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 17, 06:45 UTC
Correct comments in parallel_state.py (#1818)
explainerauthors committed
Nov 28, 2023 at 10:19 UTC
a1125ad4df7011e74bbbcf88d268e521278161fa
1 file changed
+2
-2
vllm/model_executor/parallel_utils/parallel_state.py
+2
-2
@@ -2,7 +2,7 @@
2
# Adapted from
3
# https://github.com/NVIDIA/Megatron-LM/blob/main/megatron/core/parallel_state.py
4
# Copyright (c) 2022, NVIDIA CORPORATION. All rights reserved.
5
-"""Model and data parallel groups."""
5
+"""Tensor and pipeline parallel groups."""
6
7
import torch
8
@@ -84,7 +84,7 @@ def initialize_model_parallel(
84
85
86
def model_parallel_is_initialized():
87
- """Check if model and data parallel groups are initialized."""
87
+ """Check if tensor and pipeline parallel groups are initialized."""
88
return (_TENSOR_MODEL_PARALLEL_GROUP is not None
89
and _PIPELINE_MODEL_PARALLEL_GROUP is not None)
90