SMDebug RulesConfig
Project description
SageMaker-Debugger-RulesConfig
Sagemaker-debugger is an upcoming AWS service designed to be a debugger for machine learning models. It lets you go beyond just looking at scalars like losses and accuracies during training and gives you full visibility into all tensors 'flowing through the graph' during training or inference.
SageMaker Debugger RulesConfig provides a mapping of builtin rules with default configurations. These configurations will affect both DebugHookConfig and DebugRuleConfigurations in the SageMaker Python SDK.
This library, intended to be used with SageMaker PySDK, helps you specify builtin rules without worrying about any details or tweak the configuration of builtin rules. These builtin rules are available in SageMaker.
SageMaker Debugger Rulesconfig package can be used with SageMaker Debugger or as stand-alone rule config retriever too. In addition to retrieving builtin rules, configuration for common collections can be retrieved as well.
Example: Vanilla builtin rule without customization
from sagemaker.debug import Rule
from smdebug_rulesconfig import vanishing_gradient
my_estimator = Estimator(
...
rules=[Rule.sagemaker(vanishing_gradient())]
)
Example: Builtin rule with customization
my_estimator = Estimator(
...
rules= [
Rule.sagemaker(vanishing_gradient()),
Rule.sagemaker(
base_config=weight_update_ratio(),
instance_type="t2.micro",
name="my_wup_rule_name",
container_local_path="/local/path",
s3_output_path="s3://uri",
volume_size_in_gb=100,
rule_parameters={
"param1": "value1",
"param2": "value2"
},
collection_configurations=[
CollectionConfiguration(
name="my_name", # Required
parameters= {
"param1": "value1",
"param2": "value2"
} # Required
)
],
)
],
wait=False
)
Example: Builtin rule with collection configuration specified
from smdebug_rulesconfig import get_collection
my_estimator = Estimator(
...
rules= [
Rule.sagemaker(
base_config=vanishing_gradient(),
collection_configurations=[
get_collection("weights")
],
)
],
wait=False
)
License
This project is licensed under the Apache-2.0 License.
Project details
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Hashes for smdebug_rulesconfig-0.1.2.tar.gz
Algorithm | Hash digest | |
---|---|---|
SHA256 | e61f8337f459056660603d83c03b5c34f6319cc87835713ebf4eaa00c8994bc9 |
|
MD5 | 5d7ac00b39f3d69527ccd0293f31421d |
|
BLAKE2b-256 | d30687d74c1991facada3b3ef00efed641f3a91d1f7d1561f9a9f7695d183023 |
Hashes for smdebug_rulesconfig-0.1.2-py2.py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | 1858a7ff9ddcda76e740bcba4557fc8eb89bb5eeb03395a472b4304e79b9479e |
|
MD5 | 9ca1d17fbd0aa8dd3ab2d0daaeb16199 |
|
BLAKE2b-256 | 3608b0d1d30f5c1e11a2d12af3b940cf5673dc5bbb67517d815807ae855abc5e |