Is Open Source AI Bull?

• SPELL

  1 64 0.5 Python

  Learning Long-Term Spatial-Temporal Graphs for Active Speaker Detection (ECCV 2022) (by SRA2)

  Contemporary open source AI falls short of any true or complete notion of openness. It does not come close to my definition. There are pieces of the puzzle being released independently and mostly complementarily by different groups. For example, Google led the development and release of the widely used Open Images dataset; others have released model code to use various dataset, but with limited training hyperparameters or code, such as the teamwho won the ECCV2022 Ego4D challenge. I don’t mean to single out any one group here, this is the status quo. Release a dataset. Release a model. I myself have done this too. Perhaps it is easy to find references from Google and Meta because they have done such a good job of releasing elements of open source AI.

• SaaSHub

  www.saashub.com featured

  SaaSHub - Software Alternatives and Reviews. SaaSHub helps you find the best software and product alternatives

  

• examples

  159 7,963 5.6 Jupyter Notebook

  TensorFlow examples (by tensorflow)

  

  Software Frameworks. These are the libraries and frameworks on which the system source code is built. One needs access to not only the frameworks (many of these are open source software already, such as PyTorch and Tensorflow) but also the specific versioning used in the system source code and the training source code. Details matter.

• Pytorch

  368 84,676 10.0 Python

  Tensors and Dynamic neural networks in Python with strong GPU acceleration

  

  Software Frameworks. These are the libraries and frameworks on which the system source code is built. One needs access to not only the frameworks (many of these are open source software already, such as PyTorch and Tensorflow) but also the specific versioning used in the system source code and the training source code. Details matter.

Suggest a related project