MCPcopy Create free account

hub / github.com/Jiayi-Pan/TinyZero / functions

Functions1,293 in github.com/Jiayi-Pan/TinyZero

⨍ Functions 1,293◇ Types & classes 173↳ Endpoints 30
↓ 1 callersFunctionget_nccl_id_store_by_name
(name)
verl/utils/rendezvous/ray_backend.py:34
↓ 1 callersMethodget_num_layers
(self, parallel_config: "ParallelConfig")
verl/third_party/vllm/vllm_v_0_3_1/config.py:235
↓ 1 callersFunctionget_open_port
()
verl/third_party/vllm/vllm_v_0_3_1/llm_engine_sp.py:762
↓ 1 callersFunctionget_open_port
()
verl/third_party/vllm/vllm_v_0_6_3/spmd_gpu_executor.py:240
↓ 1 callersFunctionget_open_port
()
verl/third_party/vllm/vllm_v_0_5_4/spmd_gpu_executor.py:237
↓ 1 callersFunctionget_open_port
()
verl/third_party/vllm/vllm_v_0_4_2/spmd_gpu_executor.py:202
↓ 1 callersMethodget_partitions
(self)
verl/utils/seqlen_balancing.py:63
↓ 1 callersFunctionget_predefined_dispatch_fn
(dispatch_mode)
verl/single_controller/base/decorator.py:300
↓ 1 callersFunctionget_predefined_execute_fn
Note that here we only asks execute_all and execute_rank_zero to be implemented Leave the choice of how these two functions handle argument '
verl/single_controller/base/decorator.py:350
↓ 1 callersFunctionget_random_string
(length: int)
verl/single_controller/ray/base.py:29
↓ 1 callersFunctionget_rm_data
()
tests/verl/utils/dataset/test_rm_dataset.py:21
↓ 1 callersFunctionget_tensor_model_parallel_rank
Return my rank for the tensor model parallel group.
verl/third_party/vllm/vllm_v_0_3_1/parallel_state.py:119
↓ 1 callersFunctionget_tensor_model_parallel_rank
Return my rank for the tensor model parallel group.
verl/third_party/vllm/vllm_v_0_4_2/parallel_state.py:284
↓ 1 callersMethodget_total_num_kv_heads
Returns the total number of KV heads.
verl/third_party/vllm/vllm_v_0_3_1/config.py:194
↓ 1 callersFunctionget_ulysses_sequence_parallel_rank
Get ulysses sequence parallel rank.
verl/utils/ulysses.py:53
↓ 1 callersFunctionget_version
(pkg)
verl/third_party/vllm/__init__.py:18
↓ 1 callersMethodget_vocab_size
(self)
verl/third_party/vllm/vllm_v_0_3_1/config.py:184
↓ 1 callersFunctionget_weight_loader
(arch: str)
verl/models/weight_loader_registry.py:16
↓ 1 callersMethodinit_cache_engine
(self)
verl/third_party/vllm/vllm_v_0_3_1/llm.py:133
↓ 1 callersMethodinit_cache_engine
(self)
verl/third_party/vllm/vllm_v_0_3_1/llm_engine_sp.py:215
↓ 1 callersMethodinit_device
(self)
verl/third_party/vllm/vllm_v_0_6_3/worker.py:140
↓ 1 callersMethodinit_device
(self)
verl/third_party/vllm/vllm_v_0_5_4/worker.py:134
↓ 1 callersMethodinit_device
(self)
verl/third_party/vllm/vllm_v_0_4_2/worker.py:105
↓ 1 callersMethodinit_megatron
(self, default_megatron_kwargs: Optional[Dict] = None)
verl/single_controller/ray/megatron.py:58
↓ 1 callersMethodinit_model
(self)
tests/ray/detached_worker/server.py:73
↓ 1 callersFunctioninit_worker_distributed_environment
Initialize the distributed environment.
