From a046930b43274d1d647be51c2dcc17cc69cd0fa7 Mon Sep 17 00:00:00 2001
From: SZCJW <792430652@qq.com>
Date: Mon, 30 Mar 2026 22:43:23 +0800
Subject: [PATCH] x

---
 Dockerfile         |  2 +-
 README.md          | 23 +++++++++++++++++++----
 config.json        |  6 +++---
 docker-compose.yml |  2 ++
 4 files changed, 25 insertions(+), 8 deletions(-)

diff --git a/Dockerfile b/Dockerfile
index 2fb9b1b..13f0278 100644
--- a/Dockerfile
+++ b/Dockerfile
@@ -1,4 +1,4 @@
-FROM docker.1ms.run/vllm/vllm-openai-rocm:latest
+FROM rocm/vllm:rocm7.12.0_gfx120X-all_ubuntu24.04_py3.12_pytorch_2.9.1_vllm_0.16.0
 
 WORKDIR /workspace
 
diff --git a/README.md b/README.md
index cd2f336..b176a20 100644
--- a/README.md
+++ b/README.md
@@ -1,6 +1,6 @@
 # ROCm vLLM 容器化推理项目
 
-基于镜像 `docker.1ms.run/vllm/vllm-openai-rocm:latest` 的 Python + vLLM 推理服务，适配双 AMD R9700 32G GPU。
+基于镜像 `rocm/vllm:rocm7.12.0_gfx120X-all_ubuntu24.04_py3.12_pytorch_2.9.1_vllm_0.16.0` 的 Python + vLLM 推理服务，适配双 AMD R9700 32G GPU。
 
 ## 项目目标
 
@@ -65,20 +65,35 @@
 
 ## 部署步骤
 
-1. 修改 `config.json` 中的 `models.selected` 与服务参数。
+1. 预拉取基础镜像（与官方文档一致）：
 
-2. 构建并启动容器：
+```bash
+docker pull rocm/vllm:rocm7.12.0_gfx120X-all_ubuntu24.04_py3.12_pytorch_2.9.1_vllm_0.16.0
+```
+
+2. 修改 `config.json` 中的 `models.selected` 与服务参数。
+
+3. 构建并启动容器：
 
 ```bash
 docker compose up -d --build
 ```
 
-3. 验证 OpenAI 协议服务：
+4. 验证 OpenAI 协议服务：
 
 ```bash
 curl http://localhost:<services.openai.port>/v1/models
 ```
 
+当前 `docker-compose.yml` 已按官方运行参数适配：
+
+- `--group-add=video` → `group_add: [video]`
+- `--ipc=host` → `ipc: host`
+- `--cap-add=SYS_PTRACE` → `cap_add: [SYS_PTRACE]`
+- `--security-opt seccomp=unconfined` → `security_opt: [seccomp=unconfined]`
+- `--device /dev/kfd` 与 `--device /dev/dri` → `devices`
+- `-e HF_HOME=/app/models` 在本项目等效为 `HF_HOME=/opt/model`
+
 ## OpenAI 协议示例（8001）
 
 ```bash
diff --git a/config.json b/config.json
index adb92d0..5d0a2ce 100644
--- a/config.json
+++ b/config.json
@@ -75,10 +75,10 @@
         "quantization": "gptq",
         "ctx": "32768",
         "trust_remote": true,
-        "valid_tp": [1],
+        "valid_tp": [2],
         "max_num_seqs": "1",
-        "max_tokens": "1024",
-        "gpu_util": "0.88",
+        "max_tokens": "2048",
+        "gpu_util": "0.90",
         "enforce_eager": true,
         "tool_call_parser": "qwen3_xml",
         "reasoning_parser": "qwen3",
diff --git a/docker-compose.yml b/docker-compose.yml
index 141a252..0507e43 100644
--- a/docker-compose.yml
+++ b/docker-compose.yml
@@ -12,6 +12,8 @@ services:
     volumes:
       - /opt/model:/opt/model:ro
       - ./config.json:/workspace/config.json:ro
+    environment:
+      HF_HOME: /opt/model
     devices:
       - /dev/kfd
       - /dev/dri