Functions — github.com/Alibaba-NLP/VRAG

CodeHub MCPcopy Create free account

hub / github.com/Alibaba-NLP/VRAG / functions

Functions1,037 in github.com/Alibaba-NLP/VRAG

⨍ Functions 1,037◇ Types & classes 150↳ Endpoints 26

(self, config: PretrainedConfig)

VRAG-RL/verl/utils/flops_counter.py:64

(self, project_name, experiment_name, default_backend: Union[str, List[str]] = 'console', config=None)

VRAG-RL/verl/utils/tracking.py:27

VRAG-RL/verl/utils/tracking.py:110

VRAG-RL/verl/utils/seqlen_balancing.py:29

(self, items: List[Tuple[int, int]], k: int)

VRAG-RL/verl/utils/seqlen_balancing.py:51

(self, dictionary, **kwargs)

VRAG-RL/verl/utils/py_functional.py:50

(self, module: nn.Module)

VRAG-RL/verl/utils/memory_buffer.py:149

(self, transform_memory_param_fn)

VRAG-RL/verl/utils/memory_buffer.py:178

VRAG-RL/verl/utils/model.py:30

(self, numel, numel_padded, dtype)

VRAG-RL/verl/utils/megatron/memory.py:20

(self, remote_logger=None, enable_wandb=False, print_to_console=False)

VRAG-RL/verl/utils/logger/aggregate_logger.py:32

(self, parquet_files: Union[str, List[str]], tokenizer, pro

VRAG-RL/verl/utils/dataset/rm_dataset.py:42

(self, parquet_files: Union[str, List[str]], tokenizer, pro

VRAG-RL/verl/utils/dataset/sft_dataset.py:39

(self, parquet_files: Union[str, List[str]], tokenizer: PreTrainedTokenizer,

VRAG-RL/verl/utils/dataset/rl_dataset.py:80

(self, model: FSDP, optimizer: torch.optim.Optimizer, lr_sc

VRAG-RL/verl/utils/checkpoint/fsdp_checkpoint_manager.py:47

(self, model: FSDP, optimizer: torch.optim.Optimizer, lr_scheduler: torch.optim.lr_scheduler.

VRAG-RL/verl/utils/checkpoint/checkpoint_manager.py:42

(self, hdfs_dir, verbose)

VRAG-RL/verl/utils/debug/trajectory_tracker.py:52

(self, nccl_id)

VRAG-RL/verl/utils/rendezvous/ray_backend.py:27

(self, process_on_nodes: List[int] = None, use_gpu: bool = True,

VRAG-RL/verl/single_controller/ray/base.py:71

(self, cls, *args, **kwargs)

VRAG-RL/verl/single_controller/ray/base.py:150

(self, resource_pool: RayResourcePool = None, ray_cls_with_init: RayClassWit

VRAG-RL/verl/single_controller/ray/base.py:198

(self, resource_pool: RayResourcePool, ray_cls_with_init: RayClassWithInitArgs, **kwargs)

VRAG-RL/verl/single_controller/ray/megatron.py:31

(self, process_on_nodes=None, max_collocate_count: int = 10, n_gpus_per_node=8)

VRAG-RL/verl/single_controller/base/worker_group.py:29

(self, cls, *args, **kwargs)

VRAG-RL/verl/single_controller/base/worker_group.py:67

(self, resource_pool: ResourcePool, **kwargs)

VRAG-RL/verl/single_controller/base/worker_group.py:95

VRAG-RL/verl/single_controller/base/worker.py:73

(self, cuda_visible_devices=None)

VRAG-RL/verl/single_controller/base/worker.py:118

(self, resource_pool: ResourcePool, **kwargs)

VRAG-RL/verl/single_controller/base/megatron/worker_group.py:23

(self, cuda_visible_devices=None)

VRAG-RL/verl/single_controller/base/megatron/worker.py:20

(self, rank_zero_info)

VRAG-RL/verl/single_controller/base/register_center/ray.py:21

(self, config, device_mesh: DeviceMesh, ulysses_device_mesh: DeviceMesh)

VRAG-RL/verl/trainer/fsdp_sft_trainer.py:80

(self, config, tokenizer, role_worker_mapping: dict[Role, W

VRAG-RL/verl/trainer/ppo/ray_trainer.py:255

(self, init_kl_coef, target_kl, horizon)

VRAG-RL/verl/trainer/ppo/core_algos.py:34

(self, kl_coef)

VRAG-RL/verl/trainer/ppo/core_algos.py:49

(self, config: DictConfig, role: str)

VRAG-RL/verl/workers/megatron_workers.py:73

VRAG-RL/verl/workers/megatron_workers.py:668

(self, config: DictConfig, role: str)