verl/third_party/vllm/vllm_v_0_6_3/worker.py:294
↓ 1 callersFunctioninit_worker_distributed_environment
Initialize the distributed environment.
verl/third_party/vllm/vllm_v_0_5_4/worker.py:286
↓ 1 callersFunctioninit_worker_distributed_environment
Initialize the distributed environment.
verl/third_party/vllm/vllm_v_0_4_2/worker.py:257
↓ 1 callersFunctioninitialize_cluster
Initialize the distributed cluster probably with Ray. Args: parallel_config: The configurations for parallel execution. engine_us
verl/third_party/vllm/vllm_v_0_3_1/llm_engine_sp.py:734
↓ 1 callersFunctioninitialize_cluster
Initialize the distributed cluster probably with Ray. Args: parallel_config: The configurations for parallel execution. Returns:
verl/third_party/vllm/vllm_v_0_6_3/spmd_gpu_executor.py:216
↓ 1 callersFunctioninitialize_cluster
Initialize the distributed cluster probably with Ray. Args: parallel_config: The configurations for parallel execution. Returns:
verl/third_party/vllm/vllm_v_0_5_4/spmd_gpu_executor.py:213
↓ 1 callersFunctioninitialize_cluster
Initialize the distributed cluster probably with Ray. Args: parallel_config: The configurations for parallel execution. Returns:
verl/third_party/vllm/vllm_v_0_4_2/spmd_gpu_executor.py:178
↓ 1 callersFunctioninitialize_model_parallel_for_vllm
( tensor_model_parallel_size: int, num_tensor_model_parallel_groups_per_train_tp: int = 1, pipelin
verl/third_party/vllm/vllm_v_0_6_3/parallel_state.py:104
↓ 1 callersFunctioninitialize_model_parallel_for_vllm
(tensor_model_parallel_size: int, num_tensor_model_parallel_groups_per_
verl/third_party/vllm/vllm_v_0_5_4/parallel_state.py:101
↓ 1 callersFunctioninitialize_model_parallel_for_vllm
(tensor_model_parallel_size: int, num_tensor_model_parallel_groups_per_
verl/third_party/vllm/vllm_v_0_4_2/parallel_state.py:98
↓ 1 callersFunctioninitialize_model_parallel_from_megatron
( tensor_model_parallel_size=None # we set None for backward compatibility to set infer_tp = train_tp
verl/third_party/vllm/vllm_v_0_3_1/parallel_state.py:26
↓ 1 callersFunctionis_equiv
(str1, str2, verbose=False)
verl/utils/reward_score/math.py:32
↓ 1 callersFunctionis_megatron_core_available
()
verl/utils/import_utils.py:24
↓ 1 callersFunctionis_transformers_version_in_range
(min_version: str, max_version: str)
verl/models/transformers/monkey_patch.py:66
↓ 1 callersFunctionkarmarkar_karp
(seqlen_list: List[int], k_partitions: int, equal_size: bool)
verl/utils/seqlen_balancing.py:25
↓ 1 callersFunctionlevenshtein
(s1, s2)
tests/rollout/test_vllm_hf_loader.py:30
↓ 1 callersMethodload_from_disk
(filepath)
verl/protocol.py:220
↓ 1 callersFunctionload_hf_weights
(actor_weights: Dict, vllm_model: nn.Module)
verl/third_party/vllm/vllm_v_0_6_3/hf_weight_loader.py:27
↓ 1 callersFunctionload_hf_weights
(actor_weights: Dict, vllm_model: nn.Module)
verl/third_party/vllm/vllm_v_0_5_4/hf_weight_loader.py:30
↓ 1 callersFunctionload_hf_weights
(actor_weights: Dict, vllm_model: nn.Module)
verl/third_party/vllm/vllm_v_0_4_2/hf_weight_loader.py:79
↓ 1 callersFunctionload_megatron_param_and_grad
(module_list: nn.ModuleList, device_id, load_grad=False, hybrid_engine=None)
verl/utils/megatron_utils.py:241
↓ 1 callersMethodload_model
(self)
verl/third_party/vllm/vllm_v_0_3_1/worker.py:117
↓ 1 callersMethodload_params_to_cuda
(self)
