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
bias_geglu_back
Computes the gradient for the biased GEGLU activation. Args: g (torch.Tensor): Gradient tensor from the subsequent layer. y (torc
Megatron-LM/megatron/core/fusions/fused_bias_geglu.py:69
↓ 1 callers
Function
bias_gelu
(bias, y)
Megatron-LM/megatron/core/fusions/fused_bias_gelu.py:17
↓ 1 callers
Function
bias_gelu_back
(g, bias, y)
Megatron-LM/megatron/legacy/model/fused_bias_gelu.py:24
↓ 1 callers
Function
bias_gelu_back
(g, bias, y)
Megatron-LM/megatron/core/fusions/fused_bias_gelu.py:26
↓ 1 callers
Function
bias_swiglu_back
Computes the gradient for the biased SwiGLU activation function. Args: g (torch.Tensor): Gradient tensor from the subsequent layer.
Megatron-LM/megatron/core/fusions/fused_bias_swiglu.py:73
↓ 1 callers
Function
block_fp8
(weight, block_size)
slime/tools/convert_hf_to_fp8.py:39
↓ 1 callers
Function
blockwise_cast_to_fp8_triton
(x: torch.Tensor, block_size=None)
slime/slime/backends/megatron_utils/kernels/fp8_kernel.py:61
↓ 1 callers
Function
bool_generator
()
Megatron-LM/tests/unit_tests/dist_checkpointing/test_replication.py:41
↓ 1 callers
Method
bpe
(self, token)
Megatron-LM/megatron/training/tokenizer/gpt2_tokenization.py:198
↓ 1 callers
Function
broadcast_data
Broadcast data from rank zero of each model parallel group to the members of the same model parallel group. Args: keys: list of keys
Megatron-LM/megatron/core/tensor_parallel/data.py:64
↓ 1 callers
Function
broadcast_nested_data_batch
Recursively broadcast nested dictionaries of tensors using each tensor's own dtype.
Megatron-LM/examples/mimo/utils/data_helpers.py:43
↓ 1 callers
Method
broadcast_params
Syncs parameters across all DP ranks.
Megatron-LM/megatron/core/distributed/distributed_data_parallel.py:555
↓ 1 callers
Function
buildTargetUrl
(target)
slime/docs/_static/js/lang-toggle.js:30
↓ 1 callers
Function
build_batch
Convert TrainingSamples into tinker Datums + associated metadata.
openclaw-fireworks/fireworks_training_loop.py:62
↓ 1 callers
Function
build_block_db
Split each document within block into consecutive retro_gpt_chunk_length size chunks. Args: config (RetroPreprocessingConfig): For DB bui
Megatron-LM/megatron/core/datasets/retro/db/build.py:177
↓ 1 callers
Method
build_checkpoint_metadata
Construct a simple namespace for all relevant model metadata.
Megatron-LM/tools/checkpoint/loader_base.py:428
↓ 1 callers
Method
build_comm_map
Get src/dest tp leaders and populate comm_map for each rank. This method analyzes the source and destination grids to determine which
Megatron-LM/megatron/core/pipeline_parallel/bridge_communicator.py:227
↓ 1 callers
Method
build_cpu_optimizer_list
Build several cpu optimizers to enable overlap. Currently we naively assign each parameter to an individual optimizer. Args:
Megatron-LM/megatron/core/optimizer/cpu_offloading/hybrid_optimizer.py:227
↓ 1 callers
Function
build_data_loader
Data loader. Note that batch-size is the local (per GPU) batch-size.
Megatron-LM/megatron/legacy/model/vision/knn_monitor.py:12
↓ 1 callers
Function
build_data_parallel_buffer_index
Assuming that all input tensor elements contiguously compose a global buffer, give the index range of every tensor, the bucket in the buffer,
Megatron-LM/megatron/core/distributed/fsdp/src/megatron_fsdp/param_and_grad_buffer.py:252
↓ 1 callers
Function
build_dataset
(base_dir: Path, meta_path: Path, output_path: Path)
gui-rl/build_gui_prompt_dataset.py:16
↓ 1 callers
Function
build_datasets
(idir_txt, idir_img, odir, extra_args=[])
Megatron-LM/tests/unit_tests/data/test_preprocess_mmdata.py:32
↓ 1 callers
Function
build_db
Extract token chunks from each indexed dataset. Iterate each document of each indexed dataset, extract that document's chunks, and save t
Megatron-LM/megatron/core/datasets/retro/db/build.py:614
↓ 1 callers
Function
build_docker_image
(task: dict[str, Any], timeout: float = 1200.0)
terminal-rl/remote/docker_compose_utils.py:106
↓ 1 callers
Function
build_eval_dataset_configs
( args: Any, raw_config: Iterable[dict[str, Any]], defaults: dict[str, Any], )
slime/slime/utils/eval_config.py:195
↓ 1 callers
Method
build_fn
(key, tensor, replica_id, flattened_range)
Megatron-LM/tests/unit_tests/dist_checkpointing/test_mapping.py:138
↓ 1 callers
Function
build_individual_db
Process a single indexed dataset & extract chunks. Args: config (RetroPreprocessingConfig): Retro preprocessing config. dataset_i
Megatron-LM/megatron/core/datasets/retro/db/build.py:287
↓ 1 callers
Function
build_individual_dbs
Iterate each indexed dataset & process its chunks. Args: config (RetroPreprocessingConfig): Retro preprocessing config. indexed_d
Megatron-LM/megatron/core/datasets/retro/db/build.py:386
↓ 1 callers
Method
build_instruction_prompt
(self, instruction: str, previous_actions_text: List[str])
gui-rl/agents/qwen35_agent.py:268
↓ 1 callers
Method
build_instruction_prompt
(self, instruction: str, actions_text: List[str])
gui-rl/agents/qwen3vl_agent.py:219
↓ 1 callers
Method
build_low_level_dataset
Abstract method implementation Args: dataset_path (str): The real path prefix to the IndexedDataset .bin and .idx files
Megatron-LM/megatron/core/datasets/gpt_dataset.py:143
↓ 1 callers
Function
build_merged_dbs
Merge individual dataset components into single database. This method merges databases for DB types: - 'sampled': used for training the vecto
Megatron-LM/megatron/core/datasets/retro/db/build.py:596
↓ 1 callers
Function
build_model
(config)
Megatron-LM/tests/unit_tests/fusions/test_fused_linear_cross_entropy.py:145
↓ 1 callers
Function
build_model
(config)
Megatron-LM/tests/unit_tests/a2a_overlap/test_schedule_chunk_1f1b.py:26
↓ 1 callers
Method
build_model
()
Megatron-LM/tests/functional_tests/test_cases/common/ckpt_converter/__main__.py:229
↓ 1 callers
Method
build_model
()
Megatron-LM/tests/functional_tests/test_cases/common/ckpt_converter/__main__.py:748
↓ 1 callers
Function
build_mtp_layer_callables
Callables for multi-token prediction layer nodes. This class contains the callable functions for different types of multi-token prediction la
Megatron-LM/megatron/core/models/gpt/fine_grained_callables.py:507
↓ 1 callers
Function
build_sample
Convert to numpy and return a sample consumed by the batch producer.
Megatron-LM/megatron/legacy/data/orqa_wiki_dataset.py:105
↓ 1 callers
Method
build_sample
(cls, tokenizer, token_ids)
Megatron-LM/tools/bert_embedding/dataset.py:27
↓ 1 callers
Function
build_samples_from_outcome
( sample: Sample, outcome: RolloutOutcome | None, *, reward: float, rollout_error: str | N
terminal-rl/sample_builders.py:43
↓ 1 callers
Method
build_sys_argv
Construct a sys.argv list for Megatron's argument parser. This centralizes the hack of overwriting sys.argv.
Megatron-LM/tools/checkpoint/loader_base.py:462
↓ 1 callers
Method
build_sys_argv
Construct a sys.argv list for Megatron's argument parser. This centralizes the hack of overwriting sys.argv.
Megatron-LM/tools/checkpoint/saver_base.py:214
↓ 1 callers
Function
build_tokens_types_paddings_from_ids
Build token types and paddings, trim if needed, and pad if needed.
Megatron-LM/megatron/legacy/data/orqa_wiki_dataset.py:68
↓ 1 callers
Function
build_tokens_types_paddings_from_ids
Build token types and paddings, trim if needed, and pad if needed.
Megatron-LM/tasks/data_utils.py:49
↓ 1 callers
Function
build_tokens_types_paddings_from_text
Build token types and paddings, trim if needed, and pad if needed.
Megatron-LM/megatron/legacy/data/orqa_wiki_dataset.py:51
↓ 1 callers
Function
build_train_valid_test_data_loaders
Build pretraining data loaders.
Megatron-LM/megatron/training/training.py:2973
↓ 1 callers
Function
build_train_valid_test_datasets
(data_prefix, splits_string, train_valid_test_num_samples,
Megatron-LM/megatron/legacy/data/dataset_utils.py:448
↓ 1 callers
Method
bulk_offload
Offload a group of tensors and optionally release their GPU memory.
Megatron-LM/megatron/core/pipeline_parallel/fine_grained_activation_offload.py:404
↓ 1 callers
Method
bulk_offload_group
offload a group of tensors recorded in tensor_push().
