Moving to uv to enable dependency override and cleaner locks

.dockerignore

@@ -1,2 +1,4 @@
 .idea
 target
+.venv/
+backends/python/server/.venv

.gitignore

@@ -1,2 +1,4 @@
 .idea
 target
+.venv
+backends/python/server/.venv

Dockerfile-intel

@@ -76,10 +76,8 @@ COPY backends backends
 COPY backends/python/server/text_embeddings_server/models/__init__.py backends/python/server/text_embeddings_server/models/__init__.py
 COPY backends/python/server/pyproject.toml backends/python/server/pyproject.toml
 
-RUN python -m pip install torch==2.6.0 torchvision torchaudio==2.6.0 --index-url https://download.pytorch.org/whl/cpu --no-cache-dir
-
 RUN cd backends/python/server && \
-    make install
+    make install-cpu
 
 FROM vault.habana.ai/gaudi-docker/1.19.0/ubuntu22.04/habanalabs/pytorch-installer-2.5.1:latest AS hpu
 ENV HUGGINGFACE_HUB_CACHE=/data \
@@ -99,7 +97,10 @@ COPY backends/python/server/text_embeddings_server/models/__init__.py backends/p
 COPY backends/python/server/pyproject.toml backends/python/server/pyproject.toml
 
 RUN cd backends/python/server && \
-    make install
+    pip install -U pip uv && \ 
+    uv venv && \
+    cp -r /usr/local/lib/python3.10/dist-packages/* /usr/src/backends/python/server/.venv/lib/python3.10/site-packages/ && \
+    make install-hpu
 
 FROM intel/intel-extension-for-pytorch:2.6.10-xpu AS xpu
 
@@ -115,9 +116,6 @@ RUN wget -O- https://apt.repos.intel.com/intel-gpg-keys/GPG-PUB-KEY-INTEL-SW-PRO
 
 RUN apt-get update && apt install -y intel-basekit cmake python3-dev ninja-build pciutils
 WORKDIR /usr/src
-RUN pip install torch==2.6.0 torchvision torchaudio --index-url https://download.pytorch.org/whl/xpu --no-cache-dir
-RUN pip install intel-extension-for-pytorch==2.6.10+xpu oneccl_bind_pt==2.6.0+xpu --extra-index-url https://pytorch-extension.intel.com/release-whl/stable/xpu/us/ --no-cache-dir
-
 ENV CCL_ROOT=/opt/intel/oneapi/ccl/latest
 ENV I_MPI_ROOT=/opt/intel/oneapi/mpi/latest
 ENV FI_PROVIDER_PATH=/opt/intel/oneapi/mpi/latest/opt/mpi/libfabric/lib/prov:/usr/lib/x86_64-linux-gnu/libfabric
@@ -132,7 +130,7 @@ COPY backends backends
 COPY backends/python/server/text_embeddings_server/models/__init__.py backends/python/server/text_embeddings_server/models/__init__.py
 COPY backends/python/server/pyproject.toml backends/python/server/pyproject.toml
 RUN cd backends/python/server && \
-    make install
+    make install-xpu
 
 FROM ${PLATFORM} AS grpc
 
@@ -144,6 +142,8 @@ CMD ["--json-output"]
 FROM ${PLATFORM}
 
 COPY --from=http-builder /usr/src/target/release/text-embeddings-router /usr/local/bin/text-embeddings-router
+ENV VIRTUAL_ENV=/usr/src/backends/python/server/.venv
+ENV PATH=$VIRTUAL_ENV/bin/:$PATH
 
 ENTRYPOINT ["text-embeddings-router"]
 CMD ["--json-output"]

backends/python/server/Makefile

@@ -14,12 +14,24 @@ gen-server:
 	touch text_embeddings_server/pb/__init__.py
 
 install: gen-server
-	pip install pip poetry --upgrade
-	poetry install
+	pip install pip uv --upgrade
+	uv sync
 
