MCPcopy Create free account

hub / github.com/Xuekai-Zhu/FlowRL / functions

Functions4,141 in github.com/Xuekai-Zhu/FlowRL

↓ 4 callersFunctionnormalize_answer
(s)
verl_Test/verl/utils/reward_score/search_r1_like_qa_em.py:23
↓ 4 callersFunctionnormalize_answer
(s)
verl_FlowRL/verl/utils/reward_score/search_r1_like_qa_em.py:23
↓ 4 callersFunctionpatch_vlm_for_ulysses_input_slicing
Applies a monkey patch to the forward method of a given model class to enable Ulysses sequence parallelism input slicing.
verl_Test/verl/models/transformers/monkey_patch.py:110
↓ 4 callersFunctionpatch_vlm_for_ulysses_input_slicing
Applies a monkey patch to the forward method of a given model class to enable Ulysses sequence parallelism input slicing.
verl_FlowRL/verl/models/transformers/monkey_patch.py:110
↓ 4 callersFunctionprepare_inputs
(tokenizer, prompts, max_prompt_length)
verl_FlowRL/tests/workers/rollout/utils_sglang.py:95
↓ 4 callersFunctionprocess_validation_metrics
Process validation metrics into a structured format with statistical analysis. This function organizes validation metrics by data source
verl_FlowRL/verl/trainer/ppo/metric_utils.py:323
↓ 4 callersFunctionremove_boxed
Remove the LaTeX boxed command from a string. Args: s: String with format "\\boxed{content}" Returns: The content inside the
verl_Test/recipe/r1/tasks/math_dapo.py:51
↓ 4 callersFunctionrotate_half
Rotates half the hidden dims of the input.
verl_Test/verl/models/qwen2/megatron/layers/parallel_attention.py:119
↓ 4 callersFunctionrotate_half
Rotates half the hidden dims of the input.
verl_Test/verl/models/llama/megatron/layers/parallel_attention.py:142
↓ 4 callersFunctionrotate_half
Rotates half the hidden dims of the input.
verl_FlowRL/verl/models/qwen2/megatron/layers/parallel_attention.py:119
↓ 4 callersFunctionrotate_half
Rotates half the hidden dims of the input.
verl_FlowRL/verl/models/llama/megatron/layers/parallel_attention.py:142
↓ 4 callersMethodsave_checkpoint
(self, step)
verl_Test/verl/trainer/fsdp_sft_trainer.py:450
↓ 4 callersMethodsave_checkpoint
(self, step)
verl_FlowRL/verl/trainer/fsdp_sft_trainer.py:450
↓ 4 callersMethodset_state
(self, state)
verl_FlowRL/tests/e2e/envs/digit_completion/task.py:62
↓ 4 callersFunctionset_ulysses_sequence_parallel_group
Set ulysses sequence parallel process group.
verl_FlowRL/verl/utils/ulysses.py:30
↓ 4 callersMethodshared_embedding_or_output_weight
(self)
verl_Test/verl/models/qwen2/megatron/modeling_qwen2_megatron.py:601
↓ 4 callersMethodshared_embedding_or_output_weight
(self)
verl_FlowRL/verl/models/qwen2/megatron/modeling_qwen2_megatron.py:601
↓ 4 callersMethodstop
(self)
verl_Test/verl/utils/debug/profile.py:70
↓ 4 callersMethodstop
(self)
verl_FlowRL/verl/utils/debug/profile.py:70
↓ 4 callersMethodupdate_critic
(self, data: DataProto)
verl_Test/verl/workers/fsdp_workers.py:1063
↓ 4 callersMethodupdate_critic
(self, data: DataProto)
verl_FlowRL/verl/workers/fsdp_workers.py:1087
↓ 4 callersFunctionupdate_model_config
Update the module config with the override_config_kwargs. Args: module_config: The module config from Huggingface Transformers. ov
verl_FlowRL/verl/utils/model.py:49
↓ 3 callersMethod__init__
(self, dim, max_position_embeddings=2048, base=10000, device=None)
verl_Test/verl/models/qwen2/megatron/layers/parallel_attention.py:43
↓ 3 callersMethod__init__
(self, hidden_size: int, num_layers: int = 3, dropout: float = 0.1)
verl_FlowRL/verl/workers/fsdp_workers.py:101
↓ 3 callersMethod__init__
(self, dim, max_position_embeddings=2048, base=10000, device=None)
verl_FlowRL/verl/models/qwen2/megatron/layers/parallel_attention.py:43
↓ 3 callersFunction__targspec
(func, specs, attr='__orig_arg__')
