software-engineering/spark-vllm-docker
2
0
Fork 0
Code Issues Actions Packages Projects Releases Wiki Activity

Removed expert-parallel from 3x-node Qwen

Browse Source
This commit is contained in:
eugr
2026-03-26 22:44:48 -07:00
parent 0fa585f909
commit 47a896d722
1 changed files with 2 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
recipes/3x-spark-cluster/qwen3.5-397b-int4-autoround.yaml
View File

@@ -28,9 +28,9 @@ defaults:
port: 8000 port: 8000
host: 0.0.0.0 host: 0.0.0.0
pipeline_parallel: 3 pipeline_parallel: 3
gpu_memory_utilization: 0.6 gpu_memory_utilization: 0.7
max_model_len: 262144 max_model_len: 262144
max_num_batched_tokens: 4176 max_num_batched_tokens: 16384
# Environment variables # Environment variables
env: env:
@@ -55,7 +55,6 @@ command: |
--chat-template unsloth.jinja \ --chat-template unsloth.jinja \
-tp 1 \ -tp 1 \
-pp {pipeline_parallel} \ -pp {pipeline_parallel} \
--enable-expert-parallel \
--distributed-executor-backend ray --distributed-executor-backend ray