Code
Hub
Workspaces
Following
Trending
Connect
MCP
copy
Create free account
hub
/
github.com/Gen-Verse/OpenClaw-RL
/ functions
Functions
10,561 in github.com/Gen-Verse/OpenClaw-RL
⨍
Functions
10,561
◇
Types & classes
1,405
↳
Endpoints
144
↓ 1 callers
Function
_init_sequence_parallel_cache
Initialize the cache of modules with sequence parallel attributes. Only needs to be called once, subsequent calls have no effect. Args:
Megatron-LM/megatron/core/transformer/utils.py:261
↓ 1 callers
Method
_initalize
(self, vocab_extra_ids)
Megatron-LM/megatron/training/tokenizer/tokenizer.py:388
↓ 1 callers
Method
_initialize
Actual initialization logic
slime/slime/utils/tensorboard_utils.py:41
↓ 1 callers
Function
_initialize_distributed
Initialize torch.distributed and core model parallel.
Megatron-LM/megatron/training/initialize.py:311
↓ 1 callers
Function
_initialize_distributed
Initialize torch.distributed and core model parallel.
slime/slime/backends/megatron_utils/initialize.py:33
↓ 1 callers
Method
_initialize_environment
Initialize the environment and get initial observation. Args: env: Tau-bench environment instance task_index
slime/examples/tau-bench/trainable_agents.py:128
↓ 1 callers
Method
_initialize_language_model
Initialize the language model.
Megatron-LM/megatron/core/models/mimo/model/base.py:153
↓ 1 callers
Method
_initialize_metadata
Initialize the routing map and probs to a unified format covering the TPxEP group. This design decouples the communication group from
Megatron-LM/megatron/core/transformer/moe/token_dispatcher.py:1379
↓ 1 callers
Function
_initialize_resources
(args: Any, sample: Sample)
slime/examples/geo3k_vlm_multi_turn/rollout.py:141
↓ 1 callers
Method
_initialize_submodules
Initialize modality submodules from the ModuleSpec configurations. Only modalities present in the config will be instantiated. For ea
Megatron-LM/megatron/core/models/mimo/model/base.py:137
↓ 1 callers
Method
_inject_experience_to_messages
Inject accumulated experience into the last user message.
openclaw-rl/oel/openclaw_oel_api_server.py:1013
↓ 1 callers
Function
_inject_implicit_mixed_number
Automatically make a mixed number evalable e.g. 7 3/4 => 7+3/4
slime/slime/rollout/rm_hub/math_utils.py:229
↓ 1 callers
Method
_insert
Insert tokens - skip tokens for existing nodes just like we skip text.
slime/slime/router/middleware_hub/radix_tree.py:276
↓ 1 callers
Method
_insert_sharded_data
( self, fully_parallel, sharded_part, parallelization_group, exchange_algo )
Megatron-LM/megatron/core/dist_checkpointing/tensor_aware_state_dict.py:276
↓ 1 callers
Function
_install_vm
(vm_name, vms_dir, downloaded_file_name, original_vm_name="Ubuntu", bridged_adapter_name=None)
gui-rl/desktop_env/providers/virtualbox/manager.py:47
↓ 1 callers
Function
_install_vm
(vm_name, vms_dir, downloaded_file_name, os_type, original_vm_name="Ubuntu")
gui-rl/desktop_env/providers/vmware/manager.py:119
↓ 1 callers
Function
_intersection
(s1, s2)
Megatron-LM/megatron/core/distributed/fsdp/src/megatron_fsdp/uneven_dtensor.py:383
↓ 1 callers
Method
_is_chinese_char
Checks whether CP is the codepoint of a CJK character.
Megatron-LM/megatron/training/tokenizer/bert_tokenization.py:296
↓ 1 callers
Function
_is_config_like_path
(path: str)
swe-rl/generate_with_swe_trajectory.py:234
↓ 1 callers
Function
_is_control
Checks whether `chars` is a control character.
Megatron-LM/megatron/training/tokenizer/bert_tokenization.py:406
↓ 1 callers
Function
_is_cuda
Check if a tensor is not none and is cuda.
