Merge branch 'apply-pr' into pytorch-base

Dockerfile

@@ -190,6 +190,16 @@ RUN --mount=type=cache,id=repo-cache,target=/repo-cache \
 
 WORKDIR $VLLM_BASE_DIR/vllm
 
+ARG VLLM_PRS=""
+
+RUN if [ -n "$VLLM_PRS" ]; then \
+        echo "Applying PRs: $VLLM_PRS"; \
+        for pr in $VLLM_PRS; do \
+            echo "Fetching and applying PR #$pr..."; \
+            curl -fL "https://github.com/vllm-project/vllm/pull/${pr}.diff" | git apply -v; \
+        done; \
+    fi
+
 ARG PRE_TRANSFORMERS=0
 
 # Prepare build requirements

README.md

@@ -66,6 +66,7 @@ This error happens if vLLM nightly build fails for aarch64 platform, but succeed
 **On DGX Spark cluster:**
 
 Make sure you connect your Sparks together and enable passwordless SSH as described in NVidia's [Connect Two Sparks Playbook](https://build.nvidia.com/spark/connect-two-sparks/stacked-sparks). 
+You can also check out our new [Networking Guide](docs/NETWORKING.md).
 
 Then run the following command that will build and distribute image across the cluster.
 

build-and-copy.sh

@@ -21,6 +21,7 @@ PRE_TRANSFORMERS=false
 EXP_MXFP4=false
 TRITON_REF_SET=false
 VLLM_REF_SET=false
+VLLM_PRS=""
 
 cleanup() {
     if [ -n "$TMP_IMAGE" ] && [ -f "$TMP_IMAGE" ]; then
@@ -77,6 +78,7 @@ usage() {
     echo "  --pre-flashinfer          : Use pre-release versions of FlashInfer"
     echo "  --pre-tf, --pre-transformers : Install transformers 5.0.0rc0 or higher"
     echo "  --exp-mxfp4, --experimental-mxfp4 : Build with experimental native MXFP4 support"
+    echo "  --apply-vllm-pr <pr-num>  : Apply a specific PR patch to vLLM source code. Can be specified multiple times."
     echo "  --no-build                : Skip building, only copy image (requires --copy-to)"
     echo "  -h, --help                : Show this help message"
     exit 1
@@ -140,8 +142,26 @@ while [[ "$#" -gt 0 ]]; do
         --pre-flashinfer) PRE_FLASHINFER=true ;;
         --pre-tf|--pre-transformers) PRE_TRANSFORMERS=true ;;
         --exp-mxfp4|--experimental-mxfp4) EXP_MXFP4=true ;;
+        --apply-vllm-pr)
+            if [ -n "$2" ] && [[ "$2" != -* ]]; then
+               if [ -n "$VLLM_PRS" ]; then
+                   VLLM_PRS="$VLLM_PRS $2"
+               else
+                   VLLM_PRS="$2"
+               fi
+               shift
+            else
+               echo "Error: --apply-vllm-pr requires a PR number."
+               exit 1
+            fi
+            ;;
         --no-build) NO_BUILD=true ;;
         -h|--help) usage ;;
+if [ -n "$VLLM_PRS" ]; then
+    if [ "$EXP_MXFP4" = true ]; then echo "Error: --apply-vllm-pr is incompatible with --exp-mxfp4"; exit 1; fi
+    if [ -n "$USE_WHEELS_MODE" ]; then echo "Error: --apply-vllm-pr is incompatible with --use-wheels"; exit 1; fi
+fi
+
         *) echo "Unknown parameter passed: $1"; usage ;;
     esac
     shift
@@ -204,6 +224,11 @@ if [ "$NO_BUILD" = false ]; then
         CMD+=("--build-arg" "FLASHINFER_PRE=--pre")
     fi
 
+    if [ -n "$VLLM_PRS" ]; then
+        echo "Applying vLLM PRs: $VLLM_PRS"
+        CMD+=("--build-arg" "VLLM_PRS=$VLLM_PRS")
+    fi
+
     if [ "$PRE_TRANSFORMERS" = true ]; then
         echo "Using transformers>=5.0.0..."
         CMD+=("--build-arg" "PRE_TRANSFORMERS=1")