MCPcopy Create free account

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

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

↓ 2 callersFunctionfind_yellow_triangle
(image)
gui-rl/desktop_env/evaluators/metrics/gimp.py:109
↓ 2 callersFunctionfinish_embedding_wgrad_compute
Finish embedding wgrad compute.
Megatron-LM/megatron/core/pipeline_parallel/schedules.py:714
↓ 2 callersMethodfinish_grad_sync
Finishes grad sync (all-reduce or reduce-scatter) communication operations for all model gradients. Call prior to the optimization st
Megatron-LM/megatron/core/distributed/fsdp/src/megatron_fsdp/megatron_fsdp.py:1096
↓ 2 callersMethodfinish_init
Required post-init of quantization configuration.
Megatron-LM/megatron/core/extensions/kitchen.py:239
↓ 2 callersMethodflash_decode_and_prefill
Flash attention kernel for mixed decode and prefill samples. Args: q (Tensor): Query tensor. k (Tensor): Key tensor.
Megatron-LM/megatron/core/transformer/attention.py:559
↓ 2 callersFunctionflatten
(obj, parent_key="", sep=".")
Megatron-LM/tools/checkpoint/checkpoint_inspector.py:274
↓ 2 callersFunctionflatten_range
Apply flattened range to a tensor.
Megatron-LM/megatron/core/dist_checkpointing/strategies/zarr.py:291
↓ 2 callersFunctionfloat16_to_fp32
Converts floating-point values from fp16 to fp32. Args: val: The value to convert. Can be a single number, a tuple, or a list.
Megatron-LM/megatron/core/transformer/module.py:361
↓ 2 callersFunctionfloor_by_factor
返回小于等于 number 的且能被 factor 整除的整数
gui-rl/agents/utils/qwen_vl_utils.py:14
↓ 2 callersMethodfoo
(self)
Megatron-LM/tests/unit_tests/utils/test_experimental_log_once.py:65
↓ 2 callersFunctionformat_example
Format an example into a multi-choices problem.
Megatron-LM/examples/post_training/modelopt/mmlu.py:100
↓ 2 callersMethodforward
Run a forward pass for each of the models and return the respective embeddings.
Megatron-LM/megatron/legacy/model/realm_model.py:67
↓ 2 callersMethodforward
(ctx, inputs, scale)
Megatron-LM/megatron/legacy/model/fused_softmax.py:91
↓ 2 callersMethodforward
Forward.
Megatron-LM/megatron/core/extensions/kitchen.py:346
↓ 2 callersMethodforward
Forward pass, does nothing but registers an autograd node.
Megatron-LM/megatron/core/transformer/cuda_graphs.py:415
↓ 2 callersFunctionforward_step_calc_loss
Calculate the loss and number of tokens for forward_step()
Megatron-LM/megatron/core/pipeline_parallel/schedules.py:212
↓ 2 callersFunctionforward_step_helper_postprocess
Postprocess for forward_step_helper
Megatron-LM/megatron/core/pipeline_parallel/schedules.py:1225
↓ 2 callersFunctionforward_step_helper_preprocess
Preprocess for forward_step_helper
Megatron-LM/megatron/core/pipeline_parallel/schedules.py:1189
↓ 2 callersMethodforward_with_attention
(self, x)
Megatron-LM/megatron/legacy/model/vision/esvit_swin_backbone.py:445
↓ 2 callersFunctionfp8_discard_transpose_cache
Discard the transpose cache of a FP8 tensor.
Megatron-LM/megatron/core/distributed/fsdp/src/megatron_fsdp/mixed_precision.py:177
↓ 2 callersFunctionfp8_quantize
Quantize sharded parameters to FP8.
Megatron-LM/megatron/core/distributed/fsdp/src/megatron_fsdp/mixed_precision.py:247
↓ 2 callersMethodfree_slots
Frees the Mamba state slots associated with the given request indices. Args: request_indices (Tensor): A 1D tensor of re
Megatron-LM/megatron/core/inference/contexts/attention_context/mamba_metadata.py:104
↓ 2 callersMethodfrom_payload
( cls, payload: dict[str, Any], )
terminal-rl/custom_types.py:28
↓ 2 callersMethodfrom_request
Initialize record from a single request. Args: request (DynamicInferenceRequest): Initial request. Returns:
Megatron-LM/megatron/core/inference/inference_request.py:379
↓ 2 callersMethodfrom_spec
Create a modality submodule from ModuleSpec configuration. Args: module_spec (ModuleSpec): The module specification for this moda
Megatron-LM/megatron/core/models/mimo/submodules/base.py:62
↓ 2 callersMethodfrom_str
Parse the pipeline model parallel layout from a string.