verl_Test/rllm/rewards/code_utils/pyext2.py:31
↓ 3 callersFunction__targspec
(func, specs, attr='__orig_arg__')
verl_FlowRL/rllm/rewards/code_utils/pyext2.py:31
↓ 3 callersMethod_bind_worker_method
Binds worker methods to the WorkerGroup based on registered attributes. Args: user_defined_cls (type): The class containing metho
verl_Test/verl/single_controller/base/worker_group.py:178
↓ 3 callersMethod_bind_worker_method
Binds worker methods to the WorkerGroup based on registered attributes. Args: user_defined_cls (type): The class containing metho
verl_FlowRL/verl/single_controller/base/worker_group.py:178
↓ 3 callersFunction_broadcast_tp_shard_tensor
broadcast tensor in tp shards across mp_group
verl_Test/verl/models/qwen2/megatron/checkpoint_utils/qwen2_loader_depracated.py:172
↓ 3 callersFunction_broadcast_tp_shard_tensor
broadcast tensor in tp shards across mp_group
verl_Test/verl/models/mcore/loader.py:177
↓ 3 callersFunction_broadcast_tp_shard_tensor
broadcast tensor in tp shards across mp_group
verl_Test/verl/models/llama/megatron/checkpoint_utils/llama_loader_depracated.py:174
↓ 3 callersFunction_broadcast_tp_shard_tensor
broadcast tensor in tp shards across mp_group
verl_FlowRL/verl/models/qwen2/megatron/checkpoint_utils/qwen2_loader_depracated.py:172
↓ 3 callersFunction_broadcast_tp_shard_tensor
broadcast tensor in tp shards across mp_group
verl_FlowRL/verl/models/mcore/loader.py:177
↓ 3 callersFunction_broadcast_tp_shard_tensor
broadcast tensor in tp shards across mp_group
verl_FlowRL/verl/models/llama/megatron/checkpoint_utils/llama_loader_depracated.py:174
↓ 3 callersMethod_build_param_references
(self, pp_rank, maintain_weight=False)
verl_Test/verl/workers/sharding_manager/megatron_vllm.py:122
↓ 3 callersMethod_build_param_references
(self, pp_rank, maintain_weight=False)
verl_FlowRL/verl/workers/sharding_manager/megatron_vllm.py:122
↓ 3 callersFunction_check_dispatch_mode
(dispatch_mode)
verl_Test/verl/single_controller/base/decorator.py:478
↓ 3 callersFunction_concat_data_proto_or_future
(output: List)
verl_Test/verl/single_controller/base/decorator.py:207
↓ 3 callersFunction_concat_data_proto_or_future
(output: List)
verl_FlowRL/verl/single_controller/base/decorator.py:207
↓ 3 callersMethod_execute_remote_single_worker
Execute a method on a single worker remotely. Args: worker: The worker actor handle method_name: Name of the method t
verl_Test/verl/single_controller/ray/base.py:482
↓ 3 callersMethod_execute_remote_single_worker
Execute a method on a single worker remotely. Args: worker: The worker actor handle method_name: Name of the method t
verl_FlowRL/verl/single_controller/ray/base.py:482
↓ 3 callersFunction_fetch_tp_shard_tensor
fetch tensor in tp shards
verl_Test/verl/models/qwen2/megatron/checkpoint_utils/qwen2_loader.py:114
↓ 3 callersFunction_fetch_tp_shard_tensor
fetch tensor in tp shards
verl_Test/verl/models/llama/megatron/checkpoint_utils/llama_loader.py:116
↓ 3 callersFunction_fetch_tp_shard_tensor
fetch tensor in tp shards
verl_FlowRL/verl/models/qwen2/megatron/checkpoint_utils/qwen2_loader.py:114
↓ 3 callersFunction_fetch_tp_shard_tensor
fetch tensor in tp shards
verl_FlowRL/verl/models/llama/megatron/checkpoint_utils/llama_loader.py:116
↓ 3 callersMethod_forward_micro_batch
(self, micro_batch)
verl_Test/recipe/spin/fsdp_workers.py:374
↓ 3 callersMethod_forward_micro_batch
(self, micro_batch)
verl_FlowRL/recipe/spin/fsdp_workers.py:374
↓ 3 callersFunction_get_attr
(attr_name, default_value=None)
verl_Test/verl/utils/fsdp_utils.py:78
↓ 3 callersFunction_get_attr
(attr_name, default_value=None)
verl_FlowRL/verl/utils/fsdp_utils.py:78
↓ 3 callersFunction_get_current_mem_info
Get current memory usage.
verl_Test/verl/utils/debug/performance.py:31
↓ 3 callersFunction_get_current_mem_info
Get current memory usage.
verl_FlowRL/verl/utils/debug/performance.py:31
↓ 3 callersFunction_get_gpt_model
(model)
verl_Test/verl/models/qwen2/megatron/checkpoint_utils/qwen2_saver.py:83
↓ 3 callersFunction_get_gpt_model
(model)
verl_Test/verl/models/mcore/saver.py:92
↓ 3 callersFunction_get_gpt_model
(model)
verl_Test/verl/models/llama/megatron/checkpoint_utils/llama_saver.py:83
↓ 3 callersFunction_get_gpt_model
(model)
verl_FlowRL/verl/models/qwen2/megatron/checkpoint_utils/qwen2_saver.py:83
↓ 3 callersFunction_get_gpt_model
(model)
verl_FlowRL/verl/models/mcore/saver.py:92
↓ 3 callersFunction_get_gpt_model
(model)
verl_FlowRL/verl/models/llama/megatron/checkpoint_utils/llama_saver.py:83
↓ 3 callersFunction_gettypes
(args)
verl_Test/rllm/rewards/code_utils/pyext2.py:114
↓ 3 callersFunction_gettypes
(args)
verl_FlowRL/rllm/rewards/code_utils/pyext2.py:114
↓ 3 callersMethod_init_hf_config_and_tf_config
( self, model_path, dtype, override_model_config, override_transformer
verl_Test/verl/single_controller/base/megatron/worker.py:42
↓ 3 callersMethod_init_hf_config_and_tf_config
( self, model_path, dtype, override_model_config, override_transformer
verl_FlowRL/verl/single_controller/base/megatron/worker.py:42
↓ 3 callersMethod_optimizer_step
(self)
verl_Test/recipe/prime/prime_dp_rm.py:199
↓ 3 callersMethod_optimizer_step
(self)
verl_FlowRL/recipe/prime/prime_dp_rm.py:199
↓ 3 callersFunction_pre_process_inputs
( pad_token_id, prompt_token_ids: torch.Tensor, )
verl_Test/verl/workers/rollout/sglang_rollout/sglang_rollout.py:77
↓ 3 callersFunction_pre_process_inputs
( pad_token_id, prompt_token_ids: torch.Tensor, )
verl_FlowRL/verl/workers/rollout/sglang_rollout/sglang_rollout.py:77
↓ 3 callersFunction_pre_process_inputs
(pad_token_id, prompt_token_ids: torch.Tensor)
verl_FlowRL/verl/workers/rollout/vllm_rollout/vllm_rollout.py:59
↓ 3 callersMethod_set_cos_sin_cache
(self, seq_len, device, dtype)
verl_Test/verl/models/llama/megatron/layers/parallel_attention.py:51
↓ 3 callersMethod_set_cos_sin_cache
(self, seq_len, device, dtype)
verl_FlowRL/verl/models/llama/megatron/layers/parallel_attention.py:51
↓ 3 callersFunction_split_args_kwargs_data_proto
(chunks, *args, **kwargs)
verl_Test/verl/single_controller/base/decorator.py:77
↓ 3 callersFunction_split_args_kwargs_data_proto
(chunks, *args, **kwargs)
verl_FlowRL/verl/single_controller/base/decorator.py:77
↓ 3 callersFunction_str_is_int
(x: str)
verl_Test/rllm/rewards/math_utils/utils.py:222
↓ 3 callersFunction_str_is_int
(x: str)
verl_Test/recipe/entropy_math/__init__.py:747
↓ 3 callersFunction_str_is_int
(x: str)
verl_Test/verl/utils/reward_score/prime_math/__init__.py:90
↓ 3 callersFunction_str_is_int
(x: str)
verl_FlowRL/rllm/rewards/math_utils/utils.py:222
↓ 3 callersFunction_str_is_int
(x: str)
verl_FlowRL/verl/utils/reward_score/prime_math/__init__.py:90
↓ 3 callersFunction_unwrap_ray_remote
(cls)
verl_Test/verl/single_controller/ray/base.py:664
↓ 3 callersFunction_unwrap_ray_remote
(cls)
verl_FlowRL/verl/single_controller/ray/base.py:664
↓ 3 callersMethod_validate
(self)
verl_Test/verl/utils/debug/profile.py:49
↓ 3 callersMethod_validate
(self)
verl_FlowRL/verl/utils/debug/profile.py:49
↓ 3 callersFunctionbroadcast_dict_tensor
TODO: optimize this. Technically, we only need one broadcast
verl_Test/verl/utils/torch_functional.py:221
↓ 3 callersFunctionbroadcast_dict_tensor
TODO: optimize this. Technically, we only need one broadcast
verl_FlowRL/verl/utils/torch_functional.py:221
↓ 3 callersFunctionbuild_memory_buffer
Build the memory buffer given weight_buffer_meta Args: weight_buffer_meta: contains mapping from name to a dictionary containing shape an
verl_Test/verl/utils/memory_buffer.py:70
↓ 3 callersFunctionbuild_memory_buffer
Build the memory buffer given weight_buffer_meta Args: weight_buffer_meta: contains mapping from name to a dictionary containing shape an
verl_FlowRL/verl/utils/memory_buffer.py:70
↓ 3 callersFunctioncalc_padded_numel
for cuda memory alignment, make sure alignment by 128-bits
verl_Test/verl/utils/memory_buffer.py:53
↓ 3 callersFunctioncalc_padded_numel
for cuda memory alignment, make sure alignment by 128-bits
verl_FlowRL/verl/utils/memory_buffer.py:53
↓ 3 callersMethodcheck_storage
(self, method_name, run_forward)
verl_FlowRL/tests/kernels/test_linear_cross_entropy.py:225
↓ 3 callersMethodcollective_rpc
( self, method: Union[str, Callable], timeout: Optional[float] = None, args: T
verl_Test/verl/workers/rollout/vllm_rollout/vllm_async_server.py:83
↓ 3 callersMethodcollective_rpc
( self, method: Union[str, Callable], timeout: Optional[float] = None, args: T
verl_FlowRL/verl/workers/rollout/vllm_rollout/vllm_async_server.py:83
↓ 3 callersFunctioncompile_code
(code: str, timeout: int)
verl_Test/rllm/rewards/code_utils/livecodebench.py:163
↓ 3 callersFunctioncompile_code
(code: str, timeout: int)
verl_FlowRL/rllm/rewards/code_utils/livecodebench.py:163
↓ 3 callersFunctioncompute_advantage
(data: DataProto, beta=1.0)
verl_Test/recipe/sppo/sppo_ray_trainer.py:60
↓ 3 callersFunctioncompute_advantage
(data: DataProto, beta=1.0)
verl_FlowRL/recipe/sppo/sppo_ray_trainer.py:60
↓ 3 callersFunctioncompute_detach_dpo_loss_rm
(token_level_scores, acc, Q_bc, acc_bc, response_mask, beta, bon_mode="none")
verl_Test/recipe/prime/prime_core_algos.py:81
↓ 3 callersFunctioncompute_detach_dpo_loss_rm
(token_level_scores, acc, Q_bc, acc_bc, response_mask, beta, bon_mode="none")
verl_FlowRL/recipe/prime/prime_core_algos.py:81
↓ 3 callersFunctioncompute_response_mask
(data: DataProto)
verl_Test/recipe/prime/prime_ray_trainer.py:114
↓ 3 callersFunctioncompute_response_mask
(data: DataProto)
verl_FlowRL/recipe/prime/prime_ray_trainer.py:114
↓ 3 callersFunctioncompute_reward
Compute reward for a batch of data. Args: data: DataProto object containing the input data. reward_fn: Reward function to com
verl_Test/verl/trainer/ppo/reward.py:107
↓ 3 callersFunctioncompute_reward
Compute reward for a batch of data. Args: data: DataProto object containing the input data. reward_fn: Reward function to com
verl_FlowRL/verl/trainer/ppo/reward.py:107
↓ 3 callersMethodcompute_rm_score
(self, data: DataProto)
verl_Test/recipe/prime/prime_dp_rm.py:215
← previousnext →401–500 of 4,141, ranked by callers