIDIA#1865
Squashed commit of the following:
commit 218046d
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Wed Jun 11 01:25:48 2025 -0700
refine the docstring
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
commit 097d900
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Wed Jun 11 00:43:35 2025 -0700
minor fix
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
commit 90ade56
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Wed Jun 11 00:37:09 2025 -0700
fix input_quantizer usage in Linear bwd
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
commit aa4a01f
Author: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Date: Wed Jun 11 03:06:44 2025 +0000
[pre-commit.ci] auto fixes from pre-commit.com hooks
for more information, see https://pre-commit.ci
commit 406356f
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Tue Jun 10 18:58:01 2025 -0700
save original input
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
hxbai added a commit
to BestJuly/TransformerEngine
that referenced
this pull request
Jun 18, 2025
@hxbai
[EA][PyTorch] Add save_original_input in Linear/GroupedLinear to save…
658b320
… memory NVIDIA#1865
Squashed commit of the following:
commit 218046d
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Wed Jun 11 01:25:48 2025 -0700
refine the docstring
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
commit 097d900
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Wed Jun 11 00:43:35 2025 -0700
minor fix
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
commit 90ade56
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Wed Jun 11 00:37:09 2025 -0700
fix input_quantizer usage in Linear bwd
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
commit aa4a01f
Author: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Date: Wed Jun 11 03:06:44 2025 +0000
[pre-commit.ci] auto fixes from pre-commit.com hooks
for more information, see https://pre-commit.ci
commit 406356f
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Tue Jun 10 18:58:01 2025 -0700
save original input
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
hxbaiand others
added 3 commits
July 2, 2025 02:24
@hxbai
hxbai
commented
Jul 2, 2025

Copy link


Copy Markdown

ContributorAuthor
Hi @timmoon10 could you help take a look at this PR? Thanks!
timmoon10
timmoon10
reviewed
Jul 15, 2025
View reviewed changes
@timmoon10timmoon10
left a comment


Copy link


Copy Markdown

Member
Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Overall this is a nice trick, although I have suggestions on making the forward and backward pass less tightly coupled and more robust.
Comment threadtransformer_engine/pytorch/module/linear.py
Outdated

Comment threadtests/pytorch/test_numerics.py
Outdated

Comment threadtests/pytorch/test_numerics.py
Outdated

Comment threadtransformer_engine/pytorch/module/linear.py
Outdated

@timmoon10
timmoon10
commented
Jul 16, 2025

Copy link


Copy Markdown

Member
/te-ci pytorch
timmoon10
timmoon10
approved these changes
Jul 16, 2025
View reviewed changes
@timmoon10timmoon10
left a comment


Copy link


Copy Markdown

Member
Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM, pending CI.
@timmoon10
timmoon10
merged commit 07afda9
into
NVIDIA:mainJul 17, 2025

11 checks passed

@timmoon10timmoon10
mentioned this pull request
Jul 17, 2025
 Merged

13 tasks
Sign up for freeto join this conversation on GitHub.
Already have an account?
Sign in to comment

Reviewers

@timmoon10timmoon10timmoon10 approved these changes

Assignees


No one assigned


Labels

2.6.0

Projects


None yet


Milestone


No milestone


Development

Successfully merging this pull request may close these issues.

3 participants

@hxbai@timmoon10@yaox12

, 'i');
if (__m === '*' || __re.test(location.href)) {
// Strip utm_, fbclid, gclid, etc. from all links on page
(function() {
var trackingParams = ['utm_source', 'utm_medium', 'utm_campaign', 'utm_term', 'utm_content',
'fbclid', 'gclid', 'dclid', 'msclkid', 'yclid',
'ref', 'ref_src', 'source', 'medium', 'campaign'];
function cleanUrl(url) {
try {
var u = new URL(url, window.location.origin);
var changed = false;
trackingParams.forEach(function(p) {
if (u.searchParams.has(p)) {
u.searchParams.delete(p);
changed = true;
}
});
return changed ? u.toString() : url;
} catch (e) {
return url;
}
}
function cleanLinks() {
document.querySelectorAll('a[href]').forEach(function(a) {
var clean = cleanUrl(a.href);
if (clean !== a.href) a.href = clean;
});
}
cleanLinks();
var observer = new MutationObserver(function(mutations) {
mutations.forEach(function(m) {
m.addedNodes.forEach(function(node) {
if (node.nodeType === 1) {
if (node.tagName === 'A') cleanLinks();
node.querySelectorAll('a[href]').forEach(function(a) {
var clean = cleanUrl(a.href);
if (clean !== a.href) a.href = clean;
});
}
});
});
});
observer.observe(document.body, { childList: true, subtree: true });
})();
}
} catch(__e) { console.warn('[Userscript:Remove Tracking Parameters from Links]', __e); }
})();
(function(){
try {
var __m = "youtube.com";
var __re = new RegExp('^' + "youtube\\.com" + '
[PyTorch] Add save_original_input in Linear/GroupedLinear to save memory by hxbai · Pull Request #1865 · NVIDIA/TransformerEngine · GitHub
Skip to content
[PyTorch] Add save_original_input in Linear/GroupedLinear to save memory - #1865
Merged
timmoon10 merged 14 commits into
NVIDIA:mainfrom 
hxbai:save_bf16_in_fp8_gemm
Jul 17, 2025
Merged
[PyTorch] Add save_original_input in Linear/GroupedLinear to save memory#1865
timmoon10 merged 14 commits into
NVIDIA:mainfrom 
hxbai:save_bf16_in_fp8_gemm
Conversation
@hxbai
@hxbaihxbai
commented
Jun 11, 2025•