VRAG-RL/verl/workers/fsdp_workers.py:76

VRAG-RL/verl/workers/fsdp_workers.py:904

(self, module: FSDP, inference_engine: LLM, model_config,

VRAG-RL/verl/workers/sharding_manager/fsdp_vllm.py:39

(self, module: FSDP, inference_engine: VerlEngine, model_co

VRAG-RL/verl/workers/sharding_manager/fsdp_sglang.py:49

(self, device_mesh: DeviceMesh)

VRAG-RL/verl/workers/sharding_manager/fsdp_ulysses.py:37

(self, model_provider)

VRAG-RL/verl/workers/sharding_manager/megatron_vllm.py:39

(self, module: AllGatherPPModel, inference_engine: LLM, model_config, layer_name_mapping)

VRAG-RL/verl/workers/sharding_manager/megatron_vllm.py:256

(self, tokenizer, num_examine, compute_score=None, eval_mode=False, rm_workers_num=10, rm_url="https://dashsco

VRAG-RL/verl/workers/reward_manager/rm.py:127

(self, tokenizer, num_examine, compute_score=None)

VRAG-RL/verl/workers/reward_manager/prime.py:89

(self, tokenizer, num_examine, compute_score=None)

VRAG-RL/verl/workers/reward_manager/naive.py:24

VRAG-RL/verl/workers/critic/base.py:28

(self, config, model_config, megatron_config, critic_module: nn.ModuleList, critic_optimizer:

VRAG-RL/verl/workers/critic/megatron_critic.py:45

(self, config, critic_module: nn.Module, critic_optimizer: optim.Optimizer)

VRAG-RL/verl/workers/critic/dp_critic.py:41

Args: dataloader: an Iterable of TensorDict that consistently generates prompts. Note that the dataloader should han

VRAG-RL/verl/workers/rollout/base.py:25

(self, module: nn.Module, config)

VRAG-RL/verl/workers/rollout/hf_rollout.py:37

A SGLang rollout. It requires the module is supported by the SGLang. Args: actor_module: module here follows huggingface APIs

VRAG-RL/verl/workers/rollout/sglang_rollout/sglang_rollout.py:86

A naive rollout. It requires the module to be compatible with huggingface APIs. That is: The module should define __call__ to receive input_id

VRAG-RL/verl/workers/rollout/naive/naive_rollout.py:38

A vLLM rollout. It requires the module is supported by the vllm. Args: module: module here follows huggingface APIs c

VRAG-RL/verl/workers/rollout/vllm_rollout/fire_vllm_rollout.py:60

A vLLM rollout. It requires the module is supported by the vllm. Args: module: module here follows huggingface APIs c

VRAG-RL/verl/workers/rollout/vllm_rollout/vllm_rollout.py:59

A vLLM rollout. It requires the module is supported by the vllm. Args: module: module here follows huggingface APIs c

VRAG-RL/verl/workers/rollout/vllm_rollout/vllm_rollout_spmd.py:67

MeagtronPPOActor class. This class implements the simple PPO logics when the model is built with Megatron. Args: config (OmegaCon

VRAG-RL/verl/workers/actor/megatron_actor.py:55

The base class for PPO actor Args: config (DictConfig): a config passed to the PPOActor. We expect the type to be

VRAG-RL/verl/workers/actor/base.py:28

When optimizer is None, it is Reference Policy

VRAG-RL/verl/workers/actor/dp_actor.py:41

VRAG-RL/verl/workers/reward_model/base.py:25

(self, config, model_config, reward_model_module: torch.nn.

VRAG-RL/verl/workers/reward_model/megatron/reward_model.py:34

(self, config: Qwen2Config, megatron_config: ModelParallelConfig)

VRAG-RL/verl/models/qwen2/megatron/modeling_qwen2_megatron.py:159

(self, config: Qwen2Config, megatron_config: ModelParallelConfig)

VRAG-RL/verl/models/qwen2/megatron/modeling_qwen2_megatron.py:226

(self, config: Qwen2Config, megatron_config: ModelParallelConfig)

VRAG-RL/verl/models/qwen2/megatron/modeling_qwen2_megatron.py:285

(self, config: Qwen2Config, megatron_config: ModelParallelConfig, pre_process, post_process)

VRAG-RL/verl/models/qwen2/megatron/modeling_qwen2_megatron.py:414

(self, config: Qwen2Config, megatron_config: ModelParallelConfig, pre_process, post_process,

VRAG-RL/verl/models/qwen2/megatron/modeling_qwen2_megatron.py:521

(self, config: Qwen2Config, megatron_config: ModelParallelConfig)

