MCPcopy Create free account

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

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

↓ 1 callersMethodadd_document
Add an entire document to the dataset Args: tensor (torch.Tensor): The document to add lengths (List[int]): The leng
Megatron-LM/megatron/core/datasets/indexed_dataset.py:871
↓ 1 callersMethodadd_dummy_requests_for_cudagraph_capture
Adds dummy requests to reflect the number of prefill and decode requests in the graph config. These are using during cuda graph captu
Megatron-LM/megatron/core/inference/contexts/dynamic_context.py:1118
↓ 1 callersMethodadd_earliest_waiting_request_to_active_pool
Utility to add the waiting request to active pool This method will add the earliest request (FIFO) that is in the waiting request poo
Megatron-LM/megatron/core/inference/scheduler.py:138
↓ 1 callersMethodadd_embed_data
Add the embedding of each block to the underlying FAISS index
Megatron-LM/megatron/legacy/data/realm_index.py:187
↓ 1 callersFunctionadd_eval_arguments
(parser)
slime/slime/utils/arguments.py:689
↓ 1 callersMethodadd_event_fail
Add 'fail' event.
Megatron-LM/megatron/core/inference/inference_request.py:349
↓ 1 callersMethodadd_event_finish
Add 'finish' event.
Megatron-LM/megatron/core/inference/inference_request.py:345
↓ 1 callersMethodadd_event_pause
Add 'pause' event.
Megatron-LM/megatron/core/inference/inference_request.py:341
↓ 1 callersFunctionadd_fault_tolerance_arguments
(parser)
slime/slime/utils/arguments.py:505
↓ 1 callersFunctionadd_file_handler
(path: Path | str, level: int = logging.DEBUG, *, print_path: bool = True)
swe-rl/mini-swe-agent/src/minisweagent/utils/log.py:21
↓ 1 callersFunctionadd_megatron_arguments
Add Megatron-LM arguments to the given parser.
Megatron-LM/megatron/training/arguments.py:52
↓ 1 callersMethodadd_message
(self, role: str, content: str, **kwargs)
swe-rl/mini-swe-agent/src/minisweagent/agents/interactive.py:42
↓ 1 callersMethodadd_message
(self, role: str, content: str, **kwargs)
swe-rl/mini-swe-agent/src/minisweagent/agents/interactive_textual.py:47
↓ 1 callersFunctionadd_mtp_training_arguments
Add MTP training specific arguments.
slime/slime/utils/arguments.py:1588
↓ 1 callersFunctionadd_multimodal_extra_args
Extra arguments.
Megatron-LM/examples/multimodal/multimodal_args.py:5
↓ 1 callersFunctionadd_network_arguments
(parser)
slime/slime/utils/arguments.py:1285
↓ 1 callersFunctionadd_ppl_metrics
( train_log_prob: torch.Tensor, rollout_log_prob: torch.Tensor, loss_mask: torch.Tensor, metri
slime/examples/train_infer_mismatch_helper/mis.py:383
↓ 1 callersFunctionadd_prefill_decode_disaggregation_arguments
(parser)
slime/slime/utils/arguments.py:1601
↓ 1 callersMethodadd_request
Add request to context. At this stage, we assume that the request is valid and can be added, as the checks are done in the schedule function.
Megatron-LM/megatron/core/inference/contexts/dynamic_context.py:1412
↓ 1 callersMethodadd_result
(self, filename, q_weights, module_names)
slime/tools/convert_hf_to_fp8.py:106
↓ 1 callersMethodadd_result
(self, filename, q_weights)
slime/tools/convert_hf_to_int4_direct.py:160
↓ 1 callersFunctionadd_reward_model_arguments
(parser)
slime/slime/utils/arguments.py:1290
↓ 1 callersFunctionadd_rollout_arguments
(parser)
slime/slime/utils/arguments.py:220
↓ 1 callersFunctionadd_rollout_buffer_arguments
(parser)
slime/slime/utils/arguments.py:1492
↓ 1 callersFunctionadd_router_arguments
(parser)
slime/slime/utils/arguments.py:1078
↓ 1 callersFunctionadd_sglang_arguments
Add arguments to the parser for the SGLang server.
slime/slime/backends/sglang_utils/arguments.py:31
↓ 1 callersFunctionadd_sglang_router_arguments
Add arguments to the parser for the SGLang router.
slime/slime/backends/sglang_utils/arguments.py:6
↓ 1 callersFunctionadd_sglang_tp_size
()
slime/slime/utils/arguments.py:1641
↓ 1 callersMethodadd_special_tokens
Adds a dictionary of special tokens (eos, pad, cls...). If special tokens are NOT in the vocabulary, they are added to it (in
Megatron-LM/megatron/core/tokenizers/text/libraries/huggingface_tokenizer.py:179
↓ 1 callersMethodadd_special_tokens
Adds special tokens to the tokenizer.