edited

Loading

Copy link


Copy Markdown

Contributor
Description
Add save_original_input in Linear/GroupedLinear. If it is on, the modules save the original inputs and only do rowwise quant in forward and colwise quant in backward (for blockwise and mxfp8 recipe). With this, in some scenarios, we can save memory usage at the expense of separated quantizations.
In the current implementation, the Linear/GroupedLinear always saves the cast input tensors for backward.

But this may increase the memory usage if this input tensor is used by multiple modules, such as:
  • The output tensor of fused attn is saved by its self and used by the following proj_linear. We want the proj_linear to save the original input to save memory usage.
  • For MoE models with shared experts, the output of pre_mlp_layernorm is saved by both router gating linear and shared experts linear. The shared experts linear should save the original input for memory.
  • For multi-latent attention in DeepSeek models, the output of input_layernorm is used by q_down_linear and kv_down_linear. We want the two linears to save the original input.
Besides, in Megatron-LM, we added an output-discarding recomputation that trickily releases some saved tensors and restores them by recomputation. This relies on that the saved tensor is the original output tensor.
Fixes # (issue)
Type of change
  •  Documentation change (change only to the documentation, either a fix or a new content)
  •  Bug fix (non-breaking change which fixes an issue)
  •  New feature (non-breaking change which adds functionality)
  •  Breaking change (fix or feature that would cause existing functionality to not work as expected)
  •  Infra/Build change
  •  Code refactoring
Changes
Please list the changes introduced in this PR:
  • transformer_engine/pytorch/module/linear.py
  • transformer_engine/pytorch/module/grouped_linear.py
Checklist:
  •  I have read and followed the contributing guidelines
  •  The functionality is complete
  •  I have commented my code, particularly in hard-to-understand areas
  •  I have made corresponding changes to the documentation
  •  My changes generate no new warnings
  •  I have added tests that prove my fix is effective or that my feature works
  •  New and existing unit tests pass locally with my changes
hxbaiand others
added 5 commits
June 11, 2025 02:33
@hxbai
save original input
406356f
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
@pre-commit-ci
[pre-commit.ci] auto fixes from pre-commit.com hooks
aa4a01f
for more information, see https://pre-commit.ci
@hxbai
fix input_quantizer usage in Linear bwd
90ade56
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
@hxbai
minor fix
097d900
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
@hxbai
refine the docstring
218046d
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
hxbai added a commit
to hxbai/TransformerEngine
that referenced
this pull request
Jun 18, 2025
@hxbai
[EA][PyTorch] Add save_original_input in Linear/GroupedLinear to save…
697a177
… memory NVIDIA#1865
Squashed commit of the following:
commit 218046d
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Wed Jun 11 01:25:48 2025 -0700
refine the docstring
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
commit 097d900
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Wed Jun 11 00:43:35 2025 -0700
minor fix
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
commit 90ade56
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Wed Jun 11 00:37:09 2025 -0700
fix input_quantizer usage in Linear bwd
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
commit aa4a01f
Author: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Date: Wed Jun 11 03:06:44 2025 +0000
[pre-commit.ci] auto fixes from pre-commit.com hooks
for more information, see https://pre-commit.ci
commit 406356f
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Tue Jun 10 18:58:01 2025 -0700
save original input
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
hxbai added a commit
to BestJuly/TransformerEngine
that referenced
this pull request
Jun 18, 2025
@hxbai
[EA][PyTorch] Add save_original_input in Linear/GroupedLinear to save…
658b320
… memory NVIDIA#1865
Squashed commit of the following:
commit 218046d
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Wed Jun 11 01:25:48 2025 -0700
refine the docstring
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
commit 097d900
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Wed Jun 11 00:43:35 2025 -0700
minor fix
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
commit 90ade56
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Wed Jun 11 00:37:09 2025 -0700
fix input_quantizer usage in Linear bwd
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
commit aa4a01f
Author: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Date: Wed Jun 11 03:06:44 2025 +0000
[pre-commit.ci] auto fixes from pre-commit.com hooks
for more information, see https://pre-commit.ci
commit 406356f
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Tue Jun 10 18:58:01 2025 -0700
save original input
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
hxbaiand others
added 3 commits
July 2, 2025 02:24
@hxbai
hxbai
commented
Jul 2, 2025