Megatron-LM/megatron/core/pipeline_parallel/fine_grained_activation_offload.py:323
↓ 1 callers
Method
bulk_reload
Reload the next group of tensors from CPU to GPU.
Megatron-LM/megatron/core/pipeline_parallel/fine_grained_activation_offload.py:427
↓ 1 callers
Function
bytes_to_unicode
Returns list of utf-8 byte and a corresponding list of unicode strings. The reversible bpe codes work on unicode strings. This means you
Megatron-LM/megatron/training/tokenizer/gpt2_tokenization.py:55
↓ 1 callers
Function
cache_index_file
Download a file from object storage to a local path with distributed training support. The download only happens on Rank 0, and other ranks will w
Megatron-LM/megatron/core/datasets/object_storage_utils.py:243
↓ 1 callers
Function
calc_dtensor_params_l2_norm
Calculate l2 norm of DTensor parameters.
Megatron-LM/megatron/training/utils.py:197
↓ 1 callers
Function
calculate_attention_flops
(args, seqlen, num_attention_heads)
slime/slime/utils/flops_utils.py:35
↓ 1 callers
Function
calculate_correct_answers
Calculate correct over total answers and return prediction if the `output_predictions` is true.
Megatron-LM/tasks/eval_utils.py:65
↓ 1 callers
Function
calculate_cross_entropy_loss
Calculates the final cross entropy loss for the tokens.
Megatron-LM/megatron/core/fusions/fused_cross_entropy.py:48
↓ 1 callers
Function
calculate_fwd_flops
( seqlens, args, )
slime/slime/utils/flops_utils.py:66
↓ 1 callers
Function
calculate_gradients
Calculate the logits gradients scaled based on the CE loss
Megatron-LM/megatron/core/fusions/fused_cross_entropy.py:65
↓ 1 callers
Function
calculate_grpo_loss
Get GRPO loss, the kl term of the loss and the pi/pi_{old} ratios. Args: current_logprobs: pi logprobs, [batch, seq] for unpacked or [1,
Megatron-LM/megatron/rl/rl_utils.py:2137
↓ 1 callers
Function
calculate_layer_counts
Calculate the number of attention, Mamba, and MLP layers.
Megatron-LM/megatron/training/training.py:169
↓ 1 callers
Function
calculate_lm_head_flops
(seqlen, hidden_size, vocab_size)
slime/slime/utils/flops_utils.py:5
↓ 1 callers
Function
calculate_log_probs_and_entropy
(logits, tokens, tp_group, with_entropy: bool = False, chunk_size: int = -1)
slime/slime/utils/ppo_utils.py:689
↓ 1 callers
Function
calculate_logits_max
Calculates the maximum logits of the predicted tokens.
Megatron-LM/megatron/core/fusions/fused_cross_entropy.py:13
↓ 1 callers
Function
calculate_mlp_flops
(seqlen, hidden_size, ffn_hidden_size)
slime/slime/utils/flops_utils.py:53
↓ 1 callers
Function
calculate_num_mel_frames
Calculate the number of mel frames from an audio signal. Parameters: - audio_length (int): Total number of audio samples. - sample_r
Megatron-LM/tests/unit_tests/models/test_mimo_audio_submodules.py:89
↓ 1 callers
Function
calculate_num_mel_frames
Calculate the number of mel frames from an audio signal. Parameters: - audio_length (int): Total number of audio samples. - sample_r
Megatron-LM/examples/mimo/data/utils/calculate_audio_tokens.py:18
↓ 1 callers
Function
calculate_output_flops
(seqlen, hidden_size)
slime/slime/utils/flops_utils.py:49
↓ 1 callers
Function
calculate_predicted_logits
Calculates the predicted logits for the tokens.
Megatron-LM/megatron/core/fusions/fused_cross_entropy.py:26
↓ 1 callers
Function
calculate_qkv_projection_flops
(args, seqlen, hidden_size, num_attention_heads, num_query_groups)
slime/slime/utils/flops_utils.py:9
↓ 1 callers
Function
calculate_veto_mask
( log_ratio: torch.Tensor, loss_mask: torch.Tensor, veto_threshold: float | None, metrics: dic
slime/examples/train_infer_mismatch_helper/mis.py:82
↓ 1 callers
Function
call_dynamic_filter
(fn, *args, **kwargs)
slime/slime/rollout/filter_hub/base_types.py:11
↓ 1 callers
Function
call_model
Call training server (port 30000) with side turn.
openclaw-rl/oel/eval/select_hard_problems.py:65
↓ 1 callers
Function
call_to_action_sglang
Convert sglang response message to Action, similar to original message_to_action but adapted for sglang response format.
