[Doc]: fix typos in Python comments (#24093)

Signed-off-by: Didier Durand <durand.didier@gmail.com>
This commit is contained in:
Didier Durand
2025-09-02 21:05:45 -07:00
committed by GitHub
parent c4ed78b14f
commit d7e1e59972
15 changed files with 23 additions and 23 deletions
+1 -1
View File
@@ -83,7 +83,7 @@ def test_ngram_correctness(
model_name: str,
):
'''
Compare the outputs of a original LLM and a speculative LLM
Compare the outputs of an original LLM and a speculative LLM
should be the same when using ngram speculative decoding.
'''
with monkeypatch.context() as m:
@@ -42,7 +42,7 @@ def test_basic_lifecycle():
engine_core_outputs = scheduler.update_from_output(scheduler_output,
model_runner_output)
# Ensure the request is finished after 1 tokens.
# Ensure the request is finished after 1 token.
assert request.is_finished()
assert request.status == RequestStatus.FINISHED_LENGTH_CAPPED
output = engine_core_outputs[0].outputs[0]
@@ -141,7 +141,7 @@ def test_short_prompt_lifecycle():
def test_prefix_cache_lifecycle():
"""Test that remote decode params still works with a prefix cache hit."""
"""Test that remote decode params still work with a prefix cache hit."""
vllm_config = create_vllm_config()
scheduler = create_scheduler(vllm_config)
+2 -2
View File
@@ -187,7 +187,7 @@ def test_tree_attn_correctness() -> None:
dtype=torch.bfloat16,
)
# Setup the block table and KV cache for paged KV.
# Set up the block table and KV cache for paged KV.
assert max_sequence_length % block_size == 0
max_blocks_per_batch = max_sequence_length // block_size
kv_cache = torch.randn(
@@ -222,7 +222,7 @@ def test_tree_attn_correctness() -> None:
num_alloc_blocks_per_batch] = block_ids.view(
-1, num_alloc_blocks_per_batch)
# Setup the slot mapping for the input KVs.
# Set up the slot mapping for the input KVs.
tree_positions = sequence_position + torch.arange(
0,
tree_size_q,