MCPcopy Create free account

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

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

↓ 3 callersMethodhave_requests_pending
Method to check if there are requests pending. This method returns False only when there are no active requests or waiting requests.
Megatron-LM/megatron/core/inference/scheduler.py:131
↓ 3 callersMethodincrement_sequence_len_offset
Update sequence length offset. No-op for dynamic batching.
Megatron-LM/megatron/core/inference/contexts/base_context.py:30
↓ 3 callersFunctioninit_method_normal
Init method based on N(0, sigma).
Megatron-LM/megatron/core/utils.py:738
↓ 3 callersFunctioninit_process_group
(message)
Megatron-LM/tools/checkpoint/checkpoint_inspector.py:973
↓ 3 callersFunctioninit_process_group
( backend: str | Backend = None, init_method: str | None = None, timeout: timedelta | None = None,
slime/slime/utils/distributed_utils.py:38
↓ 3 callersMethodinitialize_word_embeddings
(self)
Megatron-LM/megatron/legacy/model/module.py:47
↓ 3 callersFunctionisPerfectSquare
(x)
Megatron-LM/megatron/legacy/model/vision/vit_backbone.py:50
↓ 3 callersFunctionis_deep_ep_available
()
Megatron-LM/tests/unit_tests/transformer/moe/test_token_dispatcher.py:408
↓ 3 callersFunctionis_experimental_enabled
Return the experimental flag.
Megatron-LM/megatron/core/config.py:12
↓ 3 callersFunctionis_first_microbatch_for_model_chunk
Check if an iteration is the first for a model chunk.
Megatron-LM/megatron/core/pipeline_parallel/schedules.py:1126
↓ 3 callersFunctionis_float8tensor
Check if a tensor is a Transformer Engine Float8Tensor. Note that in TE2.x, in order to support more recipes, the design of the fp8 tensor class
Megatron-LM/megatron/core/fp8_utils.py:96
↓ 3 callersFunctionis_gated_activation
Check whether the model is gated activation
Megatron-LM/megatron/core/export/trtllm/trtllm_weights_converter/utils.py:6
↓ 3 callersFunctionis_hybrid_ep_available
()
Megatron-LM/tests/unit_tests/transformer/moe/test_token_dispatcher.py:414
↓ 3 callersFunctionis_mcore_tensor_model_parallel
Check if the given parameter is Megatron-Core tensor model parallel.
Megatron-LM/megatron/core/distributed/fsdp/src/megatron_fsdp/utils.py:718
↓ 3 callersFunctionis_model_parallel_rank_0
()
Megatron-LM/tests/functional_tests/test_cases/common/ckpt_converter/__main__.py:44
↓ 3 callersFunctionis_object_storage_path
Ascertain whether a path is in object storage Args: path (str): The path Returns: bool: True if the path is in object storag
Megatron-LM/megatron/core/datasets/object_storage_utils.py:158
↓ 3 callersFunctionis_qwen35_model_path
(model_path: str)
slime/slime/backends/sglang_utils/qwen3_5.py:14
↓ 3 callersFunctionis_vp_last_stage
Return True if in the last virtual pipeline model-parallel stage, False otherwise.