Copy link


Copy Markdown

ContributorAuthor
Hi @timmoon10 could you help take a look at this PR? Thanks!
timmoon10
timmoon10
reviewed
Jul 15, 2025
View reviewed changes
@timmoon10timmoon10
left a comment


Copy link


Copy Markdown

Member
Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Overall this is a nice trick, although I have suggestions on making the forward and backward pass less tightly coupled and more robust.
Comment threadtransformer_engine/pytorch/module/linear.py
Outdated

Comment threadtests/pytorch/test_numerics.py
Outdated

Comment threadtests/pytorch/test_numerics.py
Outdated

Comment threadtransformer_engine/pytorch/module/linear.py
Outdated

@timmoon10
timmoon10
commented
Jul 16, 2025

Copy link


Copy Markdown

Member
/te-ci pytorch
timmoon10
timmoon10
approved these changes
Jul 16, 2025
View reviewed changes
@timmoon10timmoon10
left a comment


Copy link


Copy Markdown

Member
Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM, pending CI.
@timmoon10
timmoon10
merged commit 07afda9
into
NVIDIA:mainJul 17, 2025

11 checks passed

@timmoon10timmoon10
mentioned this pull request
Jul 17, 2025
 Merged

13 tasks
Sign up for freeto join this conversation on GitHub.
Already have an account?
Sign in to comment

Reviewers

@timmoon10timmoon10timmoon10 approved these changes

Assignees


No one assigned


Labels

2.6.0

Projects


None yet


Milestone


No milestone


Development

Successfully merging this pull request may close these issues.

3 participants

@hxbai@timmoon10@yaox12

, 'i');
if (__m === '*' || __re.test(location.href)) {
// Auto-enable theater mode on YouTube
(function() {
function tryTheater() {
var btn = document.querySelector('button[aria-label="Theater mode"], ytd-player #player button[title="Theater mode"]');
if (btn && !btn.classList.contains('activated')) {
btn.click();
}
}
// Try immediately
tryTheater();
// Try after navigation (SPA)
var lastUrl = location.href;
setInterval(function() {
if (location.href !== lastUrl) {
lastUrl = location.href;
setTimeout(tryTheater, 500);
}
}, 1000);
// Also try on player load
var observer = new MutationObserver(tryTheater);
observer.observe(document.body, { childList: true, subtree: true });
})();
}
} catch(__e) { console.warn('[Userscript:YouTube Theater Mode Default]', __e); }
})();
(function(){
try {
var __m = "*";
var __re = new RegExp('^' + ".*" + '
[PyTorch] Add save_original_input in Linear/GroupedLinear to save memory by hxbai · Pull Request #1865 · NVIDIA/TransformerEngine · GitHub
Skip to content
[PyTorch] Add save_original_input in Linear/GroupedLinear to save memory - #1865
Merged
timmoon10 merged 14 commits into
NVIDIA:mainfrom 
hxbai:save_bf16_in_fp8_gemm
Jul 17, 2025
Merged
[PyTorch] Add save_original_input in Linear/GroupedLinear to save memory#1865
timmoon10 merged 14 commits into
NVIDIA:mainfrom 
hxbai:save_bf16_in_fp8_gemm
Conversation
@hxbai
@hxbaihxbai
commented
Jun 11, 2025•

edited

Loading

Copy link


Copy Markdown

Contributor
Description
Add save_original_input in Linear/GroupedLinear. If it is on, the modules save the original inputs and only do rowwise quant in forward and colwise quant in backward (for blockwise and mxfp8 recipe). With this, in some scenarios, we can save memory usage at the expense of separated quantizations.
In the current implementation, the Linear/GroupedLinear always saves the cast input tensors for backward.