VRAG-RL/verl/models/qwen2/megatron/layers/parallel_decoder.py:104

(self, config, megatron_config: ModelParallelConfig = None)

VRAG-RL/verl/models/qwen2/megatron/layers/parallel_mlp.py:33

(self, dim, max_position_embeddings=2048, base=10000, device=None, scaling_factor=1.0)

VRAG-RL/verl/models/qwen2/megatron/layers/parallel_attention.py:75

(self, dim, max_position_embeddings=2048, base=10000, device=None, scaling_factor=1.0)

VRAG-RL/verl/models/qwen2/megatron/layers/parallel_attention.py:94

(self, config: Qwen2Config, megatron_config: ModelParallelConfig)

VRAG-RL/verl/models/qwen2/megatron/layers/parallel_attention.py:146

(self, input_size, gate_ouput_size, up_output_size,

VRAG-RL/verl/models/qwen2/megatron/layers/parallel_linear.py:54

Qwen2RMSNorm is equivalent to T5LayerNorm

VRAG-RL/verl/models/qwen2/megatron/layers/parallel_rmsnorm.py:27

(self, config: LlamaConfig, megatron_config: ModelParallelConfig)

VRAG-RL/verl/models/llama/megatron/modeling_llama_megatron.py:160

(self, config: LlamaConfig, megatron_config: ModelParallelConfig)

VRAG-RL/verl/models/llama/megatron/modeling_llama_megatron.py:227

(self, config: LlamaConfig, megatron_config: ModelParallelConfig)

VRAG-RL/verl/models/llama/megatron/modeling_llama_megatron.py:286

(self, config: LlamaConfig, megatron_config: ModelParallelConfig, pre_process, post_process)

VRAG-RL/verl/models/llama/megatron/modeling_llama_megatron.py:415

(self, config: LlamaConfig, megatron_config: ModelParallelConfig, pre_process, post_process,

VRAG-RL/verl/models/llama/megatron/modeling_llama_megatron.py:522

(self, config: LlamaConfig, megatron_config: ModelParallelConfig)

VRAG-RL/verl/models/llama/megatron/layers/parallel_decoder.py:104

(self, config, megatron_config: ModelParallelConfig = None)

VRAG-RL/verl/models/llama/megatron/layers/parallel_mlp.py:33

(self, dim, max_position_embeddings=2048, base=10000, device=None, scaling_factor=1.0)

VRAG-RL/verl/models/llama/megatron/layers/parallel_attention.py:75

(self, dim, max_position_embeddings=2048, base=10000, device=None, scaling_factor=1.0)

VRAG-RL/verl/models/llama/megatron/layers/parallel_attention.py:94

(self, config: LlamaConfig, megatron_config: ModelParallelConfig)

VRAG-RL/verl/models/llama/megatron/layers/parallel_attention.py:146

(self, input_size, gate_ouput_size, up_output_size,

VRAG-RL/verl/models/llama/megatron/layers/parallel_linear.py:54

LlamaRMSNorm is equivalent to T5LayerNorm

VRAG-RL/verl/models/llama/megatron/layers/parallel_rmsnorm.py:27

(self, config: TensorConfig)

VRAG-RL/vrag_agent/tensor_helper.py:10

( self, processor, actor_rollout_wg, config: GenerationConfig, is_vali

VRAG-RL/vrag_agent/generation.py:56

Initialize VimRAG agent. Args: base_url: vLLM service base URL (e.g., http://localhost:8000/v1) sear

demo/vimrag_agent.py:22

(self, base_url='http://0.0.0.0:8002/v1', search_url='http://0.0.0.0:8001/se

demo/vrag_agent.py:15

(self, embed_model_name='GVE')

search_engine/search_engine.py:11

( self, model_name_or_path: str, pooling: str = 'last', normalize: bool = True

search_engine/models/GVE/models.py:35

(self, model_name_or_path, eod_token=None)

search_engine/models/GVE/processor.py:14

(self, dims=None, dim_step=32, **kwargs)

search_engine/models/GVE/model_utils.py:10

(self, image_processor=None, tokenizer=None, chat_template=None, eod_token='<|endoftext|>', **kwargs)

search_engine/models/GVE/qwen25vl/qwen25vl_processor.py:46

search_engine/models/Qwen3_VL_Embedding/qwen3_vl_embedding.py:47

VRAG-RL/verl/protocol.py:191

VRAG-RL/verl/utils/dataset/rm_dataset.py:96

VRAG-RL/verl/utils/dataset/sft_dataset.py:107

VRAG-RL/verl/utils/dataset/rl_dataset.py:153

VRAG-RL/verl/utils/seqlen_balancing.py:42

← previous next →601–700 of 1,037, ranked by callers