Add modeldownload

launch.sh

@@ -1,4 +1,5 @@
 #!/bin/sh
+python modeldownload.py
 uvicorn tts_server:app --host 0.0.0.0 --port 7581 --workers 1 --proxy-headers &
 sleep 5
 python script.py

nvidia.Dockerfile

@@ -13,7 +13,8 @@ ENV CUDA_DOCKER_ARCH=all
 
 RUN pip install --no-cache-dir -r requirements_nvidia.txt && \
     pip install --no-cache-dir -r requirements_finetune.txt && \
-    pip install --no-cache-dir -r requirements_other.txt
+    pip install --no-cache-dir -r requirements_other.txt && \
+    pip install --no-cache-dir deepspeed
 EXPOSE 7851
 RUN chmod +x launch.sh
 ENTRYPOINT ["sh", "-c", "./launch.sh"]