# # PyTorch documentation build configuration file, created by # sphinx-quickstart on Fri Dec 23 13:31:47 2016. # # This file is execfile()d with the current directory set to its # containing dir. # # Note that not all possible configuration values are present in this # autogenerated file. # # All configuration values have a default; values that are commented out # serve to show the default. import inspect import os # import sys import pkgutil import re from os import path # source code directory, relative to this file, for sphinx-autobuild # sys.path.insert(0, os.path.abspath('../..')) import torch # If extensions (or modules to document with autodoc) are in another directory, # add these directories to sys.path here. If the directory is relative to the # documentation root, use os.path.abspath to make it absolute, like shown here. try: import torchvision # noqa: F401 except ImportError: import warnings warnings.warn('unable to load "torchvision" package') RELEASE = os.environ.get("RELEASE", False) import pytorch_sphinx_theme2 html_theme = "pytorch_sphinx_theme2" html_theme_path = [pytorch_sphinx_theme2.get_html_theme_path()] # -- General configuration ------------------------------------------------ # Add any Sphinx extension module names here, as strings. They can be # extensions coming with Sphinx (named 'sphinx.ext.*') or your custom # ones. extensions = [ "sphinx.ext.autodoc", "sphinx.ext.autosummary", "sphinx.ext.doctest", "sphinx.ext.intersphinx", "sphinx.ext.todo", "sphinx.ext.coverage", "sphinx.ext.napoleon", "sphinx.ext.autosectionlabel", "sphinxcontrib.katex", "sphinx_copybutton", "sphinx_design", "myst_nb", "sphinx.ext.linkcode", "sphinxcontrib.mermaid", "sphinx_sitemap", ] myst_enable_extensions = [ "colon_fence", "deflist", "html_image", ] html_baseurl = "https://docs.pytorch.org/docs/stable/" # needed for sphinx-sitemap sitemap_locales = [None] sitemap_excludes = [ "search.html", "genindex.html", ] sitemap_url_scheme = "{link}" html_additional_pages = { "404": "404.html", } # build the templated autosummary files autosummary_generate = True numpydoc_show_class_members = False # autosectionlabel throws warnings if section names are duplicated. # The following tells autosectionlabel to not throw a warning for # duplicated section names that are in different documents. autosectionlabel_prefix_document = True # katex options # # katex_prerender = True # General information about the project. project = "PyTorch" copyright = "PyTorch Contributors" author = "PyTorch Contributors" torch_version = str(torch.__version__) # The version info for the project you're documenting, acts as replacement for # |version| and |release|, also used in various other places throughout the # built documents. # # The short X.Y version. # TODO: change to [:2] at v1.0 version = "main (" + torch_version + " )" # The full version, including alpha/beta/rc tags. release = "main" # Customized html_title here. # Default is " ".join(project, release, "documentation") if not set if RELEASE: # Turn 1.11.0aHASH into 1.11 # Note: the release candidates should no longer have the aHASH suffix, but in any # case we wish to leave only major.minor, even for rc builds. version = ".".join(torch_version.split(".")[:2]) html_title = " ".join((project, version, "documentation")) release = version switcher_version = "main" if not RELEASE else version html_static_path = ["_static"] html_theme_options = { "logo": {"text": "Home"}, "analytics_id": "GTM-T8XT4PS", "canonical_url": "https://pytorch.org/docs/stable/", "switcher": { "json_url": "https://docs.pytorch.org/docs/pytorch-versions.json", "version_match": switcher_version, }, "show_toc_level": 2, "navigation_with_keys": False, "external_links": [ { "name": "Tutorials", "url": "https://pytorch.org/tutorials/", }, ], "show_version_warning_banner": True, "icon_links": [ { "name": "X", "url": "https://x.com/PyTorch", "icon": "fa-brands fa-x-twitter", }, { "name": "GitHub", "url": "https://github.com/pytorch/pytorch", "icon": "fa-brands fa-github", }, { "name": "PyTorch Forum", "url": "https://discuss.pytorch.org/", "icon": "fa-brands fa-discourse", }, { "name": "PyPi", "url": "https://pypi.org/project/torch/", "icon": "fa-brands fa-python", }, ], "navbar_align": "left", "navbar_start": ["version-switcher", "navbar-logo"], "navbar_center": ["navbar-nav"], "navbar_end": ["search-field-custom", "theme-switcher", "navbar-icon-links"], "header_links_before_dropdown": 6, "navbar_persistent": [], "use_edit_page_button": True, "pytorch_project": "docs", } theme_variables = pytorch_sphinx_theme2.get_theme_variables() html_context = { "github_url": "https://github.com", "github_user": "pytorch", "github_repo": "pytorch", "feedback_url": "https://github.com/pytorch/pytorch", "github_version": "main", "pytorch_project": "docs", "doc_path": "docs/source", "theme_variables": theme_variables, # library links are defined in # pytorch_sphinx_theme2/pytorch_sphinx_theme2/links.json "library_links": theme_variables.get("library_links", []), "version": version, "date_info": { "paths_to_skip": ["generated/", "index"], }, } napoleon_use_ivar = True # Add any paths that contain templates here, relative to this directory. templates_path = [ "_templates", os.path.join(os.path.dirname(pytorch_sphinx_theme2.__file__), "templates"), ] # TODO: document these and remove them from here. coverage_ignore_functions = [ # torch "typename", # torch.cuda "check_error", "cudart", "is_bf16_supported", # torch.cuda._sanitizer "zip_arguments", "zip_by_key", # torch.distributed.autograd "is_available", # torch.distributed.checkpoint.state_dict "gc_context", "state_dict", # torch.distributed.elastic.events "construct_and_record_rdzv_event", "record_rdzv_event", # torch.distributed.elastic.metrics "initialize_metrics", # torch.distributed.elastic.rendezvous.registry "get_rendezvous_handler", # torch.distributed.launch "launch", "main", "parse_args", # torch.distributed.rpc "is_available", # torch.distributed.run "config_from_args", "determine_local_world_size", "get_args_parser", "get_rdzv_endpoint", "get_use_env", "main", "parse_args", "parse_min_max_nnodes", "run", "run_script_path", # torch.distributions.constraints "is_dependent", # torch.hub "import_module", # torch.jit "export_opnames", # torch.jit.unsupported_tensor_ops "execWrapper", # torch.onnx "unregister_custom_op_symbolic", # torch.ao.quantization "default_eval_fn", # torch.backends "disable_global_flags", "flags_frozen", # torch.distributed.algorithms.ddp_comm_hooks "register_ddp_comm_hook", # torch.nn.parallel "DistributedDataParallelCPU", # torch.utils "set_module", "burn_in_info", "get_info_and_burn_skeleton", "get_inline_skeleton", "get_model_info", "get_storage_info", "hierarchical_pickle", # torch.amp.autocast_mode "autocast_decorator", # torch.ao.nn.quantized.dynamic.modules.rnn "apply_permutation", "pack_weight_bias", # torch.ao.nn.quantized.reference.modules.rnn "get_quantized_weight", # torch.ao.ns.fx.graph_matcher "get_matching_subgraph_pairs", # torch.ao.ns.fx.graph_passes "add_loggers_to_model", "create_a_shadows_b", # torch.ao.ns.fx.mappings "add_op_to_sets_of_related_ops", "get_base_name_for_op", "get_base_name_to_sets_of_related_ops", "get_node_type_to_io_type_map", "get_unmatchable_types_map", # torch.ao.ns.fx.n_shadows_utils "create_add_loggers_graph", "create_n_transformed_and_logged_copies_of_subgraph", "create_one_transformed_and_logged_copy_of_subgraph", "create_results_comparison", "create_submodule_from_subgraph", "extract_weight_comparison", "group_results_by_subgraph", "print_n_shadows_summary", # torch.ao.ns.fx.pattern_utils "end_node_matches_reversed_fusion", "get_reversed_fusions", "get_type_a_related_to_b", # torch.ao.ns.fx.utils "get_arg_indices_of_inputs_to_log", "get_node_first_input_and_output_type", "get_node_input_qparams", "get_normalized_nth_input", "get_number_of_non_param_args", "get_target_type_str", "maybe_add_missing_fqns", "maybe_dequantize_first_two_tensor_args_and_handle_tuples", "op_type_supports_shadowing", "rekey_logger_info_on_node_name_of_model", "return_first_non_observer_node", # torch.ao.ns.fx.weight_utils "extract_weight_from_node", "get_conv_fun_weight", "get_conv_mod_weight", "get_linear_fun_weight", "get_linear_mod_weight", "get_lstm_mod_weights", "get_lstm_weight", "get_op_to_type_to_weight_extraction_fn", "get_qconv_fun_weight", "get_qlinear_fun_weight", "get_qlstm_weight", "mod_0_weight_detach", "mod_weight_bias_0", "mod_weight_detach", # torch.ao.pruning.sparsifier.utils "fqn_to_module", "get_arg_info_from_tensor_fqn", "module_contains_param", "module_to_fqn", "swap_module", # torch.ao.quantization.backend_config.executorch "get_executorch_backend_config", # torch.ao.quantization.backend_config.fbgemm "get_fbgemm_backend_config", # torch.ao.quantization.backend_config.native "get_native_backend_config", "get_native_backend_config_dict", "get_test_only_legacy_native_backend_config", "get_test_only_legacy_native_backend_config_dict", # torch.ao.quantization.backend_config.onednn "get_onednn_backend_config", # torch.ao.quantization.backend_config.qnnpack "get_qnnpack_backend_config", # torch.ao.quantization.backend_config.tensorrt "get_tensorrt_backend_config", "get_tensorrt_backend_config_dict", # torch.ao.quantization.backend_config.utils "entry_to_pretty_str", "get_fused_module_classes", "get_fuser_method_mapping", "get_fusion_pattern_to_extra_inputs_getter", "get_fusion_pattern_to_root_node_getter", "get_module_to_qat_module", "get_pattern_to_dtype_configs", "get_pattern_to_input_type_to_index", "get_qat_module_classes", "get_root_module_to_quantized_reference_module", "pattern_to_human_readable", "remove_boolean_dispatch_from_name", # torch.ao.quantization.backend_config.x86 "get_x86_backend_config", # torch.ao.quantization.fuse_modules "fuse_known_modules", "fuse_modules_qat", # torch.ao.quantization.fuser_method_mappings "fuse_conv_bn", "fuse_conv_bn_relu", "fuse_convtranspose_bn", "fuse_linear_bn", "get_fuser_method", "get_fuser_method_new", # torch.ao.quantization.fx.convert "convert", "convert_custom_module", "convert_standalone_module", "convert_weighted_module", # torch.ao.quantization.fx.fuse "fuse", # torch.ao.quantization.fx.lower_to_fbgemm "lower_to_fbgemm", # torch.ao.quantization.fx.lower_to_qnnpack "lower_to_qnnpack", # torch.ao.quantization.fx.pattern_utils "get_default_fusion_patterns", "get_default_output_activation_post_process_map", "get_default_quant_patterns", # torch.ao.quantization.fx.prepare "insert_observers_for_model", "prepare", "propagate_dtypes_for_known_nodes", # torch.ao.quantization.fx.utils "all_node_args_except_first", "all_node_args_have_no_tensors", "assert_and_get_unique_device", "collect_producer_nodes", "create_getattr_from_value", "create_node_from_old_node_preserve_meta", "get_custom_module_class_keys", "get_linear_prepack_op_for_dtype", "get_new_attr_name_with_prefix", "get_non_observable_arg_indexes_and_types", "get_qconv_prepack_op", "get_skipped_module_name_and_classes", "graph_module_from_producer_nodes", "maybe_get_next_module", "node_arg_is_bias", "node_arg_is_weight", "return_arg_list", # torch.ao.quantization.pt2e.graph_utils "bfs_trace_with_node_process", "find_sequential_partitions", "get_equivalent_types", "update_equivalent_types_dict", # torch.ao.quantization.pt2e.prepare "prepare", # torch.ao.quantization.pt2e.representation.rewrite "reference_representation_rewrite", # torch.ao.quantization.pt2e.utils "fold_bn_weights_into_conv_node", "remove_tensor_overload_for_qdq_ops", # torch.ao.quantization.qconfig "get_default_qat_qconfig", "get_default_qat_qconfig_dict", "get_default_qconfig", "get_default_qconfig_dict", "qconfig_equals", # torch.ao.quantization.quantization_mappings "get_default_compare_output_module_list", "get_default_dynamic_quant_module_mappings", "get_default_dynamic_sparse_quant_module_mappings", "get_default_float_to_quantized_operator_mappings", "get_default_qat_module_mappings", "get_default_qconfig_propagation_list", "get_default_static_quant_module_mappings", "get_default_static_quant_reference_module_mappings", "get_default_static_sparse_quant_module_mappings", "get_dynamic_quant_module_class", "get_embedding_qat_module_mappings", "get_embedding_static_quant_module_mappings", "get_quantized_operator", "get_static_quant_module_class", "no_observer_set", # torch.ao.quantization.quantize "get_default_custom_config_dict", # torch.ao.quantization.quantize_fx "attach_preserved_attrs_to_model", "convert_to_reference_fx", # torch.ao.quantization.quantize_jit "convert_dynamic_jit", "convert_jit", "fuse_conv_bn_jit", "prepare_dynamic_jit", "prepare_jit", "quantize_dynamic_jit", "quantize_jit", "script_qconfig", "script_qconfig_dict", # torch.ao.quantization.quantize_pt2e "convert_pt2e", "prepare_pt2e", "prepare_qat_pt2e", # torch.ao.quantization.quantizer.embedding_quantizer "get_embedding_operators_config", # torch.ao.quantization.quantizer.xnnpack_quantizer_utils "get_bias_qspec", "get_input_act_qspec", "get_output_act_qspec", "get_weight_qspec", "propagate_annotation", "register_annotator", # torch.ao.quantization.utils "activation_dtype", "activation_is_dynamically_quantized", "activation_is_int32_quantized", "activation_is_int8_quantized", "activation_is_statically_quantized", "calculate_qmin_qmax", "check_min_max_valid", "check_node", "determine_qparams", "get_combined_dict", "get_fqn_to_example_inputs", "get_qconfig_dtypes", "get_qparam_dict", "get_quant_type", "get_swapped_custom_module_class", "getattr_from_fqn", "has_no_children_ignoring_parametrizations", "is_per_channel", "is_per_tensor", "op_is_int8_dynamically_quantized", "to_underlying_dtype", "validate_qmin_qmax", "weight_dtype", "weight_is_quantized", "weight_is_statically_quantized", # torch.backends.cudnn.rnn "get_cudnn_mode", "init_dropout_state", # torch.backends.xeon.run_cpu "create_args", # torch.cuda.amp.autocast_mode "custom_bwd", "custom_fwd", # torch.cuda.amp.common "amp_definitely_not_available", # torch.cuda.graphs "graph_pool_handle", "is_current_stream_capturing", "make_graphed_callables", # torch.mtia.memory "reset_peak_memory_stats", # torch.cuda.nccl "all_gather", "all_reduce", "broadcast", "init_rank", "reduce", "reduce_scatter", "unique_id", "version", # torch.cuda.nvtx "range", "range_end", "range_start", # torch.cuda.profiler "init", "profile", "start", "stop", # torch.cuda.random "get_rng_state", "get_rng_state_all", "initial_seed", "manual_seed", "manual_seed_all", "seed", "seed_all", "set_rng_state", "set_rng_state_all", # torch.distributed.algorithms.ddp_comm_hooks.ddp_zero_hook "hook_with_zero_step", "hook_with_zero_step_interleaved", # torch.distributed.algorithms.ddp_comm_hooks.post_localSGD_hook "post_localSGD_hook", # torch.distributed.algorithms.ddp_comm_hooks.quantization_hooks "quantization_perchannel_hook", "quantization_pertensor_hook", # torch.distributed.algorithms.model_averaging.utils "average_parameters", "average_parameters_or_parameter_groups", "get_params_to_average", # torch.distributed.checkpoint.default_planner "create_default_global_load_plan", "create_default_global_save_plan", "create_default_local_load_plan", "create_default_local_save_plan", # torch.distributed.checkpoint.optimizer "load_sharded_optimizer_state_dict", # torch.distributed.checkpoint.planner_helpers "create_read_items_for_chunk_list", # torch.distributed.checkpoint.state_dict_loader "load_state_dict", # torch.distributed.checkpoint.state_dict_saver "save_state_dict", # torch.distributed.checkpoint.utils "find_state_dict_object", "find_tensor_shard", # torch.distributed.collective_utils "all_gather", "all_gather_object_enforce_type", "broadcast", # torch.distributed.distributed_c10d "all_gather", "all_gather_coalesced", "all_gather_into_tensor", "all_gather_object", "all_reduce", "all_reduce_coalesced", "all_to_all", "all_to_all_single", "barrier", "batch_isend_irecv", "broadcast", "broadcast_object_list", "destroy_process_group", "gather", "gather_object", "get_backend", "get_backend_config", "get_global_rank", "get_group_rank", "get_process_group_ranks", "get_rank", "get_world_size", "init_process_group", "irecv", "is_backend_available", "is_gloo_available", "is_initialized", "is_mpi_available", "is_nccl_available", "is_torchelastic_launched", "is_ucc_available", "isend", "monitored_barrier", "new_group", "new_subgroups", "new_subgroups_by_enumeration", "recv", "reduce", "reduce_scatter", "reduce_scatter_tensor", "scatter", "scatter_object_list", "send", "supports_complex", # torch.distributed.elastic.events.handlers "get_logging_handler", # torch.distributed.elastic.metrics.api "configure", "getStream", "get_elapsed_time_ms", "prof", "profile", "publish_metric", "put_metric", # torch.distributed.elastic.multiprocessing.api "get_std_cm", "to_map", # torch.distributed.elastic.multiprocessing.errors.handlers "get_error_handler", # torch.distributed.elastic.multiprocessing.redirects "get_libc", "redirect", # torch.distributed.elastic.multiprocessing.tail_log "tail_logfile", # torch.distributed.elastic.rendezvous.dynamic_rendezvous "get_method_name", # torch.distributed.elastic.rendezvous.etcd_rendezvous "create_rdzv_handler", # torch.distributed.elastic.rendezvous.etcd_server "find_free_port", "stop_etcd", # torch.distributed.elastic.rendezvous.etcd_store "cas_delay", # torch.distributed.elastic.rendezvous.static_tcp_rendezvous "create_rdzv_handler", # torch.distributed.elastic.rendezvous.utils "parse_rendezvous_endpoint", # torch.distributed.elastic.timer.api "configure", "expires", # torch.distributed.elastic.utils.api "get_env_variable_or_raise", "get_socket_with_port", # torch.distributed.elastic.utils.distributed "create_c10d_store", "get_free_port", "get_socket_with_port", # torch.distributed.elastic.utils.log_level "get_log_level", # torch.distributed.elastic.utils.logging "get_logger", # torch.distributed.elastic.utils.store "barrier", "get_all", "synchronize", "store_timeout", # torch.distributed.fsdp.wrap "always_wrap_policy", "enable_wrap", "lambda_auto_wrap_policy", "size_based_auto_wrap_policy", "transformer_auto_wrap_policy", "wrap", # torch.distributed.nn.functional "all_gather", "all_reduce", "all_to_all", "all_to_all_single", "broadcast", "gather", "reduce", "reduce_scatter", "scatter", # torch.distributed.nn.jit.instantiator "get_arg_return_types_from_interface", "instantiate_non_scriptable_remote_module_template", "instantiate_scriptable_remote_module_template", # torch.distributed.nn.jit.templates.remote_module_template "get_remote_module_template", # torch.distributed.optim.utils "as_functional_optim", "register_functional_optim", # torch.distributed.rendezvous "register_rendezvous_handler", "rendezvous", # torch.distributed.rpc.api "get_worker_info", "method_factory", "new_method", "remote", "rpc_async", "rpc_sync", "shutdown", # torch.distributed.rpc.backend_registry "backend_registered", "construct_rpc_backend_options", "init_backend", "register_backend", # torch.distributed.rpc.internal "deserialize", "serialize", # torch.distributed.tensor.parallel.api "parallelize_module", # torch.distributed.tensor.parallel.input_reshard "input_reshard", # torch.distributed.tensor.parallel.loss "loss_parallel", # torch.distributed.tensor.parallel.style "make_sharded_output_tensor", # torch.distributions.utils "broadcast_all", "clamp_probs", "logits_to_probs", "probs_to_logits", "tril_matrix_to_vec", "vec_to_tril_matrix", # torch.fx.annotate "annotate", # torch.fx.experimental.accelerator_partitioner "check_dependency", "combine_two_partitions", "get_bfs_level_partition", "get_device_partition_stats", "get_device_to_partitions_mapping", "get_logical_id_to_device", "get_node_to_partition_mapping", "reorganize_partitions", "reset_partition_device", "set_parents_and_children", # torch.fx.experimental.const_fold "get_unique_attr_name_in_module", "split_const_subgraphs", # torch.fx.experimental.debug "set_trace", # torch.fx.experimental.graph_gradual_typechecker "adaptiveavgpool2d_check", "adaptiveavgpool2d_inference_rule", "add_inference_rule", "all_eq", "bn2d_inference_rule", "broadcast_types", "calculate_out_dimension", "conv2d_inference_rule", "conv_refinement_rule", "conv_rule", "element_wise_eq", "expand_to_tensor_dim", "first_two_eq", "flatten_check", "flatten_inference_rule", "flatten_refinement_rule", "get_attr_inference_rule", "get_greatest_upper_bound", "get_parameter", "linear_check", "linear_inference_rule", "linear_refinement_rule", "maxpool2d_check", "maxpool2d_inference_rule", "register_algebraic_expressions_inference_rule", "register_inference_rule", "register_refinement_rule", "relu_inference_rule", "reshape_inference_rule", "transpose_inference_rule", # torch.fx.experimental.merge_matmul "are_nodes_independent", "may_depend_on", "merge_matmul", "split_result_tensors", # torch.fx.experimental.meta_tracer "embedding_override", "functional_relu_override", "gen_constructor_wrapper", "nn_layernorm_override", "proxys_to_metas", "symbolic_trace", "torch_abs_override", "torch_nn_relu_override", "torch_relu_override", "torch_where_override", # torch.fx.experimental.migrate_gradual_types.constraint "is_algebraic_expression", "is_bool_expr", "is_dim", # torch.fx.experimental.migrate_gradual_types.constraint_generator "adaptive_inference_rule", "add_layer_norm_constraints", "add_linear_constraints", "arange_inference_rule", "assert_inference_rule", "batchnorm_inference_rule", "bmm_inference_rule", "broadcasting_inference_rule", "conv2d_inference_rule", "cumsum_inference_rule", "embedding_inference_rule", "embedding_inference_rule_functional", "eq_inference_rule", "equality_inference_rule", "expand_inference_rule", "flatten_inference_rule", "full_inference_rule", "gen_broadcasting_constraints", "gen_embedding_rules", "gen_layer_norm_constraints", "generate_flatten_constraints", "get_attr_inference_rule", "getitem_inference_rule", "gt_inference_rule", "index_select_inference_rule", "layer_norm_functional", "layer_norm_inference_rule", "linear_constraints", "linear_inference_rule", "lt_inference_rule", "masked_fill_inference_rule", "maxpool_inference_rule", "neq_inference_rule", "range_check", "register_inference_rule", "relu_inference_rule", "reshape_inference_rule", "size_inference_rule", "tensor_inference_rule", "torch_dim_inference_rule", "torch_linear_inference_rule", "transpose_inference_rule", "type_inference_rule", "view_inference_rule", # torch.fx.experimental.migrate_gradual_types.constraint_transformation "apply_padding", "broadcast_dim", "calc_last_two_dims", "create_equality_constraints_for_broadcasting", "gen_all_reshape_possibilities", "gen_broadcasting_constraints", "gen_consistency_constraints", "gen_greatest_upper_bound", "gen_lists_of_dims", "generate_all_broadcasting_possibilities_no_padding", "generate_all_int_dyn_dim_possibilities", "generate_binconstraint_d", "generate_binconstraint_t", "generate_broadcasting", "generate_calc_conv", "generate_calc_maxpool", "generate_calc_product", "generate_conj", "generate_d_gub", "generate_disj", "generate_gub", "generate_reshape", "is_dim_div_by_target", "is_target_div_by_dim", "no_broadcast_dim_with_index", "register_transformation_rule", "transform_constraint", "transform_get_item", "transform_get_item_tensor", "transform_index_select", "transform_transpose", "valid_index", "valid_index_tensor", # torch.fx.experimental.migrate_gradual_types.transform_to_z3 "evaluate_conditional_with_constraints", # torch.fx.experimental.migrate_gradual_types.util "gen_bvar", "gen_dvar", "gen_nat_constraints", "gen_tensor_dims", "gen_tvar", # torch.fx.experimental.optimization "extract_subgraph", "fuse", "gen_mkl_autotuner", "matches_module_pattern", "modules_to_mkldnn", "optimize_for_inference", "remove_dropout", "replace_node_module", "reset_modules", "use_mkl_length", # torch.fx.experimental.partitioner_utils "get_comm_latency_between", "get_extra_size_of", "get_latency_of_one_partition", "get_latency_of_partitioned_graph", "get_partition_to_latency_mapping", # torch.fx.experimental.proxy_tensor "decompose", "disable_autocast_cache", "disable_proxy_modes_tracing", "dispatch_trace", "extract_val", "fake_signature", "fetch_sym_proxy", "fetch_object_proxy", "get_innermost_proxy_mode", "get_isolated_graphmodule", "get_proxy_slot", "get_torch_dispatch_modes", "has_proxy_slot", "is_sym_node", "maybe_handle_decomp", "proxy_call", "set_meta", "set_original_aten_op", "set_proxy_slot", "snapshot_fake", "thunkify", "track_tensor", "track_tensor_tree", "wrap_key", "wrapper_and_args_for_make_fx", # torch.fx.experimental.recording "record_shapeenv_event", "replay_shape_env_events", "shape_env_check_state_equal", # torch.fx.experimental.sym_node "ceil_impl", "floor_ceil_helper", "floor_impl", "method_to_operator", "sympy_is_channels_last_contiguous_2d", "sympy_is_channels_last_contiguous_3d", "sympy_is_channels_last_strides_2d", "sympy_is_channels_last_strides_3d", "sympy_is_channels_last_strides_generic", "sympy_is_contiguous", "sympy_is_contiguous_generic", "to_node", "wrap_node", "sym_sqrt", # torch.fx.experimental.symbolic_shapes "bind_symbols", "cast_symbool_to_symint_guardless", "create_contiguous", "error", "eval_guards", "eval_is_non_overlapping_and_dense", "expect_true", "find_symbol_binding_fx_nodes", "free_symbols", "free_unbacked_symbols", "fx_placeholder_targets", "fx_placeholder_vals", "guard_bool", "guard_float", "guard_int", "guard_scalar", "has_hint", "has_symbolic_sizes_strides", "is_channels_last_contiguous_2d", "is_channels_last_contiguous_3d", "is_channels_last_strides_2d", "is_channels_last_strides_3d", "is_contiguous", "is_non_overlapping_and_dense_indicator", "is_nested_int", "is_symbol_binding_fx_node", "is_symbolic", # torch.fx.experimental.unification.core "reify", # torch.fx.experimental.unification.match "edge", "match", "ordering", "supercedes", # torch.fx.experimental.unification.more "reify_object", "unifiable", "unify_object", # torch.fx.experimental.unification.multipledispatch.conflict "ambiguities", "ambiguous", "consistent", "edge", "ordering", "super_signature", "supercedes", # torch.fx.experimental.unification.multipledispatch.core "dispatch", "ismethod", # torch.fx.experimental.unification.multipledispatch.dispatcher "ambiguity_warn", "halt_ordering", "restart_ordering", "source", "str_signature", "variadic_signature_matches", "variadic_signature_matches_iter", "warning_text", # torch.fx.experimental.unification.multipledispatch.utils "expand_tuples", "groupby", "raises", "reverse_dict", # torch.fx.experimental.unification.multipledispatch.variadic "isvariadic", # torch.fx.experimental.unification.unification_tools "assoc", "assoc_in", "dissoc", "first", "get_in", "getter", "groupby", "itemfilter", "itemmap", "keyfilter", "keymap", "merge", "merge_with", "update_in", "valfilter", "valmap", # torch.fx.experimental.unification.utils "freeze", "hashable", "raises", "reverse_dict", "transitive_get", "xfail", # torch.fx.experimental.unification.variable "var", "vars", # torch.fx.experimental.unify_refinements "check_for_type_equality", "convert_eq", "infer_symbolic_types", "infer_symbolic_types_single_pass", "substitute_all_types", "substitute_solution_one_type", "unify_eq", # torch.fx.experimental.validator "bisect", "translation_validation_enabled", "translation_validation_timeout", "z3op", "z3str", # torch.fx.graph_module "reduce_graph_module", "reduce_package_graph_module", # torch.fx.node "has_side_effect", "map_aggregate", "map_arg", # torch.fx.operator_schemas "check_for_mutable_operation", "create_type_hint", "get_signature_for_torch_op", "normalize_function", "normalize_module", "type_matches", # torch.fx.passes.annotate_getitem_nodes "annotate_getitem_nodes", # torch.fx.passes.backends.cudagraphs "partition_cudagraphs", # torch.fx.passes.dialect.common.cse_pass "get_CSE_banned_ops", # torch.fx.passes.graph_manipulation "get_size_of_all_nodes", "get_size_of_node", "get_tensor_meta", "replace_target_nodes_with", # torch.fx.passes.infra.pass_manager "pass_result_wrapper", "this_before_that_pass_constraint", # torch.fx.passes.operator_support "any_chain", "chain", "create_op_support", # torch.fx.passes.param_fetch "default_matching", "extract_attrs_for_lowering", "lift_lowering_attrs_to_nodes", # torch.fx.passes.pass_manager "inplace_wrapper", "log_hook", "loop_pass", "these_before_those_pass_constraint", "this_before_that_pass_constraint", # torch.fx.passes.reinplace "reinplace", # torch.fx.passes.split_module "split_module", # torch.fx.passes.split_utils "getattr_recursive", "setattr_recursive", "split_by_tags", # torch.fx.passes.splitter_base "generate_inputs_for_submodules", # torch.fx.passes.tools_common "get_acc_ops_name", "get_node_target", "is_node_output_tensor", "legalize_graph", # torch.fx.passes.utils.common "compare_graphs", "lift_subgraph_as_module", # torch.fx.passes.utils.fuser_utils "erase_nodes", "fuse_as_graphmodule", "fuse_by_partitions", "insert_subgm", "topo_sort", "validate_partition", # torch.fx.passes.utils.source_matcher_utils "check_subgraphs_connected", "get_source_partitions", # torch.fx.proxy "assert_fn", # torch.fx.subgraph_rewriter "replace_pattern", "replace_pattern_with_filters", # torch.fx.tensor_type "is_consistent", "is_more_precise", # torch.fx.traceback "format_stack", "get_current_meta", "has_preserved_node_meta", "preserve_node_meta", "reset_grad_fn_seq_nr", "set_current_meta", "set_grad_fn_seq_nr", "set_stack_trace", # torch.jit.annotations "ann_to_type", "check_fn", "get_enum_value_type", "get_param_names", "get_signature", "get_type_line", "is_function_or_method", "is_tensor", "is_vararg", "parse_type_line", "split_type_line", "try_ann_to_type", "try_real_annotations", # torch.jit.frontend "build_class_def", "build_def", "build_ignore_context_manager", "build_param", "build_param_list", "build_stmts", "build_withitems", "find_before", "get_class_assigns", "get_class_properties", "get_default_args", "get_default_args_for_class", "get_jit_class_def", "get_jit_def", "is_reserved_name", "is_torch_jit_ignore_context_manager", # torch.jit.generate_bytecode "format_bytecode", "generate_upgraders_bytecode", # torch.jit.quantized "apply_permutation", "quantize_linear_modules", "quantize_rnn_cell_modules", "quantize_rnn_modules", # torch.library "define", "get_ctx", "impl", "impl_abstract", # torch.masked.maskedtensor.core "is_masked_tensor", # torch.masked.maskedtensor.creation "as_masked_tensor", "masked_tensor", # torch.multiprocessing.pool "clean_worker", # torch.multiprocessing.reductions "fd_id", "init_reductions", "rebuild_cuda_tensor", "rebuild_meta_tensor", "rebuild_event", "rebuild_nested_tensor", "rebuild_sparse_coo_tensor", "rebuild_sparse_compressed_tensor", "rebuild_storage_empty", "rebuild_storage_fd", "rebuild_storage_filename", "rebuild_tensor", "rebuild_typed_storage", "rebuild_typed_storage_child", "reduce_event", "reduce_storage", "reduce_tensor", "reduce_typed_storage", "reduce_typed_storage_child", "storage_from_cache", # torch.multiprocessing.spawn "start_processes", # torch.nn.functional "adaptive_max_pool1d_with_indices", # documented as adaptive_max_pool1d "adaptive_max_pool2d_with_indices", # documented as adaptive_max_pool2d "adaptive_max_pool3d_with_indices", # documented as adaptive_max_pool3d "assert_int_or_pair", # looks unintentionally public "fractional_max_pool2d_with_indices", # documented as fractional_max_pool2d "fractional_max_pool3d_with_indices", # documented as fractional_max_pool3d "max_pool1d_with_indices", # documented as max_pool1d "max_pool2d_with_indices", # documented as max_pool2d "max_pool3d_with_indices", # documented as max_pool3d "multi_head_attention_forward", # torch.nn.grad "conv1d_input", # legacy helper for gradient computation "conv1d_weight", # legacy helper for gradient computation "conv2d_input", # legacy helper for gradient computation "conv2d_weight", # legacy helper for gradient computation "conv3d_input", # legacy