Refactor the positional emebdding config code #4920
Merged
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
mrwyattii
approved these changes
Jan 9, 2024
mrwyattii
added a commit
that referenced
this pull request
Jan 23, 2024
follow PR #4920 on Qwen inference code Co-authored-by: Michael Wyatt <[email protected]>
mauryaavinash95
pushed a commit
to mauryaavinash95/DeepSpeed
that referenced
this pull request
Feb 17, 2024
The Mixtral PR microsoft#4828 has introduced the positional embedding config class which is a required argument of `make_attn_layer()` function. This has forced the user to override and duplicate the `make_attn_layer()` call for new model implementations using RoPE (This has also broken the Falcon model implementations). This PR: - refactors the inference transformer base class to avoid code duplication by adding a new abstract `positional_embedding_config` property - Fixes the Falcon model implementation to use positional embedding config. The models `llama_v2`, `OPT`, `Mistral 7B`, `Mixtral`, `Falcon` and `Phi-2` are tested with the PR! --------- Co-authored-by: Logan Adams <[email protected]>
mauryaavinash95
pushed a commit
to mauryaavinash95/DeepSpeed
that referenced
this pull request
Feb 17, 2024
follow PR microsoft#4920 on Qwen inference code Co-authored-by: Michael Wyatt <[email protected]>
rraminen
pushed a commit
to ROCm/DeepSpeed
that referenced
this pull request
May 9, 2024
follow PR microsoft#4920 on Qwen inference code Co-authored-by: Michael Wyatt <[email protected]>
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Outdated suggestions cannot be applied.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.
The Mixtral PR #4828 has introduced the positional embedding config class which is a required argument of
make_attn_layer()function. This has forced the user to override and duplicate themake_attn_layer()call for new model implementations using RoPE (This has also broken the Falcon model implementations). This PR:positional_embedding_configpropertyThe models
llama_v2,OPT,Mistral 7B,Mixtral,FalconandPhi-2are tested with the PR!