fix: correct FusedAdam copy-paste in FusedSGD error messages

[Mr-Neutr0n]

Bug

FusedSGD contains error messages that reference 'FusedAdam' instead of 'FusedSGD', a copy-paste artifact from duplicating the FusedAdam implementation.

Fix

Updated error message strings to reference the correct class name 'FusedSGD'.

[greptile-apps]

Greptile Overview

Greptile Summary

This PR fixes two error messages in FusedSGD that incorrectly referenced FusedAdam instead of FusedSGD, a copy-paste artifact from when the code was duplicated from the FusedAdam implementation.

• Fixed deprecation warning message on line 126 from "set_grad_none kwarg in FusedAdam constructor" to "set_grad_none kwarg in FusedSGD constructor"
• Fixed ValueError message on line 150 from "but FusedAdam was initialized" to "but FusedSGD was initialized"

The changes are purely cosmetic string corrections that improve error message accuracy. The logic and behavior remain unchanged.

Confidence Score: 5/5

• This PR is completely safe to merge with zero risk
• The changes are limited to correcting two string literals in error messages. No logic, functionality, or API changes. The fix is trivial, correct, and improves code quality by removing confusing copy-paste artifacts.
• No files require special attention

Important Files Changed

Filename	Overview
transformer_engine/pytorch/optimizers/fused_sgd.py	Fixed two error messages that incorrectly referenced FusedAdam instead of FusedSGD in deprecation warning and ValueError

[greptile-apps]

_{1 file reviewed, no comments}

_{Edit Code Review Agent Settings | Greptile}

[ptrendx]

@Mr-Neutr0n Please sign your commit, see https://github.com/NVIDIA/TransformerEngine/blob/main/CONTRIBUTING.rst#sign-your-work

[greptile-apps]

_{1 file reviewed, no comments}

_{Edit Code Review Agent Settings | Greptile}