verl/workers/reward_model/megatron/reward_model.py:270
↓ 1 callersFunctionlog_probs_from_logits_response
Compute the response log_probs from full logits. Note that logits = model(input_ids) Args: input_ids: [batch_size, seqlen] lo
verl/utils/torch_functional.py:284
↓ 1 callersFunctionlog_probs_from_logits_response_rmpad
Compute the log_probs from logits with rmpad logits and pad input. Note that logits_rmpad = model(input_ids_rmpad). For each sentences, there is a
verl/utils/torch_functional.py:300
↓ 1 callersFunctionlog_seqlen_unbalance
(seqlen_list: List[int], partitions: List[List[int]], prefix)
verl/utils/seqlen_balancing.py:186
↓ 1 callersFunctionlogprobs_from_logits_flash_attn
(logits, labels)
verl/utils/torch_functional.py:65
↓ 1 callersFunctionmain
()
tests/rollout/run_fsdp_vllm.py:27
↓ 1 callersFunctionmain
(config)
tests/e2e/arithmetic_sequence/rl/main_trainer.py:92
↓ 1 callersFunctionmain
(config)
verl/trainer/main_ppo.py:98
↓ 1 callersFunctionmain
(config)
verl/trainer/main_generation.py:40
↓ 1 callersFunctionmain
(config)
verl/trainer/main_eval.py:35
↓ 1 callersFunctionmain
(config)
verl/trainer/fsdp_sft_trainer.py:360
↓ 1 callersFunctionmain
(config)
examples/split_placement/main_ppo_split.py:93
↓ 1 callersFunctionmake_map_fn
(split)
examples/data_preprocess/full_hh_rlhf.py:93
↓ 1 callersFunctionmake_prefix
(dp, template_type)
examples/data_preprocess/countdown.py:53
↓ 1 callersFunctionmake_prefix
(dp)
examples/data_preprocess/multiply.py:69
↓ 1 callersFunctionmake_prefix
(dp)
examples/data_preprocess/arth.py:79
↓ 1 callersFunctionmake_reward_function
(tokenizer, num_examine)
tests/e2e/arithmetic_sequence/rl/main_trainer.py:35
↓ 1 callersFunctionmasked_var
Compute variance of tensor with masked values.
verl/utils/torch_functional.py:112
↓ 1 callersFunctionmerge_resource_pool
(rp1: RayResourcePool, rp2: RayResourcePool)
verl/single_controller/ray/base.py:114
↓ 1 callersFunctionmodel_parallel_is_initialized
Check if tensor and pipeline parallel groups are initialized.
verl/third_party/vllm/vllm_v_0_6_3/parallel_state.py:98
↓ 1 callersFunctionmodel_parallel_is_initialized
Check if tensor and pipeline parallel groups are initialized.
verl/third_party/vllm/vllm_v_0_5_4/parallel_state.py:95
↓ 1 callersFunctionmodel_parallel_is_initialized
Check if tensor and pipeline parallel groups are initialized.
verl/third_party/vllm/vllm_v_0_4_2/parallel_state.py:92
↓ 1 callersFunctionnormalize_model_name
Transform the model name in each model_chunk in each pp stage into the name in inference engine
verl/utils/model.py:192
↓ 1 callersFunctionoffload_fsdp_grad
(module)
verl/utils/fsdp_utils.py:79
↓ 1 callersFunctionoffload_megatron_param_and_grad
(module_list: nn.ModuleList, offload_grad=False, hybrid_engine=None)
verl/utils/megatron_utils.py:226
↓ 1 callersMethodoffload_params_to_cpu
offload params of models that are not of current pp rank to cpu
verl/workers/sharding_manager/megatron_vllm.py:157
↓ 1 callersFunctionpad_to_sequence_parallel
pad the tokens such that the total length is a multiple of sp world size Args: unpad_tokens: (total_nnz, ...). Tokens after removing padd
verl/utils/megatron/sequence_parallel.py:29
↓ 1 callersFunctionparallel_put
(data_list, max_workers=None)
verl/utils/ray_utils.py:23
↓ 1 callersMethodping
Check if the tokenizer group is alive.
