Skip to content

🐛 use 512 tokens instead of 256 #509

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 2 commits into from
Oct 7, 2025
Merged

🐛 use 512 tokens instead of 256 #509

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
6315085
:bug: use 512 tokens instead of 256
joerunde Oct 7, 2025
b191492
:bug: fixup test ordering
joerunde Oct 7, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion tests/conftest.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -31,7 +31,7 @@ def pytest_generate_tests(metafunc):
# default parameterizations
default_warmup_shape = [[(64, 20, 4)]]
default_max_num_seqs = [4]
default_max_model_len = [256]
default_max_model_len = [512]

existing_markers = [
marker.name if marker.name != "parametrize" else marker.args[0]
Expand Down
4 changes: 2 additions & 2 deletions tests/llm_cache.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -180,7 +180,7 @@ def get_engine(

# 🌶️🌶️🌶️
# Messing with the blocks and context length by either:
# - setting context < 256 tokens
# - setting context < 512 tokens
# - setting available blocks != (context * batch size // 64)
# can cause compilation failures on spyre hardware.

Expand All @@ -195,7 +195,7 @@ def get_engine(
engine_args = EngineArgs(
model=model_name,
tokenizer=model_name,
max_model_len=max(max_model_len, 256),
max_model_len=max(max_model_len, 512),
max_num_seqs=max_num_seqs_compiled,
num_gpu_blocks_override=None,
revision=revision,
Expand Down
Loading