frozenleaves/vllm
1
0
Fork 0
mirror of https://github.com/vllm-project/vllm.git synced 2025-10-20 14:53:52 +08:00
Code Issues Packages Projects Releases Wiki Activity

fix typo (#1184)

Browse Source 
Co-authored-by: Zhuohan Li <zhuohan123@gmail.com>
This commit is contained in:
Wang Ran (汪然)
2023-09-28 07:22:45 +08:00
committed by GitHub
parent 21877b0d75
commit 30e775281d
2 changed files with 5 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
vllm/engine/llm_engine.py
View File

@ -54,8 +54,8 @@ class LLMEngine:
scheduler_config: The configuration related to the request scheduler.
distributed_init_method: The initialization method for distributed
execution. See `torch.distributed.init_process_group` for details.
stage_devices: The list of devices for each stage. Each stage is a list
of (rank, node_resource, device) tuples.
placement_group: Ray placement group for distributed execution.
Required for distributed execution.
log_stats: Whether to log statistics.
"""

7
vllm/engine/ray_utils.py
View File

@ -63,11 +63,10 @@ def initialize_cluster(
the default Ray cluster address.
Returns:
A tuple of (`distributed_init_method`, `all_stage_devices`). The
A tuple of (`distributed_init_method`, `placement_group`). The
`distributed_init_method` is the address for initializing the
distributed backend. `all_stage_devices` includes device IDs for
each worker in each pipeline stage. Each device ID is a tuple of
(rank, node resource, device id).
distributed backend. `placement_group` includes the specification
of the resources for each distributed worker.
"""
if parallel_config.worker_use_ray or engine_use_ray:
if ray is None: