MCPcopy Create free account

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

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

↓ 1 callersFunction_merge_partitioned_rollout_refs
Merge actor-DP rollout refs for single-DP PRM teacher passes.
openclaw-rl/oel/train_async.py:10
↓ 1 callersFunction_merge_partitioned_rollout_refs
Merge actor-DP rollout refs for single-DP PRM teacher passes.
slime/train_async.py:10
↓ 1 callersFunction_merge_student_topk_payloads
Merge one student top-k payload per actor DP rank into original order.
openclaw-rl/oel/train_async.py:40
↓ 1 callersFunction_merge_student_topk_payloads
Merge one student top-k payload per actor DP rank into original order.
slime/train_async.py:40
↓ 1 callersMethod_min_max
Helper function to find the min/max values Args: ptime (float): avg per iteration gpu time btime (float): avg per ite
Megatron-LM/megatron/core/utils.py:1725
↓ 1 callersMethod_mock_tokenize
Generate a mock token sequence consisting of ``image_seq_length`` image tokens followed by randomly generated text tokens such that t
Megatron-LM/examples/mimo/data/mock.py:134
↓ 1 callersFunction_model_provider
Build the model.
Megatron-LM/tests/unit_tests/export/trtllm/test_single_device_fp8.py:27
↓ 1 callersFunction_model_provider
Build the model.
Megatron-LM/tests/unit_tests/export/trtllm/test_distributed_fp8.py:29
↓ 1 callersFunction_modify_state_dict
Modify state dict items in a Megatron Core Distributed Checkpoint.
Megatron-LM/tools/checkpoint/checkpoint_inspector.py:765
↓ 1 callersFunction_modify_underlying_storage_impl
( fp8_tensor: QuantizedTensor, new_raw_data: torch.Tensor )
Megatron-LM/megatron/core/fp8_utils.py:226
↓ 1 callersMethod_move_new_state_to_right_device
(self)
Megatron-LM/megatron/core/optimizer/cpu_offloading/hybrid_optimizer.py:357
↓ 1 callersFunction_named_params_and_buffers_global
Yield (global_name, param/buffer) with consistent names across PP/EP. Adjusts indices for virtual PP + EP offsets. Handles decoder.layers, mt
slime/slime/backends/megatron_utils/update_weight/common.py:174
↓ 1 callersFunction_named_params_and_buffers_vanilla
(model: Sequence[torch.nn.Module])
slime/slime/backends/megatron_utils/update_weight/common.py:158
↓ 1 callersMethod_new_env
(self)
terminal-rl/remote/pool_server.py:108
↓ 1 callersFunction_next_actor
()
slime/slime/utils/http_utils.py:156
↓ 1 callersMethod_norm
(self, x)
Megatron-LM/megatron/legacy/model/rms_norm.py:27
↓ 1 callersMethod_norm
Performs the actual L2 normalization. Args: x (torch.Tensor): The input tensor to normalize. Returns:
Megatron-LM/megatron/core/transformer/torch_norm.py:73
↓ 1 callersMethod_norm
(self, x, var)
Megatron-LM/examples/multimodal/nvlm/internvit.py:89
↓ 1 callersFunction_normalize_instruction_ids
Ensure instruction identifiers are clean strings.
slime/slime/rollout/rm_hub/ifbench.py:86
↓ 1 callersFunction_normalize_tool_call
Ensure tool_call.function.arguments is a dict so Jinja2 |items works.
openclaw-rl/openclaw_api_server.py:68
↓ 1 callersFunction_normalize_tool_call
(tc: dict)
openclaw-rl/oel/openclaw_oel_api_server.py:153
↓ 1 callersFunction_normalize_tool_call
Ensure tool_call.function.arguments is a dict so Jinja2 |items works.
openclaw-opd/openclaw_opd_api_server.py:63
↓ 1 callersFunction_normalize_tool_call
(tc: dict)
openclaw-fireworks/fireworks_server.py:93
↓ 1 callersFunction_nvtx_decorator_get_func_path
Get the path of a function. Args: func (Callable): Function to get path for. Returns: str: Module path and function name joi
Megatron-LM/megatron/core/utils.py:2186
↓ 1 callersFunction_offset_slice
(s, offset)
Megatron-LM/megatron/core/distributed/fsdp/src/megatron_fsdp/uneven_dtensor.py:392
↓ 1 callersMethod_opd_evaluate
(self, session_id: str, turn_num: int, turn_data: dict[str, Any], next_state: dict[str, Any])
openclaw-opd/openclaw_opd_api_server.py:586
↓ 1 callersMethod_opd_evaluate
( self, session_id: str, turn_num: int, turn_data: dict[str, Any], nex
openclaw-fireworks/fireworks_server.py:503
↓ 1 callersFunction_opd_teacher_source
()
openclaw-combine/combine_loss.py:40
↓ 1 callersFunction_os_execvp
(args)
slime/tests/ci/gpu_lock_exec.py:31
↓ 1 callersMethod_packed_data
Pack variable-length sequences for efficient processing. Parameters: rollout_data: Dictionary of lists containing sequence-level
slime/slime/backends/fsdp_utils/actor.py:396
↓ 1 callersFunction_pad_if_needed
(data_index: int)
Megatron-LM/megatron/core/distributed/fsdp/src/megatron_fsdp/param_and_grad_buffer.py:279
↓ 1 callersFunction_pad_param_if_needed
(numel_unpadded)
Megatron-LM/tests/unit_tests/distributed/test_param_and_grad_buffer.py:101
↓ 1 callersMethod_pad_routing_map
Pad the routing map to the nearest multiple of the pad_multiple.
Megatron-LM/megatron/core/transformer/moe/token_dispatcher.py:1254
↓ 1 callersMethod_param2group_meta_to_param_groups
Convert a mapping of parameter names to group metadata to a list of parameter groups.
Megatron-LM/megatron/core/optimizer/distrib_optimizer.py:1295
↓ 1 callersMethod_param_groups_to_param2group_meta
Convert a parameter group to a mapping of parameter names to group metadata.
Megatron-LM/megatron/core/optimizer/distrib_optimizer.py:1283
↓ 1 callersFunction_parse_args
()
slime/tests/ci/gpu_lock_exec.py:38
↓ 1 callersFunction_parse_bash_action
Extract the bash command from a response containing ```bash ... ```.
swe-rl/generate_with_swe_trajectory.py:178
↓ 1 callersFunction_parse_devices
(s: str)
slime/tests/ci/gpu_lock_exec.py:190
↓ 1 callersFunction_parse_extra_env_vars
(text: str)
slime/slime/utils/external_utils/command_utils.py:188
↓ 1 callersFunction_parse_generalized_path
(s: str)
slime/slime/utils/data.py:71
↓ 1 callersFunction_parse_judge_result
(text: str)
openclaw-opd/openclaw_opd_api_server.py:120
↓ 1 callersFunction_parse_judge_result
(text: str)
openclaw-fireworks/fireworks_server.py:186
↓ 1 callersFunction_parse_latex
Attempts to parse latex to an expression sympy can read.
slime/slime/rollout/rm_hub/math_utils.py:177
↓ 1 callersFunction_parse_prm_eval_score
(text: str)
openclaw-rl/oel/openclaw_oel_api_server.py:209
↓ 1 callersFunction_parse_prm_eval_score
(text: str)
openclaw-opd/openclaw_opd_api_server.py:204
↓ 1 callersFunction_parse_prm_eval_score
(text: str)
openclaw-fireworks/fireworks_server.py:196
↓ 1 callersFunction_parse_prm_score
(text: str)
openclaw-rl/openclaw_api_server.py:139
↓ 1 callersFunction_parse_timeout_overrides
( base: TaskTimeouts, payload: dict[str, Any] | None )
terminal-rl/remote/pool_server.py:25
↓ 1 callersMethod_parse_tool
Parse tool calls from LLM response string. Args: response: Raw response text from sglang Returns: P
slime/examples/tau-bench/trainable_agents.py:103
↓ 1 callersMethod_parse_tool_payload
Parse tool payload strictly as JSON. Malformed payloads are rejected.
slime/examples/geo3k_vlm_multi_turn/env_geo3k.py:240
↓ 1 callersFunction_passages2string
Convert retrieval results to a formatted string. This function works with both google_search and local_search results.
slime/examples/search-r1/generate_with_search.py:44
↓ 1 callersMethod_permute
Take in a sample (np array w/ size (0,chunklength)) and perform a FIM transformation on it. Maintain the same sample length (if trans
Megatron-LM/megatron/training/datasets/fim_dataset.py:233
↓ 1 callersMethod_pick_and_reserve_node
Atomically pick the least-loaded healthy node and increment its counter. The counter is incremented optimistically BEFORE the container is
swe-rl/server/swe_env_pool_server.py:77
↓ 1 callersFunction_populate_shadow_model_dir
(source_dir: Path, target_dir: Path, hf_config)
slime/slime/backends/sglang_utils/qwen3_5.py:340
↓ 1 callersMethod_post_process_rewards
(self, samples: list[Sample] | list[list[Sample]])
slime/slime/ray/rollout.py:339
↓ 1 callersMethod_post_process_step_wise_rewards
Build and normalize step-wise rewards metadata for training. Returns: step_wise_step_rewards, step_wise_step_token_spans, step_wi
slime/slime/ray/rollout.py:453
↓ 1 callersMethod_postprocess
Postprocesses the output of the transformer layers.
Megatron-LM/megatron/core/transformer/multi_token_prediction.py:817
↓ 1 callersMethod_postprocess
( self, hidden_states, input_ids, position_ids, labels, rotary
slime/slime_plugins/megatron_bridge/qwen35_vl.py:442
↓ 1 callersMethod_prec_recall_f1_score
Compute precision, recall and f1 given a set of gold and prediction items. :param pred_items: iterable of predicted values :p
Megatron-LM/tools/retro/text_generation/metrics.py:36
↓ 1 callersFunction_prepare_initial_inputs
(sample: Sample, processor, tokenizer)
slime/examples/geo3k_vlm_multi_turn/rollout.py:154
↓ 1 callersMethod_prepare_inputs
(self, cp_size, tp_size, sequence_parallel, padding)
Megatron-LM/tests/unit_tests/models/test_llava_model.py:615
↓ 1 callersMethod_prepare_prompt_tokens
Prepare prompt text and tokenize it. Args: state: GenerateState instance with tokenizer messages: Conversati
slime/examples/tau-bench/trainable_agents.py:157
↓ 1 callersFunction_prepare_start_state
(sample: Sample, state, args: Any, sampling_params: dict)
slime/examples/geo3k_vlm_multi_turn/rollout.py:170
↓ 1 callersMethod_prewarm
(self, count: int)
gui-rl/env_pool_server.py:90
↓ 1 callersFunction_print_args
Print arguments.
Megatron-LM/megatron/training/arguments.py:1318
↓ 1 callersFunction_print_args
Print arguments.
Megatron-LM/megatron/training/yaml_arguments.py:365
↓ 1 callersMethod_print_node
Recursively print node structure.
slime/slime/router/middleware_hub/radix_tree.py:561
↓ 1 callersFunction_prm_eval_majority_vote
(scores: list[int | None])
openclaw-rl/oel/openclaw_oel_api_server.py:219
↓ 1 callersFunction_prm_eval_majority_vote
(scores: list[int | None])
openclaw-fireworks/fireworks_server.py:206
↓ 1 callersMethod_prm_evaluate
(self, session_id: str, turn_num: int, response_text: str, next_state)
openclaw-rl/openclaw_api_server.py:424
↓ 1 callersMethod_prm_record_turn
( self, turn_idx: int, interaction: Any, turn_result: TurnResult, )
terminal-rl/agent_runner.py:216
↓ 1 callersFunction_prm_vote
(args, judge_prompt: str, m: int)
toolcall-rl/generate_with_retool.py:474
↓ 1 callersMethod_prm_vote
( self, args: Any, messages_for_prm: List[Dict[str, Any]], step_index: int,
gui-rl/agents/qwen3vl_reward_agent.py:312
↓ 1 callersMethod_prm_vote
( self, args: Any, messages_for_prm: List[Dict[str, Any]], step_index: int,
swe-rl/swe_prm.py:319
↓ 1 callersMethod_process_and_pack_example
Process multiple raw data and pack them into fixed sequence length.
Megatron-LM/examples/post_training/modelopt/finetune.py:236
↓ 1 callersFunction_process_conversion_tasks
(vanilla_conversion_tasks, new_weight_dict)
slime/slime/backends/megatron_utils/update_weight/hf_weight_iterator_bridge.py:44
↓ 1 callersFunction_process_env_step
(env: BaseInteractionEnv, response_text: str, tokenizer, processor, args, sample_metadata)
slime/examples/geo3k_vlm_multi_turn/rollout.py:212
↓ 1 callersMethod_process_example
Apply the chat template and compute the answer-only loss mask.
Megatron-LM/examples/post_training/modelopt/finetune.py:264
↓ 1 callersFunction_process_fc1_weight_param
(param)
Megatron-LM/megatron/core/transformer/moe/upcycling_utils.py:165
↓ 1 callersFunction_process_xlsx_cf_operator
(operator: str, value: Any, ref: List[Any])
gui-rl/desktop_env/evaluators/metrics/utils.py:414
↓ 1 callersFunction_produce_e2e_metrics
Generate APP metrics for E2E tracking NOTE: always call this function after barrier call Args: log_throughput (bool, optional): if l
Megatron-LM/megatron/training/one_logger_utils.py:76
↓ 1 callersMethod_proj_and_transformer_layer
Concatenates embeddings with hidden states and then applies transformer layer forward.
Megatron-LM/megatron/core/transformer/multi_token_prediction.py:757
↓ 1 callersMethod_proxy_setup
Setup system-wide proxy configuration using proxy pool Args: client_password (str): Password for sudo operations, default
gui-rl/desktop_env/controllers/setup.py:529
↓ 1 callersMethod_qa_template
(self, data)
Megatron-LM/examples/multimodal/evaluation/evaluation_datasets.py:1107
↓ 1 callersFunction_quantize_param_shard_impl
( model_params: List[QuantizedTensor], main_params: List[torch.Tensor], start_offsets:
Megatron-LM/megatron/core/fp8_utils.py:233
↓ 1 callersMethod_query_eval_once
(self, messages: list[dict], vote_id: int)
openclaw-tinker/scorers.py:401
↓ 1 callersMethod_query_eval_once
(self, messages: list[dict], vote_id: int)
openclaw-tinker/scorers.py:485
↓ 1 callersMethod_query_judge_once
(self, judge_prompt: str, vote_id: int)
openclaw-opd/openclaw_opd_api_server.py:416
↓ 1 callersMethod_query_judge_once
(self, judge_messages: list[dict], vote_id: int)
openclaw-fireworks/fireworks_server.py:430
↓ 1 callersMethod_query_judge_once
(self, messages: list[dict], vote_id: int)
openclaw-tinker/scorers.py:473
↓ 1 callersMethod_query_once
(self, messages: list[dict], vote_id: int)
openclaw-tinker/scorers.py:319
↓ 1 callersMethod_query_prm_eval_once
(self, prompt_text: str, vote_id: int)
openclaw-rl/oel/openclaw_oel_api_server.py:1360
↓ 1 callersMethod_query_prm_eval_once
(self, eval_messages: list[dict], vote_id: int)
openclaw-fireworks/fireworks_server.py:447
↓ 1 callersFunction_query_prm_once
(args, judge_prompt: str, vote_id: int)
toolcall-rl/generate_with_retool.py:402
↓ 1 callersMethod_query_prm_once
(self, judge_prompt: str, vote_id: int)
openclaw-rl/openclaw_api_server.py:395
↓ 1 callersMethod_query_prm_once
( self, args: Any, messages_for_prm: List[Dict[str, Any]], step_index: int,
gui-rl/agents/qwen3vl_reward_agent.py:277
↓ 1 callersMethod_query_prm_once
( self, args: Any, messages_for_prm: List[Dict[str, Any]], step_index: int,
swe-rl/swe_prm.py:275
↓ 1 callersMethod_reached_iteration_limit
(self)
terminal-rl/agent_runner.py:42
↓ 1 callersMethod_reached_parse_error_limit
(self)
terminal-rl/agent_runner.py:45
↓ 1 callersFunction_read_checkpoint_metadata
(path: Path)
slime/slime/backends/fsdp_utils/checkpoint.py:62
← previousnext →3,101–3,200 of 10,561, ranked by callers