[Gemma.cpp] Allows non-owned arguments for attention methods. #745
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
copybara-service
bot
force-pushed
the
test_823702392
branch
from
9887582 to
693486b
Compare
copybara-service
bot
changed the title
* Adds and uses a new `AttentionActivationPtrs` that holds non-owning `MatPtrs`. Acts as a view into `AttentionActivations`. * Updates `QBatch` to hold non-owning `MatPtr`s to the kv caches. * Enables the `MatPtrT` default constructor for simpler initializations. * Pulls out and passes `LayerWeightsPtrs::query_norm_scale` directly. While `LayerWeightsPtrs` already held non-owning `MatPtr`s, this change avoids the need to find and construct several empty weight tensors just to construct one `query_norm_scale` tensor. PiperOrigin-RevId: 824584177
copybara-service
bot
force-pushed
the
test_823702392
branch
from
693486b to
5a05857
Compare
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Outdated suggestions cannot be applied.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.
[Gemma.cpp] Allows non-owned arguments for attention methods.
AttentionActivationPtrsthat holds non-owningMatPtrs. Acts as a view intoAttentionActivations.QBatchto hold non-owningMatPtrs to the kv caches.MatPtrTdefault constructor for simpler initializations.LayerWeightsPtrs::query_norm_scaledirectly. WhileLayerWeightsPtrsalready held non-owningMatPtrs, this change avoids the need to find and construct several empty weight tensors just to construct onequery_norm_scaletensor.