sudo apt-get install gcc python3-dev openssl libmariadb3 libmariadb-dev
git clone https://github.com/mariadb-corporation/mariadb-connector-python/
cd mariadb-connector-python
git checkout remotes/origin/1.0
python3 setup.py build
python3 -m pip install .
python3 -m pip install hyperopt
Add server IP and address to ~/.bashrc
export CRAVE_RESULTS_HOST=localhost
export CRAVE_RESULTS_PORT=65099
Add crave results as submodule
git submodule add <repo>
Usage example
from CraveResults import CraveResults, CraveResultsException
db = CraveResults()
# Get dataset file
file_contents = results_db.get_dataset("dataset1/file1.mmap")
with open("dataset1/file1.mmap", "wb") as f:
f.write(file_contents)
config = dict(
learning_rate=0.01,
momentum=0.2,
architecture="CNN",
dataset_id="peds-0192",
infra="AWS",
)
db.init({
"project": "detect_pedestrians",
"notes": "tweak baseline",
"tags": ["baseline", "paper1"],
"config": config
})
for epoch in epochs:
db.log(
{
"loss": loss,
"acc": acc,
"some_text": "Some text %d" % it,
"augmentation image": db.binary("assets/1.jpeg", "%d.jpg" % it)
}
)
db.log_summary({"acc": accuracy})
db.log_artifact({
"confusion matrix": db.binary("assets/1.jpeg")
})
db.log_artifact({"log": db.binary("test.log")})
If float
is being recorded in database, column with same name will not be converted to another type e.g. dict
.