fix tokenizer

vllm_wrapper.sh

@@ -0,0 +1,18 @@
+#!/bin/bash
+set -e
+
+# 只拦截 "serve <model_dir>" 子命令，其他子命令直接透传
+# 适配 Biren 166m / vLLM 0.11.0 环境
+if [ "$1" = "serve" ] && [ -n "$2" ]; then
+    MODEL_DIR="$2"
+    shift 2
+
+    FIXED_DIR=$(python3 /opt/fix_tokenizer.py "$MODEL_DIR")
+    if [ -n "$FIXED_DIR" ]; then
+        exec /usr/local/bin/vllm_real serve "$MODEL_DIR" --tokenizer "$FIXED_DIR" "$@"
+    else
+        exec /usr/local/bin/vllm_real serve "$MODEL_DIR" "$@"
+    fi
+fi
+
+exec /usr/local/bin/vllm_real "$@"