tatp22/multidim-positional-encoding
A Python library providing 1D, 2D, and 3D sinusoidal positional encoding implementations for PyTorch and TensorFlow.
Not currently ranked — collecting fresh signals.
star history
This library implements sinusoidal positional encodings for deep learning models, supporting PyTorch and TensorFlow. It provides PositionalEncoding1D, PositionalEncoding2D, and PositionalEncoding3D classes along with a Summer class to add encodings to input tensors. The encodings work on tensors of various shapes and are a key building block for transformer-based architectures.
Frequently asked
- What is tatp22/multidim-positional-encoding?
- A Python library providing 1D, 2D, and 3D sinusoidal positional encoding implementations for PyTorch and TensorFlow.
- Is multidim-positional-encoding open source?
- Yes — tatp22/multidim-positional-encoding is open source, released under the MIT license.
- What language is multidim-positional-encoding written in?
- tatp22/multidim-positional-encoding is primarily written in Python.
- How popular is multidim-positional-encoding?
- tatp22/multidim-positional-encoding has 616 stars on GitHub.
- Where can I find multidim-positional-encoding?
- tatp22/multidim-positional-encoding is on GitHub at https://github.com/tatp22/multidim-positional-encoding.