MCPcopy Create free account

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

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

↓ 1 callersMethodsharded_state_dict
Generate a sharded state dictionary for the transformer layer. Args: prefix (str, optional): Prefix to be added to all k
Megatron-LM/megatron/core/transformer/transformer_layer.py:755
↓ 1 callersMethodsharded_state_dict
Maps local expert to global experts. The sharded_state_dict for the weight parts are compatible with the SequentialMLP, where
Megatron-LM/megatron/core/transformer/moe/experts.py:296
↓ 1 callersFunctionsharded_tensor_to_torch_sharded_tensor
Convert MCore ShardedTensor to PyT ShardedTensor. PyT requires information about all chunks. On high-level, this function follows the logic of
Megatron-LM/megatron/core/dist_checkpointing/strategies/torch.py:135
↓ 1 callersFunctionshould_allow_eval
(expr: str)
slime/slime/rollout/rm_hub/math_utils.py:330
↓ 1 callersMethodshould_ask_confirmation
(self, action: str)
swe-rl/mini-swe-agent/src/minisweagent/agents/interactive.py:99
↓ 1 callersMethodshould_bulk_offload
Determine if the current group should be offloaded.
Megatron-LM/megatron/core/pipeline_parallel/fine_grained_activation_offload.py:389
↓ 1 callersMethodshould_checkpoint_and_exit
Method instructing whether to checkpoint and/or abort the job. Args: None Returns: A tuple formed of:
Megatron-LM/megatron/core/rerun_state_machine.py:375
↓ 1 callersFunctionshould_free_input
Determine if the node should free its input memory. Args: name: Node name is_moe: Whether it's a MoE model enable_deepep:
Megatron-LM/megatron/core/models/gpt/fine_grained_callables.py:45
↓ 1 callersFunctionshould_keep_expert_key
Determine if this rank should keep this expert key based on expert index
Megatron-LM/megatron/core/transformer/fsdp_dtensor_checkpoint.py:119
↓ 1 callersFunctionshould_skip_change
Determine if a breaking change should be skipped. A change is skipped if: - The change kind is in IGNORED_BREAKAGE_KINDS (not a signature
Megatron-LM/scripts/check_api_backwards_compatibility.py:224
↓ 1 callersMethodshutdown
(self)
terminal-rl/router_server.py:61
↓ 1 callersMethodshutdown
(self)
terminal-rl/remote/pool_server.py:335
↓ 1 callersMethodshutdown
Shutdown the queue for Python < 3.13. Note that the listening side of the queue can continue to get old data off the queue ev
Megatron-LM/megatron/core/inference/utils.py:204
↓ 1 callersMethodshutdown
(self)
slime/slime/backends/sglang_utils/sglang_engine.py:316
↓ 1 callersMethodsignals_received
(self)
Megatron-LM/megatron/training/dist_signal_handler.py:54
↓ 1 callersFunctionsimilarity_function
(prediction: str, gold_label: str, threshold: float)
Megatron-LM/examples/multimodal/evaluation/evaluate_vqav2.py:34
↓ 1 callersFunctionsimple_speculative_generate
A simple generate function without using KV-cache.
Megatron-LM/megatron/post_training/generate.py:102
↓ 1 callersFunctionsinkhorn
(cost, tol=0.0001)
Megatron-LM/megatron/legacy/model/transformer.py:179
↓ 1 callersFunctionsinkhorn
Sinkhorn based MoE routing function
Megatron-LM/megatron/core/transformer/moe/moe_utils.py:138
↓ 1 callersMethodsinkhorn_load_balancing
Apply sinkhorn routing to the logits tensor. Args: logits (torch.Tensor): The logits tensor. Returns: Tuple[
Megatron-LM/megatron/core/transformer/moe/router.py:218
↓ 1 callersFunctionslide_fill_distance_to_rgb
(_slide, _rgb, _original_rgb)
gui-rl/desktop_env/evaluators/metrics/slides.py:780
↓ 1 callersFunctionslime_validate_args
(args)
slime/slime/utils/arguments.py:1801
↓ 1 callersFunctionsolver_worker
单组 solver 流程。
slime/examples/multi_agent/agent_system.py:171
↓ 1 callersFunctionsplit_expert_weights
Split expert weights into separate tensors for each expert.
Megatron-LM/tools/checkpoint/checkpoint_inspector.py:411
↓ 1 callersFunctionsplit_layers
Split layers into separate tensors.
Megatron-LM/tools/checkpoint/checkpoint_inspector.py:391
↓ 1 callersFunctionsplit_lines
(lines, split)
Megatron-LM/examples/academic_paper_scripts/detxoify_lm/annotations/perspective_api_annotate.py:79
↓ 1 callersMethodsplit_state_dict_if_needed
When "--fp8-param-gather" is disabled, weights and biases are stored in the same `_ParamAndGradBuffer`. So, when saving a checkpoint,
Megatron-LM/megatron/core/optimizer/distrib_optimizer.py:2104
↓ 1 callersFunctionsplit_swiglu_weight
Split SwiGLU weights into separate tensors.
Megatron-LM/tools/checkpoint/checkpoint_inspector.py:471
↓ 1 callersFunctionsplit_te_layernorm_column_parallel_linear
Split a TELayerNormColumnParallelLinear into separate TENorm and TEColumnParallelLinear layers. Args: fused_layer: The fused TELayer
Megatron-LM/megatron/core/extensions/transformer_engine.py:84
↓ 1 callersFunctionsplit_tensor_along_last_dim
Split a tensor along its last dimension. Args: tensor: input tensor. num_partitions: number of partitions to split the tensor
Megatron-LM/megatron/core/tensor_parallel/utils.py:22
↓ 1 callersFunctionsplit_tensor_for_tp
(params, key, dim, tensor)
Megatron-LM/tools/checkpoint/hybrid_conversion.py:114
↓ 1 callersFunctionsplit_tensor_into_1d_equal_chunks
Break a tensor into equal 1D chunks across tensor parallel ranks. Returns a Tensor or View with this rank's portion of the data. Args:
Megatron-LM/megatron/core/tensor_parallel/utils.py:48
↓ 1 callersMethodssm_training
Performs SSM computation for training step. Uses the memory-efficient kernel `mamba_split_conv1d_scan_combined` which reduces the si
Megatron-LM/megatron/core/ssm/mamba_mixer.py:597
↓ 1 callersMethodstart
(self)
openclaw-rl/openclaw_rollout.py:52
↓ 1 callersMethodstart
(self)
openclaw-rl/oel/openclaw_oel_rollout.py:65
↓ 1 callersMethodstart
(self)
openclaw-opd/openclaw_opd_rollout.py:56
↓ 1 callersMethodstart
(self)
openclaw-fireworks/fireworks_server.py:923
↓ 1 callersMethodstart
(self)
openclaw-combine/openclaw_combine_rollout.py:56
↓ 1 callersMethodstart
(self)
openclaw-combine/openclaw_combine_select_rollout.py:70
↓ 1 callersMethodstart
(self)
openclaw-tinker/rollout.py:61
↓ 1 callersMethodstart
Starts the main event loop for the coordinator. This method runs an infinite loop, continuously listening for incoming messa
Megatron-LM/megatron/core/inference/data_parallel_inference_coordinator.py:129
↓ 1 callersMethodstart_emulator
Method to start the emulator.
gui-rl/desktop_env/providers/base.py:12
↓ 1 callersMethodstart_grad_sync
Initiates grad sync (all-reduce or reduce-scatter) communication operations for all model gradients. When overlap_grad_reduc
Megatron-LM/megatron/core/distributed/fsdp/src/megatron_fsdp/megatron_fsdp.py:1052
↓ 1 callersMethodstart_param_sync
Initiates param sync (all-gather) communication operations for all model parameters. By default, when overlap_param_gather is set to
Megatron-LM/megatron/core/distributed/distributed_data_parallel.py:460
↓ 1 callersMethodstart_recording
(self, lease_id: str)
gui-rl/env_client.py:80
↓ 1 callersMethodstart_recording
(self, lease_id: str)
gui-rl/env_pool_server.py:251
↓ 1 callersFunctionstart_rollout
(api_base_url: str, args, metadata)
slime/slime_plugins/rollout_buffer/rollout_buffer_example.py:173
↓ 1 callersMethodstart_turn_loop
(self, input_message: Any)
terminal-rl/rollout_agent.py:15
↓ 1 callersFunctionstart_vm
(vm_name, max_retries=20)
gui-rl/desktop_env/providers/virtualbox/manager.py:216
↓ 1 callersFunctionstart_vm
(vm_path, max_retries=20)
gui-rl/desktop_env/providers/vmware/manager.py:207
↓ 1 callersMethodstartup
(self)
terminal-rl/router_server.py:55
↓ 1 callersMethodstate_dict
Return the state dict.
Megatron-LM/megatron/core/optimizer_param_scheduler.py:209
↓ 1 callersMethodstate_dict
Returns a dictionary containing references to the whole state of the wrapped module. Both parameters and persistent buffers
Megatron-LM/megatron/core/distributed/data_parallel_base.py:73
↓ 1 callersMethodstate_dict
(self, is_loading: bool = False)
Megatron-LM/megatron/core/optimizer/optimizer.py:789
↓ 1 callersMethodstatus
(self)
terminal-rl/remote/pool_server.py:300
↓ 1 callersMethodstep
(self, lease_id: str, action: Any, sleep_after_execution: float = 0.0)
gui-rl/env_pool_server.py:231
↓ 1 callersMethodstep
Query the LM, execute the action, return the observation.
swe-rl/mini-swe-agent/src/minisweagent/agents/default.py:88
↓ 1 callersMethodstep
step function for layer-wise optimizer.
Megatron-LM/megatron/core/optimizer/layer_wise_optimizer.py:222
↓ 1 callersMethodstep_with_ready_grads
Step the optimizer with ready gradients, return successful.
Megatron-LM/megatron/core/optimizer/optimizer.py:560
↓ 1 callersMethodstep_with_ready_grads
Step the optimizer with ready gradients, return successful.
Megatron-LM/megatron/core/optimizer/optimizer.py:1242
↓ 1 callersMethodstop
(self)
openclaw-rl/openclaw_rollout.py:58
↓ 1 callersMethodstop
(self)
openclaw-rl/oel/openclaw_oel_rollout.py:71
↓ 1 callersMethodstop
(self)
openclaw-opd/openclaw_opd_rollout.py:62
↓ 1 callersMethodstop
(self)
openclaw-fireworks/fireworks_server.py:952
↓ 1 callersMethodstop
(self)
openclaw-combine/openclaw_combine_rollout.py:62
↓ 1 callersMethodstop
(self)
openclaw-combine/openclaw_combine_select_rollout.py:78
↓ 1 callersMethodstop
Stops the inference coordinator, performing any necessary cleanup operations.
Megatron-LM/megatron/core/inference/data_parallel_inference_coordinator.py:306
↓ 1 callersMethodstop
Stops the inference engine by terminating the inference coordinator process if it exists, and destroys the model parallel state.
Megatron-LM/megatron/core/inference/engines/dynamic_engine.py:1393
↓ 1 callersMethodstop_emulator
(self, path_to_vm: str, region=None, *args, **kwargs)
gui-rl/desktop_env/providers/docker/provider.py:156
↓ 1 callersFunctionstr_dtype_to_torch
Get torch datatype from input datatype
Megatron-LM/megatron/core/export/trtllm/trtllm_weights_converter/single_device_trtllm_model_weights_converter.py:33
↓ 1 callersFunctionstr_dtype_to_torch
Get torch datatype from input datatype
Megatron-LM/megatron/core/export/trtllm/trtllm_weights_converter/distributed_trtllm_model_weights_converter.py:23
↓ 1 callersFunctionstraggler_detector_disabled
()
Megatron-LM/tests/unit_tests/test_utils.py:298
↓ 1 callersFunctionstraggler_detector_enable
()
Megatron-LM/tests/unit_tests/test_utils.py:306
↓ 1 callersFunctionstraggler_detector_enabled
()
Megatron-LM/tests/unit_tests/test_utils.py:302
↓ 1 callersFunctionstraggler_detector_exception_propagate
()
Megatron-LM/tests/unit_tests/test_utils.py:336
↓ 1 callersFunctionstraggler_detector_report
()
Megatron-LM/tests/unit_tests/test_utils.py:347
↓ 1 callersFunctionstraggler_detector_timeit
()
Megatron-LM/tests/unit_tests/test_utils.py:314
↓ 1 callersFunctionstraggler_value_error
()
Megatron-LM/tests/unit_tests/test_utils.py:332
↓ 1 callersFunctionstrip_thinking
Remove <think>...</think> blocks from model output.
