Code
Hub
Workspaces
Following
Trending
Connect
MCP
copy
Create free account
hub
/
github.com/PRIME-RL/SimpleVLA-RL
/ functions
Functions
1,816 in github.com/PRIME-RL/SimpleVLA-RL
⨍
Functions
1,816
◇
Types & classes
232
↳
Endpoints
23
↓ 1 callers
Method
_build_model_optimizer
(self, config, enable_gradient_checkpointing=False)
verl/workers/fsdp_workers.py:1491
↓ 1 callers
Method
_build_param_buffer
Build the parameter buffer in each pp rank
verl/workers/hybrid_engine/megatron_vllm.py:82
↓ 1 callers
Method
_build_rm_model
(self, model_path, megatron_config: ModelParallelConfig, override_model_config)
verl/workers/megatron_workers.py:614
↓ 1 callers
Method
_build_rollout
(self)
verl/workers/megatron_workers.py:216
↓ 1 callers
Method
_build_rollout
(self)
verl/workers/fsdp_workers.py:325
↓ 1 callers
Method
_build_rollout
(self)
verl/workers/fsdp_workers.py:826
↓ 1 callers
Function
_check_and_sort_partitions
(partitions)
verl/utils/seqlen_balancing.py:170
↓ 1 callers
Method
_check_beam_search_early_stopping
( self, early_stopping: Union[bool, str], sampling_params: SamplingParams, bes
verl/third_party/vllm/vllm_v_0_3_1/llm_engine_sp.py:348
↓ 1 callers
Function
_check_dispatch_mode
(dispatch_mode)
verl/single_controller/base/decorator.py:366
↓ 1 callers
Function
_check_execute_mode
(execute_mode)
verl/single_controller/base/decorator.py:375
↓ 1 callers
Method
_check_stop
Stop the finished sequences.
verl/third_party/vllm/vllm_v_0_3_1/llm_engine_sp.py:681
↓ 1 callers
Method
_configure_before_init
(self, register_center_name: str, rank: int)
verl/single_controller/base/worker.py:102
↓ 1 callers
Method
_configure_with_meta
This function should only be called inside by WorkerGroup
verl/single_controller/base/worker.py:147
↓ 1 callers
Function
_copy
(from_path: str, to_path: str, timeout: int = None)
verl/utils/hdfs_io.py:113
↓ 1 callers
Method
_create_dataloader
(self)
verl/trainer/ppo/ray_trainer.py:292
↓ 1 callers
Method
_detect_robotwin_version
Detect which version of robotwin to use based on config
verl/workers/rollout/rob_rollout.py:465
↓ 1 callers
Method
_download
(self)
verl/utils/dataset/rm_dataset.py:70
↓ 1 callers
Method
_download
(self)
verl/utils/dataset/sft_dataset.py:69
↓ 1 callers
Method
_download
(self)
verl/utils/dataset/rl_dataset.py:103
↓ 1 callers
Function
_exists
hdfs capable to check whether a file_path is exists
verl/utils/hdfs_io.py:43
↓ 1 callers
Function
_expand_mask
Expands attention_mask from `[bsz, seq_len]` to `[bsz, 1, tgt_seq_len, src_seq_len]`.
verl/utils/torch_functional.py:512
↓ 1 callers
Function
_expand_mask
Expands attention_mask from `[bsz, seq_len]` to `[bsz, 1, tgt_seq_len, src_seq_len]`.
