MCPcopy Create free account

hub / github.com/deepspeedai/DeepSpeed / functions

Functions10,922 in github.com/deepspeedai/DeepSpeed

⨍ Functions 10,922◇ Types & classes 1,850↳ Endpoints 34
↓ 2 callersFunction_Wait_bwd_comm
(input_, dic_, h_id)
deepspeed/runtime/domino/transformer.py:50
↓ 2 callersMethod__getattr__
Pass through attributes defined in the model if they are not overridden by ds-engine.
deepspeed/runtime/engine.py:873
↓ 2 callersMethod__getstate__
(self)
deepspeed/linear/quantization.py:85
↓ 2 callersMethod__init__
(self, out=True, err=True, replay=True)
tests/unit/util.py:230
↓ 2 callersMethod__init__
(self, attn: nn.Module)
tests/unit/sequence_parallelism/test_autosp_integration.py:89
↓ 2 callersMethod__init__
(self, result: str)
tests/unit/inference/v2/model_implementations/test_exaone4_5.py:101
↓ 2 callersMethod__init__
(self)
tests/unit/module_inject/test_tp_partition_config_path.py:19
↓ 2 callersMethod__init__
(self, batch_size, hidden_size, intermediate_size, heads, attn_dropout_ratio, hidden_dropout_ratio,
deepspeed/ops/transformer/transformer.py:21
↓ 2 callersMethod__init__
(self, config, q_scales=None, q_groups=1, merge_count=1, mlp_extra_grouping=False, mp_group=None)
deepspeed/ops/transformer/inference/moe_inference.py:133
↓ 2 callersMethod__init__
(self, chunk: torch.Tensor, device=None, is_in_use=False)
deepspeed/sequence/fpdt_layer.py:499
↓ 2 callersMethod__init__
A context manager to partition the model parameters during the model construction with MiCS partition strategy. Model states are partitioned
deepspeed/runtime/zero/mics.py:65
↓ 2 callersMethod__init__
Create a mapping of n-dimensional tensor coordinates to linear indices. Arguments: axes (list): the names of the tensor axes
deepspeed/runtime/pipe/topology.py:26
↓ 2 callersMethod__init__
(self, typename, *module_args, **module_kwargs)
deepspeed/runtime/pipe/module.py:53
↓ 2 callersMethod__init__
(self, path)
deepspeed/runtime/data_pipeline/data_sampling/indexed_dataset.py:143
↓ 2 callersMethod__init__
( self, init_optimizer, param_names, timers, optimizer_params,
deepspeed/runtime/zenflow/zenflow_stage_1_and_2.py:49
↓ 2 callersMethod__init__
(self, scale=1)
deepspeed/runtime/fp16/loss_scaler.py:175
↓ 2 callersMethod__init__
(self, vocab_size, hidden_size, num_layers, num_heads, max_seq_len, dropout=0.1)
examples/sdma_allgather/train_zero3.py:82
↓ 2 callersMethod__init__
(self, attn_cls, hidden: int)
benchmarks/autosp/bench_multimodal_sp.py:90
↓ 2 callersMethod__new__
( cls, data: Optional[torch.Tensor] = None, requires_grad: bool = False, # quantized
deepspeed/linear/quantization.py:37
↓ 2 callersMethod__reduce_and_partition_ipg_grads
(self, communication_data_type: torch.dtype)
deepspeed/runtime/zero/stage3.py:1488
↓ 2 callersMethod__setstate__
(self, state)
deepspeed/linear/quantization.py:92
↓ 2 callersMethod_add_tensors
(self, tensor_list)
deepspeed/runtime/swap_tensor/optimizer_utils.py:66
↓ 2 callersFunction_all_real_if_tensor
(args)
deepspeed/compile/profilers/graph_profile.py:26
↓ 2 callersMethod_all_source_files
(self)
ci/tests_fetcher.py:270
↓ 2 callersMethod_all_test_files
(self)
ci/tests_fetcher.py:262
↓ 2 callersMethod_allocate_and_return_buffers_for_swap_in
(self, params)
deepspeed/runtime/swap_tensor/partitioned_param_swapper.py:178
↓ 2 callersMethod_allocate_or_extend_buffers
(self, idx, shape, dtype)
deepspeed/runtime/pipe/engine.py:1291
↓ 2 callersFunction_assert_grad_maps_close
(actual, expected, *, lhs_name, rhs_name)
tests/unit/v1/moe/test_autoep_grad_parity.py:140
↓ 2 callersFunction_assert_module_params_match_universal
(engine, universal_dir)
tests/unit/v1/moe/test_autoep_checkpoint.py:271
↓ 2 callersFunction_assert_same_dtype_device
(actual, expected)
tests/unit/v1/moe/test_autoep_unit.py:96
↓ 2 callersMethod_assert_zero3_universal_weights_only_load
(self, tmpdir, load_kwargs)
tests/unit/v1/moe/test_autoep_checkpoint.py:478
↓ 2 callersMethod_assign_resources_to_tensor_slice
(self, slice_resources, my_slice_index, my_dp_ranks)
