setrconsumer.blogg.se - Evaluating logs

Evaluating logs update#
Evaluating logs code#

Our understanding is that it is a license more permissive than the MIT license and allows for removing of the copyright headers. This project retains the license of the aws-deepracer-workshops project from which this was based.

AWS DeepRacer Community members who have shared ideas and contributed in one way or another.

Evaluating logs code#

The AWS DeepRacer product and data science teams who had shared the original Log Analysis code as part of the AWS DeepRacer Workshops repository.Run all cells to test the log analysis on your logs!.Edit the path to your logs in the Notebook.ipynb file up on Jupyter Notebook (works in an Amazon SageMaker Notebook instance too) Extract the RoboMaker and SageMaker log files (found in logs/training/ in the.Download your own model training log (.tar.gz file) from the AWS DeepRacer console.To take this Notebook on a quick test drive, just do the following: In this Deep Dive video, I explain how I used some of the visualizations in this Notebook to optimize racing lines and determine the performance envelope of my models.

In this Machine Learning blog, I explain how I used this Notebook to drive experiments and win the AWS DeepRacer F1 ProAm Race.

does not require access to any AWS Services (hence no awscli or boto3 required) when analysing the log data files.

is maintained such that it can be run directly from a Amazon SageMaker Notebook instance, as long as the relative paths of the RoboMaker and SageMaker log files are specified correctly.

should be backwards compatible with DeepRacer logs previously downloaded from CloudWatch Logs too.

is compatible with downloaded logs from the new console (after Aug 2020).Instead, they are downloadable from the model page in the AWS DeepRacer console, after training has terminated.

Evaluating logs update#

With the new AWS DeepRacer console update in Aug 2020, these logs are no longer streamed to CloudWatch Logs during training. The log analysis here parses log data from AWS RoboMaker (SIM_TRACE_LOG data) and Amazon SageMaker (policy training data), and introduces some analyses that are not present in the AWS samples. This Notebook, which is compatible with logs from the AWS DeepRacer Console after Aug 2020, is a redo of Log Analysis solutions provided in the AWS DeepRacer Workshops repository. Deepracer-log-analysis Adventures in a data-driven approach to training, evaluating and tuning AWS DeepRacer reinforcement learning models.