verl/models/llama/megatron/modeling_llama_megatron.py:59
↓ 1 callers
Method
_expand_to_token_level
(self, data: DataProto, scores: torch.Tensor)
verl/workers/fsdp_workers.py:1374
↓ 1 callers
Function
_fix_a_slash_b
(string)
verl/utils/reward_score/evaluation_utils/math_util/math_normalize.py:56
↓ 1 callers
Function
_fix_fracs
(string)
verl/utils/reward_score/evaluation_utils/math_util/math_normalize.py:24
↓ 1 callers
Function
_fix_sqrt
(string)
verl/utils/reward_score/evaluation_utils/math_util/math_normalize.py:81
↓ 1 callers
Method
_forward_head
(self, hidden_states)
verl/models/llama/megatron/modeling_llama_megatron.py:302
↓ 1 callers
Method
_forward_head
(self, hidden_states)
verl/models/llama/megatron/modeling_llama_megatron.py:559
↓ 1 callers
Method
_forward_micro_batch
(self, micro_batch)
verl/workers/fsdp_workers.py:1365
↓ 1 callers
Method
_forward_micro_batch
micro_batch: Returns: entropy: # (bs, response_len) log_probs: # (bs, response_len)
verl/workers/actor/dp_rob.py:105
↓ 1 callers
Method
_forward_micro_batch_entropy
(self, micro_batch, temperature)
verl/workers/actor/dp_rob.py:259
↓ 1 callers
Method
_forward_micro_batch_entropy
(self, micro_batch, temperature)
verl/workers/actor/dp_actor.py:143
↓ 1 callers
Method
_forward_micro_batch_update
(self, input_ids, attention_mask, pixel_values, responses, temperature, proprio)
verl/workers/actor/dp_rob.py:202
↓ 1 callers
Method
_generate_minibatch
Generate minibatch - routes to appropriate implementation based on task suite
verl/workers/rollout/rob_rollout.py:597
↓ 1 callers
Method
_generate_minibatch
(self, prompts: DataProto)
verl/workers/rollout/hf_rollout.py:51
↓ 1 callers
Method
_generate_minibatch_libero
Generate minibatch for Libero using multiprocessing
verl/workers/rollout/rob_rollout.py:772
↓ 1 callers
Method
_generate_minibatch_robotwin
Generate minibatch for Robotwin using threading
verl/workers/rollout/rob_rollout.py:604
↓ 1 callers
Method
_generate_one_step_oft
Generate one step for OpenVLA-OFT
verl/workers/rollout/rob_rollout.py:925
↓ 1 callers
Method
_generate_one_step_openvla
Generate one step for OpenVLA
verl/workers/rollout/rob_rollout.py:980
↓ 1 callers
Method
_get_executor_cls
(cls, engine_config: EngineConfig)
verl/third_party/vllm/vllm_v_0_6_3/llm_engine_sp.py:355
↓ 1 callers
Method
_get_executor_cls
(cls, engine_config: EngineConfig)
verl/third_party/vllm/vllm_v_0_5_4/llm_engine_sp.py:281
↓ 1 callers
Method
_get_free_port
(self)
verl/single_controller/base/worker.py:58
↓ 1 callers
Function
_get_model_architecture
(config: PretrainedConfig)
verl/third_party/vllm/vllm_v_0_3_1/model_loader.py:46
↓ 1 callers
Function
_get_model_weight_loader
(arch: str)
verl/third_party/vllm/vllm_v_0_3_1/model_loader.py:123
↓ 1 callers
Function
_get_model_weight_loader
(arch: str)
verl/third_party/vllm/vllm_v_0_6_3/dtensor_weight_loaders.py:371
↓ 1 callers
Function
_get_model_weight_loader
(arch: str)
verl/third_party/vllm/vllm_v_0_6_3/megatron_weight_loaders.py:299
↓ 1 callers
Function
_get_model_weight_loader
(arch: str)
verl/third_party/vllm/vllm_v_0_5_4/dtensor_weight_loaders.py:331
↓ 1 callers
Function
_get_model_weight_loader
(arch: str)
verl/third_party/vllm/vllm_v_0_5_4/megatron_weight_loaders.py:298
↓ 1 callers
Function
_get_model_weight_loader
(arch: str)
verl/third_party/vllm/vllm_v_0_4_2/dtensor_weight_loaders.py:260
↓ 1 callers
Function
_get_model_weight_loader
(arch: str)
verl/third_party/vllm/vllm_v_0_4_2/megatron_weight_loaders.py:298
↓ 1 callers
Method
_get_node_ip
(self)
verl/single_controller/base/worker.py:39
↓ 1 callers
Function
_get_parallel_model_architecture_from_config
(config: PretrainedConfig, value=False)
verl/utils/model.py:243
↓ 1 callers
Method
_get_weights_iterator
(self, actor_model: Union[PreTrainedModel, Dict])
verl/third_party/vllm/vllm_v_0_6_3/model_loader.py:199
↓ 1 callers
Method
_get_weights_iterator
(self, actor_model: Union[PreTrainedModel, Dict])
verl/third_party/vllm/vllm_v_0_5_4/model_loader.py:181
↓ 1 callers
Method
_get_weights_iterator
(self, actor_model: Union[PreTrainedModel, Dict])
verl/third_party/vllm/vllm_v_0_4_2/model_loader.py:170
↓ 1 callers
Function
_handle_file_sync
Handle syncing of files between current directory and checkpoint. Creates backups if files exist but differ, and copies current versions to
verl/utils/openvla_utils.py:90
↓ 1 callers
Method
_init_cache_engine
(self)
verl/third_party/vllm/vllm_v_0_6_3/worker.py:235
↓ 1 callers
Method
_init_cache_engine
(self)
verl/third_party/vllm/vllm_v_0_5_4/worker.py:229
↓ 1 callers
Method
_init_cache_engine
(self)
verl/third_party/vllm/vllm_v_0_4_2/worker.py:199
↓ 1 callers
Method
_init_cache_sp
Profiles the memory usage and initializes the KV cache.
verl/third_party/vllm/vllm_v_0_3_1/llm_engine_sp.py:176
↓ 1 callers
Function
_init_distributed_environment
Initialize the distributed environment.
verl/third_party/vllm/vllm_v_0_3_1/worker.py:270
↓ 1 callers
Method
_init_executor
(self, model, distributed_init_method)
verl/third_party/vllm/vllm_v_0_6_3/spmd_gpu_executor.py:74
↓ 1 callers
Method
_init_executor
(self, model, distributed_init_method)
verl/third_party/vllm/vllm_v_0_5_4/spmd_gpu_executor.py:66
↓ 1 callers
Method
_init_executor
(self, model, distributed_init_method)
verl/third_party/vllm/vllm_v_0_4_2/spmd_gpu_executor.py:63
↓ 1 callers
Method
_init_head
(self)
verl/models/llama/megatron/modeling_llama_megatron.py:290
↓ 1 callers
Method
_init_head
(self)
verl/models/llama/megatron/modeling_llama_megatron.py:547
↓ 1 callers
Method
_init_rope
(self)
verl/models/llama/megatron/layers/parallel_attention.py:204
↓ 1 callers
Method
_init_tokenizer
(self, tokenizer, **tokenizer_init_kwargs)
verl/third_party/vllm/vllm_v_0_3_1/llm_engine_sp.py:138
↓ 1 callers
Method
_init_tokenizer
(self, tokenizer, **tokenizer_init_kwargs)
verl/third_party/vllm/vllm_v_0_6_3/llm_engine_sp.py:337
↓ 1 callers
Method
_init_tokenizer
(self, tokenizer, **tokenizer_init_kwargs)
verl/third_party/vllm/vllm_v_0_5_4/llm_engine_sp.py:263
↓ 1 callers
Method
_init_tokenizer
(self, tokenizer, **tokenizer_init_kwargs)
verl/third_party/vllm/vllm_v_0_4_2/llm_engine_sp.py:229
↓ 1 callers
Method
_init_with_detached_workers
(self, worker_names)
verl/single_controller/ray/base.py:209
↓ 1 callers
Method
_init_with_resource_pool
(self, resource_pool, ray_cls_with_init, bin_pack, detached)
verl/single_controller/ray/base.py:214
↓ 1 callers
Method
_init_workers_sp
(self, model, distributed_init_method: str)
verl/third_party/vllm/vllm_v_0_3_1/llm_engine_sp.py:149
↓ 1 callers
Method
_init_workers_sp
(self, model, distributed_init_method: str)
verl/third_party/vllm/vllm_v_0_6_3/spmd_gpu_executor.py:80
↓ 1 callers
Method
_init_workers_sp
(self, model, distributed_init_method: str)
verl/third_party/vllm/vllm_v_0_5_4/spmd_gpu_executor.py:72
↓ 1 callers
Method
_init_workers_sp
(self, model, distributed_init_method: str)
verl/third_party/vllm/vllm_v_0_4_2/spmd_gpu_executor.py:69
↓ 1 callers
Function
_inject_implicit_mixed_number
Automatically make a mixed number evalable e.g. 7 3/4 => 7+3/4
verl/utils/reward_score/evaluation_utils/math_util/__init__.py:118
↓ 1 callers
Function
_inner_call_method
(_method)
verl/utils/reward_score/evaluation_utils/code_util/testing_utils_org.py:661
↓ 1 callers
Function
_inner_call_method
(_method)
verl/utils/reward_score/evaluation_utils/code_util/testing_util.py:728
↓ 1 callers
Function
_inner_call_method
(_method)
verl/utils/reward_score/evaluation_utils/code_util/testing_util_org.py:661
↓ 1 callers
Function
_is_float
(num: str)
verl/utils/reward_score/evaluation_utils/math_util/__init__.py:84
↓ 1 callers
Function
_is_int
(x: float)
verl/utils/reward_score/evaluation_utils/math_util/__init__.py:92
↓ 1 callers
Method
_is_worker_alive
(self, worker)
verl/single_controller/base/worker_group.py:110
↓ 1 callers
Function
_last_boxed_only_string
(string)
verl/utils/reward_score/evaluation_utils/math_util/__init__.py:338
↓ 1 callers
Method
_load_params_to_cuda
(self, pp_rank, to_empty=False)
verl/workers/hybrid_engine/megatron_vllm.py:92
↓ 1 callers
Function
_make_causal_mask
Make causal mask used for bi-directional self-attention.
verl/utils/torch_functional.py:499
↓ 1 callers
Function
_make_causal_mask
Make causal mask used for bi-directional self-attention.
verl/models/llama/megatron/modeling_llama_megatron.py:46
↓ 1 callers
Method
_make_minibatch_iterator
(self, data: DataProto)
verl/workers/critic/dp_critic.py:56
↓ 1 callers
Method
_make_minibatch_iterator
Make minibatch iterator for updating the actor See PPO paper for details. https://arxiv.org/abs/1707.06347
verl/workers/actor/dp_prime.py:63
↓ 1 callers
Function
_materialize_futures
(*args, **kwargs)
verl/single_controller/base/decorator.py:379
↓ 1 callers
Function
_megatron_calc_layer_map
Calculate the mapping of global layer_idx to local layer_idx Returns: layer_map (Dict: int -> tuple(int, int, int)): mapping f
verl/models/llama/megatron/checkpoint_utils/llama_saver.py:45
↓ 1 callers
Function
_megatron_calc_layer_map
Calculate the mapping of global layer_idx to local layer_idx Returns: layer_map (Dict: int -> tuple(int, int, int)): mapping f
verl/models/llama/megatron/checkpoint_utils/llama_loader.py:21
↓ 1 callers
Function
_mkdir
hdfs mkdir
verl/utils/hdfs_io.py:75
↓ 1 callers
Method
_normalize_config_bsz
(self)
verl/trainer/fsdp_sft_trainer.py:81
↓ 1 callers
Method
_optimizer_step
(self)
verl/workers/critic/dp_critic.py:63
↓ 1 callers
Method
_optimizer_step
(self)
verl/workers/actor/dp_rob.py:338
↓ 1 callers
Method
_optimizer_step
(self)
verl/workers/actor/dp_prime.py:72
↓ 1 callers
Method
_optimizer_step
(self)
verl/workers/actor/dp_actor.py:214
↓ 1 callers
Function
_parse_latex
Attempts to parse latex to an expression sympy can read.
verl/utils/reward_score/evaluation_utils/math_util/__init__.py:66
← previous
next →
401–500 of 1,816, ranked by callers