1 回答
- 最新
- 投票最多
- 评论最多
0
- Go to the SageMaker console and select the 'Model' tab. You should see your model listed here.
- Select the model and then choose the 'Monitoring' tab. Here you can configure monitoring schedules and view any alerts.
- For a new model, you may need to wait some time for baseline metrics to be collected before monitoring can be enabled
- You can set monitoring schedules to check for concept drift at regular intervals. Schedules can be custom frequency like every 6 hours.
- Alerts will be triggered if the model performance deviates from the baseline by a specified threshold.
- You can drill down on any alerts to analyze root cause using various plots and metrics provided in SageMaker Studio notebooks.
- Logs and metrics from the model endpoint are continuously exported to CloudWatch for long term retention and analysis.
相关内容
- AWS 官方已更新 2 年前
- AWS 官方已更新 2 年前
- AWS 官方已更新 1 年前
I have mentioned earlier I hosted the LLM model using Sagemaker studio jump start. I did nothing manual configuration at the time of hosting just select the model and click on deploy button. So, after deployment I have the endpoint.
I checked on the Sagemaker console and found Governance option is there under the Governance - Model Dashboard, I can see my model but there is no option for enabling monitoring service.
I checked on Sagemaker studio console I can list my endpoint, but no option is there for Monitoring.