Sacred VS pytorch-lightning

🔗 https://github.com/IDSIA/sacred

I'm using Omniboard (https://github.com/vivekratnavel/omniboard) with Sacred (https://github.com/IDSIA/sacred) for tracking experiments. You can specify custom Observers in Sacred so the model metrics and logs will be saved to a local directory or to a remote DB (e.g., MongoDB). I use a MongoDB database hosted on Atlas. Unlike other suggested options, Sacred and Omniboard are free. Atlas free tier comes with 512MB of free storage which is a huge amount if you're uploading only log files to it.

I'm using Omniboard (https://github.com/vivekratnavel/omniboard) with Sacred (https://github.com/IDSIA/sacred) for tracking experiments. You can specify custom Observers in Sacred so the model metrics and logs will be saved to a local directory or to a remote DB (e.g., MongoDB). I use a MongoDB database hosted on Atlas. Unlike other suggested options, Sacred and Omniboard are free. Atlas free tier comes with 512MB of free storage which is a huge amount if you're uploading only log files to it. ex = Experiment() ex.observers.append(FileStorageObserver(EXPERIMENTS_ROOT)) ex.observers.append(MongoObserver(url=MONGODB_URL, db_name='sacred'))

sacred helped me log my experiments. I did setup my environment only once 4 years ago, and since then I have a list of all my training runs with the hyperparameters and results.

For 1, setup an experiment tracking framework. I found Sacred to be helpful https://github.com/IDSIA/sacred.

def on_validation_end(self, trainer: Trainer, pl_module: LightningModule) -> None: # Trainer calls `on_validation_end` for sanity check. Therefore, it is necessary to avoid # calling `trial.report` multiple times at epoch 0. For more details, see # https://github.com/PyTorchLightning/pytorch-lightning/issues/1391. if trainer.sanity_checking: return

Under the hood, the ModelWrapper object will create a ML model based on the config (so far, an XGBoost model and a PyTorch Lightning model). Each of those will have a wrapper that conducts training and evaluation (since from my understanding of Lightning, Trainers are required to be outside of the class). In lack of a better name, I call these wrappers Fitters. For uniformity, I thought about adding a common interface IFitter, which is inherited by all model wrappers as outlined below.

Join their Slack to ask the community questions and check out the GitHub here.

Pytorch lightning benchmarks against pytorch on every PR (benchmarks to make sure that it is mot slower.

There is already a ton of momentum around automating ML workflows. I would suggest you contribute to a preexisting project like, for instance, PyTorch Lightning or fast.ai.

Is the problem the sheer number of options, or the fact that they are all together in one place? Would it be better if they were organized into the different trainer entrypoints (fit, validate, ...)? If that is the case, there was an RFC proposing this which you might find interesting, feel free to drop by and comment on the issue: https://github.com/PyTorchLightning/pytorch-lightning/issues/10444

I wanted to make a quick performance comparison between the GPU (Tesla K80) and TPU (v2-8) available in Google Colab with PyTorch. To do so quickly, I used an MNIST example from pytorch-lightning that trains a simple CNN.

We've noticed GPU 0 on our 3 GPU system is sometimes idle (which would explain performance differences). However its unclear to us why that may be. Similar to this issue

If you want a good example for reference, take a look at Pytorch Lightning's readme (https://github.com/PyTorchLightning/pytorch-lightning) It answers the 3 questions of "what is this", "why should I care", and "how do i use it" almost instantly