verl/third_party/vllm/vllm_v_0_4_2/tokenizer.py:35
↓ 1 callersMethodprepare_input_tensors
( self, seq_group_metadata_list: Optional[List[SequenceGroupMetadata]], )
verl/third_party/vllm/vllm_v_0_3_1/model_runner.py:174
↓ 1 callersMethodprepare_input_tensors
( self, seq_group_metadata_list: List[SequenceGroupMetadata], )
verl/third_party/vllm/vllm_v_0_4_2/model_runner.py:147
↓ 1 callersMethodprofile_num_available_blocks
( self, block_size: int, gpu_memory_utilization: float, cpu_swap_space: int,
verl/third_party/vllm/vllm_v_0_3_1/worker.py:121
↓ 1 callersMethodre_encode_by_rm_tokenizer
(self, data: DataProto)
verl/workers/reward_model/megatron/reward_model.py:58
↓ 1 callersFunctionremove_right_units
(string)
verl/utils/reward_score/math.py:140
↓ 1 callersMethodsave_to_disk
(self, filepath)
verl/protocol.py:215
↓ 1 callersFunctionselect_reward_fn
(data_source)
verl/trainer/main_eval.py:27
↓ 1 callersFunctionset_basic_config
This function sets the global logging format and level. It will be called when import verl
verl/utils/logging_utils.py:18
↓ 1 callersFunctionset_pad_token_id
Set pad_token_id to eos_token_id if it is None. Args: tokenizer (transformers.PreTrainedTokenizer): The tokenizer to be set.
verl/utils/tokenizer.py:20
↓ 1 callersFunctionslice_input_tensor
(x: Tensor, dim: int, padding: bool = True, group: ProcessGroup = None)
verl/utils/ulysses.py:116
↓ 1 callersMethodstart_worker_aliveness_check
(self, every_n_seconds=1)
verl/single_controller/base/worker_group.py:121
↓ 1 callersFunctiontest
()
tests/ray/test_check_worker_alive.py:20
↓ 1 callersFunctiontest_basics
()
tests/ray/test_ray_local_envs.py:36
↓ 1 callersFunctiontest_basics
()
tests/ray/test_worker_group_basics.py:90
↓ 1 callersFunctiontest_hf_casual_fwd
()
tests/model/test_transformers_ulysses.py:50
↓ 1 callersFunctiontest_hf_casual_fwd_bwd
()
tests/model/test_transformers_ulysses.py:128
↓ 1 callersFunctiontest_hf_casual_models
()
tests/model/test_transformer.py:32
↓ 1 callersFunctiontest_hf_value_models
()
tests/model/test_transformer.py:91
↓ 1 callersFunctiontest_memory_buffers
()
tests/gpu_utility/test_memory_buffers.py:27
↓ 1 callersMethodtrain_model
(self, data: DataProto)
tests/ray/detached_worker/server.py:120
↓ 1 callersMethodtraining_step
(self, batch: TensorDict)
verl/trainer/fsdp_sft_trainer.py:252
↓ 1 callersFunctionunfold_batch_dim
Unfold the first n dims as new batch dim
verl/protocol.py:124
↓ 1 callersFunctionunion_two_dict
Union two dict. Will throw an error if there is an item not the same object with the same key. Args: dict1: dict2: Returns:
verl/utils/py_functional.py:22
↓ 1 callersFunctionunit_convert
(number, level)
verl/utils/flops_counter.py:23
← previousnext →501–600 of 1,293, ranked by callers