Megatron-LM/megatron/core/pipeline_parallel/utils.py:34
↓ 3 callersMethodlast_token_logits
Last tokens of logits. Args: logits (Tensor): Output logits of forward pass. Return: (Tensor) Last token log
Megatron-LM/megatron/core/inference/contexts/dynamic_context.py:1366
↓ 3 callersFunctionlinear_with_frozen_weight
Linear layer execution with weight.requires_grad == False. This function handles linear layers with weight frozen (untrainable). In the forwa
Megatron-LM/megatron/core/tensor_parallel/layers.py:368
↓ 3 callersFunctionload
Load checkpoint from disk. Loads model weights and optionally optimizer state from separate directories. This allows loading weights without
slime/slime/backends/fsdp_utils/checkpoint.py:78
↓ 3 callersFunctionload_args_from_checkpoint
Set required arguments from the checkpoint specified in the arguments. Will overwrite arguments that have a non-None default value, but w
Megatron-LM/megatron/training/checkpointing.py:1262
↓ 3 callersFunctionload_docs
(corpus, doc_idxs)
slime/examples/search-r1/local_dense_retriever/retrieval_server.py:38
↓ 3 callersMethodload_parameter_state_from_dp_zero
Load parameter state (i.e., parameter & optimizer tensors) from DP 0 rank, using the new checkpoint format with coalesced state across buckets
Megatron-LM/megatron/core/optimizer/distrib_optimizer.py:1949
↓ 3 callersFunctionload_tensors_metadata
Load tensors metadata from the checkpoint. Returns a dictionary similar to a sharded state dict, but note that the dictionary keys are simply
Megatron-LM/megatron/core/dist_checkpointing/serialization.py:192
↓ 3 callersFunctionload_zarr_based_sharded_metadata
Load metadata of Zarr arrays. Args: checkpoint_dir (str): checkpoint root directory get_shape_dtype_fn (str -> ((int, ...), np.dt
Megatron-LM/megatron/core/dist_checkpointing/strategies/zarr.py:330
↓ 3 callersMethodlocate_item_in_global_item
Return the coordinates of the slice of the item that is contained in this buffer shard. In other words, this returns the coordinates
Megatron-LM/megatron/core/distributed/fsdp/src/megatron_fsdp/param_and_grad_buffer.py:1021
↓ 3 callersMethodlock
(self)
slime/tests/ci/gpu_lock_exec.py:161
↓ 3 callersFunctionmain
Run mini-SWE-agent on a GitHub issue
swe-rl/mini-swe-agent/src/minisweagent/run/github_issue.py:42
↓ 3 callersFunctionmain
Runs the text generation server with the specified model type.
Megatron-LM/tools/run_text_generation_server.py:129
↓ 3 callersFunctionmajority_vote
Return majority-voted score; ties or all-None -> 0.0.
openclaw-tinker/scorers.py:143
↓ 3 callersFunctionmake_test_packed_seq_params
(sequence_length)
Megatron-LM/tests/unit_tests/transformer/test_attention_packed_seq.py:16
↓ 3 callersFunctionmerge_dbs
Merge individual DBs into single DB. Args: project_dir (str): Retro project dir. indexed_dataset_infos (List[Dict]): Preprocessin
Megatron-LM/megatron/core/datasets/retro/db/build.py:468
↓ 3 callersFunctionmodel_provider
Build the model.
Megatron-LM/pretrain_bert.py:31
↓ 3 callersFunctionnd_flattened_tensor_reformulated_global_shape
Reformulated global shape of the flattened N-D ShardedTensor. N-D tensor global shape [X, Y, Z] and local shape [x, y, z] is reformulated int
Megatron-LM/megatron/core/dist_checkpointing/strategies/resharding.py:60
↓ 3 callersFunctionnormalize_answer
(s)
Megatron-LM/tools/retro/text_generation/evaluate.py:17
↓ 3 callersFunctionnormalize_final_answer
Normalize a final answer to a quantitative reasoning question. Args: final_answer: The answer string to normalize Returns: N
slime/slime/rollout/rm_hub/math_dapo_utils.py:143
↓ 3 callersFunctionnormalize_qwen3vl_key
(global_name: str)
slime/tools/convert_hf_to_torch_dist.py:31
↓ 3 callersFunctionnormalize_str
Normalize the str to lower case and make them float numbers if possible.
Megatron-LM/examples/multimodal/evaluation/mmmu_utils.py:218
↓ 3 callersMethodnum_requests_pending
Get the number of requests pending. This method returns the number of active + waiting requests.
