Docker Image for running distilabel CLI

docker/CUDA.Dockerfile

@@ -0,0 +1,34 @@
+# Arguments
+ARG PYTORCH_VERSION=2.1.1
+ARG CUDA_VERSION=12.1.1
+ARG UBUNTU_VERSION=22.04
+ARG TORCH_VERSION=2.2.0
+
+# Using runpod base image and installing torch, cuda and ubuntu
+FROM runpod/pytorch:${PYTORCH_VERSION}-py3.10-cuda${CUDA_VERSION}-devel-ubuntu${UBUNTU_VERSION} AS base
+
+# Run system updates and clean up
+RUN apt-get update && \
+    apt-get install python3 python3-pip -y
+
+# Set python3 as default python
+RUN ln -s /usr/bin/python3 /usr/bin/python
+ENV PYTHON=/usr/bin/python
+
+# Install torch
+RUN python -m pip install --no-cache-dir --upgrade pip && \
+    python -m pip install --no-cache-dir torch==${TORCH_VERSION}
+
+# Set the working directory to /
+WORKDIR /
+
+FROM runpod/pytorch:${PYTORCH_VERSION}-py3.10-cuda${CUDA_VERSION}-devel-ubuntu${UBUNTU_VERSION}
+
+COPY . .
+
+# Installing distilabel with GPU-related dependencies
+RUN pip install -e ".[argilla,hf-transformers,hf-inference-endpoints,llama-cpp,vllm]"
+
+EXPOSE 80
+
+CMD ["distilabel"]

docker/local.Dockerfile

@@ -0,0 +1,11 @@
+ARG PYTHON_VERSION="3.11-slim"
+
+FROM python:${PYTHON_VERSION}
+
+WORKDIR /app
+
+RUN pip install "distilabel[argilla]"
+
+EXPOSE 80
+
+CMD ["distilabel"]

src/distilabel/pipeline/local.py

@@ -65,6 +65,14 @@
 _SUBPROCESS_EXCEPTION: Union[Exception, None] = None
 
 
+# if platform.system() != "Windows":
+#     _MULTIPROCESSING_CONTEXT = "forkserver"
+# else:
+#     _MULTIPROCESSING_CONTEXT = "spawn"
+
+_MULTIPROCESSING_CONTEXT = "spawn"
+
+
 def _init_worker(queue: "Queue[Any]") -> None:
     signal.signal(signal.SIGINT, signal.SIG_IGN)
     setup_logging(queue)