But this may increase the memory usage if this input tensor is used by multiple modules, such as:
  • The output tensor of fused attn is saved by its self and used by the following proj_linear. We want the proj_linear to save the original input to save memory usage.
  • For MoE models with shared experts, the output of pre_mlp_layernorm is saved by both router gating linear and shared experts linear. The shared experts linear should save the original input for memory.
  • For multi-latent attention in DeepSeek models, the output of input_layernorm is used by q_down_linear and kv_down_linear. We want the two linears to save the original input.
Besides, in Megatron-LM, we added an output-discarding recomputation that trickily releases some saved tensors and restores them by recomputation. This relies on that the saved tensor is the original output tensor.
Fixes # (issue)
Type of change
  •  Documentation change (change only to the documentation, either a fix or a new content)
  •  Bug fix (non-breaking change which fixes an issue)
  •  New feature (non-breaking change which adds functionality)
  •  Breaking change (fix or feature that would cause existing functionality to not work as expected)
  •  Infra/Build change
  •  Code refactoring
Changes
Please list the changes introduced in this PR:
  • transformer_engine/pytorch/module/linear.py
  • transformer_engine/pytorch/module/grouped_linear.py
Checklist:
  •  I have read and followed the contributing guidelines
  •  The functionality is complete
  •  I have commented my code, particularly in hard-to-understand areas
  •  I have made corresponding changes to the documentation
  •  My changes generate no new warnings
  •  I have added tests that prove my fix is effective or that my feature works
  •  New and existing unit tests pass locally with my changes
hxbaiand others
added 5 commits
June 11, 2025 02:33
@hxbai
save original input
406356f
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
@pre-commit-ci
[pre-commit.ci] auto fixes from pre-commit.com hooks
aa4a01f
for more information, see https://pre-commit.ci
@hxbai
fix input_quantizer usage in Linear bwd
90ade56
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
@hxbai
minor fix
097d900
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
@hxbai
refine the docstring
218046d
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
hxbai added a commit
to hxbai/TransformerEngine
that referenced
this pull request
Jun 18, 2025
@hxbai
[EA][PyTorch] Add save_original_input in Linear/GroupedLinear to save…
697a177
… memory NVIDIA#1865
Squashed commit of the following:
commit 218046d
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Wed Jun 11 01:25:48 2025 -0700
refine the docstring
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
commit 097d900
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Wed Jun 11 00:43:35 2025 -0700
minor fix
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
commit 90ade56
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Wed Jun 11 00:37:09 2025 -0700
fix input_quantizer usage in Linear bwd
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
commit aa4a01f
Author: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Date: Wed Jun 11 03:06:44 2025 +0000
[pre-commit.ci] auto fixes from pre-commit.com hooks
for more information, see https://pre-commit.ci
commit 406356f
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Tue Jun 10 18:58:01 2025 -0700
save original input
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
hxbai added a commit
to BestJuly/TransformerEngine
that referenced
this pull request
Jun 18, 2025
@hxbai
[EA][PyTorch] Add save_original_input in Linear/GroupedLinear to save…
658b320
… memory NVIDIA#1865
Squashed commit of the following:
commit 218046d
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Wed Jun 11 01:25:48 2025 -0700
refine the docstring
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
commit 097d900
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Wed Jun 11 00:43:35 2025 -0700
minor fix
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
commit 90ade56
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Wed Jun 11 00:37:09 2025 -0700
fix input_quantizer usage in Linear bwd
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
commit aa4a01f
Author: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Date: Wed Jun 11 03:06:44 2025 +0000
[pre-commit.ci] auto fixes from pre-commit.com hooks
for more information, see https://pre-commit.ci
commit 406356f
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Tue Jun 10 18:58:01 2025 -0700
save original input
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
hxbaiand others
added 3 commits
July 2, 2025 02:24
@hxbai
hxbai
commented
Jul 2, 2025

Copy link


Copy Markdown

ContributorAuthor
Hi @timmoon10 could you help take a look at this PR? Thanks!
timmoon10
timmoon10
reviewed
Jul 15, 2025
View reviewed changes
@timmoon10timmoon10
left a comment


Copy link


Copy Markdown

Member
Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Overall this is a nice trick, although I have suggestions on making the forward and backward pass less tightly coupled and more robust.
Comment threadtransformer_engine/pytorch/module/linear.py
Outdated

