Thingsboard VS pytorch-lightning

Until recently I was still using TB v3.2.1, now I've set about updating the installation to the latest version. I proceeded as described at https://github.com/thingsboard/thingsboard/tree/master/docker:

You might find something more attuned to that use case to be more helpful out of the box (like thingsboard.io) but if you're committed to django, set up an API endpoint to receive json updates ('events') sent from the arduino.

Introduction：As a 100% open source IoT platform that can be hosted as a SaaS or PaaS solution, Thingsboard can provide device management, data collection, processing and visualization for your IoT project. The standard protocols that provide device connectivity such as MQTT, CoAP, and HTTP are all available on ThingsBoard. In addition, it supports cloud and local deployment and provides more than 30 customizable components that allow you to build end-user custom dashboards for most IoT cases. GitHub: https://github.com/thingsboardThingsboard: http://thingsboard.io/ Features:

Then host a ThingsBoard server and use the HTTP API to push data from the device. You can send alerts with Pushover.

**Source** [thingsboard (a foreign iot platform)](https://thingsboard.io/)

ThingsBoard is an open-source IoT platform for data collection, processing, visualization and device management. It supports device connectivity via protocols, such as MQTT, CoAP and HTTP, and supports both cloud and private deployments. Deliver, monitor and control your IoT entities in a secure way using rich server-side APIs that define the relationships between your devices, assets, customers, or any other entities. Collect and store telemetry data in a scalable and fault-tolerant manner, visualize your data with built-in or custom widgets and flexible dashboards, and share the Dashboard interface with your customers.

The company I work for uses https://thingsboard.io and the team from Ukraine that built it are awesome and the company is still responding even though they are going through this hell of a war.

https://see.stanford.edu/Course/CS229 https://lightning.ai/ https://www.youtube.com/watch?v=00s9ireCnCw&t=57s https://towardsdatascience.com/

I've been wondering on what is the recommended method of saving logits/activations using PyTorch Lightning. I've looked at Callbacks, Loggers and ModelHooks but none of the use-cases seem to be for this kind of activity (even if I were to create my own custom variants of each utility). The ModelCheckpoint Callback in its utility makes me feel like custom Callbacks would be the way to go but I'm not quite sure. This closed GitHub issue does address my issue to some extent.

Our codebase for the diffusion models builds heavily on OpenAI's ADM codebase , lucidrains, Stable Diffusion, Lightning and Hugging Face. Thanks for open-sourcing!

For lightweight use, pytorch-lightning is too heavy, and its source code will be very difficult for beginners to read, at least for me.

For the A100 GPU, theoretical performance is the same for FP16/BF16 and both rely on the same number of bits, meaning memory should be the same. However since it's quite newly added to PyTorch, performance seems to still be dependent on underlying operators used (pytorch lightning debugging in progress here).