Update dependencies; use uv.

.github/dependabot.yml

@@ -1,6 +1,6 @@
 version: 2
 updates:
-  - package-ecosystem: "pip" # Confusingly, pip is correct for poetry
+  - package-ecosystem: "pip" # Dependabot uses pip for uv-managed pyproject files
     directory: "/" # Location of package manifests
     schedule:
       interval: "monthly"

.github/workflows/tests.yml

@@ -31,26 +31,19 @@ jobs:
         run: |
           ./start_services.sh --no-jupyter -d
 
-      - name: Install poetry
-        run: pipx install poetry
+      - name: Install uv
+        run: pipx install uv
 
       - name: Verify MLflow versions match
         run: ./scripts/check_mlflow_versions.sh
 
-      - name: Check poetry lock file
-        run: poetry check --lock
-
-      - name: Remove existing virtual environment
-        run: |
-          poetry env remove python || true
-
       - name: Install dependencies
         run: |
-          poetry cache clear --no-interaction --all .
-          poetry install --with test --no-cache
+          uv lock --check
+          uv sync --group test
 
       - name: Run all tests
-        run: MLFLOW_TRACKING_URI=http://localhost:8080 poetry run pytest
+        run: MLFLOW_TRACKING_URI=http://localhost:8080 uv run pytest
 
       - name: Stop MLflow server
         run: docker compose down
@@ -79,7 +72,7 @@ jobs:
 
       - name: Execute all notebooks
         run: |
-          docker exec modelplane-jupyter-1 poetry run python /app/test_notebooks.py
+          docker exec modelplane-jupyter-1 uv run python /app/test_notebooks.py
 
       - name: Stop MLflow server
         run: docker compose down

.gitignore

@@ -11,3 +11,4 @@ secrets.toml
 *.txt
 *.json
 *.jsonl
+modelplane.egg-info/

Dockerfile.jupyter

@@ -4,22 +4,21 @@ ENV PATH="/root/.local/bin:$PATH"
 # Used for the notebook server
 WORKDIR /app
 
-# pipx needed for poetry installation
+# pipx needed for uv installation script
 # ssh client needed for installing private modelbench dependencies
 # git needed dvc
 RUN apt-get update && apt-get install -y pipx openssh-client git && \
-    pipx install poetry
-COPY pyproject.toml poetry.lock README.md ./
+    pipx install uv
+COPY pyproject.toml uv.lock README.md ./
 
 RUN mkdir -p /root/.ssh && chmod 700 /root/.ssh
 RUN git config --global core.sshCommand "ssh -o UserKnownHostsFile=/root/.ssh/known_hosts -o ForwardAgent=yes"
 RUN ssh-keyscan github.com > /root/.ssh/known_hosts
 
-RUN poetry install --no-interaction --no-ansi --no-root
+RUN uv sync --frozen --no-install-project
 
 COPY src/ ./src/
-RUN echo $(poetry env info --path)
-RUN poetry install --no-interaction --no-ansi
+RUN uv sync --frozen
 
 EXPOSE 8888
-CMD ["poetry", "run", "jupyter", "notebook", "--ip", "0.0.0.0", "--allow-root", "--no-browser", "--ServerApp.root_dir=/app/flightpaths"]
+CMD ["uv", "run", "jupyter", "notebook", "--ip", "0.0.0.0", "--allow-root", "--no-browser", "--ServerApp.root_dir=/app/flightpaths"]

Dockerfile.mlflow

@@ -3,5 +3,5 @@ FROM ghcr.io/mlflow/mlflow:v3.7.0
 # The base image does not include various dependencies that are needed for
 # the MLflow server. We assume a postgres backend, so we need psycopg2.
 # We also need boto3 for S3 support, and google-cloud-storage for GCS support.
-RUN pip install mlflow[auth]==3.7.0 psycopg2-binary==2.9.11 boto3==1.42.43 \
+RUN pip install mlflow[auth]==3.7.0 psycopg2-binary==2.9.11 boto3==1.42.51 \
     google-cloud-storage==3.9.0

Dockerfile.mockvllm

@@ -4,8 +4,9 @@ WORKDIR /app
 
 COPY tests/notebooks/mock_vllm_server.py .
 
-# versions chosen to match whats in poetry.lock as of 2025-07-09
-RUN pip install fastapi==0.115.12 uvicorn==0.34.3
+# versions chosen to match whats in uv.lock as of 2026-02-18
+# TODO: maybe match the uv.lock, though it's not functionally important here
+RUN pip install fastapi==0.129.0 uvicorn==0.41.0
 
 EXPOSE ${VLLM_PORT}
 

README.md

@@ -79,31 +79,31 @@ Annotator and SUT responses will be cached (locally) unless you pass the
 
 ## CLI
 
-You can also interact with modelplane via CLI. Run `poetry run modelplane --help`
+You can also interact with modelplane via CLI. Run `uv run modelplane --help`
 for more details.
 
 *Important:* You must set the `MLFLOW_TRACKING_URI` environmental variable.
 For example, if you've brought up MLFlow using the fully local docker compose process above,
 you could run:
 ```
-MLFLOW_TRACKING_URI=http://localhost:8080 poetry run modelplane get-sut-responses --sut_id {sut_id} --prompts tests/data/prompts.csv --experiment expname
+MLFLOW_TRACKING_URI=http://localhost:8080 uv run modelplane get-sut-responses --sut_id {sut_id} --prompts tests/data/prompts.csv --experiment expname
 ```
 After running the command, you'd see the `run_id` in the output from mlflow, 
 or you can get the `run_id` via the MLFlow UI.
 
 ### Basic Annotations
 ```
-MLFLOW_TRACKING_URI=http://localhost:8080 poetry run modelplane annotate --annotator_id {annotator_id} --experiment expname --response_run_id {run_id}
+MLFLOW_TRACKING_URI=http://localhost:8080 uv run modelplane annotate --annotator_id {annotator_id} --experiment expname --response_run_id {run_id}
 ```
 
 #### Private Ensemble
 If you have access to the private annotator, you can run directly with:
 ```
-MLFLOW_TRACKING_URI=http://localhost:8080 poetry run modelplane annotate --annotator_id safety-v1.1 --experiment expname --response_run_id {run_id}
+MLFLOW_TRACKING_URI=http://localhost:8080 uv run modelplane annotate --annotator_id safety-v1.1 --experiment expname --response_run_id {run_id}
 ```
 
 
 ### Custom Ensembles
 ```
-MLFLOW_TRACKING_URI=http://localhost:8080 poetry run modelplane annotate --annotator_id {annotator_id1} --annotator_id {annotator_id2} --ensemble_strategy {ensemble_strategy} --experiment expname --response_file path/to/response.csv
+MLFLOW_TRACKING_URI=http://localhost:8080 uv run modelplane annotate --annotator_id {annotator_id1} --annotator_id {annotator_id2} --ensemble_strategy {ensemble_strategy} --experiment expname --response_file path/to/response.csv
 ```