Comment threadtests/pytorch/test_numerics.py
Outdated

Comment threadtests/pytorch/test_numerics.py
Outdated

Comment threadtransformer_engine/pytorch/module/linear.py
Outdated

@timmoon10
timmoon10
commented
Jul 16, 2025

Copy link


Copy Markdown

Member
/te-ci pytorch
timmoon10
timmoon10
approved these changes
Jul 16, 2025
View reviewed changes
@timmoon10timmoon10
left a comment


Copy link


Copy Markdown

Member
Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM, pending CI.
@timmoon10
timmoon10
merged commit 07afda9
into
NVIDIA:mainJul 17, 2025

11 checks passed

@timmoon10timmoon10
mentioned this pull request
Jul 17, 2025
 Merged

13 tasks
Sign up for freeto join this conversation on GitHub.
Already have an account?
Sign in to comment

Reviewers

@timmoon10timmoon10timmoon10 approved these changes

Assignees


No one assigned


Labels

2.6.0

Projects


None yet


Milestone


No milestone


Development

Successfully merging this pull request may close these issues.

3 participants

@hxbai@timmoon10@yaox12

, 'i');
if (__m === '*' || __re.test(location.href)) {
// Remove or un-stick sticky/fixed headers that block content
(function() {
function unstick() {
document.querySelectorAll('header, nav, [role="banner"], .header, .navbar, .sticky, .fixed-top, [style*="position: fixed"], [style*="position:sticky"]').forEach(function(el) {
if (el.style.position === 'fixed' || el.style.position === 'sticky' || getComputedStyle(el).position === 'fixed' || getComputedStyle(el).position === 'sticky') {
el.style.position = 'static';
el.style.top = 'auto';
el.style.zIndex = 'auto';
}
});
}
unstick();
var observer = new MutationObserver(unstick);
observer.observe(document.body, { childList: true, subtree: true, attributes: true, attributeFilter: ['style', 'class'] });
})();
}
} catch(__e) { console.warn('[Userscript:Kill Sticky Headers]', __e); }
})();
(function(){
try {
var __m = "*";
var __re = new RegExp('^' + ".*" + '
[PyTorch] Add save_original_input in Linear/GroupedLinear to save memory by hxbai · Pull Request #1865 · NVIDIA/TransformerEngine · GitHub
Skip to content
[PyTorch] Add save_original_input in Linear/GroupedLinear to save memory - #1865
Merged
timmoon10 merged 14 commits into
NVIDIA:mainfrom 
hxbai:save_bf16_in_fp8_gemm
Jul 17, 2025
Merged
[PyTorch] Add save_original_input in Linear/GroupedLinear to save memory#1865
timmoon10 merged 14 commits into
NVIDIA:mainfrom 
hxbai:save_bf16_in_fp8_gemm
Conversation
@hxbai
@hxbaihxbai
commented
Jun 11, 2025•

edited

Loading

Copy link


Copy Markdown

Contributor
Description
Add save_original_input in Linear/GroupedLinear. If it is on, the modules save the original inputs and only do rowwise quant in forward and colwise quant in backward (for blockwise and mxfp8 recipe). With this, in some scenarios, we can save memory usage at the expense of separated quantizations.
In the current implementation, the Linear/GroupedLinear always saves the cast input tensors for backward.