Megatron-LM/megatron/core/tokenizers/text/libraries/sentencepiece_tokenizer.py:275
↓ 1 callersFunctionadd_tensorboard_arguments
(parser)
slime/slime/utils/arguments.py:1185
↓ 1 callersMethodadd_tokens
Add tokens to the tokenizer. No-op for this mock tokenizer.
Megatron-LM/tests/unit_tests/test_tokenizer.py:230
↓ 1 callersFunctionadd_train_arguments
(parser)
slime/slime/utils/arguments.py:112
↓ 1 callersMethodadd_vm
Add the path of new VM to the registration.
gui-rl/desktop_env/providers/base.py:58
↓ 1 callersMethodadd_vm
(self, vm_path, region)
gui-rl/desktop_env/providers/azure/manager.py:32
↓ 1 callersFunctionadd_wandb_arguments
(parser)
slime/slime/utils/arguments.py:1113
↓ 1 callersMethodadetect_tool_calls_parse_error
r""" Asynchronously detect tool calls in the response content using Qwen25Detector. if the model is Qwen 2.5 or Qwen 3. if the
terminal-rl/agent/camel_agent.py:436
↓ 1 callersFunctionadjust_non_strict_load
Adjusts sharded state dict removing keys not existing in the checkpoint. Args: sharded_state_dict (ShardedStateDict): sharded state dict
Megatron-LM/megatron/core/dist_checkpointing/validation.py:262
↓ 1 callersFunctionagent_factory
( tools_info: list[dict[str, Any]], wiki, config: RunConfig, rollout_args: dict[str, Any] | No
slime/examples/tau-bench/trainable_agents.py:456
↓ 1 callersFunctionalign_unpacked_inference_logprobs
Align inference logprobs with old_logprobs for unpacked sequences and compute statistics. Args: inference_logprobs: List of inference log
Megatron-LM/megatron/rl/rl_utils.py:278
↓ 1 callersFunctionall_gather_item
(item, dtype, group=None, async_op=False, local_rank=None)
Megatron-LM/megatron/training/dist_signal_handler.py:28
↓ 1 callersMethodall_gather_params
All-gather the params. If prefetch is enabled, prefetch next buckets in the order of `prefetch_order`. Args: params (List
Megatron-LM/megatron/core/distributed/fsdp/src/megatron_fsdp/param_and_grad_buffer.py:3348
↓ 1 callersFunctionall_gather_params_async
Parallel TP all-gather for multiple params. Loop 1: for each TP param, allocate buffers + dist.all_gather(async_op=True) on expert-TP/regular
slime/slime/backends/megatron_utils/update_weight/common.py:76
↓ 1 callersMethodall_reduce_gradients
All reduce the gradients. Args: async_op (bool, optional): Whether to do the all-reduce asynchronously. Defaults t
Megatron-LM/megatron/core/distributed/fsdp/src/megatron_fsdp/param_and_grad_buffer.py:2841
↓ 1 callersMethodallgather_params
All-gather updated params from all ranks.
Megatron-LM/megatron/core/optimizer/layer_wise_optimizer.py:147
↓ 1 callersMethodallocate
( self, task_key: str, request_id: str | None = None, )
terminal-rl/env_client.py:30
↓ 1 callersMethodallocate
( self, task_key: str, request_id: str | None = None )
terminal-rl/remote/pool_server.py:181
↓ 1 callersMethodallocate
(self, episode_id: str)
gui-rl/env_client.py:24
↓ 1 callersMethodallocate
(self, episode_id: str | None = None)
gui-rl/env_pool_server.py:173
↓ 1 callersMethodallocate
(self, image: str, instance_id: str, cwd: str = "/testbed")
swe-rl/server/swe_env_pool_server.py:106
↓ 1 callersFunctionanls_score
( prediction: str, gold_labels: List[str], threshold: float = 0.5 )
Megatron-LM/examples/multimodal/evaluation/evaluate_vqav2.py:38
↓ 1 callersFunctionappend_hint_to_messages
Append a hindsight hint to the last user message.
openclaw-tinker/scorers.py:164
↓ 1 callersMethodappend_key_value_cache
Append to KV cache. Args: layer_number (int): Layer number. key (Tensor): Key tensor. value (Tensor): Val
Megatron-LM/megatron/core/inference/contexts/dynamic_context.py:775
↓ 1 callersFunctionapply_fsdp_moe_patch
()
slime/slime/backends/fsdp_utils/models/qwen3_moe_hf.py:5
↓ 1 callersMethodapply_fused_qk_rotary_emb
Apply rotary embedding to query and key tensors using flashinfer's fused rope. Args: query (Tensor): Query tensor.