helper for gradient computation "conv3d_weight", # legacy helper for gradient computation # torch.nn.init "constant", # deprecated "dirac", # deprecated "eye", # deprecated "kaiming_normal", # deprecated "kaiming_uniform", # deprecated "normal", # deprecated "orthogonal", # deprecated "sparse", # deprecated "uniform", # deprecated "xavier_normal", # deprecated "xavier_uniform", # deprecated # torch.nn.modules.rnn "apply_permutation", # deprecated # torch.nn.modules.utils "consume_prefix_in_state_dict_if_present", # torch.nn.parallel.comm "broadcast", "broadcast_coalesced", "gather", "reduce_add", "reduce_add_coalesced", "scatter", # torch.nn.parallel.data_parallel "data_parallel", # torch.nn.parallel.parallel_apply "get_a_var", "parallel_apply", # torch.nn.parallel.replicate "replicate", # torch.nn.parallel.scatter_gather "gather", "is_namedtuple", "scatter", "scatter_kwargs", # torch.nn.utils.rnn "bind", # looks unintentionally public # torch.onnx.operators "reshape_from_tensor_shape", "shape_as_tensor", # torch.onnx.symbolic_caffe2 "add", "avg_pool2d", "cat", "conv2d", "conv2d_relu", "conv_prepack", "dequantize", "linear", "linear_prepack", "max_pool2d", "nchw2nhwc", "nhwc2nchw", "quantize_per_tensor", "register_quantized_ops", "relu", "reshape", "sigmoid", "slice", "upsample_nearest2d", # torch.onnx.symbolic_helper "args_have_same_dtype", "check_training_mode", "dequantize_helper", "is_complex_value", "quantize_helper", "quantized_args", "requantize_bias_helper", # torch.onnx.symbolic_opset10 "dequantize", "div", "embedding_bag", "fake_quantize_per_tensor_affine", "flip", "fmod", "isfinite", "isinf", "nan_to_num", "quantize_per_tensor", "quantized_add", "quantized_add_relu", "quantized_cat", "quantized_conv1d", "quantized_conv1d_relu", "quantized_conv2d", "quantized_conv2d_relu", "quantized_conv3d", "quantized_conv3d_relu", "quantized_conv_transpose1d", "quantized_conv_transpose2d", "quantized_conv_transpose3d", "quantized_group_norm", "quantized_hardswish", "quantized_instance_norm", "quantized_layer_norm", "quantized_leaky_relu", "quantized_linear", "quantized_linear_relu", "quantized_mul", "quantized_sigmoid", "slice", "sort", "topk", # torch.onnx.symbolic_opset11 "Delete", "add", "append", "arange", "argsort", "atleast_1d", "atleast_2d", "atleast_3d", "cat", "chunk", "clamp", "clamp_max", "clamp_min", "constant_pad_nd", "cumsum", "embedding_bag", "embedding_renorm", "flatten", "gather", "hardtanh", "hstack", "im2col", "index", "index_copy", "index_fill", "index_put", "insert", "linalg_det", "linalg_vector_norm", "logdet", "masked_scatter", "masked_select", "mm", "narrow", "normal", "pad", "pixel_shuffle", "pop", "prim_constant_chunk", "reflection_pad", "relu6", "remainder", "replication_pad", "round", "scatter", "select", "size", "sort", "split", "split_with_sizes", "squeeze", "stack", "topk", "unbind", "unique_dim", "unsqueeze", "vstack", # torch.onnx.symbolic_opset12 "argmax", "argmin", "binary_cross_entropy_with_logits", "celu", "cross_entropy_loss", "dropout", "einsum", "ge", "le", "native_dropout", "nll_loss", "nll_loss2d", "nll_loss_nd", "outer", "pow", "tensordot", "unfold", # torch.onnx.symbolic_opset13 "diagonal", "fake_quantize_per_channel_affine", "fake_quantize_per_tensor_affine", "frobenius_norm", "log_softmax", "nonzero_numpy", "quantized_conv1d", "quantized_conv1d_relu", "quantized_conv2d", "quantized_conv2d_relu", "quantized_conv3d", "quantized_conv3d_relu", "quantized_conv_transpose1d", "quantized_conv_transpose2d", "quantized_conv_transpose3d", "quantized_linear", "quantized_linear_relu", "repeat_interleave", "softmax", "split", "split_with_sizes", "tensor_split", "tile", "unbind", "unflatten", "unsafe_chunk", "unsafe_split", "unsafe_split_with_sizes", "where", # torch.onnx.symbolic_opset14 "batch_norm", "hardswish", "quantized_hardswish", "reshape", "scaled_dot_product_attention", "tril", "triu", # torch.onnx.symbolic_opset15 "aten__is_", "aten__isnot_", "bernoulli", "prim_unchecked_cast", # torch.onnx.symbolic_opset16 "grid_sampler", "scatter_add", "scatter_reduce", # torch.onnx.symbolic_opset17 "layer_norm", "stft", # torch.onnx.symbolic_opset18 "col2im", # torch.onnx.symbolic_opset7 "max", "min", # torch.onnx.symbolic_opset8 "addmm", "bmm", "empty", "empty_like", "flatten", "full", "full_like", "gt", "lt", "matmul", "mm", "ones", "ones_like", "prelu", "repeat", "zeros", "zeros_like", # torch.onnx.symbolic_opset9 "abs", "acos", "adaptive_avg_pool1d", "adaptive_avg_pool2d", "adaptive_avg_pool3d", "adaptive_max_pool1d", "adaptive_max_pool2d", "adaptive_max_pool3d", "add", "addcmul", "addmm", "alias", "amax", "amin", "aminmax", "arange", "argmax", "argmin", "as_strided", "as_tensor", "asin", "atan", "atan2", "avg_pool1d", "avg_pool2d", "avg_pool3d", "baddbmm", "batch_norm", "bernoulli", "bitwise_not", "bitwise_or", "bmm", "broadcast_tensors", "broadcast_to", "bucketize", "cat", "cdist", "ceil", "clamp", "clamp_max", "clamp_min", "clone", "constant_pad_nd", "contiguous", "conv1d", "conv2d", "conv3d", "conv_tbc", "conv_transpose1d", "conv_transpose2d", "conv_transpose3d", "convert_element_type", "convolution", "cos", "cosine_similarity", "cross", "cumsum", "detach", "dim", "div", "dot", "dropout", "elu", "embedding", "embedding_bag", "empty", "empty_like", "eq", "erf", "exp", "expand", "expand_as", "eye", "fill", "flatten", "floor", "floor_divide", "floordiv", "frobenius_norm", "full", "full_like", "gather", "ge", "gelu", "get_pool_ceil_padding", "glu", "group_norm", "gru", "gt", "hann_window", "hardshrink", "hardsigmoid", "hardswish", "hardtanh", "index", "index_add", "index_copy", "index_fill", "index_put", "index_select", "instance_norm", "is_floating_point", "is_pinned", "isnan", "item", "kl_div", "layer_norm", "le", "leaky_relu", "lerp", "lift", "linalg_cross", "linalg_matrix_norm", "linalg_norm", "linalg_vector_norm", "linear", "linspace", "log", "log10", "log1p", "log2", "log_sigmoid", "log_softmax", "logical_and", "logical_not", "logical_or", "logical_xor", "logit", "logsumexp", "lstm", "lstm_cell", "lt", "masked_fill", "masked_fill_", "matmul", "max", "max_pool1d", "max_pool1d_with_indices", "max_pool2d", "max_pool2d_with_indices", "max_pool3d", "max_pool3d_with_indices", "maximum", "meshgrid", "min", "minimum", "mish", "mm", "movedim", "mse_loss", "mul", "multinomial", "mv", "narrow", "native_layer_norm", "ne", "neg", "new_empty", "new_full", "new_ones", "new_zeros", "nonzero", "nonzero_numpy", "noop_complex_operators", "norm", "numel", "numpy_T", "one_hot", "ones", "ones_like", "onnx_placeholder", "overload_by_arg_count", "pad", "pairwise_distance", "permute", "pixel_shuffle", "pixel_unshuffle", "pow", "prelu", "prim_constant", "prim_constant_chunk", "prim_constant_split", "prim_data", "prim_device", "prim_dtype", "prim_if", "prim_layout", "prim_list_construct", "prim_list_unpack", "prim_loop", "prim_max", "prim_min", "prim_shape", "prim_tolist", "prim_tuple_construct", "prim_type", "prim_unchecked_cast", "prim_uninitialized", "rand", "rand_like", "randint", "randint_like", "randn", "randn_like", "reciprocal", "reflection_pad", "relu", "relu6", "remainder", "repeat", "repeat_interleave", "replication_pad", "reshape", "reshape_as", "rnn_relu", "rnn_tanh", "roll", "rrelu", "rsqrt", "rsub", "scalar_tensor", "scatter", "scatter_add", "select", "selu", "sigmoid", "sign", "silu", "sin", "size", "slice", "softmax", "softplus", "softshrink", "sort", "split", "split_with_sizes", "sqrt", "square", "squeeze", "stack", "std", "std_mean", "sub", "t", "take", "tan", "tanh", "tanhshrink", "tensor", "threshold", "to", "topk", "transpose", "true_divide", "type_as", "unbind", "unfold", "unsafe_chunk", "unsafe_split", "unsafe_split_with_sizes", "unsqueeze", "unsupported_complex_operators", "unused", "upsample_bilinear2d", "upsample_linear1d", "upsample_nearest1d", "upsample_nearest2d", "upsample_nearest3d", "upsample_trilinear3d", "var", "var_mean", "view", "view_as", "where", "wrap_logical_op_with_cast_to", "wrap_logical_op_with_negation", "zero", "zeros", "zeros_like", # torch.onnx.utils "disable_apex_o2_state_dict_hook", "export", "export_to_pretty_string", "exporter_context", "is_in_onnx_export", "model_signature", "register_custom_op_symbolic", "select_model_mode_for_export", "setup_onnx_logging", "unconvertible_ops", "unpack_quantized_tensor", "warn_on_static_input_change", # torch.onnx.verification "check_export_model_diff", "verify", "verify_aten_graph", # torch.optim.optimizer "register_optimizer_step_post_hook", "register_optimizer_step_pre_hook", # torch.optim.swa_utils "get_ema_avg_fn", "get_ema_multi_avg_fn", "get_swa_avg_fn", "get_swa_multi_avg_fn", "update_bn", # torch.overrides "enable_reentrant_dispatch", # torch.package.analyze.find_first_use_of_broken_modules "find_first_use_of_broken_modules", # torch.package.analyze.is_from_package "is_from_package", # torch.package.analyze.trace_dependencies "trace_dependencies", # torch.profiler.itt "range", # torch.profiler.profiler "schedule", "supported_activities", "tensorboard_trace_handler", # torch.return_types "pytree_register_structseq", # torch.serialization "check_module_version_greater_or_equal", "default_restore_location", "load", "location_tag", "mkdtemp", "normalize_storage_type", "save", "storage_to_tensor_type", "validate_cuda_device", "validate_hpu_device", # torch.signal.windows.windows "bartlett", "blackman", "cosine", "exponential", "gaussian", "general_cosine", "general_hamming", "hamming", "hann", "kaiser", "nuttall", # torch.sparse.semi_structured "to_sparse_semi_structured", # torch.utils.backend_registration "generate_methods_for_privateuse1_backend", "rename_privateuse1_backend", # torch.utils.benchmark.examples.op_benchmark "assert_dicts_equal", # torch.utils.benchmark.op_fuzzers.spectral "power_range", # torch.utils.benchmark.utils.common "ordered_unique", "select_unit", "set_torch_threads", "trim_sigfig", "unit_to_english", # torch.utils.benchmark.utils.compare "optional_min", # torch.utils.benchmark.utils.compile "bench_all", "bench_loop", "benchmark_compile", # torch.utils.benchmark.utils.cpp_jit "compile_callgrind_template", "compile_timeit_template", "get_compat_bindings", # torch.utils.benchmark.utils.fuzzer "dtype_size", "prod", # torch.utils.benchmark.utils.timer "timer", # torch.utils.benchmark.utils.valgrind_wrapper.timer_interface "wrapper_singleton", # torch.utils.bundled_inputs "augment_many_model_functions_with_bundled_inputs", "augment_model_with_bundled_inputs", "bundle_inputs", "bundle_large_tensor", "bundle_randn", # torch.utils.checkpoint "check_backward_validity", "detach_variable", "get_device_states", "noop_context_fn", "set_checkpoint_early_stop", "set_device_states", # torch.utils.collect_env "check_release_file", "get_cachingallocator_config", "get_clang_version", "get_cmake_version", "get_conda_packages", "get_cpu_info", "get_cuda_module_loading_config", "get_cudnn_version", "get_env_info", "get_gcc_version", "get_gpu_info", "get_libc_version", "get_lsb_version", "get_mac_version", "get_nvidia_driver_version", "get_nvidia_smi", "get_os", "get_pip_packages", "get_platform", "get_pretty_env_info", "get_python_platform", "get_running_cuda_version", "get_windows_version", "is_xnnpack_available", "pretty_str", # torch.utils.cpp_backtrace "get_cpp_backtrace", # torch.utils.cpp_extension "check_compiler_is_gcc", "check_compiler_ok_for_platform", "get_cxx_compiler", "get_default_build_root", "library_paths", "remove_extension_h_precompiler_headers", # torch.utils.data.backward_compatibility "worker_init_fn", # torch.utils.data.datapipes.dataframe.dataframe_wrapper "concat", "create_dataframe", "get_columns", "get_df_wrapper", "get_item", "get_len", "is_column", "is_dataframe", "iterate", "set_df_wrapper", # torch.utils.data.datapipes.dataframe.dataframes "disable_capture", "get_val", # torch.utils.data.datapipes.gen_pyi "extract_class_name", "extract_method_name", "find_file_paths", "gen_from_template", "get_method_definitions", "materialize_lines", "parse_datapipe_file", "parse_datapipe_files", "process_signature", "split_outside_bracket", # torch.utils.data.datapipes.map.callable "default_fn", # torch.utils.data.datapipes.utils.common "get_file_binaries_from_pathnames", "get_file_pathnames_from_root", "match_masks", "validate_input_col", "validate_pathname_binary_tuple", # torch.utils.data.datapipes.utils.decoder "audiohandler", "basichandlers", "extension_extract_fn", "handle_extension", "imagehandler", "mathandler", "videohandler", # torch.utils.data.dataset "random_split", # torch.utils.data.graph "traverse", "traverse_dps", # torch.utils.data.graph_settings "apply_random_seed", "apply_sharding", "apply_shuffle_seed", "apply_shuffle_settings", "get_all_graph_pipes", # torch.utils.flop_counter "addmm_flop", "baddbmm_flop", "bmm_flop", "conv_backward_flop", "conv_flop", "conv_flop_count", "convert_num_with_suffix", "get_shape", "get_suffix_str", "mm_flop", "normalize_tuple", "register_flop_formula", "sdpa_backward_flop", "sdpa_backward_flop_count", "sdpa_flop", "sdpa_flop_count", "shape_wrapper", "transpose_shape", # torch.utils.hipify.hipify_python "add_dim3", "compute_stats", "extract_arguments", "file_add_header", "file_specific_replacement", "find_bracket_group", "find_closure_group", "find_parentheses_group", "fix_static_global_kernels", "get_hip_file_path", "hip_header_magic", "hipify", "is_caffe2_gpu_file", "is_cusparse_file", "is_out_of_place", "is_pytorch_file", "is_special_file", "match_extensions", "matched_files_iter", "openf", "preprocess_file_and_save_result", "preprocessor", "processKernelLaunches", "replace_extern_shared", "replace_math_functions", "str2bool", # torch.utils.hooks "unserializable_hook", "warn_if_has_hooks", # torch.utils.jit.log_extract "extract_ir", "load_graph_and_inputs", "make_tensor_from_type", "no_fuser", "time_cpu", "time_cuda", # torch.utils.mkldnn "to_mkldnn", # torch.utils.mobile_optimizer "generate_mobile_module_lints", # torch.utils.tensorboard.summary "audio", "compute_curve", "custom_scalars", "draw_boxes", "half_to_int", "histogram", "histogram_raw", "hparams", "image", "image_boxes", "int_to_half", "make_histogram", "make_image", "make_video", "mesh", "pr_curve", "pr_curve_raw", "scalar", "tensor_proto", "text", "video", # torch.utils.throughput_benchmark "format_time", ] coverage_ignore_classes = [ # torch "FatalError", "QUInt2x4Storage", "Size", "Storage", "Stream", "Tensor", "finfo", "iinfo", "qscheme", "AggregationType", "AliasDb", "AnyType", "Argument", "ArgumentSpec", "AwaitType", "BenchmarkConfig", "BenchmarkExecutionStats", "Block", "BoolType", "BufferDict", "CallStack", "Capsule", "ClassType", "Code", "CompleteArgumentSpec", "ComplexType", "ConcreteModuleType", "ConcreteModuleTypeBuilder", "DeepCopyMemoTable", "DeserializationStorageContext", "DeviceObjType", "DictType", "DispatchKey", "DispatchKeySet", "EnumType", "ExcludeDispatchKeyGuard", "ExecutionPlan", "FileCheck", "FloatType", "FunctionSchema", "Gradient", "Graph", "GraphExecutorState", "IODescriptor", "InferredType", "IntType", "InterfaceType", "ListType", "LockingLogger", "MobileOptimizerType", "ModuleDict", "Node", "NoneType", "NoopLogger", "NumberType", "OperatorInfo", "OptionalType", "ParameterDict", "PyObjectType", "PyTorchFileReader", "PyTorchFileWriter", "RRefType", "ScriptClass", "ScriptClassFunction", "ScriptDict", "ScriptDictIterator", "ScriptDictKeyIterator", "ScriptList", "ScriptListIterator", "ScriptMethod", "ScriptModule", "ScriptModuleSerializer", "ScriptObject", "ScriptObjectProperty", "SerializationStorageContext", "StaticModule", "StringType", "SymIntType", "SymBoolType", "ThroughputBenchmark", "TracingState", "TupleType", "Type", "UnionType", "Use", "Value", # torch.cuda "BFloat16Storage", "BFloat16Tensor", "BoolStorage", "BoolTensor", "ByteStorage", "ByteTensor", "CharStorage", "CharTensor", "ComplexDoubleStorage", "ComplexFloatStorage", "CudaError", "DeferredCudaCallError", "DoubleStorage", "DoubleTensor", "FloatStorage", "FloatTensor", "HalfStorage", "HalfTensor", "IntStorage", "IntTensor", "LongStorage", "LongTensor", "ShortStorage", "ShortTensor", "cudaStatus", # torch.cuda._sanitizer "Access", "AccessType", "Await", "CUDASanitizer", "CUDASanitizerDispatchMode", "CUDASanitizerErrors", "EventHandler", "SynchronizationError", "UnsynchronizedAccessError", # torch.cuda.memory "MemPool", # torch.distributed.elastic.multiprocessing.errors "ChildFailedError", "ProcessFailure", # torch.distributions.constraints "cat", "greater_than", "greater_than_eq", "half_open_interval", "independent", "integer_interval", "interval", "less_than", "multinomial", "stack", # torch.distributions.transforms "AffineTransform", "CatTransform", "ComposeTransform", "CorrCholeskyTransform", "CumulativeDistributionTransform", "ExpTransform", "IndependentTransform", "PowerTransform", "ReshapeTransform", "SigmoidTransform", "SoftmaxTransform", "SoftplusTransform", "StackTransform", "StickBreakingTransform", "TanhTransform", "Transform", # torch.jit "CompilationUnit", "Error", "Future", "ScriptFunction", # torch.onnx "CheckerError", "ExportTypes", # torch.backends "ContextProp", "PropModule", # torch.backends.cuda "cuBLASModule", "cuFFTPlanCache", "cuFFTPlanCacheAttrContextProp", "cuFFTPlanCacheManager", # torch.distributed.algorithms.ddp_comm_hooks "DDPCommHookType", # torch.jit.mobile "LiteScriptModule", # torch.ao.nn.quantized.modules "DeQuantize", "Quantize", # torch.utils.backcompat "Warning", # torch.ao.nn.intrinsic.modules.fused "ConvAdd2d", "ConvAddReLU2d", "LinearBn1d", "LinearLeakyReLU", "LinearTanh", # torch.ao.nn.intrinsic.qat.modules.conv_fused "ConvBnReLU1d", "ConvBnReLU2d", "ConvBnReLU3d", "ConvReLU1d", "ConvReLU2d", "ConvReLU3d", # torch.ao.nn.intrinsic.qat.modules.linear_fused "LinearBn1d", # torch.ao.nn.intrinsic.qat.modules.linear_relu "LinearReLU", # torch.ao.nn.intrinsic.quantized.dynamic.modules.linear_relu "LinearReLU", # torch.ao.nn.intrinsic.quantized.modules.bn_relu "BNReLU2d", "BNReLU3d", # torch.ao.nn.intrinsic.quantized.modules.conv_add "ConvAdd2d", "ConvAddReLU2d", # torch.ao.nn.intrinsic.quantized.modules.conv_relu "ConvReLU1d", "ConvReLU2d", "ConvReLU3d", # torch.ao.nn.intrinsic.quantized.modules.linear_relu "LinearLeakyReLU", "LinearReLU", "LinearTanh", # torch.ao.nn.qat.modules.conv "Conv1d", "Conv2d", "Conv3d", # torch.ao.nn.qat.modules.embedding_ops "Embedding", "EmbeddingBag", # torch.ao.nn.qat.modules.linear "Linear", # torch.ao.nn.quantizable.modules.activation "MultiheadAttention", # torch.ao.nn.quantizable.modules.rnn "LSTM", "LSTMCell", # torch.ao.nn.quantized.dynamic.modules.conv "Conv1d", "Conv2d", "Conv3d", "ConvTranspose1d", "ConvTranspose2d", "ConvTranspose3d", # torch.ao.nn.quantized.dynamic.modules.linear "Linear", # torch.ao.nn.quantized.dynamic.modules.rnn "GRU", "GRUCell", "LSTM", "LSTMCell", "PackedParameter", "RNNBase", "RNNCell", "RNNCellBase", # torch.ao.nn.quantized.modules.activation "ELU", "Hardswish", "LeakyReLU", "MultiheadAttention", "PReLU", "ReLU6", "Sigmoid", "Softmax", # torch.ao.nn.quantized.modules.batchnorm "BatchNorm2d", "BatchNorm3d", # torch.ao.nn.quantized.modules.conv "Conv1d", "Conv2d", "Conv3d", "ConvTranspose1d", "ConvTranspose2d", "ConvTranspose3d", # torch.ao.nn.quantized.modules.dropout "Dropout", # torch.ao.nn.quantized.modules.embedding_ops "Embedding", "EmbeddingBag", "EmbeddingPackedParams", # torch.ao.nn.quantized.modules.functional_modules "FXFloatFunctional", "FloatFunctional", "QFunctional", # torch.ao.nn.quantized.modules.linear "Linear", "LinearPackedParams", # torch.ao.nn.quantized.modules.normalization "GroupNorm", "InstanceNorm1d", "InstanceNorm2d", "InstanceNorm3d", "LayerNorm", # torch.ao.nn.quantized.modules.rnn "LSTM", # torch.ao.nn.quantized.modules.utils "WeightedQuantizedModule", # torch.ao.nn.quantized.reference.modules.conv "Conv1d", "Conv2d", "Conv3d", "ConvTranspose1d", "ConvTranspose2d", "ConvTranspose3d", # torch.ao.nn.quantized.reference.modules.linear "Linear", # torch.ao.nn.quantized.reference.modules.rnn "GRU", "GRUCell", "LSTM", "LSTMCell", "RNNBase", "RNNCell", "RNNCellBase", # torch.ao.nn.quantized.reference.modules.sparse "Embedding", "EmbeddingBag", # torch.ao.nn.quantized.reference.modules.utils "ReferenceQuantizedModule", # torch.ao.nn.sparse.quantized.dynamic.linear "Linear", # torch.ao.nn.sparse.quantized.linear "Linear", "LinearPackedParams", # torch.ao.nn.sparse.quantized.utils "LinearBlockSparsePattern", # torch.ao.ns.fx.graph_matcher "SubgraphTypeRelationship", # torch.ao.ns.fx.n_shadows_utils "OutputProp", # torch.ao.ns.fx.ns_types "NSSingleResultValuesType", "NSSubgraph", # torch.ao.ns.fx.qconfig_multi_mapping "QConfigMultiMapping", # torch.ao.pruning.scheduler.base_scheduler "BaseScheduler", # torch.ao.pruning.scheduler.cubic_scheduler "CubicSL", # torch.ao.pruning.scheduler.lambda_scheduler "LambdaSL", # torch.ao.pruning.sparsifier.base_sparsifier "BaseSparsifier", # torch.ao.pruning.sparsifier.nearly_diagonal_sparsifier "NearlyDiagonalSparsifier", # torch.ao.pruning.sparsifier.utils "FakeSparsity", # torch.ao.pruning.sparsifier.weight_norm_sparsifier "WeightNormSparsifier", # torch.ao.quantization.backend_config.backend_config "BackendConfig", "BackendPatternConfig", "DTypeConfig", # torch.ao.quantization.fake_quantize "FakeQuantize", "FakeQuantizeBase", "FixedQParamsFakeQuantize", "FusedMovingAvgObsFakeQuantize", # torch.ao.quantization.fx.fuse_handler "DefaultFuseHandler", "FuseHandler", # torch.ao.quantization.fx.graph_module "FusedGraphModule", "ObservedGraphModule", "ObservedStandaloneGraphModule", # torch.ao.quantization.fx.quantize_handler "BatchNormQuantizeHandler", "BinaryOpQuantizeHandler", "CatQuantizeHandler", "ConvReluQuantizeHandler", "CopyNodeQuantizeHandler", "CustomModuleQuantizeHandler", "DefaultNodeQuantizeHandler", "EmbeddingQuantizeHandler", "FixedQParamsOpQuantizeHandler", "GeneralTensorShapeOpQuantizeHandler", "LinearReLUQuantizeHandler", "RNNDynamicQuantizeHandler", "StandaloneModuleQuantizeHandler", # torch.ao.quantization.fx.tracer "QuantizationTracer", "ScopeContextManager", # torch.ao.quantization.fx.utils "ObservedGraphModuleAttrs", # torch.ao.quantization.observer "FixedQParamsObserver", "HistogramObserver", "MinMaxObserver", "MovingAverageMinMaxObserver", "MovingAveragePerChannelMinMaxObserver", "NoopObserver", "ObserverBase", "PerChannelMinMaxObserver", "PlaceholderObserver", "RecordingObserver", "ReuseInputObserver", "UniformQuantizationObserverBase", "default_debug_observer", "default_placeholder_observer", "default_reuse_input_observer", # torch.ao.quantization.pt2e.duplicate_dq_pass "DuplicateDQPass", # torch.ao.quantization.pt2e.port_metadata_pass "PortNodeMetaForQDQ", # torch.ao.quantization.qconfig "QConfigDynamic", # torch.ao.quantization.quant_type "QuantType", # torch.ao.quantization.quantizer.composable_quantizer "ComposableQuantizer", # torch.ao.quantization.quantizer.embedding_quantizer "EmbeddingQuantizer", # torch.ao.quantization.quantizer.quantizer "DerivedQuantizationSpec", "FixedQParamsQuantizationSpec", "QuantizationAnnotation", "QuantizationSpec", "QuantizationSpecBase", "SharedQuantizationSpec", # torch.ao.quantization.quantizer.x86_inductor_quantizer "X86InductorQuantizer", # torch.ao.quantization.quantizer.xpu_inductor_quantizer "XPUInductorQuantizer", # torch.ao.quantization.quantizer.xnnpack_quantizer "XNNPACKQuantizer", # torch.ao.quantization.quantizer.xnnpack_quantizer_utils "OperatorConfig", "QuantizationConfig", # torch.ao.quantization.stubs "DeQuantStub", "QuantStub", "QuantWrapper", # torch.ao.quantization.utils "MatchAllNode", # torch.backends.cudnn.rnn "Unserializable", # torch.amp.grad_scaler "GradScaler", "OptState", # torch.cuda.graphs "CUDAGraph", # torch.cuda.streams "Event", # torch.distributed.algorithms.ddp_comm_hooks.post_localSGD_hook "PostLocalSGDState", # torch.distributed.algorithms.ddp_comm_hooks.powerSGD_hook "PowerSGDState", # torch.distributed.algorithms.join "Join", "JoinHook", "Joinable", # torch.distributed.algorithms.model_averaging.averagers "ModelAverager", "PeriodicModelAverager", # torch.distributed.algorithms.model_averaging.hierarchical_model_averager "HierarchicalModelAverager", # torch.distributed.argparse_util "check_env", "env", # torch.distributed.checkpoint.api "CheckpointException", # torch.distributed.checkpoint.default_planner "DefaultLoadPlanner", "DefaultSavePlanner", # torch.distributed.checkpoint.filesystem "FileSystemReader", "FileSystemWriter", # torch.distributed.checkpoint.hf_storage "HuggingFaceStorageReader", "HuggingFaceStorageWriter", # torch.distributed.checkpoint.metadata "BytesStorageMetadata", "ChunkStorageMetadata", "Metadata", "MetadataIndex", # torch.distributed.checkpoint.planner "LoadItemType", "LoadPlanner", "SavePlanner", "WriteItemType", # torch.distributed.checkpoint.state_dict "DistributedStateDictOptions", # torch.distributed.checkpoint.storage "WriteResult", # torch.distributed.collective_utils "SyncPayload", # torch.distributed.distributed_c10d "AllToAllOptions", "AllreduceCoalescedOptions", "AllreduceOptions", "Backend", "BackendConfig", "BarrierOptions", "BroadcastOptions", "DebugLevel", "GatherOptions", "GroupMember", "ProcessGroup", "ProcessGroupGloo", "ProcessGroupNCCL", "ReduceOptions", "ReduceScatterOptions", "ScatterOptions", "Work", "group", # torch.distributed.elastic.agent.server.api "ElasticAgent", "RunResult", "SimpleElasticAgent", "WorkerSpec", # torch.distributed.elastic.events.api "Event", "RdzvEvent", # torch.distributed.elastic.metrics.api "ConsoleMetricHandler", "MetricData", "MetricHandler", "MetricStream", "MetricsConfig", "NullMetricHandler", # torch.distributed.elastic.multiprocessing.api "MultiprocessContext", "PContext", "RunProcsResult", "SignalException", "Std", "SubprocessContext", "SubprocessHandler", # torch.distributed.elastic.multiprocessing.tail_log "TailLog", # torch.distributed.elastic.rendezvous.api "RendezvousHandler", "RendezvousHandlerRegistry", "RendezvousParameters", # torch.distributed.elastic.rendezvous.dynamic_rendezvous "DynamicRendezvousHandler", "RendezvousSettings", # torch.distributed.elastic.rendezvous.etcd_rendezvous "EtcdRendezvous", "EtcdRendezvousHandler", "EtcdRendezvousRetryImmediately", "EtcdRendezvousRetryableFailure", # torch.distributed.elastic.rendezvous.etcd_server "EtcdServer", # torch.distributed.elastic.rendezvous.static_tcp_rendezvous "StaticTCPRendezvous", # torch.distributed.elastic.timer.api "RequestQueue", "TimerClient", "TimerServer", # torch.distributed.elastic.timer.file_based_local_timer "FileTimerClient", "FileTimerRequest", "FileTimerServer", # torch.distributed.elastic.timer.local_timer "LocalTimerClient", "LocalTimerServer", "MultiprocessingRequestQueue", # torch.distributed.elastic.utils.api "macros", # torch.distributed.elastic.utils.data.cycling_iterator "CyclingIterator", # torch.distributed.elastic.utils.data.elastic_distributed_sampler "ElasticDistributedSampler", # torch.distributed.fsdp.api "StateDictType", # torch.distributed.fsdp.fully_sharded_data_parallel "FullyShardedDataParallel", "OptimStateKeyType", # torch.distributed.fsdp.sharded_grad_scaler "ShardedGradScaler", # torch.distributed.fsdp.wrap "CustomPolicy", "ModuleWrapPolicy", # torch.distributed.launcher.api "LaunchConfig", "elastic_launch", # torch.distributed.optim.optimizer "DistributedOptimizer", # torch.distributed.optim.post_localSGD_optimizer "PostLocalSGDOptimizer", # torch.distributed.optim.zero_redundancy_optimizer "ZeroRedundancyOptimizer", # torch.distributed.rpc.api "AllGatherStates", "RRef", # torch.distributed.rpc.backend_registry "BackendValue", # torch.distributed.rpc.internal "PythonUDF", "RPCExecMode", "RemoteException", # torch.distributed.rpc.rref_proxy "RRefProxy", # torch.distributed.tensor.parallel.fsdp "DTensorExtensions", # torch.distributed.tensor.parallel.style "ParallelStyle", # torch.distributions.logistic_normal "LogisticNormal", # torch.distributions.one_hot_categorical "OneHotCategoricalStraightThrough", # torch.distributions.relaxed_categorical "ExpRelaxedCategorical", # torch.distributions.utils "lazy_property", # torch.export.unflatten "UnflattenedModule", # torch.export.exported_program "ConstantArgument", "ExportedProgram", # torch.fx.experimental.accelerator_partitioner "DAG", "DAGNode", "PartitionResult", "Partitioner", # torch.fx.experimental.const_fold "FoldedGraphModule", # torch.fx.experimental.graph_gradual_typechecker "Refine", # torch.fx.experimental.meta_tracer "MetaAttribute", "MetaDeviceAttribute", "MetaProxy", "MetaTracer", # torch.fx.experimental.migrate_gradual_types.constraint "ApplyBroadcasting", "BVar", "BinConstraintD", "BinConstraintT", "BinaryConstraint", "CalcConv", "CalcMaxPool", "CalcProduct", "CanReshape", "Conj", "Constraint", "DGreatestUpperBound", "DVar", "Disj", "F", "GetItem", "GetItemTensor", "IndexSelect", "Prod", "T", "TGreatestUpperBound", "TVar", "Transpose", # torch.fx.experimental.migrate_gradual_types.constraint_generator "ConstraintGenerator", # torch.fx.experimental.normalize "NormalizeArgs", "NormalizeOperators", # torch.fx.experimental.optimization "MklSubgraph", "UnionFind", # torch.fx.experimental.partitioner_utils "Device", "Partition", "PartitionLatency", "PartitionMode", "PartitionerConfig", # torch.fx.experimental.proxy_tensor "DecompositionInterpreter", "PreDispatchTorchFunctionMode", "ProxySymDispatchMode", "ProxyTorchDispatchMode", "PythonKeyTracer", # torch.fx.experimental.recording "FakeTensorMeta", "NotEqualError", "ShapeEnvEvent", # torch.fx.experimental.refinement_types "Equality", # torch.fx.experimental.rewriter "AST_Rewriter", "RewritingTracer", # torch.fx.experimental.schema_type_annotation "AnnotateTypesWithSchema", # torch.fx.experimental.sym_node "SymNode", # torch.fx.experimental.symbolic_shapes "Constraint", "ConstraintViolationError", "DynamicDimConstraintPrinter", "GuardOnDataDependentSymNode", "PendingUnbackedSymbolNotFound", "LoggingShapeGuardPrinter", "SymExprPrinter", "RelaxedUnspecConstraint", "RuntimeAssert", "ShapeGuardPrinter", "ShapeGuardPythonPrinter", "SymDispatchMode", "SymbolicContext", # torch.fx.experimental.unification.match "Dispatcher", "VarDispatcher", # torch.fx.experimental.unification.multipledispatch.conflict "AmbiguityWarning", # torch.fx.experimental.unification.multipledispatch.dispatcher "Dispatcher", "MDNotImplementedError", "MethodDispatcher", # torch.fx.experimental.unification.multipledispatch.variadic "Variadic", "VariadicSignatureMeta", "VariadicSignatureType", # torch.fx.experimental.unification.variable "Var", # torch.fx.experimental.validator "BisectValidationException", "PopulateValidator", "SympyToZ3", "ValidationException", # torch.fx.graph "PythonCode", # torch.fx.immutable_collections "immutable_dict", "immutable_list", # torch.fx.interpreter "Interpreter", # torch.fx.operator_schemas "ArgsKwargsPair", # torch.fx.passes.backends.cudagraphs "CudaGraphsSupport", # torch.fx.passes.dialect.common.cse_pass "CSEPass", # torch.fx.passes.fake_tensor_prop "FakeTensorProp", # torch.fx.passes.graph_drawer "FxGraphDrawer", # torch.fx.passes.graph_manipulation "size_bytes", # torch.fx.passes.infra.partitioner "CapabilityBasedPartitioner", "Partition", # torch.fx.passes.infra.pass_base "PassBase", "PassResult", # torch.fx.passes.infra.pass_manager "PassManager", # torch.fx.passes.net_min_base "FxNetMinimizerBadModuleError", "FxNetMinimizerResultMismatchError", "FxNetMinimizerRunFuncError", # torch.fx.passes.operator_support "OpSupports", "OperatorSupport", "OperatorSupportBase", # torch.fx.passes.pass_manager "PassManager", # torch.fx.passes.shape_prop "ShapeProp", # torch.fx.passes.split_module "Partition", # torch.fx.passes.split_utils "Component", # torch.fx.passes.splitter_base "FxNetAccNodesFinder", "FxNetSplitterInternalError", "SplitResult", "Subgraph", # torch.fx.passes.tests.test_pass_manager "TestPassManager", # torch.fx.passes.tools_common "FxNetAccFusionsFinder", # torch.fx.passes.utils.common "HolderModule", # torch.fx.passes.utils.matcher_utils "InternalMatch", "SubgraphMatcher", # torch.fx.passes.utils.source_matcher_utils "SourcePartition", # torch.fx.proxy "Attribute", "ParameterProxy", "Proxy", "Scope", "ScopeContextManager", "TraceError", "TracerBase", # torch.fx.subgraph_rewriter "Match", "ReplacedPatterns", # torch.jit.annotations "EvalEnv", "Module", # torch.jit.frontend "Builder", "ExprBuilder", "FrontendError", "FrontendTypeError", "NotSupportedError", "StmtBuilder", "UnsupportedNodeError", "WithItemBuilder", # torch.masked.maskedtensor.core "MaskedTensor", # torch.multiprocessing.pool "Pool", # torch.multiprocessing.queue "ConnectionWrapper", "Queue", "SimpleQueue", # torch.multiprocessing.reductions "SharedCache", # torch.multiprocessing.spawn "ProcessContext", "ProcessException", "ProcessExitedException", "ProcessRaisedException", "SpawnContext", # torch.nn.cpp "ModuleWrapper", "OrderedDictWrapper", # torch.nn.modules.container "Container", # deprecated # torch.nn.modules.linear "NonDynamicallyQuantizableLinear", # torch.nn.modules.module # TODO: causes multiple sphinx warnings # WARNING: more than one target found for cross-reference 'Module' "Module", # torch.nn.modules.loss "NLLLoss2d", # deprecated # torch.nn.modules.normalization "CrossMapLRN2d", # torch.nn.parallel.data_parallel "DataParallel", # torch.nn.parallel.distributed "DistributedDataParallel", # torch.nn.parameter "UninitializedTensorMixin", # torch.nn.utils.parametrize "ParametrizationList", # torch.nn.utils.prune "CustomFromMask", "Identity", "L1Unstructured", "RandomUnstructured", # torch.nn.utils.rnn "PackedSequence", "PackedSequence_", # torch.nn.utils.spectral_norm "SpectralNorm", "SpectralNormLoadStateDictPreHook", "SpectralNormStateDictHook", # torch.nn.utils.weight_norm "WeightNorm", # torch.onnx.errors "OnnxExporterError", "OnnxExporterWarning", "SymbolicValueError", "UnsupportedOperatorError", # torch.onnx.verification "OnnxBackend", "OnnxTestCaseRepro", # torch.optim.adamax "Adamax", # torch.optim.adamw "AdamW", # torch.optim.asgd "ASGD", # torch.optim.lbfgs "LBFGS", # torch.optim.lr_scheduler "ChainedScheduler", "ConstantLR", "CosineAnnealingLR", "CosineAnnealingWarmRestarts", "CyclicLR", "ExponentialLR", "LRScheduler", "LambdaLR", "LinearLR", "MultiStepLR", "MultiplicativeLR", "OneCycleLR", "PolynomialLR", "ReduceLROnPlateau", "SequentialLR", "StepLR", # torch.optim.optimizer "Optimizer", # torch.overrides "BaseTorchFunctionMode", "TorchFunctionMode", # torch.package.file_structure_representation "Directory", # torch.package.glob_group "GlobGroup", # torch.package.importer "Importer", "ObjMismatchError", "ObjNotFoundError", "OrderedImporter", # torch.package.package_exporter "PackageExporter", "PackagingErrorReason", # torch.package.package_importer "PackageImporter", # torch.profiler.profiler "ExecutionTraceObserver", "profile", # torch.return_types "aminmax", "aminmax_out", "cummax", "cummax_out", "cummin", "cummin_out", "frexp", "frexp_out", "geqrf", "geqrf_out", "histogram", "histogram_out", "histogramdd", "kthvalue", "kthvalue_out", "linalg_cholesky_ex", "linalg_cholesky_ex_out", "linalg_eig", "linalg_eig_out", "linalg_eigh", "linalg_eigh_out", "linalg_inv_ex", "linalg_inv_ex_out", "linalg_ldl_factor", "linalg_ldl_factor_ex", "linalg_ldl_factor_ex_out", "linalg_ldl_factor_out", "linalg_lstsq", "linalg_lstsq_out", "linalg_lu", "linalg_lu_factor", "linalg_lu_factor_ex", "linalg_lu_factor_ex_out", "linalg_lu_factor_out", "linalg_lu_out", "linalg_qr", "linalg_qr_out", "linalg_slogdet", "linalg_slogdet_out", "linalg_solve_ex", "linalg_solve_ex_out", "linalg_svd", "linalg_svd_out", "lu_unpack", "lu_unpack_out", "max", "max_out", "median", "median_out", "min", "min_out", "mode", "mode_out", "nanmedian", "nanmedian_out", "qr", "qr_out", "slogdet", "slogdet_out", "sort", "sort_out", "svd", "svd_out", "topk", "topk_out", "triangular_solve", "triangular_solve_out", # torch.serialization "LoadEndianness", "SourceChangeWarning", # torch.sparse.semi_structured "SparseSemiStructuredTensor", # torch.storage "UntypedStorage", # torch.torch_version "TorchVersion", # torch.types "SymInt", # torch.utils.benchmark.examples.compare "FauxTorch", # torch.utils.benchmark.examples.spectral_ops_fuzz_test "Benchmark", # torch.utils.benchmark.op_fuzzers.binary "BinaryOpFuzzer", # torch.utils.benchmark.op_fuzzers.sparse_binary "BinaryOpSparseFuzzer", # torch.utils.benchmark.op_fuzzers.sparse_unary "UnaryOpSparseFuzzer", # torch.utils.benchmark.op_fuzzers.spectral "SpectralOpFuzzer", # torch.utils.benchmark.op_fuzzers.unary "UnaryOpFuzzer", # torch.utils.benchmark.utils.common "Measurement", "TaskSpec", # torch.utils.benchmark.utils.compare "Colorize", "Compare", "Table", # torch.utils.benchmark.utils.fuzzer "FuzzedParameter", "FuzzedTensor", "Fuzzer", "ParameterAlias", # torch.utils.benchmark.utils.sparse_fuzzer "FuzzedSparseTensor", # torch.utils.benchmark.utils.timer "CPPTimer", "Language", "Timer", # torch.utils.benchmark.utils.valgrind_wrapper.timer_interface "CallgrindStats", "CopyIfCallgrind", "FunctionCount", "FunctionCounts", "GlobalsBridge", "Serialization", # torch.utils.bundled_inputs "InflatableArg", # torch.utils.checkpoint "CheckpointError", "CheckpointFunction", "DefaultDeviceType", # torch.utils.collect_env "SystemEnv", # torch.utils.cpp_extension "BuildExtension", # torch.utils.data.dataloader "DataLoader", # torch.utils.data.datapipes.dataframe.dataframe_wrapper "PandasWrapper", "default_wrapper", # torch.utils.data.datapipes.dataframe.dataframes "Capture", "CaptureA", "CaptureAdd", "CaptureCall", "CaptureControl", "CaptureDataFrame", "CaptureDataFrameWithDataPipeOps", "CaptureF", "CaptureGetAttr", "CaptureGetItem", "CaptureInitial", "CaptureLikeMock", "CaptureMul", "CaptureSetItem", "CaptureSub", "CaptureVariable", "CaptureVariableAssign", "DataFrameTracedOps", "DataFrameTracer", # torch.utils.data.datapipes.dataframe.datapipes "ConcatDataFramesPipe", "DataFramesAsTuplesPipe", "ExampleAggregateAsDataFrames", "FilterDataFramesPipe", "PerRowDataFramesPipe", "ShuffleDataFramesPipe", # torch.utils.data.datapipes.dataframe.structures "DataChunkDF", # torch.utils.data.datapipes.datapipe "DFIterDataPipe", "DataChunk", "IterDataPipe", "MapDataPipe", # torch.utils.data.datapipes.iter.callable "CollatorIterDataPipe", "MapperIterDataPipe", # torch.utils.data.datapipes.iter.combinatorics "SamplerIterDataPipe", "ShufflerIterDataPipe", # torch.utils.data.datapipes.iter.combining "ConcaterIterDataPipe", "DemultiplexerIterDataPipe", "ForkerIterDataPipe", "MultiplexerIterDataPipe", "ZipperIterDataPipe", # torch.utils.data.datapipes.iter.filelister "FileListerIterDataPipe", # torch.utils.data.datapipes.iter.fileopener "FileOpenerIterDataPipe", # torch.utils.data.datapipes.iter.grouping "BatcherIterDataPipe", "GrouperIterDataPipe", "UnBatcherIterDataPipe", # torch.utils.data.datapipes.iter.routeddecoder "RoutedDecoderIterDataPipe", # torch.utils.data.datapipes.iter.selecting "FilterIterDataPipe", # torch.utils.data.datapipes.iter.sharding "SHARDING_PRIORITIES", "ShardingFilterIterDataPipe", # torch.utils.data.datapipes.iter.utils "IterableWrapperIterDataPipe", # torch.utils.data.datapipes.map.callable "MapperMapDataPipe", # torch.utils.data.datapipes.map.combinatorics "ShufflerIterDataPipe", # torch.utils.data.datapipes.map.combining "ConcaterMapDataPipe", "ZipperMapDataPipe", # torch.utils.data.datapipes.map.grouping "BatcherMapDataPipe", # torch.utils.data.datapipes.map.utils "SequenceWrapperMapDataPipe", # torch.utils.data.datapipes.utils.decoder "Decoder", "ImageHandler", "MatHandler", # torch.utils.data.dataset "ConcatDataset", # torch.utils.data.distributed "DistributedSampler", # torch.utils.dlpack "DLDeviceType", # torch.utils.file_baton "FileBaton", # torch.utils.flop_counter "FlopCounterMode", # torch.utils.hipify.hipify_python "CurrentState", "GeneratedFileCleaner", "HipifyResult", "InputError", "Trie", "bcolors", # torch.utils.hooks "BackwardHook", "RemovableHandle", # torch.utils.mkldnn "MkldnnBatchNorm", "MkldnnConv1d", "MkldnnConv2d", "MkldnnConv3d", "MkldnnLinear", "MkldnnPrelu", # torch.utils.mobile_optimizer "LintCode", # torch.utils.show_pickle "DumpUnpickler", "FakeClass", "FakeObject", # torch.utils.tensorboard.writer "FileWriter", "SummaryWriter", # torch.utils.throughput_benchmark "ExecutionStats", # torch.utils.weak "WeakIdKeyDictionary", "WeakIdRef", "WeakTensorKeyDictionary", ] # The suffix(es) of source filenames. # You can specify multiple suffix as a list of string: # # source_suffix = ['.rst', '.md'] source_suffix = ".rst" # The master toctree document. master_doc = "index" # Use the linkcode extension to override [SOURCE] links to point # to the repo. Use the torch_version variable defined above to # determine link def linkcode_resolve(domain, info): if domain != "py": return None if not info["module"]: return None try: module = __import__(info["module"], fromlist=[""]) obj = module for part in info["fullname"].split("."): obj = getattr(obj, part) # Get the source file and line number obj = inspect.unwrap(obj) fn = inspect.getsourcefile(obj) source, lineno = inspect.getsourcelines(obj) except Exception: return None # Determine the tag based on the torch_version if RELEASE: version_parts = torch_version.split( "." ) # For release versions, format as "vX.Y.Z" for correct path in repo patch_version = ( version_parts[2].split("+")[0].split("a")[0] ) # assuming a0 always comes after release version in versions.txt version_path = f"v{version_parts[0]}.{version_parts[1]}.{patch_version}" else: version_path = torch.version.git_version fn = os.path.relpath(fn, start=os.path.dirname(torch.