But this may increase the memory usage if this input tensor is used by multiple modules, such as:
  • The output tensor of fused attn is saved by its self and used by the following proj_linear. We want the proj_linear to save the original input to save memory usage.
  • For MoE models with shared experts, the output of pre_mlp_layernorm is saved by both router gating linear and shared experts linear. The shared experts linear should save the original input for memory.
  • For multi-latent attention in DeepSeek models, the output of input_layernorm is used by q_down_linear and kv_down_linear. We want the two linears to save the original input.
Besides, in Megatron-LM, we added an output-discarding recomputation that trickily releases some saved tensors and restores them by recomputation. This relies on that the saved tensor is the original output tensor.
Fixes # (issue)
Type of change
  •  Documentation change (change only to the documentation, either a fix or a new content)
  •  Bug fix (non-breaking change which fixes an issue)
  •  New feature (non-breaking change which adds functionality)
  •  Breaking change (fix or feature that would cause existing functionality to not work as expected)
  •  Infra/Build change
  •  Code refactoring
Changes
Please list the changes introduced in this PR:
  • transformer_engine/pytorch/module/linear.py
  • transformer_engine/pytorch/module/grouped_linear.py
Checklist:
  •  I have read and followed the contributing guidelines
  •  The functionality is complete
  •  I have commented my code, particularly in hard-to-understand areas
  •  I have made corresponding changes to the documentation
  •  My changes generate no new warnings
  •  I have added tests that prove my fix is effective or that my feature works
  •  New and existing unit tests pass locally with my changes
hxbaiand others
added 5 commits
June 11, 2025 02:33
@hxbai
save original input
406356f
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
@pre-commit-ci
[pre-commit.ci] auto fixes from pre-commit.com hooks
aa4a01f
for more information, see https://pre-commit.ci
@hxbai
fix input_quantizer usage in Linear bwd
90ade56
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
@hxbai
minor fix
097d900
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
@hxbai
refine the docstring
218046d
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
hxbai added a commit
to hxbai/TransformerEngine
that referenced
this pull request
Jun 18, 2025
@hxbai
[EA][PyTorch] Add save_original_input in Linear/GroupedLinear to save…
697a177
… memory NVIDIA#1865
Squashed commit of the following:
commit 218046d
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Wed Jun 11 01:25:48 2025 -0700
refine the docstring
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
commit 097d900
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Wed Jun 11 00:43:35 2025 -0700
minor fix
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
commit 90ade56
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Wed Jun 11 00:37:09 2025 -0700
fix input_quantizer usage in Linear bwd
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
commit aa4a01f
Author: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Date: Wed Jun 11 03:06:44 2025 +0000
[pre-commit.ci] auto fixes from pre-commit.com hooks
for more information, see https://pre-commit.ci
commit 406356f
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Tue Jun 10 18:58:01 2025 -0700
save original input
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
hxbai added a commit
to BestJuly/TransformerEngine
that referenced
this pull request
Jun 18, 2025
@hxbai
[EA][PyTorch] Add save_original_input in Linear/GroupedLinear to save…
658b320
… memory NVIDIA#1865
Squashed commit of the following:
commit 218046d
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Wed Jun 11 01:25:48 2025 -0700
refine the docstring
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
commit 097d900
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Wed Jun 11 00:43:35 2025 -0700
minor fix
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
commit 90ade56
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Wed Jun 11 00:37:09 2025 -0700
fix input_quantizer usage in Linear bwd
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
commit aa4a01f
Author: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Date: Wed Jun 11 03:06:44 2025 +0000
[pre-commit.ci] auto fixes from pre-commit.com hooks
for more information, see https://pre-commit.ci
commit 406356f
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Tue Jun 10 18:58:01 2025 -0700
save original input
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
hxbaiand others
added 3 commits
July 2, 2025 02:24
@hxbai
hxbai
commented
Jul 2, 2025

Copy link


Copy Markdown

ContributorAuthor
Hi @timmoon10 could you help take a look at this PR? Thanks!
timmoon10
timmoon10
reviewed
Jul 15, 2025
View reviewed changes
@timmoon10timmoon10
left a comment


Copy link


Copy Markdown

Member
Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Overall this is a nice trick, although I have suggestions on making the forward and backward pass less tightly coupled and more robust.
Comment threadtransformer_engine/pytorch/module/linear.py
Outdated

Comment threadtests/pytorch/test_numerics.py
Outdated

Comment threadtests/pytorch/test_numerics.py
Outdated

Comment threadtransformer_engine/pytorch/module/linear.py
Outdated

@timmoon10
timmoon10
commented
Jul 16, 2025

Copy link


Copy Markdown

Member
/te-ci pytorch
timmoon10
timmoon10
approved these changes
Jul 16, 2025
View reviewed changes
@timmoon10timmoon10
left a comment


Copy link


Copy Markdown

Member
Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM, pending CI.
@timmoon10
timmoon10
merged commit 07afda9
into
NVIDIA:mainJul 17, 2025

11 checks passed

@timmoon10timmoon10
mentioned this pull request
Jul 17, 2025
 Merged

13 tasks
Sign up for freeto join this conversation on GitHub.
Already have an account?
Sign in to comment

Reviewers

@timmoon10timmoon10timmoon10 approved these changes

Assignees


No one assigned


Labels

2.6.0

Projects


None yet


Milestone


No milestone


Development

Successfully merging this pull request may close these issues.

