flex_model package
Subpackages
- Core: Wrapper and HookFunction
- Modules
- flex_model.core.FlexModel
- flex_model.core.HookFunction
FlexModelFlexModel.forward()FlexModel.get_hook_function_groups()FlexModel.get_group_hook_functions()FlexModel.update_hook_groups()FlexModel.remove_hook_groups()FlexModel.create_hook_group()FlexModel.register_forward_hook()FlexModel.register_full_backward_hook()FlexModel.register_hook()FlexModel.register_forward_pre_hook()FlexModel.register_full_backward_pre_hook()FlexModel.register_trainable_module()FlexModel.get_module_parameter()FlexModel.wrapped_module_namesFlexModel.trainable_modules_namesFlexModel.all_modules_namesFlexModel.named_parameters()FlexModel.named_buffers()FlexModel.restore()
HookFunction
- Miscellaneous
- Modules
- Distributed: Backend, mappings and strategies
- Distributed API
- Mappings
- flex_model.distributed.broadcast_tensor_parallel
- flex_model.distributed.broadcast_data_parallel
- flex_model.distributed.all_gather_tensor_parallel
- flex_model.distributed.all_gather_data_parallel
- flex_model.distributed.scatter_tensor_parallel
- flex_model.distributed.scatter_data_parallel
- flex_model.distributed.gather_pipeline_parallel_tensor_dicts
- Strategies
- flex_model.distributed.BaseRoutingStrategy
- flex_model.distributed.ParameterTensorParallelRoutingStrategy
- flex_model.distributed.ActivationTensorAllToAllRoutingStrategy
- flex_model.distributed.BaseOffloadStrategy
- flex_model.distributed.NullMemoryOffloadStrategy
- flex_model.distributed.CPUPinnedMemoryOffloadStrategy
- flex_model.distributed.CPUPagedMemoryOffloadStrategy
- flex_model.distributed.GPUMemoryOffloadStrategy
- flex_model.distributed.BaseFunctionStrategy
- flex_model.distributed.NonValidatedFunctionStrategy
- Traverse: Pytree traversal utility
Submodules
flex_model.utils module
- flex_model.utils.setup_logger(level)