Skip to content

Commit 0e7f6e0

Browse files
committed
add comments
Signed-off-by: imkero <[email protected]>
1 parent d5eb8f4 commit 0e7f6e0

File tree

1 file changed

+7
-0
lines changed

1 file changed

+7
-0
lines changed

vllm/v1/worker/gpu_model_runner.py

Lines changed: 7 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -237,6 +237,13 @@ def __init__(
237237
dtype=torch.int64,
238238
device="cpu",
239239
pin_memory=self.pin_memory)
240+
241+
# NOTE: `mrope_positions_np` shares same
242+
# underlying data with `mrope_positions_cpu`.
243+
#
244+
# `mrope_positions_np` is created in favor of
245+
# numba accelerated func `mrope_assign_next_input_positions`
246+
# while it can operate numpy array only.
240247
self.mrope_positions_np = self.mrope_positions_cpu.numpy()
241248

242249
# Only relevant for models using ALiBi (e.g, MPT)

0 commit comments

Comments
 (0)