Fix GPT3 attention missing KV cache initialization and handling#3927
Merged
Conversation
shralex
requested review from
A9isha,
NicoGrande,
NuojCheng,
RissyRan,
SurbhiJainUSC,
abhinavclemson,
aireenmei,
bvandermoon,
darisoy,
dipannita08,
gagika,
gobbleturk,
hengtaoguo,
igorts-git,
jesselu-google,
jiangjy1982,
khatwanimohit,
parambole,
richjames0,
shuningjin,
suexu1025 and
vipannalla
as code owners
Codecov Report✅ All modified and coverable lines are covered by tests. 📢 Thoughts on this report? Let us know! |
shralex
force-pushed
the
shralex_test_5
branch
2 times, most recently
from
ddaccf2 to
12bc21b
Compare
igorts-git
approved these changes
May 18, 2026
NuojCheng
reviewed
May 18, 2026
| self.rng = jax.random.PRNGKey(1234) | ||
|
|
||
| devices_array = maxtext_utils.create_device_mesh(self.cfg) | ||
| devices_array = maxtext_utils.create_device_mesh(self.cfg, devices=[jax.devices()[0]]) |
Collaborator
There was a problem hiding this comment.
why we only use one device for testing? No sharding involved?
Collaborator
Author
There was a problem hiding this comment.
This is for the KV cache test. Running everything on a single device eliminates communication and ensures the test is fully deterministic.
NuojCheng
reviewed
May 18, 2026
| enable_checkpointing=False, | ||
| model_name="gpt3-52k", | ||
| dtype="float32", | ||
| per_device_batch_size=1.0 / jax.device_count(), |
Collaborator
Author
There was a problem hiding this comment.
Its a way to set global batch size to 1 regardless of device count.
NuojCheng
approved these changes
May 19, 2026
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
This pull request resolves an issue where
Gpt3MultiHeadAttentioncalledAttentionOpwithout passingcached_values, causing decoding to fail with anAssertionError: assert prefill_kv_cache.FIXES: b/452778717
Tests
Updated GPT-3 tests. Verified that b/452778717 is fixed on a TPU VM.
Checklist
Before submitting this PR, please make sure (put X in square brackets):
gemini-reviewlabel.