Megatron-LM/megatron/core/inference/scheduler.py:124
↓ 3 callersMethodoffload
(self)
slime/slime/ray/rollout.py:179
↓ 3 callersFunctionopen_ts_array
Opens a Zarr file array with Tensorstore with basic setting. Args: arr_path (Path): path to a Zarr (Tensorstore) array
Megatron-LM/megatron/core/dist_checkpointing/strategies/tensorstore.py:133
↓ 3 callersFunctionoptim_state_to_sharding_state
Turn optimizer state dict to sharded state dict based on model state dict *in-place*. Can be used to add sharding information to most common opti
Megatron-LM/megatron/core/dist_checkpointing/optimizer.py:109
↓ 3 callersMethodoptimal_dtype
Get the dtype to use for an index of a certain cardinality Args: cardinality (Optional[int]): The number of elements to be indexe
Megatron-LM/megatron/core/datasets/indexed_dataset.py:106
↓ 3 callersMethodpad_id
Returns id of padding token. Need for NeMo.
Megatron-LM/megatron/core/tokenizers/text/text_tokenizer.py:176
↓ 3 callersFunctionparallel_lm_logits
LM logits using word embedding weights.
Megatron-LM/megatron/legacy/model/language_model.py:22
↓ 3 callersFunctionparse_and_normalize_split
Parse the dataset split ratios from a string Args: split (str): The train valid test split string e.g. "99,1,0" Returns: Lis
Megatron-LM/megatron/core/datasets/blended_megatron_dataset_config.py:134
↓ 3 callersMethodparse_from_config
Parses quantization config for a layer or throw an error.
Megatron-LM/megatron/core/extensions/kitchen.py:131
↓ 3 callersFunctionparse_multi_choice_response
Parse the prediction from the generated response. Return the predicted index e.g., A, B, C, D.
Megatron-LM/examples/multimodal/evaluation/mmmu_utils.py:151
↓ 3 callersFunctionparse_prm_eval_score
Extract \\boxed{N} score for PRM eval (N in {+1, -1, 0}).
openclaw-tinker/scorers.py:123
↓ 3 callersFunctionparse_s3_path
Parses the given S3 path returning correspsonding bucket and key. Args: path (str): The S3 path Returns: Tuple[str, str]: A
Megatron-LM/megatron/core/datasets/object_storage_utils.py:195
↓ 3 callersMethodpause_engines
Sends a signal to pause all inference engines. The signal first propagates thru the coordinator to all engines. All engines acknowled
Megatron-LM/megatron/core/inference/inference_client.py:194
↓ 3 callersFunctionpermute
Permute the tokens and probs based on the mask. Tokens with the same designated expert will be grouped together. The shape of mask is [tokens,
Megatron-LM/megatron/core/transformer/moe/moe_utils.py:227
↓ 3 callersFunctionpost_all_gather_processing
Post-processing after all-gather for weights in distributed optimizer. - tensorwise: may need to create a transposed view to match backend GE
Megatron-LM/megatron/core/fp8_utils.py:499
↓ 3 callersFunctionprefix_product
(a: List[int], init=1)
Megatron-LM/megatron/core/parallel_state.py:299
↓ 3 callersMethodprepare_request
( self, prompts: list[str], generation_args: GenericGenerationArgs )
Megatron-LM/megatron/rl/inference/inference_interface.py:38
↓ 3 callersFunctionpreprocess_state_dict_for_uneven_dtensor
Preprocess the state_dict to prepare it for saving or loading unevenly sharded DTensors. This function modifies the DTensors in the state_dic
Megatron-LM/megatron/core/distributed/fsdp/src/megatron_fsdp/uneven_dtensor.py:234
↓ 3 callersFunctionprint_split_stats
(name, index)
Megatron-LM/megatron/legacy/data/dataset_utils.py:487
↓ 3 callersMethodpurge_record_files
(self)
openclaw-tinker/api_server.py:331
↓ 3 callersMethodreceive_lm
(self, schema)
Megatron-LM/tools/checkpoint/saver_hf_llava.py:258
↓ 3 callersMethodrecord_current_stream
Records the current CUDA stream in the event.