slime/examples/tau-bench/trainable_agents.py:38
↓ 1 callers
Method
cancel_gradients_last_layer
(self, iteration)
Megatron-LM/megatron/legacy/model/vision/dino.py:278
↓ 1 callers
Method
capture_cuda_graphs
Capture multiple cuda graphs by initializing the `DynamicInferenceEngine`. The `DynamicInferenceEngine` is used here because it is currently
Megatron-LM/tests/unit_tests/transformer/test_cuda_graphs.py:549
↓ 1 callers
Function
ceildiv
(a, b)
slime/tools/convert_hf_to_torch_dist.py:124
↓ 1 callers
Function
change_resolution
(vm_name, resolution=(1920, 1080, 32))
gui-rl/desktop_env/providers/virtualbox/manager.py:246
↓ 1 callers
Function
channel_fp8
(weight)
slime/tools/convert_hf_to_fp8.py:73
↓ 1 callers
Method
check_and_clean
(self)
gui-rl/desktop_env/providers/azure/manager.py:55
↓ 1 callers
Function
check_and_cleanup_memory
Check memory usage and perform appropriate cleanup
toolcall-rl/tool_sandbox.py:73
↓ 1 callers
Function
check_ans
(pred, gt)
Megatron-LM/examples/multimodal/evaluation/evaluate_video_mvbench.py:36
↓ 1 callers
Function
check_arguments
Checking user arguments.
Megatron-LM/examples/post_training/modelopt/validate.py:59
↓ 1 callers
Function
check_arguments
Checking user arguments.
Megatron-LM/examples/post_training/modelopt/convert_model.py:85
↓ 1 callers
Function
check_arguments
Checking user arguments.
Megatron-LM/examples/post_training/modelopt/generate.py:39
↓ 1 callers
Function
check_arguments
Checking user arguments.
Megatron-LM/examples/post_training/modelopt/prune.py:133
↓ 1 callers
Function
check_arguments
Checking user arguments.
Megatron-LM/examples/post_training/modelopt/quantize.py:88
↓ 1 callers
Function
check_checkpoint_args
Ensure fixed arguments for a model are the same for the input arguments and the one retrieved from checkpoint.
Megatron-LM/megatron/training/checkpointing.py:83
↓ 1 callers
Function
check_gpu_memory
Check if the GPU memory is over the threshold. Args: threshold (float, optional): The threshold to check if the GPU memory is over.
Megatron-LM/tools/checkpoint/checkpoint_inspector.py:246
↓ 1 callers
Function
check_gpu_memory
Check if the GPU memory is over the threshold. Args: threshold (float, optional): The threshold to check if the GPU memory is over.
Megatron-LM/megatron/core/distributed/fsdp/src/megatron_fsdp/param_and_grad_buffer.py:3653
↓ 1 callers
Method
check_grads
Make sure norm of grads in bucket are not NaN prior to data-parallel all-reduce / reduce-scatter.
Megatron-LM/megatron/core/distributed/param_and_grad_buffer.py:211
↓ 1 callers
Function
check_has_nvlink
()
slime/slime/utils/external_utils/command_utils.py:195
↓ 1 callers
Function
check_is_number
Check if the given string a number.
Megatron-LM/examples/multimodal/evaluation/mmmu_utils.py:206
↓ 1 callers
Method
check_message
(self, msg)
Megatron-LM/tools/checkpoint/saver_hf_llava.py:58
↓ 1 callers
Function
check_mtp_loss
Check that MTP loss is within expected bounds. Args: mtp_loss: The computed MTP loss value. max_mtp_loss: Maximum allowed MTP los
slime/slime/backends/megatron_utils/ci_utils.py:71
↓ 1 callers
Function
check_mtp_only_grad
Check that only MTP parameters have non-zero gradients. This is used for CI testing to verify that when all outputs are truncated, only the M
slime/slime/backends/megatron_utils/ci_utils.py:11
↓ 1 callers
Function
checkpoint_and_decide_exit
Save checkpoint and decide whether to exit based on arguments (e.g., if --exit-duration-in-mins is set). Actual exit happens in main training loop
Megatron-LM/megatron/training/training.py:1999
↓ 1 callers
Function
checkpoint_forward
(input)
Megatron-LM/tests/unit_tests/tensor_parallel/test_random.py:209
↓ 1 callers
Function
chunk_named_params_by_size
(named_params: Iterable[tuple[str, torch.Tensor]], chunk_size: int)
slime/slime/utils/misc.py:123
↓ 1 callers
Function
ci_is_busy
List all merge request pipelines created before the given pipeline that are still pending or running.
Megatron-LM/tests/test_utils/python_scripts/wait_for_resources.py:27
↓ 1 callers
Method
clean_loss_in_tracker
Clear the indexer losses.
Megatron-LM/megatron/core/transformer/experimental_attention_variant/dsa.py:83
← previous
next →
3,501–3,600 of 10,561, ranked by callers