Megatron-LM/megatron/core/inference/communication_utils.py:28
↓ 1 callers
Function
_is_dir_nonempty
(path)
slime/slime/backends/megatron_utils/checkpoint.py:171
↓ 1 callers
Function
_is_float
(num: str)
slime/slime/rollout/rm_hub/math_utils.py:195
↓ 1 callers
Method
_is_in_embd_group
(self)
Megatron-LM/megatron/core/models/common/language_module/language_module.py:66
↓ 1 callers
Function
_is_int
(x: float)
slime/slime/rollout/rm_hub/math_utils.py:203
↓ 1 callers
Function
_is_megatron_checkpoint
(path: str | Path)
slime/slime/backends/megatron_utils/checkpoint.py:127
↓ 1 callers
Function
_is_not_found
(e: Exception)
gui-rl/desktop_env/providers/volcengine/manager.py:107
↓ 1 callers
Function
_is_oom_error
(exc: BaseException)
slime/slime/utils/ppo_utils.py:649
↓ 1 callers
Function
_is_punctuation
Checks whether `chars` is a punctuation character.
Megatron-LM/megatron/training/tokenizer/bert_tokenization.py:418
↓ 1 callers
Function
_is_quantized_weight_key
Check if the key is a quantized MoE expert weight key.
slime/tools/convert_k2_thinking_int4_to_bf16.py:87
↓ 1 callers
Method
_is_sink_module
Check if a module is a sink module (has no outgoing connections).
Megatron-LM/megatron/core/pipeline_parallel/multimodule_communicator.py:146
↓ 1 callers
Method
_is_source_module
Check if a module is a source module (has no incoming connections).
Megatron-LM/megatron/core/pipeline_parallel/multimodule_communicator.py:138
↓ 1 callers
Function
_is_test_like_path
(path: str)
swe-rl/generate_with_swe_trajectory.py:223
↓ 1 callers
Method
_is_two_bucket_group_equal
(self, group_a, group_b)
Megatron-LM/megatron/core/distributed/fsdp/src/megatron_fsdp/param_and_grad_buffer.py:695
↓ 1 callers
Function
_is_valid_git_patch
(patch_text: str)
swe-rl/server/swe_exec_server.py:47
↓ 1 callers
Method
_is_valid_image_response
Quick validation for PNG/JPEG payload using magic bytes; Content-Type is advisory. Returns True only when bytes look like a real PNG or JPEG.
gui-rl/desktop_env/controllers/python.py:25
↓ 1 callers
Function
_is_valid_rl_score
(score)
openclaw-fireworks/fireworks_server.py:244
↓ 1 callers
Method
_is_valid_rl_score
(score)
openclaw-combine/openclaw_combine_api_server.py:34
↓ 1 callers
Method
_is_valid_rl_score
(score)
openclaw-tinker/api_server.py:957
↓ 1 callers
Function
_is_whitespace
Checks whether `chars` is a whitespace character.
Megatron-LM/megatron/training/tokenizer/bert_tokenization.py:394
↓ 1 callers
Function
_is_zero_std
(samples: list[Sample])
slime/slime/ray/rollout.py:1283
↓ 1 callers
Function
_item_size
Calculates size (in bytes) of a single write item. Same as torch.distributed.checkpoint.filesystem._item_size, but fixes computing chunk
Megatron-LM/megatron/core/dist_checkpointing/strategies/filesystem_async.py:613
↓ 1 callers
Function
_iter_meta
(meta_path: Path)
gui-rl/build_gui_prompt_dataset.py:8
↓ 1 callers
Function
_judge_step_with_prm
( args, sample: Sample, *, step_index: int, action: str, observation: str, history
toolcall-rl/generate_with_retool.py:496
↓ 1 callers
Method
_keepalive
(self)
openclaw-tinker/rollout.py:69
↓ 1 callers
Method
_key_config_attributes
Return all config attributes which contribute to uniquely identifying the dataset. These attributes will be used to build a uniquely identify
Megatron-LM/megatron/core/datasets/megatron_dataset.py:155
↓ 1 callers
Method
_kill_engine
(self, rollout_engine_id: int)
slime/slime/utils/health_monitor.py:161
↓ 1 callers
Method
_launch_setup
(self, command: Union[str, List[str]], shell: bool = False)
gui-rl/desktop_env/controllers/setup.py:316
↓ 1 callers
Function
_launch_sglang_engine
()
slime/tests/test_external_rollout.py:160
↓ 1 callers
Function
_launch_sglang_router
()
slime/tests/test_external_rollout.py:138
↓ 1 callers
Function
_layer_counts_match
(a: list, b: list)
Megatron-LM/megatron/core/ssm/mamba_hybrid_layer_allocation.py:88
↓ 1 callers
Function
_list_dir_contents
List the contents of a directory recursively, building a tree structure. :param directory: The path of the directory to inspect.
gui-rl/desktop_env/server/main.py:1092
↓ 1 callers
Function
_load_checkpoint
(queue, args)
Megatron-LM/tools/checkpoint/loader_mixtral_hf.py:152
↓ 1 callers
Function
_load_checkpoint
(queue, args)
Megatron-LM/tools/checkpoint/loader_llama_mistral.py:406
↓ 1 callers
Method
_load_checkpoint_args
Load arguments from checkpoint to margs.