openclaw-test/test_sglang.py:22
↓ 1 callersFunctionstrip_thinking
Remove <think>...</think> blocks from model output.
openclaw-test/teacher_chat.py:109
↓ 1 callersFunctionstrip_thinking
Remove <think>...</think> blocks from model output.
openclaw-test/TA_chat.py:104
↓ 1 callersFunctionstrip_thinking
Remove <think>...</think> blocks from model output.
openclaw-test/student_chat.py:69
↓ 1 callersFunctionstructure_check_by_mse
Check if two images are approximately the same by MSE
gui-rl/desktop_env/evaluators/metrics/gimp.py:201
↓ 1 callersMethodsubmit_generate_tasks
(self, samples: list[list[Sample]])
slime/slime/rollout/sglang_rollout.py:86
↓ 1 callersMethodsubmit_step_judge
( self, args: Any, *, instruction: str, actions_history: list[str],
gui-rl/agents/qwen3vl_reward_agent.py:379
↓ 1 callersMethodsuspend
Suspend engine by deallocating context's GPU state.
Megatron-LM/megatron/core/inference/engines/dynamic_engine.py:539
↓ 1 callersMethodswap_labels
Get filtered reviewer emails who haven't approved yet.
Megatron-LM/tests/test_utils/python_scripts/swap_pr_labels.py:52
↓ 1 callersMethodsync
Context manager that synchronizes the MegatronFSDP model parameters and gradients every training step as opposed to every optimizatio
Megatron-LM/megatron/core/distributed/fsdp/src/megatron_fsdp/megatron_fsdp.py:971
↓ 1 callersMethodsync_rng_states_across_tp_group
Synchronize the tensor parallel random number generator states.
Megatron-LM/megatron/core/distributed/fsdp/mcore_fsdp_adapter.py:296
↓ 1 callersFunctiont5_extended_attention_mask
(attention_mask_list)
Megatron-LM/megatron/legacy/model/t5_model.py:19
↓ 1 callersMethodtarget_has_trainable_tokens
(self, input_ids, num_tiles, target)
Megatron-LM/examples/multimodal/dataset_helpers.py:533
↓ 1 callersFunctionte_parallel_cross_entropy
Wrapper function for TE's Cross Entropy Loss kernel
Megatron-LM/megatron/core/extensions/transformer_engine.py:2187
↓ 1 callersMethodteardown_method
(self, method)
Megatron-LM/tests/unit_tests/inference/test_data_parallel_inference_coordinator.py:244
↓ 1 callersMethodteardown_method
(self, method)
Megatron-LM/tests/unit_tests/transformer/moe/test_grouped_mlp.py:282
↓ 1 callersMethodteardown_method
(self, method)
Megatron-LM/tests/unit_tests/transformer/moe/test_sequential_mlp.py:47
↓ 1 callersFunctiontensor_fp8
(weight)
slime/tools/convert_hf_to_fp8.py:81
↓ 1 callersMethodtensor_pop
Pop tensor from the offload handler.
Megatron-LM/megatron/core/pipeline_parallel/fine_grained_activation_offload.py:303
↓ 1 callersMethodtensor_push
Push tensor to the offload handler.
Megatron-LM/megatron/core/pipeline_parallel/fine_grained_activation_offload.py:279
↓ 1 callersFunctiontest
()
Megatron-LM/examples/academic_paper_scripts/detxoify_lm/perspective_api.py:73
↓ 1 callersMethodtest
( num_layers, hidden_size, num_attention_heads,
Megatron-LM/tests/unit_tests/transformer/test_transformer_layer.py:66
← previousnext →5,101–5,200 of 10,561, ranked by callers