Megatron-LM/megatron/core/transformer/pipeline_parallel_layer_layout.py:263
↓ 2 callersMethodfront
Get the first non-empty chunk handler without removing it from the queue.
Megatron-LM/megatron/core/pipeline_parallel/fine_grained_activation_offload.py:101
↓ 2 callersFunctionfused_apply_mla_rope_for_kv
Fused function for applying YARN RoPE to MLA's key and value. It splits the input tensor kv into key and value, and concatenates the proc
Megatron-LM/megatron/core/fusions/fused_mla_yarn_rope_apply.py:754
↓ 2 callersFunctionfused_apply_mla_rope_for_q
Fused function for applying YARN RoPE to MLA's query. This function inplace modifies the input tensor t. Along the last dimension of t, t
Megatron-LM/megatron/core/fusions/fused_mla_yarn_rope_apply.py:327
↓ 2 callersFunctionfused_indices_to_multihot
Convert moe topk indices to multihot representation. This function is an experimental feature and may change in future versions.
Megatron-LM/megatron/core/fusions/fused_indices_converter.py:282
↓ 2 callersFunctionfused_pad_routing_map
Fused version of pad_routing_map. Args: routing_map (torch.Tensor): A boolean or integer tensor of shape [num_tokens, num_expe
Megatron-LM/megatron/core/fusions/fused_pad_routing_map.py:74
↓ 2 callersFunctionfused_vocab_parallel_cross_entropy
Performs cross entropy loss when logits are split across tensor parallel ranks Args: vocab_parallel_logits: logits split across tens
Megatron-LM/megatron/core/fusions/fused_cross_entropy.py:136
↓ 2 callersMethodgc_by_weight_version
Perform garbage collection based on weight version. Remove nodes with weight_version < (current_weight_version - gc_threshold_k).
slime/slime/router/middleware_hub/radix_tree.py:439
↓ 2 callersFunctiongeglu
Performs GEGLU (GELU-Gated Linear Unit) activation. Args: y (torch.Tensor): Input tensor to be split into two halves along the last dimen
Megatron-LM/megatron/core/fusions/fused_bias_geglu.py:17
↓ 2 callersFunctiongeglu_back
Computes the gradient for the GEGLU activation. Args: g (torch.Tensor): Gradient tensor from the subsequent layer. y (torch.Tenso
Megatron-LM/megatron/core/fusions/fused_bias_geglu.py:49
↓ 2 callersMethodgen_multi_turn_loss_mask_distill_qwen
( self, messages: list[dict], tools: list[dict] = None )
slime/slime/utils/mask_utils.py:111
↓ 2 callersMethodgenerate_with_sglang
( self, *, args: Any, state: GenerateState, messages: List[Dict[str, A
gui-rl/agents/qwen35_agent.py:370
↓ 2 callersFunctionget
(url)
slime/slime/utils/http_utils.py:339
↓ 2 callersMethodget
Get number of microbatches.
Megatron-LM/megatron/core/num_microbatches_calculator.py:275
↓ 2 callersMethodget
This function is n abstraction over __getitem__ with support for slicing Args: idx (int): The index into the dataset
Megatron-LM/megatron/core/datasets/gpt_dataset.py:744
↓ 2 callersMethodgetCursorImageArrayFast
(self)
gui-rl/desktop_env/server/pyxcursor.py:125
↓ 2 callersMethodgetCursorImageData
(self)
gui-rl/desktop_env/server/pyxcursor.py:102
↓ 2 callersMethodget_active_avail
Compute number of active blocks available.
Megatron-LM/megatron/core/inference/contexts/dynamic_block_allocator.py:63
↓ 2 callersMethodget_active_used
Compute number of active blocks used.
Megatron-LM/megatron/core/inference/contexts/dynamic_block_allocator.py:47
↓ 2 callersFunctionget_addr_and_ports
(engine)
slime/slime/ray/rollout.py:1008
↓ 2 callersFunctionget_adlr_autoresume
ADLR autoresume object. It can be None so no need to check if it is initialized.
Megatron-LM/megatron/training/global_vars.py:56
↓ 2 callersFunctionget_agent
Get an agent based on environment configuration. If args.langrl_env_config is provided, uses weighted environment selection. Otherwise falls
Megatron-LM/megatron/rl/rl_utils.py:515
↓ 2 callersFunctionget_all_text_shapes
递归获取slide中所有包含文本的shapes,包括GROUP内部的
gui-rl/desktop_env/evaluators/metrics/slides.py:166
↓ 2 callersFunctionget_attention_module_spec_for_backend
Helper function to get module spec for Attention
Megatron-LM/megatron/core/models/gpt/gpt_layer_specs.py:391
↓ 2 callersFunctionget_base_path
(pattern)
Megatron-LM/tests/unit_tests/find_test_cases.py:19
↓ 2 callersFunctionget_batch
Generate a batch
Megatron-LM/pretrain_retro.py:83
↓ 2 callersFunctionget_batch
Generate a batch
Megatron-LM/tools/retro/sft/sft_retro.py:88
↓ 2 callersFunctionget_batch
Generate a CP-ready micro-batch with packed sequence parameters. Steps: - Fetch raw fields via iterator. - Save original token tenso
slime/slime/backends/megatron_utils/data.py:46
↓ 2 callersMethodget_batch
Get batch of data, from input token IDs.
Megatron-LM/tests/functional_tests/test_cases/common/ckpt_converter/__main__.py:261
↓ 2 callersMethodget_batch
(self, seq_length, micro_batch_size)
Megatron-LM/tests/unit_tests/transformer/test_multi_token_prediction.py:233
↓ 2 callersFunctionget_batch_iterator
Generator function that yields batches indefinitely or for a specified number of batches. Args: seq_length: Length of the sequence
Megatron-LM/tests/unit_tests/pipeline_parallel/test_pipeline_layout.py:324
↓ 2 callersFunctionget_blend_and_blend_per_split
Get blend and blend_per_split from passed-in arguments.
Megatron-LM/megatron/training/utils.py:470
↓ 2 callersFunctionget_causal_conv1d_version
Get causal_conv1d version from __version__; if not available use pip's. Use caching.
Megatron-LM/megatron/core/utils.py:436
↓ 2 callersFunctionget_chat_template
()
Megatron-LM/tests/unit_tests/tokenizers/test_tokenizer.py:32
↓ 2 callersFunctionget_comm_stream
Get the stream for communication
Megatron-LM/megatron/core/pipeline_parallel/utils.py:344
↓ 2 callersMethodget_config
(cls)
Megatron-LM/tests/unit_tests/transformer/test_retro_attention.py:27
↓ 2 callersFunctionget_config_logger_path
Get the path to the config logger directory.
Megatron-LM/megatron/core/config_logger.py:25
↓ 2 callersFunctionget_conversation
()
Megatron-LM/tests/unit_tests/tokenizers/test_tokenizer.py:11
↓ 2 callersFunctionget_cuda_rng_tracker
Get cuda rng tracker.
Megatron-LM/megatron/core/distributed/fsdp/src/megatron_fsdp/utils.py:398
↓ 2 callersFunctionget_current_global_batch_size
Get current global batch size.
Megatron-LM/megatron/core/num_microbatches_calculator.py:22
↓ 2 callersFunctionget_current_memory_info
Get current memory usage.
Megatron-LM/megatron/post_training/utils.py:8
↓ 2 callersFunctionget_data_parallel_rank
Return caller's rank in the data-parallel group.
Megatron-LM/megatron/core/parallel_state.py:1726
↓ 2 callersFunctionget_data_parallel_world_size
Return world size for the data parallel group.
Megatron-LM/megatron/core/parallel_state.py:1707
↓ 2 callersMethodget_dataset
(self, validation: bool = False)
Megatron-LM/examples/rl/environments/math/dapo_agent.py:30
↓ 2 callersMethodget_dataset
(self, validation: bool = False)
Megatron-LM/examples/rl/environments/math/bigmath_agent.py:19
↓ 2 callersMethodget_dataset
(self, validation: bool = False)
Megatron-LM/examples/rl/environments/math/openmath_agent.py:19
↓ 2 callersMethodget_dataset
(self, validation: bool = False)
Megatron-LM/examples/rl/environments/math/gsm8k_agent.py:37
↓ 2 callersMethodget_dataset
(self, validation: bool = False)
Megatron-LM/examples/rl/environments/countdown/countdown_agent.py:27
↓ 2 callersMethodget_db_chunk_gpt
Get DB chunk as GPT token ids.
Megatron-LM/tools/retro/cli/cli.py:115
↓ 2 callersFunctionget_deepscaler_rule_based_reward
(response, label)
slime/slime/rollout/rm_hub/deepscaler.py:4
↓ 2 callersFunctionget_default_sampling_params
(termination_id: int = None)
Megatron-LM/examples/inference/gpt/utils.py:148
↓ 2 callersFunctionget_device_arch_version
Returns GPU arch version (8: Ampere, 9: Hopper, 10: Blackwell, ...)
Megatron-LM/megatron/training/utils.py:447
↓ 2 callersFunctionget_distributed_optimizer_checkpoint_name
(model_checkpoint_name)
Megatron-LM/megatron/training/checkpointing.py:200
↓ 2 callersFunctionget_docstring
()
swe-rl/mini-swe-agent/src/minisweagent/run/mini_extra.py:17
↓ 2 callersMethodget_document_indices
Get the document indices This method is slated for deprecation. Returns: numpy.ndarray: The document indices
Megatron-LM/megatron/core/datasets/indexed_dataset.py:781
↓ 2 callersFunctionget_empty_index_path
Path of empty index. Args: config (RetroPreprocessingConfig): Retro preprocessing config. Returns: Path to the empty (traine
Megatron-LM/megatron/core/datasets/retro/index/build.py:51
↓ 2 callersFunctionget_env_or_exit
(name: str)
openclaw-test/teacher_chat.py:114
↓ 2 callersFunctionget_env_or_exit
(name: str)
openclaw-test/TA_chat.py:109
↓ 2 callersFunctionget_env_or_exit
(name: str)
openclaw-test/student_chat.py:74
↓ 2 callersFunctionget_ep_layer_offset
Get the expert layer offset for the current model. Args: num_experts: Total number of experts in the model. If None, returns 0.
Megatron-LM/megatron/core/transformer/fsdp_dtensor_checkpoint.py:50
↓ 2 callersFunctionget_evaluation_configs
Get evaluation config(s) from a config file or command-line arguments. Args: config_path: Optional path to config file. If not provided,
Megatron-LM/examples/multimodal/run_text_generation.py:365
↓ 2 callersFunctionget_expert_index_from_key
Extract expert index from various expert key formats. Supported formats: - GroupedMLP: 'mlp.experts.linear_fc1.weight0', 'mlp.experts.linear_
Megatron-LM/megatron/core/transformer/fsdp_dtensor_checkpoint.py:81
↓ 2 callersFunctionget_expert_model_parallel_group
Get the expert-model-parallel group the caller rank belongs to.
Megatron-LM/megatron/core/parallel_state.py:1772
↓ 2 callersFunctionget_expert_param
(args, name, param)
slime/tools/convert_torch_dist_to_hf.py:66
↓ 2 callersFunctionget_expert_tensor_parallel_group
Get the expert-tensor-parallel group the caller rank belongs to.
Megatron-LM/megatron/core/parallel_state.py:1813
↓ 2 callersFunctionget_fa_version
Get Flash attention version from __version__; if not available use pip's. Use caching.
Megatron-LM/megatron/core/utils.py:372
↓ 2 callersFunctionget_fp8_align_size
Get the alignment size required for fp8 GEMM.
Megatron-LM/megatron/core/fp8_utils.py:168
↓ 2 callersFunctionget_fp8_recipe
Return fp8 recipe. Arguments: config (TransformerConfig): Configuration object. Returns: FP8 recipe.
Megatron-LM/megatron/core/fp8_utils.py:536
↓ 2 callersMethodget_freqs_non_repeated
Generates matrix of frequencies based on positions in the sequence, used to create positional encodings
Megatron-LM/megatron/core/models/common/embeddings/rotary_pos_embedding.py:127
↓ 2 callersMethodget_fsdp_buffer
Get the FSDP buffer with the given bucket ID.
Megatron-LM/megatron/core/distributed/fsdp/src/megatron_fsdp/param_and_grad_buffer.py:3578
↓ 2 callersFunctionget_func_args
Get the argument names of a function.
Megatron-LM/megatron/core/nccl_allocator.py:99
↓ 2 callersFunctionget_functional_test_analytics
(pipeline_id: int, type: str)
Megatron-LM/tests/test_utils/python_scripts/dashboard.py:67
↓ 2 callersFunctionget_gitlab_handle
()
Megatron-LM/tests/test_utils/python_scripts/notify.py:21
↓ 2 callersFunctionget_global_memory_buffer
Return the global GlobalMemoryBuffer object
Megatron-LM/megatron/core/distributed/fsdp/src/megatron_fsdp/utils.py:686
↓ 2 callersFunctionget_global_unique_param_name
Get the global unique parameter name for a given model and parameter. Args: model_chunks: List of model chunks to search for the par
Megatron-LM/megatron/core/transformer/fsdp_dtensor_checkpoint.py:452
← previousnext →1,901–2,000 of 10,561, ranked by callers