Megatron-LM/tools/checkpoint/saver_base.py:55
↓ 1 callers
Function
_load_checkpoint_hf
(ddp_model, optimizer, args, load_path: str)
slime/slime/backends/megatron_utils/checkpoint.py:133
↓ 1 callers
Function
_load_config
Read config.json and return hidden_size, inter_size, and group_size.
slime/tools/convert_k2_thinking_int4_to_bf16.py:38
↓ 1 callers
Function
_load_env_module
Load the interaction environment module from a module path or a file path.
slime/examples/geo3k_vlm_multi_turn/rollout.py:27
↓ 1 callers
Function
_load_evaluation_lib
()
slime/slime/rollout/rm_hub/ifbench.py:69
↓ 1 callers
Method
_load_experience_list
Load multiple experience files from an experience_list.txt. Each line is a path to an experience file. All are loaded into the experi
openclaw-rl/oel/openclaw_oel_api_server.py:477
↓ 1 callers
Function
_load_from_array
(sharded_tensor: ShardedTensor, checkpoint_dir: Path)
Megatron-LM/megatron/core/dist_checkpointing/strategies/zarr.py:233
↓ 1 callers
Function
_load_fsdp_state_dict
(input_dir: str)
slime/tools/convert_fsdp_to_hf.py:62
↓ 1 callers
Function
_load_hf_prm
(path: str, device: torch.device)
openclaw-combine/prm_teacher_postprocess.py:68
↓ 1 callers
Method
_load_initial_experience
Load experience from file(s) on startup. Supports: - OPENCLAW_OEL_EXPERIENCE_PATH: single experience file or experience_list.txt
openclaw-rl/oel/openclaw_oel_api_server.py:448
↓ 1 callers
Function
_load_non_persistent_base_checkpoint
Load the base state_dict from a non-persistent distributed checkpoint. Depending on the non_persistent_ckpt_type, different logic may be required
Megatron-LM/megatron/training/checkpointing.py:1003
↓ 1 callers
Function
_load_regular_chunk
(sharded_tensor: ShardedTensor, checkpoint_dir: Path)
Megatron-LM/megatron/core/dist_checkpointing/strategies/tensorstore.py:111
↓ 1 callers
Function
_load_rng_state_dict
(rng_state_dict)
Megatron-LM/megatron/core/distributed/fsdp/mcore_fsdp_adapter.py:430
↓ 1 callers
Method
_load_scaling_factors
Loads scaling factors from model state dictionary. Args: model_state_dict (dict): Model state dictionary Returns:
Megatron-LM/megatron/core/export/trtllm/trtllm_helper.py:210
↓ 1 callers
Function
_load_state_if_exists
()
Megatron-LM/megatron/training/ft_integration.py:217
↓ 1 callers
Function
_load_teacher_model_config
Reads teacher config from a file. The config provided via --teacher-model-config should specify (in NEMO format) any model architecture setti
Megatron-LM/megatron/post_training/model_builder.py:45
↓ 1 callers
Method
_load_tokenizer
(self)
openclaw-tinker/api_server.py:176
↓ 1 callers
Function
_local_lse
Numerically-stable lse over the True positions of ``mask``. Both tensors are ``[R, K]``. Returns ``[R, 1]``. If the entire row's mask is all-
openclaw-combine/hint_opd_loss.py:276
↓ 1 callers
Function
_local_softmax
Softmax over the True positions of ``mask``. Both tensors are ``[R, K]``. Returns ``[R, K]``: off-mask entries are 0 and rows where the mask
openclaw-combine/hint_opd_loss.py:294
↓ 1 callers
Function
_log_eval_rollout_data
(rollout_id, args, data, extra_metrics: dict[str, Any] | None = None)
slime/slime/ray/rollout.py:1178
↓ 1 callers
Method
_log_parameter_groups
Compact log of FSDP parameter groups and their parameters.
