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
Method
rollout
(self, request: RolloutRequest)
Megatron-LM/megatron/rl/agent/reward_only_agent.py:124
Method
rollout_engines
(self)
slime/slime/ray/rollout.py:131
Function
rollout_log
(rollout_id, args, samples, rollout_extra_metrics, rollout_time)
terminal-rl/rollout_log.py:23
Method
rollouts
(request: RemoteRolloutRequest)
Megatron-LM/megatron/rl/server/agent/fastapi_env_server.py:76
Function
rotary_bwd_kv_kernel
Triton kernel of the backward pass for applying YARN RoPE to MLA's key and value. Input: dK: [seq_len, batch_size, head_num, emb_dim
Megatron-LM/megatron/core/fusions/fused_mla_yarn_rope_apply.py:496
Function
rotary_bwd_q_kernel
Triton kernel of the backward pass for applying YARN RoPE to MLA's query. This kernel inplace modifies the input tensor DO. Input:
Megatron-LM/megatron/core/fusions/fused_mla_yarn_rope_apply.py:148
Function
rotary_fwd_kv_kernel
Triton kernel of the forward pass for applying YARN RoPE to MLA's key and value. It splits the input tensor KV into key and value, and co
Megatron-LM/megatron/core/fusions/fused_mla_yarn_rope_apply.py:379
Function
rotary_fwd_q_kernel
Triton kernel of the forward pass for applying YARN RoPE to MLA's query. This kernel inplace modifies the input tensor Q. Input:
Megatron-LM/megatron/core/fusions/fused_mla_yarn_rope_apply.py:68
Method
rotate_with_fill
Define rotation transformation with fill. The input image is first rotated, then it is blended together with a gray mask of t
Megatron-LM/megatron/legacy/data/autoaugment.py:211
Method
routing
Routing function. Args: logits (torch.Tensor): Logits tensor. Returns: Tuple[torch.Tensor, torch.Tensor]: A
Megatron-LM/megatron/core/transformer/moe/router.py:103
Method
row_parallel_linear
Which row parallel linear module TE backend uses
Megatron-LM/megatron/core/extensions/transformer_engine_spec_provider.py:41
Method
row_parallel_linear
Which row parallel linear module kitchen backend uses
Megatron-LM/megatron/core/extensions/kitchen.py:1040
Method
row_parallel_linear
Which row parallel linear module the backend uses
Megatron-LM/megatron/core/models/backends.py:48
Function
rsync_simple
(path_src: str, path_dst: str)
slime/slime/utils/external_utils/command_utils.py:66
Method
rtol
(self)
Megatron-LM/tests/functional_tests/python_test_utils/common.py:53
Method
run
(self, task: str, **kwargs)
swe-rl/mini-swe-agent/src/minisweagent/agents/interactive_textual.py:62
Method
run
Model Chunk level 1f1b fine-grained scheduler. This function schedules the forward and backward passes for a model chunk, which inter
Megatron-LM/megatron/core/models/common/model_chunk_schedule_plan.py:435
Method
run
(*flat_tensor_args)
Megatron-LM/megatron/core/transformer/multi_token_prediction.py:858
Function
run_agent_system
并发运行 num_parallel 组 pipeline。
slime/examples/multi_agent/agent_system.py:187
Function
run_bash_script
()
gui-rl/desktop_env/server/main.py:1667
Method
run_bash_script
Executes a bash script on the server. :param script: The bash script content (can be multi-line) :param timeout: Exe
gui-rl/desktop_env/controllers/python.py:189
Method
run_engine
Continually steps the engine asynchronously.
Megatron-LM/megatron/core/inference/engines/dynamic_engine.py:1409
Method
run_evaluation
(request: RemoteEvaluationRequest)
Megatron-LM/megatron/rl/server/agent/fastapi_env_server.py:84
Method
run_evaluation
Run evaluation across all sub-agents.
Megatron-LM/megatron/rl/agent/weighted_multi_task.py:266
Method
run_evaluation
(self, request: EvaluationRequest)
Megatron-LM/megatron/rl/agent/reward_only_agent.py:191
Method
run_one_forward_step
The forward pass of the model for inference Appropriate utility is called for the forward pass depending on the type of model parallelism use
Megatron-LM/megatron/core/inference/model_inference_wrappers/abstract_model_inference_wrapper.py:352
Function
run_online_eval
Run an evaluation benchmark during training.
Megatron-LM/examples/multimodal/train.py:326
Function
run_python
()
gui-rl/desktop_env/server/main.py:1572
Method
run_python_script
Executes a python script on the server.
gui-rl/desktop_env/controllers/python.py:165
Function
run_rollout
(data: dict)
slime/slime_plugins/rollout_buffer/buffer.py:298
Function
run_rollout
(data: dict)
slime/slime_plugins/rollout_buffer/generator/base_generator.py:272
Function
run_router
(args)
slime/slime/utils/http_utils.py:117
Function
run_sqlite3
(result: str, rules: Dict[str, Any])
gui-rl/desktop_env/evaluators/metrics/general.py:273
Function
sample_loader
Load the sample from the raw data. Example content of wds data: - Example of wds data 100_100.img 100_100.jso
Megatron-LM/examples/mimo/data/avlm_sample_loader.py:3
Function
sample_simple_trajectory
Sample trajectory in simple format (list of messages).
swe-rl/mini-swe-agent/tests/run/test_inspector.py:32
Function
sample_swebench_trajectory
Sample trajectory in SWEBench format (dict with messages array).
swe-rl/mini-swe-agent/tests/run/test_inspector.py:47
Function
sanitize_search_query
(query: str)
slime/examples/search-r1/google_search_server.py:16
Function
save
Save checkpoint to disk. Saves model weights and optimizer state to separate directories. This allows loading weights without optimizer or de
slime/slime/backends/fsdp_utils/checkpoint.py:203
Method
save
(self, rollout_id)
gui-rl/gui_data_source.py:257
Method
save
Orchestrate the entire flow of saving the Megatron checkpoint.
Megatron-LM/tools/checkpoint/saver_base.py:309
Method
save
Each async strategy can be trivially used as a sync strategy.
Megatron-LM/megatron/core/dist_checkpointing/strategies/base.py:221
Method
save
(self, sharded_state_dict: ShardedStateDict, checkpoint_dir: Union[str, Path])
Megatron-LM/megatron/core/dist_checkpointing/strategies/zarr.py:85
Method
save
(self, rollout_id)
slime/slime/rollout/data_source.py:117
Function
save_checkpoint
(queue, args)
Megatron-LM/tools/checkpoint/saver_legacy.py:28
Function
save_checkpoint
Required top-level function that creates the saver and calls its .save().
Megatron-LM/tools/checkpoint/saver_hf_llava.py:393
Function
save_checkpoint
Required top-level function that creates the saver and calls its .save().
Megatron-LM/tools/checkpoint/saver_llava.py:398
Method
save_common
Save common part of the state dict.
Megatron-LM/megatron/core/dist_checkpointing/strategies/common.py:40
Method
save_custom_attrs
(module)
Megatron-LM/megatron/core/distributed/torch_fully_sharded_data_parallel.py:89
Function
save_debug_train_data
(args, *, rollout_id, rollout_data)
slime/slime/utils/train_dump_utils.py:9
Method
save_model
Save actor model
slime/slime/ray/actor_group.py:119
Method
save_model
(self, rollout_id: int, force_sync: bool = False)
slime/slime/backends/megatron_utils/actor.py:1415
Method
save_model
Delegate checkpoint saving to the shared checkpoint utilities.
slime/slime/backends/fsdp_utils/actor.py:323
Method
save_parameter_state
Save the distributed parameter state on DP rank 0. Args: filename (str): path to save parameter state to.
Megatron-LM/megatron/core/optimizer/distrib_optimizer.py:1117
Method
save_parameter_state
(self, *args, **kwargs)
Megatron-LM/tests/unit_tests/test_checkpointing.py:66
Method
save_pretrained
Saves tokenizer's vocabulary and other artifacts to the specified directory
Megatron-LM/megatron/core/tokenizers/text/libraries/huggingface_tokenizer.py:338
Method
save_sharded_objects
Save sharded objects from the state dict.
Megatron-LM/megatron/core/dist_checkpointing/strategies/common.py:50
Method
save_state
(self, path_to_vm: str, snapshot_name: str)
gui-rl/desktop_env/providers/azure/provider.py:85
Method
save_state
(self, path_to_vm: str, snapshot_name: str)
gui-rl/desktop_env/providers/aws/provider.py:90
Method
save_state
(self, path_to_vm: str, snapshot_name: str)
gui-rl/desktop_env/providers/docker/provider.py:150
Method
save_state
(self, path_to_vm: str, snapshot_name: str)
gui-rl/desktop_env/providers/virtualbox/provider.py:103
Method
save_state
(self, path_to_vm: str, snapshot_name: str)
gui-rl/desktop_env/providers/volcengine/provider.py:88
Method
save_state
(self, path_to_vm: str, snapshot_name: str)
gui-rl/desktop_env/providers/vmware/provider.py:87
Method
save_state
(self, path_to_vm: str, snapshot_name: str)
gui-rl/desktop_env/providers/aliyun/provider.py:145
Function
save_to_temp_file
(text: str, ext: str)
slime/slime/utils/external_utils/command_utils.py:254
Method
save_vocabulary
Saves tokenizer's vocabulary and other artifacts to the specified directory
Megatron-LM/megatron/core/tokenizers/text/libraries/huggingface_tokenizer.py:332
Method
save_vocabulary
Save the tokenizer vocabulary and merge files to a directory.
Megatron-LM/megatron/training/tokenizer/gpt2_tokenization.py:291
Method
scale
(self)
Megatron-LM/megatron/core/optimizer/grad_scaler.py:18
Method
scale_gradients
Scale the gradient data by `scaling_factor`.
Megatron-LM/megatron/core/distributed/param_and_grad_buffer.py:863
Method
scale_gradients
Scale all gradients inside the buffers by `scaling_factor`.
Megatron-LM/megatron/core/distributed/distributed_data_parallel.py:534
Method
scale_gradients
Scale all gradients inside the buffers by `scaling_factor`.
Megatron-LM/megatron/core/distributed/fsdp/src/megatron_fsdp/megatron_fsdp.py:1192
Method
scale_gradients
Scale the gradient data by `scaling_factor`.
Megatron-LM/megatron/core/distributed/fsdp/src/megatron_fsdp/param_and_grad_buffer.py:2376
Method
scale_loss
Simple scaling.
Megatron-LM/megatron/core/optimizer/optimizer.py:246
Function
scaled_loss_func
Scaled loss function Scale the loss for each conversation turn using the formula: 1 / sum_j[ sqrt(length(loss_turn_j)) ] * sum_i[ sum(l
Megatron-LM/examples/multimodal/train.py:193
Method
schedule_async_call
Spawn a process with `async_fn` as the target. This method must be called on all ranks. Args: async_fn (Callable, option
Megatron-LM/megatron/core/dist_checkpointing/strategies/async_utils.py:187
Method
schedule_async_call
Put `AsyncRequest` to the Persistent Async Caller This method must be called on all ranks. Args: async_fn (Callable, opt
Megatron-LM/megatron/core/dist_checkpointing/strategies/async_utils.py:303
Function
scope
Simple fixture returning golden values.
Megatron-LM/tests/functional_tests/python_test_utils/conftest.py:72
Method
search_mips_index
Get the top-k blocks by the index distance metric. :param reconstruct: if True: return a [num_queries x k x embed_dim]
Megatron-LM/megatron/legacy/data/realm_index.py:206
Method
select_model
(self)
swe-rl/mini-swe-agent/src/minisweagent/models/extra/roulette.py:57
Method
select_samples_to_pack
Selects which samples will be packed together. NOTE: Energon dataloader calls this method internally if packing is used. Please see h
Megatron-LM/examples/multimodal/dataset_helpers.py:869
Function
selective_log_softmax_raw
Fused version of the common `log_softmax -> gather` operation. The fused version of this operation avoids the (potentially large) memory overhead
slime/slime/backends/fsdp_utils/actor.py:855
Method
send_model_over_queue
(self)
Megatron-LM/tools/checkpoint/loader_core.py:69
Method
send_model_over_queue
(self)
Megatron-LM/tools/checkpoint/loader_llava.py:311
Method
sep
Returns id of SEP token.
Megatron-LM/megatron/core/tokenizers/text/text_tokenizer.py:236
Method
sep
Returns sep token.
Megatron-LM/megatron/core/tokenizers/text/libraries/null_tokenizer.py:70
Method
sep
The SEP token id Raises: NotImplementedError: Non-abstract, optional attribute
Megatron-LM/megatron/core/datasets/megatron_tokenizer.py:111
Method
sep
(self)
Megatron-LM/megatron/training/tokenizer/tokenizer.py:279
Method
sep
(self)
Megatron-LM/megatron/training/tokenizer/tokenizer.py:510
Method
sep
(self)
Megatron-LM/megatron/training/tokenizer/tokenizer.py:562
Method
sep
(self)
Megatron-LM/megatron/training/tokenizer/tokenizer.py:616
Method
sep
(self)
Megatron-LM/megatron/training/tokenizer/tokenizer.py:855
Method
sep
(self)
Megatron-LM/megatron/training/tokenizer/tokenizer.py:928
Method
sep_id
Returns id of SEP token. Need for NeMo.
Megatron-LM/megatron/core/tokenizers/text/text_tokenizer.py:241
Method
sep_id
Returns id of SEP token.
Megatron-LM/megatron/core/tokenizers/text/libraries/huggingface_tokenizer.py:305
Method
sep_id
Returns id of end of SEP token.
Megatron-LM/megatron/core/tokenizers/text/libraries/sentencepiece_tokenizer.py:341
Method
sep_id
Property alias to match MegatronTokenizer; returns sep_id if available.
Megatron-LM/megatron/core/tokenizers/text/libraries/abstract_tokenizer.py:103
Method
sep_id
Returns id of SEP token.
Megatron-LM/megatron/core/tokenizers/text/libraries/tiktoken_tokenizer.py:276
Method
sequence_lengths
Get the sequence lengths Returns: numpy.ndarray: The sequence lengths
Megatron-LM/megatron/core/datasets/indexed_dataset.py:764
Method
sequence_modes
Get the sequence modes Returns: numpy.ndarray: The sequence modes
Megatron-LM/megatron/core/datasets/indexed_dataset.py:802
Method
serialize
Converts the instance into a serializable dictionary. Returns: (dict) A dictionary representation of the instance suitable for
Megatron-LM/megatron/core/inference/inference_request.py:188
Method
serialize
Converts the instance into a serializable dictionary. Returns: (dict) A dictionary representation of the instance suitable for
Megatron-LM/megatron/core/inference/inference_request.py:273
← previous
next →
8,601–8,700 of 10,561, ranked by callers