Upload dyve_tts/eval/run_vllm_LM.sh with huggingface_hub

dyve_tts/eval/run_vllm_LM.sh

@@ -0,0 +1,10 @@
+export CUDA_VISIBLE_DEVICES=4,5
+vllm serve \
+    "/data/zeju/DeepSeek-R1-Distill-Qwen-14B" \
+    --served-model-name "DeepSeek-R1-Distill-Qwen-14B" \
+    --port 8016 \
+    --tensor-parallel-size 2 \
+    --dtype auto \
+    --api-key "token-abc123" \
+    --gpu_memory_utilization 0.9 \
+    # --enable-prefix-caching