Megatron-LM/megatron/core/distributed/fsdp/src/megatron_fsdp/param_and_grad_buffer.py:1718
↓ 1 callers
Function
_log_rollout_data
(rollout_id, args, samples, rollout_extra_metrics, rollout_time)
slime/slime/ray/rollout.py:1215
↓ 1 callers
Method
_log_rollout_data
(self, rollout_id: int, rollout_data, packed_batches)
slime/slime/backends/fsdp_utils/actor.py:494
↓ 1 callers
Method
_log_step
(self, batch: list[TrainingSample], step: int, method: str)
openclaw-tinker/trainer.py:217
↓ 1 callers
Function
_main_async
()
slime/slime/utils/debug_utils/send_to_sglang.py:28
↓ 1 callers
Function
_main_async
(samples, custom_rm_path)
slime/slime/utils/debug_utils/replay_reward_fn.py:34
↓ 1 callers
Function
_majority_vote
(scores: list[int | None])
openclaw-rl/openclaw_api_server.py:149
↓ 1 callers
Method
_make_activation_op
Construct activation op.
Megatron-LM/megatron/core/extensions/transformer_engine.py:1803
↓ 1 callers
Method
_make_forward_pre_hook
Create a forward pre-hook to wait on all-gather handles when necessary (i.e., when a module uses a parameter in a bucket with a still
Megatron-LM/megatron/core/distributed/distributed_data_parallel.py:383
↓ 1 callers
Method
_make_fused_impl
Construct fused module matching MLP.
Megatron-LM/megatron/core/extensions/transformer_engine.py:1677
↓ 1 callers
Method
_make_prompt_samples
(self, num_samples: int)
gui-rl/gui_data_source.py:72
↓ 1 callers
Method
_make_prompt_samples
(self, num_samples: int)
gui-rl/gui_data_source.py:212
↓ 1 callers
Function
_make_run_context
(sample: Sample)
terminal-rl/generate.py:20
↓ 1 callers
Function
_match_answer_boxed
Match 'Answer: \\boxed{...}' with arbitrary brace nesting depth. The previous regex ``(?:[^{}]|\\{[^{}]*\\})*`` only handled one level of nes
toolcall-rl/generate_with_retool.py:189
↓ 1 callers
Function
_matches
Returns true if passed-in parameter (with name) matches `param_key`. Args: param (torch.nn.Parameter): Handle to parameter object.
Megatron-LM/megatron/core/optimizer/__init__.py:58
↓ 1 callers
Function
_materialize_response
Convert streaming-like Response into a regular Response/JSONResponse safely.
slime/slime/router/middleware_hub/radix_tree_middleware.py:35
↓ 1 callers
Function
_max_cand
()
openclaw-combine/openclaw_combine_select_api_server.py:79
↓ 1 callers
Function
_maybe_clear_task_result_dir
(args: Any, domain: str, example_id: str, sample: Sample)
gui-rl/generate_with_gui.py:177
↓ 1 callers
Method
_maybe_ensure_additional_required_arguments
Can be used to ensure some expected args are present. For instance, use self.check_for_arg('some_arg')
Megatron-LM/tools/checkpoint/loader_base.py:91
↓ 1 callers
Function
_maybe_get_cpu_backup
(x: torch.Tensor)
slime/slime/backends/megatron_utils/update_weight/common.py:149
↓ 1 callers
Method
_maybe_parse_additional_megatron_args
Method used to optionally add arguments from the checkpoint to the main args. For instance, using margs.some_arg = checkpoint_args.so
Megatron-LM/tools/checkpoint/loader_base.py:30
↓ 1 callers
Method
_maybe_parse_additional_megatron_args
Method used to optionally add arguments from the checkpoint to the main args. For instance, using margs.some_arg = checkpoint_args.so
Megatron-LM/tools/checkpoint/saver_base.py:31
↓ 1 callers
Method
_maybe_report_stats
Internal method that reports stats if needed.
Megatron-LM/megatron/core/rerun_state_machine.py:919
↓ 1 callers
Function
_mcore_to_dcp_compatible_tensor
Build a PyT ShardedTensor from given shards. During loading: - if data is None, initialize it with an empty tensor (will be used to c
Megatron-LM/megatron/core/dist_checkpointing/strategies/torch.py:320
↓ 1 callers
Function
_mcore_to_torch_sharded_object
Build io.BytesIO from given sharded objects data.
Megatron-LM/megatron/core/dist_checkpointing/strategies/torch.py:367
↓ 1 callers
Function
_merge_multimodal_train_inputs
Merge per-turn multimodal_train_inputs with a single concat per key. Note: Only torch.Tensor values are merged; non-tensor fields are ignore
slime/examples/geo3k_vlm_multi_turn/rollout.py:115
← previous
next →
3,001–3,100 of 10,561, ranked by callers