Megatron-LM/megatron/core/inference/contexts/dynamic_context.py:859
↓ 1 callersMethodapply_input_jitter
Add noise to the input tensor. Refer to https://arxiv.org/abs/2101.03961. Args: input (Tensor): Input tensor. Re
Megatron-LM/megatron/core/transformer/moe/router.py:454
↓ 1 callersMethodapply_interleaved_mrope
(freqs: torch.Tensor, mrope_section: List[int])
slime/slime_plugins/megatron_bridge/qwen35_vl.py:249
↓ 1 callersMethodapply_loading_parallelization
Distributes the load across ranks by exchanging metadata. Exchanges metadata from the state dict and computes the uniform (as close a
Megatron-LM/megatron/core/dist_checkpointing/strategies/fully_parallel.py:354
↓ 1 callersFunctionapply_lora
Wrap *model* with PEFT LoRA adapters according to *args*. Returns the PeftModel wrapper. All base-model parameters are frozen; only the LoRA
slime/slime/backends/fsdp_utils/lora_utils.py:19
↓ 1 callersMethodapply_pos_enc
Apply positional encoding to patches
Megatron-LM/megatron/core/models/vision/radio.py:237
↓ 1 callersMethodapply_prompt_template
Create conversation prompt string using HF chat template. The first user turn always contains an image placeholder, later turns are text-only
Megatron-LM/examples/mimo/data/energon_avlm_task_encoder.py:99
↓ 1 callersMethodapply_prompt_template
Create conversation prompt string using HF chat template. The first user turn always contains an image placeholder, later turns are text-only
Megatron-LM/examples/mimo/data/energon_vlm_task_encoder.py:78
↓ 1 callersFunctionapply_random_logits
Apply the RandomSTE function to the logits.
Megatron-LM/megatron/core/transformer/moe/moe_utils.py:947
↓ 1 callersFunctionapply_router_token_dropping
Apply token dropping to top-k expert selection. This function enforces expert capacity limits by dropping tokens that exceed the capacity and
Megatron-LM/megatron/core/transformer/moe/moe_utils.py:667
↓ 1 callersFunctionapply_safe_qwen3vl_mapping_filter
Patch qwen3_vl bridge in-process to skip known non-parameter buffers only.
slime/tools/convert_hf_to_torch_dist.py:21
↓ 1 callersFunctionapply_true_on_policy_patch_for_qwen3_moe
()
slime/slime/backends/fsdp_utils/models/qwen3_moe.py:122
↓ 1 callersMethodapply_z_loss
Encourages the router's logits to remain small to enhance stability. Please refer to the ST-MoE paper (https://arxiv.org/pdf/2202.08906.pdf) f
Megatron-LM/megatron/core/transformer/moe/router.py:419
↓ 1 callersFunctionare_equal_under_sympy
(ground_truth_normalized: str, given_normalized: str)
slime/slime/rollout/rm_hub/math_utils.py:346
↓ 1 callersMethodargbdata_to_pixdata
(self, data, len)
gui-rl/desktop_env/server/pyxcursor.py:81
↓ 1 callersFunctionarrival
(r)
Megatron-LM/examples/inference/gpt/utils.py:225
↓ 1 callersMethodas_array
(self)
Megatron-LM/megatron/legacy/data/realm_dataset_utils.py:90
↓ 1 callersMethodas_tuple
(self)
Megatron-LM/megatron/legacy/data/realm_dataset_utils.py:93
↓ 1 callersMethodask_confirmation
(self)
swe-rl/mini-swe-agent/src/minisweagent/agents/interactive.py:102
↓ 1 callersMethodasolve
Execute async agent-environment interaction for training. This method extends the original Agent to support async interaction with L
slime/examples/tau-bench/trainable_agents.py:176
↓ 1 callersFunctionassert_grads_close
(left: torch.Tensor, right: torch.Tensor)
Megatron-LM/tests/functional_tests/python_test_utils/test_optimizer_grads_match.py:111
↓ 1 callersFunctionassert_viewless_tensor
Assert that a tensor is not a view (i.e., its '._base' field is not set).