deepspeed/runtime/model_checkpointing/data_parallel_writer_factory.py:178
↓ 2 callersFunction_autoep_expert_param_names_by_rank
(model_files)
deepspeed/checkpoint/ds_to_universal.py:479
↓ 2 callersMethod_autoep_expert_parameter_names
(autoep_layers, model)
deepspeed/runtime/engine.py:4205
↓ 2 callersFunction_autoep_modules
(engine)
tests/unit/v1/moe/test_autoep_checkpoint.py:122
↓ 2 callersMethod_autoep_sequence_parallel_world_size
(self)
deepspeed/runtime/engine.py:614
↓ 2 callersMethod_autoep_zero_optimizer_param_families
(self)
deepspeed/runtime/engine.py:3749
↓ 2 callersFunction_backfill_missing_profile_metadata
(graph: Graph, profile_complete: bool = True)
deepspeed/compile/profilers/graph_profile.py:86
↓ 2 callersMethod_backup_cpuinfo
(self)
op_builder/builder.py:438
↓ 2 callersFunction_bf16_optimizer_stub
(lp, hp_grad)
tests/unit/v1/moe/test_autoep_autotp_bf16_folding_parity.py:15
↓ 2 callersFunction_blas_linear_helper
(tokens: int, in_channels: int, out_channels: int,
tests/unit/inference/v2/modules/test_blas_linear_module.py:49
↓ 2 callersFunction_bloom_type_transpose
(input, mp_size)
deepspeed/module_inject/fusedqkv_utils.py:89
↓ 2 callersMethod_broadcast_model
(self)
deepspeed/runtime/engine.py:1593
↓ 2 callersFunction_build_average_tensor_optimizer
(monkeypatch, *, copy_streams)
tests/unit/v1/zero/test_overlap_comm_record_stream.py:137
↓ 2 callersMethod_build_indexes
(self, files: list[Path])
ci/tests_fetcher.py:318
↓ 2 callersFunction_build_writer
(file_path)
tests/unit/ops/aio/test_fast_file_writer_fd_close.py:58
↓ 2 callersFunction_byte_cast_multiple_tensors
(typed_tensor_list)
tests/unit/utils/test_byte_cast.py:30
↓ 2 callersFunction_byte_cast_single_tensor
(typed_tensor)
tests/unit/utils/test_byte_cast.py:23
↓ 2 callersFunction_capture_params
(engine)
tests/unit/v1/zero/test_zero2_offload_multi_backward.py:54
↓ 2 callersMethod_change_recovery_script_permissions
(self, dst)
deepspeed/runtime/engine.py:5193
↓ 2 callersMethod_choose_module_key
(self, sd)
deepspeed/runtime/state_dict_factory.py:140
↓ 2 callersMethod_clean_inflight_param_registry
(self)
deepspeed/runtime/zero/partitioned_param_coordinator.py:189
↓ 2 callersMethod_clear_previous_reduced_grads
(self)
deepspeed/runtime/zero/stage_1_and_2.py:1793
↓ 2 callersMethod_close_pool
(self, pool, num_procs, force=False)
tests/unit/common.py:358
↓ 2 callersFunction_collect_autoep_expert_grads
(engine)
tests/unit/v1/moe/test_autoep_grad_parity.py:114
↓ 2 callersMethod_combine_output_splits
Join the splits of the output into a single result. Args: outputs (List[Any]): The reduced outputs for each output split.
deepspeed/runtime/zero/tiling.py:195
↓ 2 callersMethod_common_checkpoint_state
(self, module_state_dict, zero_optimizer_state, save_frozen_param)
deepspeed/runtime/engine.py:4705
↓ 2 callersFunction_compare_optimizers
(model_size, param1, optimizer1, param2, optimizer2)
tests/unit/ops/adam/test_cpu_adam.py:36
↓ 2 callersFunction_compute
(module, *inputs, do_checkpoint=False)
tests/unit/runtime/activation_checkpointing/test_activation_checkpointing.py:19
↓ 2 callersMethod_configure_bf16_optimizer
(self, optimizer)
deepspeed/runtime/engine.py:2211
↓ 2 callersMethod_configure_master_weights
Common validation and dtype selection for ZeRO optimizer master-weight settings. Optionally accepts callables that enforce backend-sp
deepspeed/runtime/base_optimizer.py:467
↓ 2 callersMethod_configure_train_batch_size
(self)
deepspeed/runtime/config.py:971
↓ 2 callersMethod_configure_zenflow
Configure ZenFlow optimizer
deepspeed/runtime/zenflow/zenflow_stage_1_and_2.py:88
↓ 2 callersMethod_configure_zero_optimizer
(self, optimizer)
deepspeed/runtime/engine.py:2234
↓ 2 callersMethod_convert_to_zero_parameters
(self, param_list)
deepspeed/runtime/zero/partition_parameters.py:1129
↓ 2 callersFunction_count_deleted_fds
How many fds in /proc/self/fd point at a now-deleted file located under target_dir? Restricting to target_dir avoids false positives from unre
tests/unit/ops/aio/test_fast_file_writer_fd_close.py:42
↓ 2 callersMethod_create_checkpoint_file
(self, save_dir, tag, zero_checkpoint)
deepspeed/runtime/engine.py:5015
↓ 2 callersMethod_create_module_forward_post_hook
(self)
deepspeed/runtime/engine.py:2586
↓ 2 callersMethod_create_module_forward_pre_hook
(self)
deepspeed/runtime/engine.py:2579
↓ 2 callersMethod_decode
(self, x, return_dict=True, generator=None)
deepspeed/model_implementations/diffusers/vae.py:34
↓ 2 callersFunction_define_dc_ops
()
tests/unit/compile/test_list_schedule.py:25
↓ 2 callersMethod_diff_files
Return (changed, deleted) repo-root-relative paths for ``base_rev..HEAD``. 'changed' = added / modified / renamed-new / copied-new; 'deleted'
ci/tests_fetcher.py:217
↓ 2 callersFunction_digest_words
(words: torch.Tensor, *, salt: int, modulus: int)
deepspeed/moe/ep_tp_dispatch.py:99
↓ 2 callersMethod_do_init
(self, path, skip_warmup)
deepspeed/runtime/data_pipeline/data_sampling/indexed_dataset.py:494
↓ 2 callersFunction_do_io_complete
csrc/aio/common/deepspeed_aio_common.cpp:112
↓ 2 callersFunction_do_io_submit_block
csrc/aio/common/deepspeed_aio_common.cpp:92
↓ 2 callersFunction_do_io_submit_singles
csrc/aio/common/deepspeed_aio_common.cpp:71
↓ 2 callersFunction_do_schedule_without_allgather
(scheduled: List[Node], unscheduled: List[Node], edges: Dict[Node, List[Node]],
deepspeed/compile/list_schedule.py:132
↓ 2 callersFunction_do_set_z3_leaf_modules
(model: torch.nn.Module, leaf_module_classes: Union[List[Type], List[str]],
deepspeed/utils/z3_leaf_module.py:57
↓ 2 callersMethod_drain
(self, num_bytes, fd, file_offset, blocking=False)
deepspeed/io/base_io_buffer.py:46
↓ 2 callersMethod_drain_io_buffer
(self, num_bytes)
deepspeed/io/fast_file_writer.py:132
↓ 2 callersFunction_eigenvalue_summary_events
(block_eigenvalue, global_samples)
deepspeed/runtime/engine.py:230
↓ 2 callersFunction_elementwise_flops_compute
(input, other)
deepspeed/profiling/flops_profiler/profiler.py:841
↓ 2 callersFunction_empty_grad_buffer
(param)
deepspeed/compile/init_z1.py:18
↓ 2 callersMethod_encode
(self, x, return_dict=True)
deepspeed/model_implementations/diffusers/vae.py:77
↓ 2 callersMethod_ensure_quantized
(self, tensor: torch.Tensor)
deepspeed/linear/quantization.py:58
↓ 2 callersMethod_exec_schedule
(self, pipe_schedule)
deepspeed/runtime/pipe/engine.py:1380
↓ 2 callersMethod_fill_param_grad_accum_attribute
(self, param)
deepspeed/runtime/zero/stage_1_and_2.py:1055
↓ 2 callersMethod_fini
(self)
deepspeed/io/fast_file_writer.py:113
↓ 2 callersMethod_flush_buffers_until_complete
(self)
deepspeed/runtime/swap_tensor/async_swapper.py:120
↓ 2 callersMethod_flush_ready_buffers
(self)
deepspeed/runtime/swap_tensor/async_swapper.py:113
↓ 2 callersFunction_folded_zero2_tp2_ep4_config
()
tests/unit/v1/moe/test_autoep_autotp_grad_parity.py:211
↓ 2 callersMethod_forward
(self, sample, timestamp, encoder_hidden_states,
deepspeed/model_implementations/diffusers/unet.py:66
↓ 2 callersMethod_forward
(self, sample, timestamp, encoder_hidden_states, return_dict=True)
deepspeed/model_implementations/diffusers/vae.py:150
↓ 2 callersMethod_forward_attention
(self, layer_idx: int, qkv: torch.Tensor, kv_cache: torch.Tensor, ragged_batch_info
deepspeed/inference/v2/model_implementations/exaone4/model.py:145
↓ 2 callersFunction_fp6_quantized_linear_helper
(tokens: int, in_channels: int, out_channels
tests/unit/inference/v2/modules/test_quantized_linear_module.py:86
↓ 2 callersFunction_gather
Gather tensors and concatenate along the last dimension.
deepspeed/compression/basic_layer.py:675
↓ 2 callersMethod_gather_and_compare_params
(self, model, torch_q, torch_o, compare_values=True)
tests/unit/model_parallelism/test_tp_plan_e2e.py:104
↓ 2 callersFunction_gather_logical_tensor
(tensor, logical_shape, partition_dim,
deepspeed/module_inject/layers.py:1187
↓ 2 callersMethod_generate
(self, model, tokenizer, prompt)
tests/unit/hybrid_engine/test_he_llama.py:31
← previousnext →1,201–1,300 of 10,922, ranked by callers