A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 12, 09:45 UTC
[Doc]: fixing typos in multiple files. (#29685)
Signed-off-by: Didier Durand <durand.didier@gmail.com>
Didier Durand committed
Nov 28, 2025 at 17:41 UTC
fae6943068e05fb90705ccb5c08bbd04d771048f
6 files changed
+6
-6
vllm/compilation/sequence_parallelism.py
+1
-1
@@ -304,7 +304,7 @@ class SequenceParallelismPass(VllmPatternMatcherPass):
304
def __init__(self, config: VllmConfig):
305
super().__init__(config)
306
307
- # Used to cleanup redundant views created temporarily
307
+ # Used to clean up redundant views created temporarily
308
# to circumvent residual shape change issues
309
self.noop_cleanup = NoOpEliminationPass(config)
310
self.noop_cleanup.pass_name = f"{self.pass_name}.{self.noop_cleanup.pass_name}"
vllm/model_executor/models/nano_nemotron_vl.py
+1
-1
@@ -1472,7 +1472,7 @@ class NemotronH_Nano_VL_V2(
1472
return []
1473
1474
# # The result multimodal_embeddings is tuple of tensors, with each
1475
- # tensor correspoending to a multimodal data item (image or video).
1475
+ # tensor corresponding to a multimodal data item (image or video).
1476
multimodal_embeddings: tuple[torch.Tensor, ...] = ()
1477
1478
# NOTE: It is important to iterate over the keys in this dictionary
vllm/model_executor/models/stablelm.py
+1
-1
@@ -19,7 +19,7 @@
19
# This code is based off the following work:
20
# https://huggingface.co/stabilityai/stablelm-3b-4e1t/blob/main/modeling_stablelm_epoch.py
21
# https://huggingface.co/stabilityai/stablelm-3b-4e1t/blob/main/config.json
22
-"""Inference-only StabeLM (https://github.com/Stability-AI/StableLM)
22
+"""Inference-only StableLM (https://github.com/Stability-AI/StableLM)
23
model compatible with HuggingFace weights."""
24
25
from collections.abc import Iterable
vllm/utils/gc_utils.py
+1
-1
@@ -17,7 +17,7 @@ class GCDebugConfig:
17
"""
18
Config for GC Debugger.
19
- 0: disable GC debugger
20
- - 1: enable GC debugger with gc.collect elpased times
20
+ - 1: enable GC debugger with gc.collect elapsed times
21
- '{"top_objects":5}': enable GC debugger with top 5 collected objects
22
"""
23
vllm/v1/spec_decode/ngram_proposer.py
+1
-1
@@ -269,7 +269,7 @@ def _find_longest_matched_ngram_and_propose_tokens(
269
prev_lps = lps[max_ngram - 1]
270
i += 1
271
elif prev_lps != 0:
272
- # Token mismatch: try the second longest prefix
272
+ # Token mismatch: try the second-longest prefix
273
# among all suffix of tokens[:i],
274
# which is the longest prefix of tokens[:prev_lps]
275
prev_lps = lps[prev_lps - 1]
vllm/v1/worker/ec_connector_model_runner_mixin.py
+1
-1
@@ -59,7 +59,7 @@ class ECConnectorModelRunnerMixin:
59
)
60
61
# This context manager must be used within an active forward context.
62
- # It encapsulates the entire EC conector lifecycle within execute_model
62
+ # It encapsulates the entire EC connector lifecycle within execute_model
63
@staticmethod
64
@contextmanager
65
def _get_ec_connector_output(