Megatron-LM/megatron/core/models/common/model_chunk_schedule_plan.py:396
↓ 3 callersFunctionreduce_max_stat_across_model_parallel_group
Ranks without an optimizer will have no grad_norm or num_zeros_in_grad stats. We need to ensure the logging and writer rank has those values.
Megatron-LM/megatron/training/utils.py:243
↓ 3 callersMethodregister_grad_ready
Registers grads for the passed-in param to be "ready" for grad sync. When the number of microbatches is greater than 1, we only want
Megatron-LM/megatron/core/distributed/param_and_grad_buffer.py:523
↓ 3 callersMethodrelease_memory_blocks
Release memory blocks. Args: blocks (Tensor): Block IDs to release. Return: None
Megatron-LM/megatron/core/inference/contexts/dynamic_block_allocator.py:99
↓ 3 callersFunctionreload_process_groups
Reload all reloadable process groups.
slime/slime/utils/reloadable_process_group.py:270
↓ 3 callersFunctionremove_padding
Remove vocab padding: param[:vocab_size] for embedding/output layers, else unchanged.
slime/slime/backends/megatron_utils/megatron_to_hf/processors/padding_remover.py:6
↓ 3 callersMethodreport
Function to log the min/max metircs and the associated rank over a time period It finds the slowest and fastest rank among all ranks. It shou
Megatron-LM/megatron/core/utils.py:1577
↓ 3 callersMethodrequires_explicit_ckpt_mismatch_check
Whether a given strict flag involves mismatch check against the checkpoint.
Megatron-LM/megatron/core/dist_checkpointing/validation.py:90
↓ 3 callersMethodreset
(self, lease_id: str, task_config: dict[str, Any] | None)
gui-rl/env_client.py:35
↓ 3 callersMethodreset
Reset entire context. This method does: - Reset active/paused request/token counts to zero. - Reset available blocks to entir
Megatron-LM/megatron/core/inference/contexts/dynamic_context.py:1290
↓ 3 callersMethodreset_attention_state
Reset state used within attention, after each step.
Megatron-LM/megatron/core/inference/contexts/dynamic_context.py:964
↓ 3 callersMethodreset_eval_scores
(self)
openclaw-tinker/rollout.py:95
↓ 3 callersMethodreset_mamba_state
Reset state used within Mamba layers.
Megatron-LM/megatron/core/inference/contexts/dynamic_context.py:976
↓ 3 callersFunctionresize
(input, size=None, scale_factor=None, mode='nearest', align_corner
Megatron-LM/megatron/legacy/model/vision/utils.py:7
↓ 3 callersFunctionretry_with_backoff
Retry a function with exponential backoff.
Megatron-LM/tests/test_utils/python_scripts/auto_reminder.py:60
↓ 3 callersFunctionrouter_gating_linear
Customized linear layer for router gating. This linear layer accepts bfloat16 input and weight, and can return output with router_dtype.
Megatron-LM/megatron/core/transformer/moe/moe_utils.py:1015
↓ 3 callersMethodrun
Schedule one-forward-one-backward operations for a single transformer layer. This function interleaves forward and backward operations, overl
Megatron-LM/megatron/core/models/common/model_chunk_schedule_plan.py:211
↓ 3 callersMethodrun_engine
Main functionality to run inference Runs the engine until there are no requests in the queue. Args: dynamic_generation (
Megatron-LM/megatron/core/inference/engines/static_engine.py:341
↓ 3 callersMethodrun_engine_async
Runs the engine asynchronously using asyncio
Megatron-LM/megatron/core/inference/engines/static_engine.py:389
↓ 3 callersFunctionrun_mcore_engine
Server-compatible version of the MCore Engine, used in tools/run_text_generation_server.py.
