![]() If you wish to continue modifying your code, while a training run is going on, you may do so without worry whether it will affect the running job(s).Runx actually makes a copy of your code within each run's log directory. If you include the RUNX.TAG field in your experiment yaml or if you supply the -tag argument to the runx CLI, the names will include that tag. The use of coolname makes it much easier to refer to a given run than referring to a date code. The individual run directories are named with a combination of coolname and date. CODE_IGNORE_PATTERNS - ignore these files patterns when copying code to output directory.You can list any number of these items, the ones shown below are just examples. RESOURCES - hyperparameters passed to the SUBMIT_CMD.SUBMIT_CMD - the farm submission command.For a given farm, these fields are required:.FARM - if defined, jobs should be submitted to this farm, else run interactively. ![]() This is a path that any farm job can write to. LOGROOT - the root directory where you want your logs placed.runx file defines a number of critical fields: In order to use runx, you need to create a configuration file in the directory where you'll call the runx CLI. Create a project-specific configuration file However using sumx requires that you've used logx to record metrics. These modules are intended to be used jointly, but if you just want to use runx, that's fine. Summarize the results of training runs, showing results and unique hyperparameters.Logging of metrics, messages, checkpoints, tensorboard.For each run, create an output directory, copy your code there, and then launch the training command.Calculate cross product of all hyperparameters -> runs.Launch sweeps of training runs using a concise yaml format that allows for multiple values for each hyperparameter.The following sections will go into more details about all the various features. Notice that we used the -sortwith feature of sumx, which sorts your results so you can easily locate your best runs. Sumx is part of the runx suite, and is able to summarize the different hyperparmeters used as well as the metrics/results of your runs. ![]() > python -m runx.sumx sweep -sortwith acc In the following example, we ask sumx to summarize the sweep experiment. runx file) and so it looks within that directory for your experiment directory. sumx knows what your LOGROOT (it'll get that from the. All you need to do is tell sumx which experiment you want summarized. You summarize your runs with on the commandline with sumx. Submit_job -gpu 2 -cpu 16 -mem 128 -c "python train.py -lr 0.02 -solver adam -logdir /home/logs/vengeful-jaguar_2020.02.06_14.19 " Summarization with sumxĪfter you've run your experiment, you will likely want to summarize the results.
0 Comments
Leave a Reply. |