update the moe to support arch < 90 #123
Merged
Conversation
This file contains hidden or 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
wenlei-bao
reviewed
May 22, 2025
| @@ -162,10 +167,10 @@ def forward(self): | |||
|
|
|||
| # MLP layer 1 (GEMM1 and combine) | |||
| mlp_output = self.flux_rs_op.forward_gather_rs( | |||
| input=self.ctx.intermediate_output, | |||
There was a problem hiding this comment.
revert this part? @burness
There was a problem hiding this comment.
@wenlei-bao sorry, I debug some code and forget revert it. e6f6b3b
houqi
reviewed
May 27, 2025
examples/moe.py
Outdated
| @@ -165,7 +170,7 @@ def forward(self): | |||
| input=self.ctx.intermediate_output, | |||
| weight=self.ctx.weight1, | |||
| splits_cpu=self.ctx.splits_cpu, | |||
| routing_idx=self.ctx.scatter_index.view(-1), | |||
| rounting_idx=self.ctx.scatter_index.view(-1), | |||
There was a problem hiding this comment.
yeah， it's a typo. But when I review the code, I found the flux.GemmGroupedV2GatherRSOp and GemmGroupedV3GatherRS, the 4-th arg is scatter_idx. So the first commit, I del the args_name to adopt it like the ut case here:
Maybe, the routing_idx should be scatter_idx?
Sorry about that I have no resource to debug it now
There was a problem hiding this comment.
@houqi I think the code example in moe_usage.md should be right?
There was a problem hiding this comment.
I think v2 and v3 use different names for this field @houqi ?
wenlei-bao
approved these changes
May 28, 2025
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Outdated suggestions cannot be applied.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.
As Describe in #121, update the examples/moe.py to support arch < 90