Refactor PEER #123
Open
Refactor PEER #123
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
sordonia
reviewed
Oct 24, 2024
| @dataclass | ||
| class PEERConfig(ModifierConfig): | ||
| n_heads: int = 8 | ||
| moe__num_experts: int = 100 |
| def experts_names(self): | ||
| return self.model.experts_names | ||
|
|
||
| def get_expert_instance(self, name): |
Member
There was a problem hiding this comment.
this is only available for multiexpert, i am not sure to add this dependency here
| module = MultiExpertModule(**vars(expert.training_config)).to("cuda") | ||
| module.add_expert_instance(expert, is_default=True) | ||
|
|
||
| elif args.merge_or_route in ["peer"]: |
Member
There was a problem hiding this comment.
can you explain what you are trying to do?
| for expert_name in module.experts_names: | ||
| expert = module.get_expert_instance(expert_name) | ||
| expert: Expert = module.get_expert_instance(expert_name) | ||
| expert.expert_info.training_config = args |
Member
There was a problem hiding this comment.
remove this, not a good idea to store complex object in training_config, it will be transformed to a Dict in the next PR :)
| self.selector = selector or TaskNameSelector() | ||
| self._default_expert_name = None | ||
| self.expert_infos = {} | ||
| self.experts = nn.ModuleDict({}) |
Member
There was a problem hiding this comment.
not all containers have "experts", see my last PR on making LoRA faster
| self.experts = nn.ModuleDict({}) | ||
|
|
||
| @property | ||
| def num_experts(self): |
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Outdated suggestions cannot be applied.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.
TODO: