Merge pull request #32 from numerai/pschork/bump_data_version

Set default `--dataset` value to `v4.2/live_int8.parquet`

Makefile

@@ -16,30 +16,30 @@ build:	build_3_9 build_3_10 build_3_11 ## Build all Python containers
 
 .PHONY: build_3_9
 build_3_9: ## Build Python 3.9 container
-	docker build -t ${NAME}_py_3_9:${GIT_REF} -t ${NAME}_py_3_9:latest -f py3.9/Dockerfile .
+	docker build --build-arg GIT_REF=${GIT_REF} -t ${NAME}_py_3_9:${GIT_REF} -t ${NAME}_py_3_9:latest -f py3.9/Dockerfile .
 
 .PHONY: build_3_10
 build_3_10: ## Build Python 3.10 container
-	docker build -t ${NAME}_py_3_10:${GIT_REF} -t ${NAME}_py_3_10:latest -f py3.10/Dockerfile .
+	docker build --build-arg GIT_REF=${GIT_REF} -t ${NAME}_py_3_10:${GIT_REF} -t ${NAME}_py_3_10:latest -f py3.10/Dockerfile .
 
 .PHONY: build_3_11
 build_3_11: ## Build Python 3.11 container
-	docker build -t ${NAME}_py_3_11:${GIT_REF} -t ${NAME}_py_3_11:latest -f py3.11/Dockerfile .
+	docker build --build-arg GIT_REF=${GIT_REF} -t ${NAME}_py_3_11:${GIT_REF} -t ${NAME}_py_3_11:latest -f py3.11/Dockerfile .
 
 .PHONY: test
 test: test_3_9 test_3_10 test_3_11 ## Test all container versions
 
 .PHONY: test_3_9
 test_3_9: build_3_9 ## Test Python 3.9 pickle
-	docker run -i --rm -v ${PWD}:${PWD} -v /tmp:/tmp ${NAME}_py_3_9:latest --model ${PWD}/tests/models/model_3_9.pkl
+	docker run -i --rm -v ${PWD}:${PWD} -v /tmp:/tmp ${NAME}_py_3_9:latest --dataset=v4.1/live.parquet --model ${PWD}/tests/models/model_3_9.pkl
 
 .PHONY: test_3_10
 test_3_10: build_3_10 ## Test Python 3.10 pickle
-	docker run -i --rm -v ${PWD}:${PWD} -v /tmp:/tmp ${NAME}_py_3_10:latest --model ${PWD}/tests/models/model_3_10.pkl
+	docker run -i --rm -v ${PWD}:${PWD} -v /tmp:/tmp ${NAME}_py_3_10:latest --dataset=v4.1/live.parquet --model ${PWD}/tests/models/model_3_10.pkl
 
 .PHONY: test_3_11
 test_3_11: build_3_11 ## Test Python 3.11 pickle
-	docker run -i --rm -v ${PWD}:${PWD} -v /tmp:/tmp ${NAME}_py_3_11:latest --model ${PWD}/tests/models/model_3_11.pkl
+	docker run -i --rm -v ${PWD}:${PWD} -v /tmp:/tmp ${NAME}_py_3_11:latest --dataset=v4.1/live.parquet --model ${PWD}/tests/models/model_3_11.pkl
 
 .PHONY: push_latest
 push_latest: push_latest_3_9 push_latest_3_10 push_latest_3_11 ## Push latest docker containers

predict.py

@@ -20,7 +20,7 @@ def parse_args():
     group = parser.add_mutually_exclusive_group()
     group.add_argument(
         "--dataset",
-        default="v4.1/live.parquet",
+        default="v4.2/live_int8.parquet",
         help="Numerapi dataset path or local file.",
     )
     group.add_argument(
@@ -65,7 +65,8 @@ def py_version(separator='.'):
 
 
 def exit_with_help(error):
-    docker_image_path = f"ghcr.io/numerai/numerai_predict_py_{py_version('_')}:latest"
+    git_ref = os.getenv('GIT_REF', 'latest')
+    docker_image_path = f"ghcr.io/numerai/numerai_predict_py_{py_version('_')}:{git_ref}"
     docker_args = "--debug --model $PWD/[PICKLE_FILE]"
 
     logging.root.handlers[0].flush()
@@ -90,8 +91,7 @@ def exit_with_help(error):
 def main(args):
     logging.getLogger().setLevel(logging.DEBUG if args.debug else logging.INFO)
 
-    python_version = f"Python{py_version()}"
-    logging.info(python_version)
+    logging.info(f"Running numerai-predict:{os.getenv('GIT_REF')} Python{py_version()}")
 
     if args.model.lower().startswith("http"):
         truncated_url = args.model.split("?")[0]

py3.10/Dockerfile

@@ -1,5 +1,7 @@
 FROM python:3.10-slim
 
+ARG GIT_REF
+ENV GIT_REF=$GIT_REF
 RUN apt-get update -yqq \
     && apt-get install -yqq --no-install-recommends \
         build-essential \

py3.11/Dockerfile

@@ -1,5 +1,7 @@
 FROM python:3.11-slim
 
+ARG GIT_REF
+ENV GIT_REF=$GIT_REF
 RUN apt-get update -yqq \
     && apt-get install -yqq --no-install-recommends \
         build-essential \

py3.9/Dockerfile

@@ -1,5 +1,7 @@
 FROM python:3.9-slim
 
+ARG GIT_REF
+ENV GIT_REF=$GIT_REF
 RUN apt-get update -yqq \
     && apt-get install -yqq --no-install-recommends \
         build-essential \