Megatron-LM/megatron/core/utils.py:708
↓ 1 callersMethodasync_bookkeep
Uses `asyncio` for continuous bookkeeping. Args: step_result (Optional[Dict]): The result of the step. context_state
Megatron-LM/megatron/core/inference/engines/dynamic_engine.py:1060
↓ 1 callersMethodasync_forward
Uses `asyncio` for continuous generation. Sleeps when no requests are available, until new requests have been added. Returns:
Megatron-LM/megatron/core/inference/engines/dynamic_engine.py:992
↓ 1 callersMethodattach_grad_to_optimizer_state
Attach gradients to optimizer named parameters in preparation for optimizer.step().
Megatron-LM/megatron/core/distributed/fsdp/src/megatron_fsdp/megatron_fsdp.py:1089
↓ 1 callersFunctionattention_mask_func
(attention_scores, attention_mask)
Megatron-LM/megatron/legacy/model/utils.py:31
↓ 1 callersFunctionattn_layer_flops
Calculate FLOPs for an attention layer.
Megatron-LM/megatron/training/training.py:198
↓ 1 callersFunctionattn_mask_postprocess
(attn_mask)
Megatron-LM/megatron/legacy/model/t5_model.py:21
↓ 1 callersFunctionattn_mask_postprocess
(attn_mask)
Megatron-LM/megatron/core/models/T5/t5_model.py:513
↓ 1 callersFunctionawait_process_event
Repeatedly wait for a multiprocessing event to be set, aborting upon process failure. Note that the timeout in this function is only for checking
Megatron-LM/megatron/core/inference/utils.py:142
↓ 1 callersMethodbackward
Backward pass.
Megatron-LM/megatron/core/tensor_parallel/random.py:511
↓ 1 callersMethodbackward
Backward pass for DownProjFunction using Triton kernels. Args: grad_output: shape (num_tokens, topk, hidden_size)
slime/slime/backends/fsdp_utils/kernels/fused_experts.py:283
↓ 1 callersFunctionbackward_step_helper
Helper method to run backward step with model split into chunks
Megatron-LM/megatron/core/pipeline_parallel/schedules.py:1313
↓ 1 callersMethodbase_generate
(self, request: ChatInferenceRequest)
Megatron-LM/megatron/rl/server/inference/inference_interface_server.py:30
↓ 1 callersFunctionbaseline_get_microbatch_id_in_model_chunk
Helper method to get the microbatch_id within model chunk given the iteration number.
Megatron-LM/tests/unit_tests/pipeline_parallel/test_helpers.py:13
↓ 1 callersFunctionbaseline_is_first_microbatch_for_model_chunk
Check if an iteration is the first for a model chunk.
Megatron-LM/tests/unit_tests/pipeline_parallel/test_helpers.py:22
↓ 1 callersFunctionbaseline_is_last_microbatch_for_model_chunk
Check if an iteration is the last for a model chunk.
Megatron-LM/tests/unit_tests/pipeline_parallel/test_helpers.py:32
↓ 1 callersMethodbatch_search
(self, query_list: list[str], num: int = None, return_score: bool = False)
slime/examples/search-r1/local_dense_retriever/retrieval_server.py:141
↓ 1 callersFunctionbatch_to_datums
Convert a batch of samples + per-sample scalar advantages to Tinker Datums.
openclaw-tinker/data_formatter.py:131
↓ 1 callersFunctionbatch_to_datums_combined
Convert a batch of samples to Tinker Datums with combined advantages.
openclaw-tinker/data_formatter.py:188
↓ 1 callersMethodbert_extended_attention_mask
Creates the extended attention mask Converts the attention mask of dimension [batch size, 1, seq len] to [batch size, 1, seq len, seq
Megatron-LM/megatron/core/models/bert/bert_model.py:237
↓ 1 callersMethodbert_position_ids
Position ids for bert model
Megatron-LM/megatron/core/models/bert/bert_model.py:269
↓ 1 callersFunctionbias_dropout_add_fused_train
( x_with_bias: Tuple[torch.Tensor, Optional[torch.Tensor]], residual: torch.Tensor, prob: float )
Megatron-LM/megatron/core/fusions/fused_bias_dropout.py:68
↓ 1 callersFunctionbias_dropout_add_unfused
(training)
Megatron-LM/megatron/core/fusions/fused_bias_dropout.py:60
↓ 1 callersFunctionbias_dropout_add_unfused_layer_scaling
Bias-dropout-add as in Megatron but with added LayerScaling handling.
Megatron-LM/examples/multimodal/layer_scaling.py:22
↓ 1 callersFunctionbias_geglu
Performs GEGLU activation with bias addition. Args: bias (torch.Tensor): Bias tensor to be added to the input. y (torch.Tensor):
Megatron-LM/megatron/core/fusions/fused_bias_geglu.py:31
← previousnext →3,401–3,500 of 10,561, ranked by callers