47 lines
1.1 KiB
YAML
47 lines
1.1 KiB
YAML
# Recipe: Qwen3-Coder-Next-int4-Autoround
|
|
# Qwen3-Coder-Next model in Intel int4-Autoround format
|
|
|
|
|
|
recipe_version: "1"
|
|
name: Qwen3-Coder-Next-int4-Autoround
|
|
description: Qwen3-Coder-Next-int4-Autoround
|
|
|
|
# HuggingFace model to download (optional, for --download-model)
|
|
model: Intel/Qwen3-Coder-Next-int4-AutoRound
|
|
|
|
solo_only: true
|
|
|
|
# Container image to use
|
|
container: vllm-node
|
|
|
|
# Mod required to fix autoround weight loading issues
|
|
mods:
|
|
- mods/fix-qwen3-next-autoround
|
|
|
|
# Default settings (can be overridden via CLI)
|
|
defaults:
|
|
port: 8000
|
|
host: 0.0.0.0
|
|
tensor_parallel: 2
|
|
gpu_memory_utilization: 0.7
|
|
max_model_len: 262144
|
|
|
|
# Environment variables
|
|
env:
|
|
VLLM_MARLIN_USE_ATOMIC_ADD: 1
|
|
|
|
# The vLLM serve command template
|
|
command: |
|
|
vllm serve Intel/Qwen3-Coder-Next-int4-AutoRound \
|
|
--enable-auto-tool-choice \
|
|
--tool-call-parser qwen3_coder \
|
|
--gpu-memory-utilization {gpu_memory_utilization} \
|
|
--host {host} \
|
|
--port {port} \
|
|
--load-format fastsafetensors \
|
|
--enable-prefix-caching \
|
|
--max-model-len {max_model_len} \
|
|
-tp {tensor_parallel} \
|
|
--distributed-executor-backend ray
|
|
|