Megatron-LM/megatron/core/inference/text_generation_server/run_mcore_engine.py:12
↓ 3 callersFunctionsample
()
Megatron-LM/tests/unit_tests/utils/test_experimental_log_once.py:25
↓ 3 callersMethodsave_loss_to_tracker
Save the mtp loss for logging. Args: loss (torch.Tensor): The loss tensor. layer_number (int): Layer index of the loss
Megatron-LM/megatron/core/transformer/multi_token_prediction.py:319
↓ 3 callersFunctionscaled_init_method_normal
Init method based on N(0, sigma/sqrt(2*num_layers).
Megatron-LM/megatron/legacy/model/utils.py:21
↓ 3 callersFunctionscatter_to_sequence_parallel_region
Wrapper for autograd function: forward: split, backward: AG <last dim>
Megatron-LM/megatron/core/tensor_parallel/mappings.py:493
↓ 3 callersFunctionscatter_to_tensor_model_parallel_region
Wrapper for autograd function: forward: RS, backward: AG <last dim>
Megatron-LM/megatron/core/tensor_parallel/mappings.py:481
↓ 3 callersMethodschedule_async_request
Start a new async call and add it to a queue of active async calls. This method must be called on all ranks. Args: async
Megatron-LM/megatron/core/dist_checkpointing/strategies/async_utils.py:528
↓ 3 callersMethodsend_backward
Send backward activation tensor. Args: grad_dict: A dictionary mapping module names to tensors.
Megatron-LM/megatron/core/pipeline_parallel/multimodule_communicator.py:400
↓ 3 callersMethodsend_backward_recv_forward
Send backward activation tensor and receive forward activation tensor. Args: grad_dict: A dictionary mapping module names to tens
Megatron-LM/megatron/core/pipeline_parallel/multimodule_communicator.py:330
↓ 3 callersMethodserialize
Return a dictionary that is msgpack-serializable.
Megatron-LM/megatron/core/inference/sampling_params.py:47
↓ 3 callersFunctionset_cache_control
This messages processor adds manual cache control marks to the messages.
swe-rl/mini-swe-agent/src/minisweagent/models/utils/cache_control.py:32
↓ 3 callersMethodset_layer_number
Set the layer number for the router.
Megatron-LM/megatron/core/transformer/moe/router.py:125
↓ 3 callersFunctionset_model_to_sequence_parallel
Set sequence parallel attributes for the model. Args: set_to: Value to set for sequence_parallel attributes exclude_modules:
Megatron-LM/megatron/core/transformer/utils.py:310
↓ 3 callersMethodset_rank
(self, rank)
Megatron-LM/tools/checkpoint/utils.py:66
↓ 3 callersFunctionset_rounder
Utility function to set the DynamicInferenceContext rounder.
Megatron-LM/tests/unit_tests/inference/test_wandb_logging.py:21
↓ 3 callersFunctionset_streams
Set the streams for communication and computation
Megatron-LM/megatron/core/pipeline_parallel/utils.py:320
↓ 3 callersFunctionset_tp_pp_vpp
(tp, pp, vpp=None, pp_layout=None, destroy_first=True)
Megatron-LM/tests/unit_tests/dist_checkpointing/test_pipeline_parallel_layout.py:294
↓ 3 callersMethodsetup_method
(self, method)
Megatron-LM/tests/unit_tests/transformer/test_cuda_graphs.py:298
↓ 3 callersMethodsetup_model
(self, tensor_parallel_size, pipeline_parallel_size)
Megatron-LM/tests/unit_tests/inference/model_inference_wrappers/gpt/test_gpt_inference_wrapper.py:28
↓ 3 callersFunctionsharded_tensor_chunk_id
Id of a sharded tensor.
Megatron-LM/megatron/core/dist_checkpointing/strategies/two_stage.py:67
↓ 3 callersMethodsleep
(self)
slime/slime/backends/megatron_utils/actor.py:375
← previousnext →1,301–1,400 of 10,561, ranked by callers