-install-intel: gen-server
-	pip install pip poetry --upgrade
-	poetry install --without cuda
+install-cpu: gen-server
+	pip install pip uv --upgrade
+	uv sync --extra cpu
+
+install-xpu: gen-server
+	pip install pip uv --upgrade
+	uv sync --extra xpu
+
+install-gpu: gen-server
+	pip install pip uv --upgrade
+	uv sync --extra gpu
+
+install-hpu: gen-server
+	pip install pip uv --upgrade
+	uv sync --extra hpu --inexact
 
 run-dev:
 	python text_embeddings_server/cli.py serve BAAI/bge-small-en

backends/python/server/pyproject.toml

@@ -1,37 +1,102 @@
-[tool.poetry]
+[project]
 name = "text-embeddings-server"
 version = "0.1.0"
 description = "Text Embeddings Python gRPC Server"
-authors = ["Olivier Dehaene <[email protected]>"]
+authors = [{ name = "Olivier Dehaene" , email = "[email protected]" }]
+requires-python = ">= 3.9"
+dependencies = [
+	"protobuf>=4.25.3,<6",
+	"grpcio>=1.51.1",
+	"grpcio-status>=1.51.1",
+	"grpcio-reflection>=1.51.1",
+	"grpc-interceptor>=0.15.0",
+	"typer>=0.6.1",
+	"safetensors>=0.4",
+	"loguru>=0.6.0",
+	"opentelemetry-api>=1.25.0",
+	"opentelemetry-exporter-otlp>=1.25.0",
+	"opentelemetry-instrumentation-grpc>=0.46b0",
+	"sentence-transformers>=3.3.1",
+]
 
-[tool.poetry.scripts]
+[project.scripts]
 python-text-embeddings-server = 'text_embeddings_server.cli:app'
 
-[tool.poetry.dependencies]
-python = ">=3.9,<3.13"
-protobuf = ">=4.25.3,<6"
-grpcio = "^1.51.1"
-grpcio-status = "^1.51.1"
-grpcio-reflection = "^1.51.1"
-grpc-interceptor = "^0.15.0"
-typer = "^0.6.1"
-safetensors = "^0.4"
-loguru = "^0.6.0"
-opentelemetry-api = "^1.25.0"
-opentelemetry-exporter-otlp = "^1.25.0"
-opentelemetry-instrumentation-grpc = "^0.46b0"
-sentence-transformers = "^3.3.1"
-
-[tool.poetry.extras]
-
-[tool.poetry.group.dev.dependencies]
-grpcio-tools = "^1.51.1"
-pytest = "^7.3.0"
-
-[[tool.poetry.source]]
-name = "pytorch-gpu-src"
+[dependency-groups]
+dev = [
+ "grpcio-tools>=1.51.1",
+ "pytest>=7.3.0"
+]
+
+
+[project.optional-dependencies]
+cpu = ["torch"]
+xpu = [
+	"torch",
+	"intel-extension-for-pytorch==2.6.10+xpu",
+	"oneccl_bind_pt==2.6.0+xpu",
+
+]
+hpu = [
+	# Hpu brings its own dependencies within docker
+	# "torch",
+]
+gpu = ["torch"]
+
+[tool.uv]
+conflicts = [
+    [
+      { extra = "cpu" },
+      { extra = "gpu" },
+      { extra = "xpu" },
+      { extra = "hpu" },
+    ],
+]
+override-dependencies = [
+   "intel-sycl-rt==2025.0.4; extra == 'xpu'",
+   "intel-openmp==2025.0.4; extra == 'xpu'",
+]
+constraint-dependencies = [
+	"torch==2.5.1; extra == 'hpu'",
+	"torch==2.6.0; extra == 'cpu'",
+	"torch==2.6.0; extra == 'gpu'",
+	"torch==2.6.0+xpu; extra == 'xpu'",
+        # "torch; extra == 'hpu'",
+]
+
+[tool.uv.sources]
+torch = [
+  { index = "torch-cpu", extra = "cpu" },
+  { index = "torch-gpu", extra = "gpu" },
+  { index = "torch-xpu", extra = "xpu" },
+]
+intel-extension-for-pytorch = [ {index = "xpu" }]
+oneccl_bind_pt = [ {index = "xpu" }]
+mkl = [ {index = "pypi" }]
+
+[[tool.uv.index]]
+name = "pypi"
+url = "https://pypi.org/simple"
+
+
+[[tool.uv.index]]
+name = "torch-cpu"
+url = "https://download.pytorch.org/whl/cpu"
+explicit = true
+
+[[tool.uv.index]]
+name = "torch-gpu"
 url = "https://download.pytorch.org/whl/cu118"
-priority = "explicit"
+explicit = true
+
+[[tool.uv.index]]
+name = "xpu"
+url = "https://pytorch-extension.intel.com/release-whl/stable/xpu/us/"
+
+
+[[tool.uv.index]]
+name = "torch-xpu"
+url = "https://download.pytorch.org/whl/xpu"
 
 [tool.pytest.ini_options]
 markers = ["private: marks tests as requiring an admin hf token (deselect with '-m \"not private\"')"]