MCPcopy Create free account

hub / github.com/Gen-Verse/OpenClaw-RL / functions

Functions10,561 in github.com/Gen-Verse/OpenClaw-RL

Function_init_from_local_shards_and_global_metadata
Initialize a ShardedTensor with local shards and a global ShardedTensorMetadata built on each rank. Warning: This API is exp
slime/slime/backends/megatron_utils/checkpoint.py:35
Method_init_weights
(self, m)
Megatron-LM/megatron/legacy/model/vision/classification.py:72
Method_init_weights
(self, m)
Megatron-LM/megatron/legacy/model/vision/esvit_swin_backbone.py:580
Method_init_weights
(self, m)
Megatron-LM/megatron/legacy/model/vision/dino.py:105
Method_init_weights
(self, m)
Megatron-LM/megatron/legacy/model/vision/mit_backbone.py:31
Method_init_weights
(self, m)
Megatron-LM/megatron/legacy/model/vision/mit_backbone.py:86
Method_init_weights
(self, m)
Megatron-LM/megatron/legacy/model/vision/mit_backbone.py:143
Method_init_weights
(self, m)
Megatron-LM/megatron/legacy/model/vision/mit_backbone.py:180
Method_init_weights
(self, m)
Megatron-LM/megatron/legacy/model/vision/mit_backbone.py:260
Method_init_weights
(self, m)
Megatron-LM/megatron/legacy/model/vision/swin_backbone.py:552
Method_initalize
(self, vocab_extra_ids)
Megatron-LM/megatron/training/tokenizer/tokenizer.py:544
Method_initalize
(self, vocab_extra_ids)
Megatron-LM/megatron/training/tokenizer/tokenizer.py:588
Method_key_config_attributes
Inherited method implementation Returns: List[str]: The key config attributes
Megatron-LM/megatron/core/datasets/t5_dataset.py:85
Method_key_config_attributes
Inherited method implementation Returns: List[str]: The key config attributes
Megatron-LM/megatron/core/datasets/bert_dataset.py:64
Method_key_config_attributes
Inherited method implementation Returns: List[str]: The key config attributes
Megatron-LM/megatron/core/datasets/masked_dataset.py:126
Method_key_config_attributes
Add custom attributes for building unique dataset hash. The preprocessing split used for preprocessing will constrain the samples
Megatron-LM/megatron/core/datasets/retro/query/multi_split_gpt_dataset.py:104
Function_launch_background
()
slime/tests/test_external_rollout.py:132
Function_launch_server_entry
(server_args: ServerArgs)
slime/slime/backends/sglang_utils/sglang_engine.py:72
Function_layers_block_type
(self)
slime/slime/backends/sglang_utils/qwen3_5.py:267
Function_linear_layer_ids
(self)
slime/slime/backends/sglang_utils/qwen3_5.py:274
Method_list_free_vms
(self, region=DEFAULT_REGION)
gui-rl/desktop_env/providers/aws/manager.py:229
Method_list_free_vms
(self)
gui-rl/desktop_env/providers/volcengine/manager.py:326
Method_list_free_vms
(self)
gui-rl/desktop_env/providers/aliyun/manager.py:309
Method_load_checkpoint_args
(self, margs)
Megatron-LM/tools/checkpoint/saver_llava.py:53
Function_load_from_array
( sharded_tensor: ShardedTensor, checkpoint_dir: Path, load_directly_on_device: bool = False,
Megatron-LM/megatron/core/dist_checkpointing/strategies/tensorstore.py:96
Method_load_from_state_dict
(self, state_dict, prefix, *args, **kwargs)
Megatron-LM/megatron/core/models/common/embeddings/rotary_pos_embedding.py:207
Method_load_from_state_dict
Load the state dict of the router.
Megatron-LM/megatron/core/transformer/moe/router.py:575
Function_load_state_dict_hook_ignore_extra_state
Hook to ignore Transformer Engine _extra_state used for FP8. This is for backwards-compatibility. Newer TE versions add _extra_state keys to the
Megatron-LM/megatron/core/models/multimodal/llava_model.py:976
Function_load_state_dict_hook_ignore_param_names
Hook to ignore missing keys during checkpoint loading. By default, this should not be used to avoid accidentally missing weights in checkpoint lo
Megatron-LM/megatron/core/models/multimodal/llava_model.py:952
Method_load_state_dict_pre_hook
( state_dict, prefix, local_metadata, strict, missing_keys, unexpected_keys, error_msgs )
Megatron-LM/megatron/core/post_training/modelopt/layers.py:86
Method_login_setup
Login to a website with account and password information. @args: config(Dict[str, Any]): contain keys settings_fi
gui-rl/desktop_env/controllers/setup.py:773
Method_lora_name_transform
(name: str)
slime/slime/backends/fsdp_utils/actor.py:161
Function_loss_func
(loss_mask: torch.Tensor, output_tensor: torch.Tensor)
Megatron-LM/tests/unit_tests/export/trtllm/test_single_device_fp8.py:84
Method_make_dummy_samples
(count: int)
slime/slime/ray/rollout.py:630
Function_mamba2_cache_params
(self)
slime/slime/backends/sglang_utils/qwen3_5.py:288
Method_mark_removed_samples
(samples: list[Sample])
slime/slime/ray/rollout.py:610
Method_maybe_ensure_additional_required_arguments
Ensure that certain Megatron arguments (from checkpoint) are present. If missing, either set defaults or exit.
