This package contains an implementation for tensor parallelism in transformer models (see Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism and Reducing Activation Recomputation in Large Transformer Models for details).
.. automodule:: core.tensor_parallel.cross_entropy :members: :undoc-members: :show-inheritance:
.. automodule:: core.tensor_parallel.data :members: :undoc-members: :show-inheritance:
.. automodule:: core.tensor_parallel.layers :members: :undoc-members: :show-inheritance:
.. automodule:: core.tensor_parallel.mappings :members: :undoc-members: :show-inheritance:
.. automodule:: core.tensor_parallel.random :members: :undoc-members: :show-inheritance:
.. automodule:: core.tensor_parallel.utils :members: :undoc-members: :show-inheritance:
.. automodule:: core.tensor_parallel :members: :undoc-members: :show-inheritance: