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
↓ 2 callers
Function
get_tokenizers
(config)
Megatron-LM/tools/retro/preprocess_data.py:192
↓ 2 callers
Function
get_train_valid_test_num_samples
Train/valid/test num samples.
Megatron-LM/megatron/training/training.py:2940
↓ 2 callers
Function
get_transformer_layer_spec_for_backend
Helper function to get module spec for TransformerLayer
Megatron-LM/megatron/core/models/gpt/gpt_layer_specs.py:350
↓ 2 callers
Function
get_updated_expert_bias
Update expert bias for biased expert routing. See https://arxiv.org/abs/2408.15664v1# Args: tokens_per_expert (torch.Tensor): The number
Megatron-LM/megatron/core/transformer/moe/moe_utils.py:876
↓ 2 callers
Function
get_vicuna_language_model_config
Return a TransformerConfig tuned for **Vicuna-7B**. The hyper-parameters follow the published Vicuna-7B weights (same sizes as Llama-7B).
Megatron-LM/examples/mimo/configs/llava_vlm.py:24
↓ 2 callers
Function
get_vision_model_config
(config, apply_query_key_layer_scaling)
Megatron-LM/examples/multimodal/config.py:179
↓ 2 callers
Function
get_vision_projection_config
(config, hidden_size)
Megatron-LM/examples/multimodal/config.py:334
↓ 2 callers
Function
get_vision_submodules_spec
Get the submodule spec for the vision modality.
Megatron-LM/tests/unit_tests/models/test_mimo_model.py:44
↓ 2 callers
Function
get_vit_layer_with_transformer_engine_spec
Returns ViT layer spec with Transformer Engine layers
Megatron-LM/megatron/core/models/vision/vit_layer_specs.py:36
↓ 2 callers
Function
get_vlm_mimo_model
( hidden_size, vocab_size, seq_len, img_h, img_w, patch_dim, special_token_ids )
Megatron-LM/tests/unit_tests/models/test_mimo_model.py:158
↓ 2 callers
Method
get_vm_directory_tree
Gets the directory tree of the vm.
gui-rl/desktop_env/controllers/python.py:551
↓ 2 callers
Function
get_vm_ip
(vm_path, max_retries=20)
gui-rl/desktop_env/providers/vmware/manager.py:227
↓ 2 callers
Method
get_vm_platform
Gets the size of the vm screen.
gui-rl/desktop_env/controllers/python.py:457
↓ 2 callers
Method
get_vm_screen_size
Gets the size of the vm screen.
gui-rl/desktop_env/controllers/python.py:463
↓ 2 callers
Function
get_world_max
Get max value across ranks. Args: n (int): Value on this rank. Returns: Max value across all ranks.
Megatron-LM/megatron/core/datasets/retro/utils.py:268
↓ 2 callers
Function
gpt2_merge
(odir)
Megatron-LM/tests/unit_tests/data/test_preprocess_data.py:177
↓ 2 callers
Function
gpt2_vocab
(odir)
Megatron-LM/tests/unit_tests/data/test_preprocess_data.py:168
↓ 2 callers
Function
grade_answer_mathd
(given_answer: str, ground_truth: str)
slime/slime/rollout/rm_hub/math_utils.py:463
↓ 2 callers
Function
grade_answer_sympy
(given_answer: str, ground_truth: str)
slime/slime/rollout/rm_hub/math_utils.py:424
↓ 2 callers
Function
grade_answer_verl
(solution_str, ground_truth)
slime/slime/rollout/rm_hub/math_utils.py:479
↓ 2 callers
Function
grouped_gemm_is_available
Check if grouped_gemm is available.
Megatron-LM/megatron/core/transformer/moe/grouped_gemm_util.py:9
↓ 2 callers
Function
guard
Pass-through to callee attribute if experimental flag is enabled. Args: super (super): Parent class of callee.
Megatron-LM/megatron/core/utils.py:214
↓ 2 callers
Function
handle_swiglu_in_state_dict
Handle SWiGLU in model and optimizer state dicts.
Megatron-LM/megatron/core/transformer/fsdp_dtensor_checkpoint.py:192
↓ 2 callers
Function
has_file_allowed_extension
Checks if a file is an allowed extension. Args: filename (string): path to a file extensions (tuple of strings): extensions to con
Megatron-LM/megatron/legacy/data/image_folder.py:43
↓ 2 callers
Method
health_check
(self)
swe-rl/server/swe_env_pool_server.py:205
↓ 2 callers
Method
health_monitoring_pause
(self)
slime/slime/ray/rollout.py:223
↓ 2 callers
Method
health_monitoring_resume
(self)
slime/slime/ray/rollout.py:227
↓ 2 callers
Method
hsdp_one_step_test
(cls, num_fsdp_group)
Megatron-LM/tests/unit_tests/distributed/test_mcore_fully_sharded_data_parallel.py:375
↓ 2 callers
Method
ids_to_text
Converts ids to text.
Megatron-LM/megatron/core/tokenizers/text/libraries/null_tokenizer.py:23
↓ 2 callers
Method
ids_to_tokens
Converts list of tokens ids to list of tokens.
Megatron-LM/megatron/core/tokenizers/text/libraries/tiktoken_tokenizer.py:189
↓ 2 callers
Function
import_class
Import a class from a string path. Args: class_path: String path to the class (e.g. 'examples.rl.environments.countdown.countdown_agent.C
Megatron-LM/megatron/rl/__init__.py:15
↓ 2 callers
Method
increment_sequences
Increment the sequence counter.
Megatron-LM/megatron/rl/rl_utils.py:104
↓ 2 callers
Function
init
(args)
slime/slime/backends/megatron_utils/initialize.py:56
↓ 2 callers
Method
init_args_and_model
Initialize Megatron and build model.
Megatron-LM/tests/functional_tests/test_cases/common/ckpt_converter/__main__.py:159
↓ 2 callers
Method
init_model
(self, share_embeddings_and_output_weights: bool = False)
Megatron-LM/tests/unit_tests/distributed/test_finalize_model_grads.py:24
↓ 2 callers
Function
init_rollout_engines
(args, pg, all_rollout_engines)
slime/slime/ray/rollout.py:831
↓ 2 callers
Method
initialize
Initialize the dataset This method is called by IndexedDataset.__init__ during object creation and by IndexedDataset.__setstate__ dur
Megatron-LM/megatron/core/datasets/indexed_dataset.py:606
↓ 2 callers
Function
initialize_gpt_model
( seed, layer_spec_fn=gpt_te_spec, vocab_size=128, virtual_pipeline_model_parallel_size=None,
Megatron-LM/tests/unit_tests/pipeline_parallel/test_pipeline_layout.py:35
↓ 2 callers
Function
initialize_mamba
(seed, glu=True, **config_kwargs)
Megatron-LM/tests/unit_tests/dist_checkpointing/models/test_mamba.py:29
↓ 2 callers
Function
initialize_mlp
(glu=True)
Megatron-LM/tests/unit_tests/dist_checkpointing/models/test_mlp_glu.py:24
↓ 2 callers
Function
initialize_rerun_state_machine
Helper function to initialize the rerun machine instance. Check the RerunStateMachine class for the details.
Megatron-LM/megatron/core/rerun_state_machine.py:1279
↓ 2 callers
Function
initialize_t5_model
(seed, encoder_decoder_spec_fn, num_layers=8, **config_kwargs)
Megatron-LM/tests/unit_tests/dist_checkpointing/models/test_t5_model.py:36
↓ 2 callers
Method
initializer
(self)
Megatron-LM/tools/preprocess_data.py:54
↓ 2 callers
Function
inner_product
(a: List[int], b: List[int])
Megatron-LM/megatron/core/parallel_state.py:306
↓ 2 callers
Function
insert
(attempt=0)
slime/docs/_static/js/lang-toggle.js:82
↓ 2 callers
Method
install_optimized_model_weights
Copies optimized parameter values into the model training parameters managed by Megatron-FSDP. Should be called after the optimizer.s
Megatron-LM/megatron/core/distributed/fsdp/src/megatron_fsdp/megatron_fsdp.py:1206
↓ 2 callers
Function
intersection
(s1, s2)
Megatron-LM/megatron/core/transformer/fsdp_dtensor_checkpoint.py:201
↓ 2 callers
Function
inverse_timer
(name)
slime/slime/utils/timer.py:84
↓ 2 callers
Function
invoke_fused_moe_backward_kernel
Invoke the fused MOE backward kernels to compute gradients. Args: grad_output: Gradient of output, shape (num_tokens * topk, N) or (
slime/slime/backends/fsdp_utils/kernels/fused_moe_triton_backward_kernels.py:399
↓ 2 callers
Method
is_empty
Check if the store has any non-None tensors. Returns: bool: True if all fields are None, False otherwise.
Megatron-LM/megatron/core/transformer/moe/moe_utils.py:1138
↓ 2 callers
Function
is_empty_async_queue
Check if async calls queue is empty. This result is consistent across ranks. Returns: bool: True if there is any ongoing async call.
Megatron-LM/megatron/training/async_utils.py:60
↓ 2 callers
Method
is_empty_chunk
Check if this chunk has no tensors to manage.
Megatron-LM/megatron/core/pipeline_parallel/fine_grained_activation_offload.py:265
↓ 2 callers
Function
is_fa_min_version
Check if minimum version of `flash-attn` is installed.
Megatron-LM/megatron/core/utils.py:393
↓ 2 callers
Function
is_first_last_bf16_layer
Check if the layer is in bf16.
Megatron-LM/megatron/core/fp8_utils.py:513
↓ 2 callers
Method
is_first_last_layer
Check if this is the last layer of the first microbatch of the last vp stage. These tensors should not be offloaded to avoid unnecess
Megatron-LM/megatron/core/pipeline_parallel/fine_grained_activation_offload.py:269
↓ 2 callers
Function
is_first_or_last_stage
Check if the current pipeline parallel stage is the first or last stage.
Megatron-LM/examples/multimodal/dataloader_provider.py:71
↓ 2 callers
Function
is_flaky_failure
(concat_allranks_logs: str)
Megatron-LM/tests/test_utils/python_scripts/launch_jet_workload.py:271
↓ 2 callers
Function
is_graph_safe_cuda_rng_tracker
Check if the cuda rng tracker is graph safe version.
Megatron-LM/megatron/core/tensor_parallel/random.py:432
↓ 2 callers
Method
is_kernel_available
(self, mask, b, np, sq, sk)
Megatron-LM/megatron/legacy/model/fused_softmax.py:170
↓ 2 callers
Function
is_last_microbatch_for_model_chunk
Check if an iteration is the last for a model chunk.
Megatron-LM/megatron/core/pipeline_parallel/schedules.py:1133
↓ 2 callers
Function
is_megatron_main_rank
()
slime/slime/backends/megatron_utils/initialize.py:108
↓ 2 callers
Function
is_number
Check if input is a number.
Megatron-LM/examples/multimodal/evaluation/evaluate_vqav2.py:85
↓ 2 callers
Function
is_pipeline_last_stage
Return True if in the last pipeline-model-parallel stage, False otherwise.
Megatron-LM/megatron/core/parallel_state.py:1591
↓ 2 callers
Function
is_port_available
Return whether a port is available.
slime/slime/utils/http_utils.py:28
↓ 2 callers
Function
is_rank0
Returns true if called in the rank0, false otherwise
Megatron-LM/megatron/training/utils.py:417
↓ 2 callers
Function
is_single_shape
Check if the input is a single shape.
Megatron-LM/megatron/core/pipeline_parallel/p2p_communication.py:130
↓ 2 callers
Function
is_start_piece
Check if the current word piece is the starting piece (BERT).
Megatron-LM/megatron/legacy/data/dataset_utils.py:167
↓ 2 callers
Function
is_submodule
Check if a module is a submodule of another module.
Megatron-LM/megatron/core/distributed/fsdp/src/megatron_fsdp/utils.py:82
↓ 2 callers
Function
is_swiglu_key
Check if this key should be handled as SwiGLU linear_fc1 weight or bias.
Megatron-LM/megatron/core/transformer/fsdp_dtensor_checkpoint.py:212
↓ 2 callers
Method
key_value_cache
Read from KV cache. Args: layer_number (int): Layer number. Return: (Tuple[Tensor, Tensor]) The key and valu
Megatron-LM/megatron/core/inference/contexts/dynamic_context.py:824
↓ 2 callers
Method
lap
Returns lap (iteration) energy (J) and updates total energy.
Megatron-LM/megatron/core/energy_monitor.py:66
↓ 2 callers
Function
last_boxed_only_string
Extract the last LaTeX boxed expression from a string. Args: string: Input string containing LaTeX code Returns: The last bo
slime/slime/rollout/rm_hub/math_dapo_utils.py:20
↓ 2 callers
Function
launch_server_process
(server_args: ServerArgs)
slime/slime/backends/sglang_utils/sglang_engine.py:54
↓ 2 callers
Method
layer_norm
Which module to use for layer norm
Megatron-LM/megatron/core/models/backends.py:160
↓ 2 callers
Function
linear_with_grad_accumulation_and_async_allreduce
Linear layer execution with asynchronous communication and gradient accumulation fusion in backprop. This has the option to accumulate the re
Megatron-LM/megatron/core/tensor_parallel/layers.py:637
↓ 2 callers
Function
load
(args)
Megatron-LM/megatron/legacy/fused_kernels/__init__.py:17
↓ 2 callers
Function
load_charts
Args: xlsx_file (Workbook): concerned excel book sheet_name (str): sheet name options (Dict[str, List[str]]): dict like {
gui-rl/desktop_env/evaluators/metrics/utils.py:98
↓ 2 callers
Function
load_checkpoint
(ddp_model, optimizer, opt_param_scheduler, checkpointing_context, skip_load_to_model_and_opt)
slime/slime/backends/megatron_utils/checkpoint.py:101
↓ 2 callers
Method
load_checkpoint
Load checkpoint, and forward pass data.
Megatron-LM/tests/functional_tests/test_cases/common/ckpt_converter/__main__.py:427
↓ 2 callers
Function
load_config
Loads and parses a yaml file into a JETWorkloadManifest
Megatron-LM/tests/test_utils/python_scripts/recipe_parser.py:83
↓ 2 callers
Function
load_content_metadata
Load content metadata stored in the checkpoint with `save(..., content_metadata=...)`. Args: checkpoint_dir (str, optional): checkpoint d
Megatron-LM/megatron/core/dist_checkpointing/serialization.py:283
↓ 2 callers
Function
load_corpus
(corpus_path: str)
slime/examples/search-r1/local_dense_retriever/retrieval_server.py:33
↓ 2 callers
Function
load_filters
(xlsx_file: Workbook, sheet_name: str, **options)
gui-rl/desktop_env/evaluators/metrics/utils.py:645
↓ 2 callers
Method
load_from_file
Populate members from instance saved to file
Megatron-LM/megatron/legacy/data/realm_index.py:51
↓ 2 callers
Function
load_indexed_datasets
Loaded indexed datasets into memory-mapped datasets. Args: project_dir (str): Path to Retro project dir. indexed_dataset_infos (L
Megatron-LM/megatron/core/datasets/retro/db/utils.py:114
↓ 2 callers
Function
load_model_state_dict
Helper function to load state dict with fallback for missing extra states.
Megatron-LM/megatron/training/checkpointing.py:1684
↓ 2 callers
Method
load_other_checkpoint
(self, model_tag: str, path: str)
slime/slime/backends/megatron_utils/actor.py:1491
↓ 2 callers
Function
load_pivot_tables
Args: xlsx_file (Workbook): concerned excel book sheet_name (str): sheet name options (Dict[str, List[str]]): dict like {
gui-rl/desktop_env/evaluators/metrics/utils.py:197
↓ 2 callers
Function
load_plugin
(plugin_type, name)
Megatron-LM/tools/checkpoint/convert.py:90
↓ 2 callers
Function
load_preprocess
Preprocesses the given state dictionary by applying factories and extracting non-persistent data, without modifying the original dictionary.
Megatron-LM/megatron/core/dist_checkpointing/state_dict_utils.py:62
↓ 2 callers
Function
load_retro_config
Load Retro's config.json.
Megatron-LM/megatron/training/arguments.py:205
↓ 2 callers
Function
load_sharded_metadata
Load sharded metadata from the checkpoint. Similar to `load_tensors_metadata`, but includes also ShardedObjects. Returns a dictionary simila
Megatron-LM/megatron/core/dist_checkpointing/serialization.py:223
↓ 2 callers
Method
load_sharded_metadata
Load just the metadata from the checkpoint.
Megatron-LM/megatron/core/dist_checkpointing/strategies/base.py:128
↓ 2 callers
Function
load_sparklines
Args: xlsx_file (str): path to xlsx sheet_name (str): sheet name Returns: List[Dict[str, str]]: sparkline definition
gui-rl/desktop_env/evaluators/metrics/utils.py:50
↓ 2 callers
Method
load_state_dict
( self, state_dict, strict=True )
Megatron-LM/megatron/core/transformer/module.py:478
↓ 2 callers
Method
load_state_dict
Load the state dict. As detailed in state_dict(), the state dict contains all non- parameter-related variables. This method is notabl
Megatron-LM/megatron/core/optimizer/distrib_optimizer.py:689
↓ 2 callers
Function
load_submodule_ckpt
Load *ckpt_dir* into *module* using Megatron distributed-checkpointing.
Megatron-LM/examples/mimo/utils/model_helpers.py:10
↓ 2 callers
Method
load_tensors_metadata
Load tensors metadata from the checkpoint for ShardedTensors. Returns a dictionary similar to a sharded state dict, but note that the
Megatron-LM/megatron/core/dist_checkpointing/strategies/base.py:144
↓ 2 callers
Function
load_xlsx_styles
Args: xlsx_file (Workbook): concerned excel book sheet_name (str): sheet name book_name (str): book name options
gui-rl/desktop_env/evaluators/metrics/utils.py:461
↓ 2 callers
Function
local_step
()
Megatron-LM/megatron/core/dist_checkpointing/strategies/state_dict_saver.py:103
← previous
next →
2,101–2,200 of 10,561, ranked by callers