Megatron-LM/tools/checkpoint/loader_llava.py:88
Method_maybe_parse_additional_megatron_args
Parse Megatron arguments by forcibly overwriting sys.argv. Populates self.margs and self.checkpoint_args.
Megatron-LM/tools/checkpoint/loader_llava.py:54
Method_maybe_parse_additional_megatron_args
(self, margs)
Megatron-LM/tools/checkpoint/saver_llava.py:135
Method_maybe_submit_ready_samples
( self, session_id: str, force_drop_without_next_state: bool = False, )
openclaw-combine/openclaw_combine_api_server.py:151
Function_merge_slice
(dim_slice, dim_size)
Megatron-LM/megatron/core/dist_checkpointing/strategies/tensorstore.py:83
Method_missing_keys
(self)
Megatron-LM/tests/unit_tests/dist_checkpointing/test_strict.py:123
Function_model_builder
(args, pre_process, post_process, vp_stage=None)
Megatron-LM/train_rl.py:366
Method_modify_submodules
(submodules)
Megatron-LM/tests/unit_tests/transformer/test_transformer_block_custom_pgs.py:71
Function_monkeypatch_offload_deps
(monkeypatch)
Megatron-LM/tests/unit_tests/pipeline_parallel/test_fine_grained_activation_offloading.py:72
Function_msc_download_file
(remote_path, local_path)
Megatron-LM/tests/unit_tests/data/test_bin_reader.py:124
Function_msc_resolve_storage_client
(path)
Megatron-LM/tests/unit_tests/data/test_bin_reader.py:130
Function_multihot_to_indices_kernel
Triton kernel for converting multihot representation to indices. Input: probs_in_multihot: [num_of_tokens, num_of_local_experts]
Megatron-LM/megatron/core/fusions/fused_indices_converter.py:112
Method_occupy_vm
(self, vm_path, pid, region=DEFAULT_REGION)
gui-rl/desktop_env/providers/aws/manager.py:217
Method_occupy_vm
(self, vm_path, pid)
gui-rl/desktop_env/providers/volcengine/manager.py:314
Method_occupy_vm
(self, vm_path, pid)
gui-rl/desktop_env/providers/aliyun/manager.py:297
Function_on_shutdown
()
terminal-rl/router_server.py:371
Function_on_shutdown
()
terminal-rl/remote/pool_server.py:565
Function_on_startup
()
terminal-rl/router_server.py:365
Function_on_startup
()
terminal-rl/remote/pool_server.py:558
Method_opd_evaluate
( self, session_id: str, turn_num: int, turn_data: dict[str, Any], nex
openclaw-combine/openclaw_combine_select_api_server.py:96
Method_open_setup
(self, path: str)
gui-rl/desktop_env/controllers/setup.py:296
Function_p2p_ops
( *, tensor_send_prev: Optional[torch.Tensor], tensor_recv_prev: Optional[torch.Tensor], tenso
Megatron-LM/megatron/core/pipeline_parallel/p2p_communication.py:54
Method_pack_sample_by_key
(key: str)
Megatron-LM/megatron/core/datasets/data_schedule.py:208
Method_pad
(number_to_be_padded: int, divisor: int)
Megatron-LM/megatron/core/distributed/param_and_grad_buffer.py:614
Method_pad_end_of_bucket_if_needed
Pads end index of bucket if using distributed optimizer (to ensure uniform sharding).
Megatron-LM/megatron/core/distributed/param_and_grad_buffer.py:617
Function_pad_routing_map_kernel
( routing_map_ptr, output_ptr, num_tokens, pad_multiple: tl.constexpr, BLOCK_SIZE: tl.constexpr )
Megatron-LM/megatron/core/fusions/fused_pad_routing_map.py:31
Method_pad_start_of_param_if_needed
Pads start index of param if using distributed optimizer (to ensure "good" alignment).
Megatron-LM/megatron/core/distributed/param_and_grad_buffer.py:637
Function_parse_sheet_idx
( sheet_idx: Union[int, str], result: BOOK, expected: BOOK, result_sheet_names: List[str],
gui-rl/desktop_env/evaluators/metrics/table.py:42
Function_patched_forward
(self, *args, packed_seq_params=None, **kwargs)
slime/slime/backends/megatron_utils/__init__.py:32
Function_periodic_health_check
Background thread: re-check all exec nodes every `interval` seconds.
swe-rl/server/swe_env_pool_server.py:359
Method_post_backward
Deallocate the module parameters after the backward pass, and reduce-scatter the gradients before the optimizer step.
Megatron-LM/megatron/core/distributed/fsdp/src/megatron_fsdp/megatron_fsdp.py:559
Method_post_forward
(module: nn.Module, input: Any, output: Any)
Megatron-LM/megatron/core/distributed/fsdp/src/megatron_fsdp/megatron_fsdp.py:793
Method_pre_backward_param_unshard
Sub-module pre-backward hook to all-gather the module parameters before the backward pass.