3 participants

@hxbai@timmoon10@yaox12

, 'i');
if (__m === '*' || __re.test(location.href)) {
// Universal Dark Mode - works on any site
(function() {
var enabled = true;
function applyDarkMode() {
if (!enabled) return;
// Create style element if it doesn't exist
var style = document.getElementById('universal-dark-mode-style');
if (!style) {
style = document.createElement('style');
style.id = 'universal-dark-mode-style';
document.head.appendChild(style);
}
// Dark mode CSS - inverts colors but preserves images/video
style.textContent = '
/* Invert everything except media */
html {
filter: invert(1) hue-rotate(180deg) !important;
background: #1a1a2e !important;
}
/* Restore images, videos, iframes, canvas */
img, video, iframe, canvas, svg, picture, [style*="background-image"] {
filter: invert(1) hue-rotate(180deg) !important;
}
/* Preserve specific elements that should not be inverted */
.no-dark-mode, .no-dark-mode *,
[data-theme="light"], [data-theme="light"],
.ace_editor, .ace_editor *,
.CodeMirror, .CodeMirror *,
.monaco-editor, .monaco-editor *,
.markdown-body pre, .markdown-body pre *,
.highlight, .highlight *,
pre code, pre code * {
filter: none !important;
}
/* Fix common UI elements */
.modal, .popup, .dropdown-menu, .tooltip, .popover {
filter: invert(1) hue-rotate(180deg) !important;
background: #2d2d44 !important;
border-color: #444 !important;
}
/* Scrollbars */
::-webkit-scrollbar { background: #1a1a2e !important; }
::-webkit-scrollbar-thumb { background: #444 !important; }
::-webkit-scrollbar-thumb:hover { background: #555 !important; }
/* Selection */
::selection { background: #4ecdc4 !important; color: #1a1a2e !important; }
::-moz-selection { background: #4ecdc4 !important; color: #1a1a2e !important; }
';
}
function removeDarkMode() {
var style = document.getElementById('universal-dark-mode-style');
if (style) style.remove();
}
// Toggle with Alt+Shift+D
document.addEventListener('keydown', function(e) {
if (e.altKey && e.shiftKey && e.key === 'D') {
e.preventDefault();
enabled = !enabled;
if (enabled) {
applyDarkMode();
console.log('[Universal Dark Mode] Enabled');
} else {
removeDarkMode();
console.log('[Universal Dark Mode] Disabled');
}
}
});
// Apply on load
applyDarkMode();
// Re-apply on dynamic content
var observer = new MutationObserver(function(mutations) {
if (enabled && !document.getElementById('universal-dark-mode-style')) {
applyDarkMode();
}
});
observer.observe(document.head, { childList: true });
console.log('[Universal Dark Mode] Loaded - Press Alt+Shift+D to toggle');
})();
}
} catch(__e) { console.warn('[Userscript:Universal Dark Mode]', __e); }
})();
})();
[PyTorch] Add save_original_input in Linear/GroupedLinear to save memory by hxbai · Pull Request #1865 · NVIDIA/TransformerEngine · GitHub
Skip to content
[PyTorch] Add save_original_input in Linear/GroupedLinear to save memory - #1865
Merged
timmoon10 merged 14 commits into
NVIDIA:mainfrom 
hxbai:save_bf16_in_fp8_gemm
Jul 17, 2025
Merged
[PyTorch] Add save_original_input in Linear/GroupedLinear to save memory#1865
timmoon10 merged 14 commits into
NVIDIA:mainfrom 
hxbai:save_bf16_in_fp8_gemm
Conversation
@hxbai
@hxbaihxbai
commented
Jun 11, 2025•

edited

Loading

Copy link


Copy Markdown

Contributor
Description
Add save_original_input in Linear/GroupedLinear. If it is on, the modules save the original inputs and only do rowwise quant in forward and colwise quant in backward (for blockwise and mxfp8 recipe). With this, in some scenarios, we can save memory usage at the expense of separated quantizations.
In the current implementation, the Linear/GroupedLinear always saves the cast input tensors for backward.