__file__)) return ( f"https://github.com/pytorch/pytorch/blob/{version_path}/torch/{fn}#L{lineno}" ) # The language for content autogenerated by Sphinx. Refer to documentation # for a list of supported languages. # # This is also used if you do content translation via gettext catalogs. # Usually you set "language" from the command line for these cases. language = "en" # List of patterns, relative to source directory, that match files and # directories to ignore when looking for source files. # This patterns also effect to html_static_path and html_extra_path exclude_patterns = [] # The name of the Pygments (syntax highlighting) style to use. pygments_style = "sphinx" # If true, `todo` and `todoList` produce output, else they produce nothing. # Disable docstring inheritance autodoc_inherit_docstrings = False # Show type hints in the description autodoc_typehints = "description" # Add parameter types if the parameter is documented in the docstring autodoc_typehints_description_target = "documented_params" # Type aliases for common types # Sphinx type aliases only works with Postponed Evaluation of Annotations # (PEP 563) enabled (via `from __future__ import annotations`), which keeps the # type annotations in string form instead of resolving them to actual types. # However, PEP 563 does not work well with JIT, which uses the type information # to generate the code. Therefore, the following dict does not have any effect # until PEP 563 is supported by JIT and enabled in files. autodoc_type_aliases = { "_size_1_t": "int or tuple[int]", "_size_2_t": "int or tuple[int, int]", "_size_3_t": "int or tuple[int, int, int]", "_size_4_t": "int or tuple[int, int, int, int]", "_size_5_t": "int or tuple[int, int, int, int, int]", "_size_6_t": "int or tuple[int, int, int, int, int, int]", "_size_any_opt_t": "int or None or tuple", "_size_2_opt_t": "int or None or 2-tuple", "_size_3_opt_t": "int or None or 3-tuple", "_ratio_2_t": "float or tuple[float, float]", "_ratio_3_t": "float or tuple[float, float, float]", "_ratio_any_t": "float or tuple", "_tensor_list_t": "Tensor or tuple[Tensor]", } # Enable overriding of function signatures in the first line of the docstring. autodoc_docstring_signature = True # -- katex javascript in header # # def setup(app): # app.add_javascript("https://cdn.jsdelivr.net/npm/katex@0.10.0-beta/dist/katex.min.js") # -- Options for HTML output ---------------------------------------------- # # The theme to use for HTML and HTML Help pages. See the documentation for # a list of builtin themes. # # # # Theme options are theme-specific and customize the look and feel of a theme # further. For a list of options available for each theme, see the # documentation. # Add any paths that contain custom static files (such as style sheets) here, # relative to this directory. They are copied after the builtin static files, # so a file named "default.css" will overwrite the builtin "default.css". html_css_files = [ "css/jit.css", "css/custom.css", "https://cdn.jsdelivr.net/npm/katex@0.10.0-beta/dist/katex.min.css", ] from sphinx.ext.coverage import CoverageBuilder # NB: Due to some duplications of the following modules/functions, we keep # them as expected failures for the time being instead of return 1 ignore_duplicated_modules = { "torch.nn.utils.weight_norm", "torch.nn.utils.spectral_norm", "torch.nn.parallel.data_parallel", "torch.ao.quantization.quantize", } def coverage_post_process(app, exception): if exception is not None: return # Only run this test for the coverage build if not isinstance(app.builder, CoverageBuilder): return if not torch.distributed.is_available(): raise RuntimeError( "The coverage tool cannot run with a version " "of PyTorch that was built with USE_DISTRIBUTED=0 " "as this module's API changes." ) # These are all the modules that have "automodule" in an rst file # These modules are the ones for which coverage is checked # Here, we make sure that no module is missing from that list modules = app.env.domaindata["py"]["modules"] # We go through all the torch submodules and make sure they are # properly tested missing = set() def is_not_internal(modname): split_name = modname.split(".") for name in split_name: if name[0] == "_": return False return True # The walk function does not return the top module if "torch" not in modules: missing.add("torch") for _, modname, ispkg in pkgutil.walk_packages( path=torch.__path__, prefix=torch.__name__ + "." ): if is_not_internal(modname): if modname not in modules and modname not in ignore_duplicated_modules: missing.add(modname) output = [] if missing: mods = ", ".join(missing) output.append( f"\nYou added the following module(s) to the PyTorch namespace '{mods}' " "but they have no corresponding entry in a doc .rst file. You should " "either make sure that the .rst file that contains the module's documentation " "properly contains either '.. automodule:: mod_name' (if you do not want " "the paragraph added by the automodule, you can simply use '.. py:module:: mod_name') " " or make the module private (by appending an '_' at the beginning of its name)." ) # The output file is hard-coded by the coverage tool # Our CI is setup to fail if any line is added to this file output_file = path.join(app.outdir, "python.txt") if output: with open(output_file, "a") as f: for o in output: f.write(o) def process_docstring(app, what_, name, obj, options, lines): """ Custom process to transform docstring lines Remove "Ignore" blocks Args: app (sphinx.application.Sphinx): the Sphinx application object what (str): the type of the object which the docstring belongs to (one of "module", "class", "exception", "function", "method", "attribute") name (str): the fully qualified name of the object obj: the object itself options: the options given to the directive: an object with attributes inherited_members, undoc_members, show_inheritance and noindex that are true if the flag option of same name was given to the auto directive lines (List[str]): the lines of the docstring, see above References: https://www.sphinx-doc.org/en/master/usage/extensions/autodoc.html """ import re remove_directives = [ # Remove all xdoctest directives re.compile(r"\s*>>>\s*#\s*x?doctest:\s*.*"), re.compile(r"\s*>>>\s*#\s*x?doc:\s*.*"), ] filtered_lines = [ line for line in lines if not any(pat.match(line) for pat in remove_directives) ] # Modify the lines inplace lines[:] = filtered_lines # make sure there is a blank line at the end if lines and lines[-1].strip(): lines.append("") def setup(app): app.connect("build-finished", coverage_post_process) app.connect("autodoc-process-docstring", process_docstring) app.connect("html-page-context", hide_edit_button_for_pages) app.config.add_last_updated = True return {"version": "0.1", "parallel_read_safe": True} def hide_edit_button_for_pages(app, pagename, templatename, context, doctree): if pagename.startswith("generated/"): context["theme_use_edit_page_button"] = False # From PyTorch 1.5, we now use autogenerated files to document classes and # functions. This breaks older references since # https://pytorch.org/docs/stable/torch.html#torch.flip # moved to # https://pytorch.org/docs/stable/generated/torch.flip.html # which breaks older links from blog posts, stack overflow answers and more. # To mitigate that, we add an id="torch.flip" in an appropriated place # in torch.html by overriding the visit_reference method of html writers. # Someday this can be removed, once the old links fade away from sphinx.writers import html, html5 def replace(Klass): old_call = Klass.visit_reference def visit_reference(self, node): if "refuri" in node and "generated" in node.get("refuri"): ref = node.get("refuri") ref_anchor = ref.split("#") if len(ref_anchor) > 1: # Only add the id if the node href and the text match, # i.e. the href is "torch.flip#torch.flip" and the content is # "torch.flip" or "flip" since that is a signal the node refers # to autogenerated content anchor = ref_anchor[1] txt = node.parent.astext() if txt == anchor or txt == anchor.split(".")[-1]: self.body.append(f'

') return old_call(self, node) Klass.visit_reference = visit_reference replace(html.HTMLTranslator) replace(html5.HTML5Translator) # -- Options for HTMLHelp output ------------------------------------------ # Output file base name for HTML help builder. htmlhelp_basename = "PyTorchdoc" # -- Options for LaTeX output --------------------------------------------- latex_engine = "lualatex" latex_show_urls = "footnote" latex_elements = { "papersize": "letterpaper", "pointsize": "10pt", "tableofcontents": r"\pdfbookmark[0]{Contents}{toc}\tableofcontents", "preamble": r""" \usepackage{tocloft} \setcounter{tocdepth}{3} \setcounter{secnumdepth}{3} % Fix table column widths \renewenvironment{tabulary}{\begin{longtable}{p{0.3\linewidth}p{0.7\linewidth}}}{\end{longtable}} % Ensure tables don't overflow \AtBeginEnvironment{tabular}{\sloppy} """, "fncychap": r"\usepackage[Bjornstrup]{fncychap}", "extraclassoptions": "oneside", } # Grouping the document tree into LaTeX files. List of tuples # (source start file, target name, title, # author, documentclass [howto, manual, or own class]). latex_documents = [ ( master_doc, "pytorch.tex", "PyTorch Documentation", "Torch Contributors", "manual", ), ] latex_use_xindy = False # -- Options for manual page output --------------------------------------- # One entry per manual page. List of tuples # (source start file, name, description, authors, manual section). man_pages = [(master_doc, "PyTorch", "PyTorch Documentation", [author], 1)] # -- Options for Texinfo output ------------------------------------------- # Grouping the document tree into Texinfo files. List of tuples # (source start file, target name, title, author, # dir menu entry, description, category) texinfo_documents = [ ( master_doc, "PyTorch", "PyTorch Documentation", author, "PyTorch", "One line description of project.", "Miscellaneous", ), ] # Example configuration for intersphinx: refer to the Python standard library. intersphinx_mapping = { "python": ("https://docs.python.org/3", None), "numpy": ("https://numpy.org/doc/stable", None), } import sphinx.ext.doctest # -- A patch that prevents Sphinx from cross-referencing ivar tags ------- # See http://stackoverflow.com/a/41184353/3343043 from docutils import nodes from sphinx import addnodes from sphinx.util.docfields import TypedField # Without this, doctest adds any example with a `>>>` as a test doctest_test_doctest_blocks = "" doctest_default_flags = sphinx.ext.doctest.doctest.ELLIPSIS doctest_global_setup = """ import torch try: import torchvision except ImportError: torchvision = None """ def patched_make_field(self, types, domain, items, **kw): # `kw` catches `env=None` needed for newer sphinx while maintaining # backwards compatibility when passed along further down! # type: (List, unicode, Tuple) -> nodes.field def handle_item(fieldarg, content): par = nodes.paragraph() par += addnodes.literal_strong("", fieldarg) # Patch: this line added # par.extend(self.make_xrefs(self.rolename, domain, fieldarg, # addnodes.literal_strong)) if fieldarg in types: par += nodes.Text(" (") # NOTE: using .pop() here to prevent a single type node to be # inserted twice into the doctree, which leads to # inconsistencies later when references are resolved fieldtype = types.pop(fieldarg) if len(fieldtype) == 1 and isinstance(fieldtype[0], nodes.Text): typename = fieldtype[0].astext() builtin_types = ["int", "long", "float", "bool", "type"] for builtin_type in builtin_types: pattern = rf"(?>> |\.\.\. " copybutton_prompt_is_regexp = True