Megatron-LM/megatron/core/distributed/fsdp/src/megatron_fsdp/megatron_fsdp.py:730
Method_pre_forward_param_unshard
( module: nn.Module, args: Tuple[Any, ...], kwargs: Dict[str, Any] )
Megatron-LM/megatron/core/distributed/fsdp/src/megatron_fsdp/megatron_fsdp.py:608
Function_preprocess_data_for_llava
Preprocess data sample to the format expected by a LLaVA model. Note: This doesn't support all the different modes in the official LLaVA repo yet
Megatron-LM/pretrain_vlm.py:244
Method_print_ready
(self)
openclaw-tinker/api_server.py:447
Method_print_ready
(self)
openclaw-tinker/api_server.py:857
Function_process_fc1_bias_param
(param)
Megatron-LM/megatron/core/transformer/moe/upcycling_utils.py:179
Function_process_fc2_bias_param
(param)
Megatron-LM/megatron/core/transformer/moe/upcycling_utils.py:193
Function_process_fc2_weight_param
(param)
Megatron-LM/megatron/core/transformer/moe/upcycling_utils.py:185
Function_process_router_param
(value)
Megatron-LM/megatron/core/transformer/moe/upcycling_utils.py:137
Method_query_document_sample_shuffle_indices
Get the text (token ids) and document ids for a given index Args: idx (int): The index into the dataset Returns:
Megatron-LM/megatron/training/datasets/fim_dataset.py:104
Function_qwen3_patched_init
(self, *args, **kwargs)
slime/slime_plugins/megatron_bridge/__init__.py:9
Method_readiness_check
()
openclaw-fireworks/fireworks_server.py:931
Method_recompute
Used as a hook to recompute the output.
Megatron-LM/megatron/core/tensor_parallel/random.py:628
Method_reduce_scatter_base
(self, *a, **kw)
slime/slime/utils/reloadable_process_group.py:220
Method_register_grad_acc_and_reduce_hook
Register the post-backward hook to deallocate model parameters and reduce-scatter gradients immediately after the module back
Megatron-LM/megatron/core/distributed/fsdp/src/megatron_fsdp/megatron_fsdp.py:871
Method_register_post_backward_hook
Pre-forward hook utilized to attach a gradient reduction post-backward hook to the module.
Megatron-LM/megatron/core/distributed/fsdp/src/megatron_fsdp/megatron_fsdp.py:639
Method_register_pre_backward_param_unshard_hook
Register the backward pre-hook to unshard FSDP unit module parameters immediately before the backward pass via attaching a gr
Megatron-LM/megatron/core/distributed/fsdp/src/megatron_fsdp/megatron_fsdp.py:861
Method_register_pre_forward_param_unshard_hook
Register the forward pre-hook to unshard parameters before the forward pass. If we are not sharding anything, we do not have
Megatron-LM/megatron/core/distributed/fsdp/src/megatron_fsdp/megatron_fsdp.py:848
Method_release_module_fp8_transpose_cache
(module: nn.Module, *unused)
Megatron-LM/megatron/core/distributed/fsdp/src/megatron_fsdp/megatron_fsdp.py:812
Function_remove_prefix_state_dict_pre_hook
Pytorch state_dict pre_hook to remove prefix of the state_dict keys.
Megatron-LM/megatron/post_training/checkpointing.py:153
Function_replace_prefix
(x)
Megatron-LM/megatron/core/dist_checkpointing/utils.py:200
Function_replace_prefixes
(x)
Megatron-LM/megatron/core/dist_checkpointing/utils.py:222
Method_replay_setup
Args: trajectory (str): path to the replay trajectory file
gui-rl/desktop_env/controllers/setup.py:480
Function_resolve_eval_script
(instance: dict)
swe-rl/generate_with_swe_trajectory.py:330
Method_ring_exchange_wrapper
(**kwargs)
Megatron-LM/megatron/core/pipeline_parallel/p2p_communication.py:348
Method_root_post_backward
(*unused)
Megatron-LM/megatron/core/distributed/fsdp/src/megatron_fsdp/megatron_fsdp.py:690
Method_root_pre_backward
Marks the module's training state as 'pre_backward' before the backprop, this function is registered on the root module. This
Megatron-LM/megatron/core/distributed/fsdp/src/megatron_fsdp/megatron_fsdp.py:755
Method_run
( self, messages: list[OpenAIMessage], response_format: type[BaseModel] | None = None,
terminal-rl/agent/camel_agent.py:277
Method_safe_div
(x, y)
Megatron-LM/megatron/core/dist_checkpointing/mapping.py:398
Method_sanity_check_server_args
(actual_server_args, expect_server_args)
slime/slime/backends/sglang_utils/sglang_engine.py:187
Method_save_state
(self, snapshot_name=None)
gui-rl/desktop_env/desktop_env.py:234
Method_scale_scroll_for_windows
(self, code: str, factor: int = 50)
gui-rl/agents/qwen3vl_agent.py:93
← previousnext →6,401–6,500 of 10,561, ranked by callers