dmx.compressor.modeling.nn.torch\_modules.ScaledDotProductAttention =================================================================== .. currentmodule:: dmx.compressor.modeling.nn.torch_modules .. autoclass:: ScaledDotProductAttention .. automethod:: __init__ .. rubric:: Methods .. autosummary:: ~ScaledDotProductAttention.__init__ ~ScaledDotProductAttention.add_module ~ScaledDotProductAttention.align_device ~ScaledDotProductAttention.apply ~ScaledDotProductAttention.approx_forward ~ScaledDotProductAttention.approximator_wrapper ~ScaledDotProductAttention.bfloat16 ~ScaledDotProductAttention.buffers ~ScaledDotProductAttention.calibrating_quantizers ~ScaledDotProductAttention.calibrating_smoothquant ~ScaledDotProductAttention.check_format_dim_consistency ~ScaledDotProductAttention.check_input_format_dim_consistency ~ScaledDotProductAttention.check_residual_format_dim_consistency ~ScaledDotProductAttention.check_sparseness_dim_consistency ~ScaledDotProductAttention.check_weight_format_dim_consistency ~ScaledDotProductAttention.children ~ScaledDotProductAttention.compile ~ScaledDotProductAttention.configure ~ScaledDotProductAttention.count_flops ~ScaledDotProductAttention.counting_flops ~ScaledDotProductAttention.cpu ~ScaledDotProductAttention.cuda ~ScaledDotProductAttention.dmx_config ~ScaledDotProductAttention.double ~ScaledDotProductAttention.enable_approximation_function_tuning ~ScaledDotProductAttention.enable_flop_counter ~ScaledDotProductAttention.enable_optimal_brain_compression ~ScaledDotProductAttention.enable_quantizer_calib ~ScaledDotProductAttention.enable_smoothquant_calib ~ScaledDotProductAttention.eval ~ScaledDotProductAttention.extra_repr ~ScaledDotProductAttention.float ~ScaledDotProductAttention.fold_weight_and_bias ~ScaledDotProductAttention.forward ~ScaledDotProductAttention.get_buffer ~ScaledDotProductAttention.get_extra_state ~ScaledDotProductAttention.get_parameter ~ScaledDotProductAttention.get_submodule ~ScaledDotProductAttention.half ~ScaledDotProductAttention.infer_ch_axis ~ScaledDotProductAttention.init_casts ~ScaledDotProductAttention.init_smoothquant ~ScaledDotProductAttention.init_sparsifier ~ScaledDotProductAttention.ipu ~ScaledDotProductAttention.load_state_dict ~ScaledDotProductAttention.load_state_dict_and_register_url ~ScaledDotProductAttention.measuring_runtime ~ScaledDotProductAttention.module_graph ~ScaledDotProductAttention.modules ~ScaledDotProductAttention.monitoring ~ScaledDotProductAttention.mtia ~ScaledDotProductAttention.named_buffers ~ScaledDotProductAttention.named_children ~ScaledDotProductAttention.named_modules ~ScaledDotProductAttention.named_parameters ~ScaledDotProductAttention.optimal_brain_compressing ~ScaledDotProductAttention.parameters ~ScaledDotProductAttention.register_backward_hook ~ScaledDotProductAttention.register_buffer ~ScaledDotProductAttention.register_forward_hook ~ScaledDotProductAttention.register_forward_pre_hook ~ScaledDotProductAttention.register_full_backward_hook ~ScaledDotProductAttention.register_full_backward_pre_hook ~ScaledDotProductAttention.register_load_state_dict_post_hook ~ScaledDotProductAttention.register_load_state_dict_pre_hook ~ScaledDotProductAttention.register_module ~ScaledDotProductAttention.register_parameter ~ScaledDotProductAttention.register_state_dict_post_hook ~ScaledDotProductAttention.register_state_dict_pre_hook ~ScaledDotProductAttention.requires_grad_ ~ScaledDotProductAttention.save_state_dict_and_register_url ~ScaledDotProductAttention.set_extra_state ~ScaledDotProductAttention.set_submodule ~ScaledDotProductAttention.share_memory ~ScaledDotProductAttention.slanc_tuning ~ScaledDotProductAttention.state_dict ~ScaledDotProductAttention.to ~ScaledDotProductAttention.to_compiler_graph ~ScaledDotProductAttention.to_empty ~ScaledDotProductAttention.train ~ScaledDotProductAttention.transform ~ScaledDotProductAttention.tuning_approximation_function ~ScaledDotProductAttention.type ~ScaledDotProductAttention.update_params_with_raw ~ScaledDotProductAttention.update_smoothquant_scale ~ScaledDotProductAttention.xpu ~ScaledDotProductAttention.zero_flop_counter ~ScaledDotProductAttention.zero_grad .. rubric:: Attributes .. autosummary:: ~ScaledDotProductAttention.T_destination ~ScaledDotProductAttention.accum_format ~ScaledDotProductAttention.approximation_function ~ScaledDotProductAttention.bias_format ~ScaledDotProductAttention.bops ~ScaledDotProductAttention.call_super_init ~ScaledDotProductAttention.dump_patches ~ScaledDotProductAttention.effective_weight ~ScaledDotProductAttention.flop_counter ~ScaledDotProductAttention.flop_counter_enabled ~ScaledDotProductAttention.flops ~ScaledDotProductAttention.functional_forward ~ScaledDotProductAttention.input_formats ~ScaledDotProductAttention.input_precision ~ScaledDotProductAttention.is_compound ~ScaledDotProductAttention.last_input_shape ~ScaledDotProductAttention.last_output_shape ~ScaledDotProductAttention.multiplier_format ~ScaledDotProductAttention.output_formats ~ScaledDotProductAttention.plugins ~ScaledDotProductAttention.residual_format ~ScaledDotProductAttention.weight_elem_count ~ScaledDotProductAttention.weight_format ~ScaledDotProductAttention.weight_hypernet ~ScaledDotProductAttention.weight_precision ~ScaledDotProductAttention.weight_scale ~ScaledDotProductAttention.weight_size_in_bytes ~ScaledDotProductAttention.weight_sparseness ~ScaledDotProductAttention.weight_storage_format ~ScaledDotProductAttention.weight_storage_precision ~ScaledDotProductAttention.weight_storage_scale ~ScaledDotProductAttention.weight_storage_zero_point ~ScaledDotProductAttention.weight_zero_point ~ScaledDotProductAttention.training