[CI] Fix proxy & fix vllm

.github/workflows/workflow_finetune.yml

@@ -11,10 +11,10 @@ on:
         default: '10.1.2.13:5000/llmray-build'
       http_proxy:
         type: string
-        default: 'http://proxy-chain.intel.com:911'
+        default: 'http://10.24.221.149:911'
       https_proxy:
         type: string
-        default: 'http://proxy-chain.intel.com:911'
+        default: 'http://10.24.221.149:911'
       runner_config_path:
         type: string
         default: '/home/ci/llm-ray-actions-runner'

.github/workflows/workflow_inference.yml

@@ -11,10 +11,10 @@ on:
         default: '10.1.2.13:5000/llmray-build'
       http_proxy:
         type: string
-        default: 'http://proxy-chain.intel.com:911'
+        default: 'http://10.24.221.149:911'
       https_proxy:
         type: string
-        default: 'http://proxy-chain.intel.com:911'
+        default: 'http://10.24.221.149:911'
       runner_config_path:
         type: string
         default: '/home/ci/llm-ray-actions-runner'

dev/docker/Dockerfile.vllm

@@ -38,6 +38,3 @@ RUN --mount=type=cache,target=/root/.cache/pip pip install -e .[cpu] -f https://
 RUN --mount=type=cache,target=/root/.cache/pip \
     source /opt/conda/bin/activate base && ./install-vllm-cpu.sh
 
-# TODO: workaround, remove this when fixed in vllm-cpu upstream
-RUN --mount=type=cache,target=/root/.cache/pip \
-    pip install xformers