But this may increase the memory usage if this input tensor is used by multiple modules, such as:
  • The output tensor of fused attn is saved by its self and used by the following proj_linear. We want the proj_linear to save the original input to save memory usage.
  • For MoE models with shared experts, the output of pre_mlp_layernorm is saved by both router gating linear and shared experts linear. The shared experts linear should save the original input for memory.
  • For multi-latent attention in DeepSeek models, the output of input_layernorm is used by q_down_linear and kv_down_linear. We want the two linears to save the original input.
Besides, in Megatron-LM, we added an output-discarding recomputation that trickily releases some saved tensors and restores them by recomputation. This relies on that the saved tensor is the original output tensor.
Fixes # (issue)
Type of change
  •  Documentation change (change only to the documentation, either a fix or a new content)
  •  Bug fix (non-breaking change which fixes an issue)
  •  New feature (non-breaking change which adds functionality)
  •  Breaking change (fix or feature that would cause existing functionality to not work as expected)
  •  Infra/Build change
  •  Code refactoring
Changes
Please list the changes introduced in this PR:
  • transformer_engine/pytorch/module/linear.py
  • transformer_engine/pytorch/module/grouped_linear.py
Checklist:
  •  I have read and followed the contributing guidelines
  •  The functionality is complete
  •  I have commented my code, particularly in hard-to-understand areas
  •  I have made corresponding changes to the documentation
  •  My changes generate no new warnings
  •  I have added tests that prove my fix is effective or that my feature works
  •  New and existing unit tests pass locally with my changes
hxbaiand others
added 5 commits
June 11, 2025 02:33
@hxbai
save original input
406356f
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
@pre-commit-ci
[pre-commit.ci] auto fixes from pre-commit.com hooks
aa4a01f
for more information, see https://pre-commit.ci
@hxbai
fix input_quantizer usage in Linear bwd
90ade56
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
@hxbai
minor fix
097d900
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
@hxbai
refine the docstring
218046d
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
hxbai added a commit
to hxbai/TransformerEngine
that referenced
this pull request
Jun 18, 2025
@hxbai
[EA][PyTorch] Add save_original_input in Linear/GroupedLinear to save…
697a177
… memory NVIDIA#1865
Squashed commit of the following:
commit 218046d
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Wed Jun 11 01:25:48 2025 -0700
refine the docstring
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
commit 097d900
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Wed Jun 11 00:43:35 2025 -0700
minor fix
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
commit 90ade56
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Wed Jun 11 00:37:09 2025 -0700
fix input_quantizer usage in Linear bwd
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
commit aa4a01f
Author: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Date: Wed Jun 11 03:06:44 2025 +0000
[pre-commit.ci] auto fixes from pre-commit.com hooks
for more information, see https://pre-commit.ci
commit 406356f
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Tue Jun 10 18:58:01 2025 -0700
save original input
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
hxbai added a commit
to BestJuly/TransformerEngine
that referenced
this pull request
Jun 18, 2025
@hxbai
[EA][PyTorch] Add save_original_input in Linear/GroupedLinear to save…
658b320
… memory NVIDIA#1865
Squashed commit of the following:
commit 218046d
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Wed Jun 11 01:25:48 2025 -0700
refine the docstring
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
commit 097d900
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Wed Jun 11 00:43:35 2025 -0700
minor fix
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
commit 90ade56
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Wed Jun 11 00:37:09 2025 -0700
fix input_quantizer usage in Linear bwd
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
commit aa4a01f
Author: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Date: Wed Jun 11 03:06:44 2025 +0000
[pre-commit.ci] auto fixes from pre-commit.com hooks
for more information, see https://pre-commit.ci
commit 406356f
Author: Hongxiao Bai <hongxiaob@nvidia.com>
Date: Tue Jun 10 18:58:01 2025 -0700
save original input
Signed-off-by: Hongxiao Bai <hongxiaob@nvidia.com>
hxbaiand others
added 3 commits
July 2, 2025 02:24
@hxbai
hxbai
commented
Jul 2, 2025

Copy link


Copy Markdown

ContributorAuthor
Hi @timmoon10 could you help take a look at this PR? Thanks!
timmoon10
timmoon10
reviewed
Jul 15, 2025
View reviewed changes
@timmoon10timmoon10
left a comment


Copy link


Copy Markdown

Member
Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Overall this is a nice trick, although I have suggestions on making the forward and backward pass less tightly coupled and more robust.
Comment threadtransformer_engine/pytorch/module/linear.py
Outdated

Comment threadtests/pytorch/test_numerics.py
Outdated

Comment threadtests/pytorch/test_numerics.py
Outdated

Comment threadtransformer_engine/pytorch/module/linear.py
Outdated

@timmoon10
timmoon10
commented
Jul 16, 2025

Copy link


Copy Markdown

Member
/te-ci pytorch
timmoon10
timmoon10
approved these changes
Jul 16, 2025
View reviewed changes
@timmoon10timmoon10
left a comment


Copy link


Copy Markdown

Member
Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM, pending CI.
@timmoon10
timmoon10
merged commit 07afda9
into
NVIDIA:mainJul 17, 2025

11 checks passed

@timmoon10timmoon10
mentioned this pull request
Jul 17, 2025
 Merged

13 tasks
Sign up for freeto join this conversation on GitHub.
Already have an account?
Sign in to comment

Reviewers

@timmoon10timmoon10timmoon10 approved these changes

Assignees


No one assigned


Labels

2.6.0

Projects


None yet


Milestone


No milestone


Development

Successfully merging this pull request may close these issues.

3 participants

@hxbai@timmoon10@yaox12