Functions — github.com/Agent-RL/ReCall

hub / github.com/Agent-RL/ReCall / functions

Functions1,539 in github.com/Agent-RL/ReCall

⨍ Functions 1,539◇ Types & classes 237↳ Endpoints 28

↓ 5 callersMethodtruncate_prompt

(self, prompt)

src/flashrag/prompt/base_prompt.py:59

↓ 5 callersFunctionupdate_model_config

(module_config, override_config_kwargs)

src/verl/utils/model.py:42

↓ 5 callersFunctionvalidate_ulysses_config

(num_heads, ulysses_sequence_size)

src/verl/utils/ulysses.py:291

↓ 4 callersMethod__compress

( self, context_list: list, reduce_rate: float = 0.5, token_to_word: str = "me

src/flashrag/refiner/llmlingua_compressor.py:2227

↓ 4 callersMethod__init__

(self, config)

src/flashrag/refiner/refiner.py:12

↓ 4 callersMethod__init__

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

src/verl/models/llama/megatron/layers/parallel_attention.py:37

↓ 4 callersFunction_broadcast_tensor

broadcast tensor across mp_group

src/verl/models/mcore/saver.py:137

↓ 4 callersFunction_broadcast_tp_shard_tensor

broadcast tensor in tp shards across mp_group

src/verl/models/qwen2/megatron/checkpoint_utils/qwen2_saver.py:162

↓ 4 callersFunction_broadcast_tp_shard_tensor

broadcast tensor in tp shards across mp_group

src/verl/models/mcore/saver.py:176

↓ 4 callersFunction_broadcast_tp_shard_tensor

broadcast tensor in tp shards across mp_group

src/verl/models/llama/megatron/checkpoint_utils/llama_saver.py:162

↓ 4 callersFunction_get_cpu_tensor

(tensor: torch.Tensor)

src/verl/models/qwen2/megatron/checkpoint_utils/qwen2_saver.py:116

↓ 4 callersFunction_get_cpu_tensor

(tensor: torch.Tensor)

src/verl/models/mcore/saver.py:130

↓ 4 callersFunction_get_cpu_tensor

(tensor: torch.Tensor)

src/verl/models/llama/megatron/checkpoint_utils/llama_saver.py:116

↓ 4 callersFunction_get_gpt_model

(model)

src/verl/models/qwen2/megatron/checkpoint_utils/qwen2_loader_depracated.py:66

↓ 4 callersFunction_get_gpt_model

(model)

src/verl/models/qwen2/megatron/checkpoint_utils/qwen2_loader.py:66

↓ 4 callersFunction_get_gpt_model

(model)

src/verl/models/mcore/loader.py:64

↓ 4 callersFunction_get_gpt_model

(model)

src/verl/models/llama/megatron/checkpoint_utils/llama_loader.py:70

↓ 4 callersFunction_get_gpt_model

(model)

src/verl/models/llama/megatron/checkpoint_utils/llama_loader_depracated.py:70

↓ 4 callersFunction_is_non_local

(path: str)

src/verl/utils/hdfs_io.py:143

↓ 4 callersFunction_repeat_interleave

(value: Union[torch.Tensor, np.ndarray], repeats: int)

src/verl/workers/rollout/vllm_rollout/vllm_rollout_spmd.py:59

↓ 4 callersFunction_split_args_kwargs_data_proto

(chunks, *args, **kwargs)

src/verl/single_controller/base/decorator.py:45

↓ 4 callersMethod_update_dict

(old_dict: dict, new_dict: dict)

src/flashrag/config/config.py:56

↓ 4 callersFunctionapply_monkey_patch

Replace _flash_attention_forward to _ulysses_flash_attention_forward

src/verl/models/transformers/monkey_patch.py:110

↓ 4 callersFunctionbroadcast_dict_tensor

TODO: optimize this. Technically, we only need one broadcast

src/verl/utils/torch_functional.py:182

↓ 4 callersFunctionbuild_memory_reference_from_module

(module: torch.nn.Module, memory_buffers: Dict[torch.dtype, MemoryBuffe

src/verl/utils/memory_buffer.py:100

↓ 4 callersMethodcompress_prompt

Compresses the given context. Args: context (List[str]): List of context strings that form the basis of the prompt.

src/flashrag/refiner/llmlingua_compressor.py:552

↓ 4 callersFunctioncompute_position_id_with_mask

(mask)

src/verl/utils/model.py:194

↓ 4 callersMethodestimate_flops

Estimate the FLOPS based on the number of valid tokens in the current batch and the time taken. Args: batch_seqlens (Lis

src/verl/utils/flops_counter.py:164

↓ 4 callersMethodexecute_rank_zero_async

(self, method_name: str, *args, **kwargs)

src/verl/single_controller/ray/base.py:362

↓ 4 callersMethodfinish

(self)

src/verl/utils/tracking.py:134

↓ 4 callersMethodgenerate_sequences

Generate sequences

src/verl/workers/rollout/base.py:35

↓ 4 callersMethodget

(self)

src/verl/utils/rendezvous/ray_backend.py:30

↓ 4 callersFunctionget_fsdp_wrap_policy

Get FSDP wrap policy for the module. Args: module: The module to get wrap policy for config: Configuration for wrap policy

src/verl/utils/fsdp_utils.py:54

↓ 4 callersFunctionget_micro_data_parallel_group

()

src/verl/workers/sharding_manager/megatron_vllm.py:476

↓ 4 callersMethodget_placement_groups

(self, strategy="STRICT_PACK", name=None)

src/verl/single_controller/ray/base.py:85

↓ 4 callersMethodget_resource_pool

Get the resource pool of the worker_cls

src/verl/trainer/ppo/ray_trainer.py:101

↓ 4 callersFunctionis_digit

(s)

src/verl/utils/reward_score/prime_math/grader.py:107

↓ 4 callersFunctionprint_model_size

(model: nn.Module, name: str = None)

src/verl/utils/model.py:146

↓ 4 callersFunctionresolve_max_tokens

Resolve and validate max_tokens parameters from both params and generation_params. Args: params: Dictionary containing user-provided

src/flashrag/generator/utils.py:4

↓ 4 callersFunctionrotate_half

Rotates half the hidden dims of the input.

src/verl/models/qwen2/megatron/layers/parallel_attention.py:116

↓ 4 callersFunctionrotate_half

Rotates half the hidden dims of the input.

src/verl/models/llama/megatron/layers/parallel_attention.py:148

↓ 4 callersMethodrun

(self, env, func_schemas, question)

src/re_call/inference/re_call.py:141

↓ 4 callersMethodsave_checkpoint

(self, step)

src/verl/trainer/fsdp_sft_trainer.py:434

↓ 4 callersMethodsave_faiss_index

( self, all_embeddings, faiss_type, index_save_path, )

src/flashrag/retriever/index_builder.py:336

↓ 4 callersMethodshared_embedding_or_output_weight

(self)

src/verl/models/qwen2/megatron/modeling_qwen2_megatron.py:603

↓ 3 callersMethod__init__

(self, config)

src/flashrag/generator/fid.py:13

↓ 3 callersMethod__init__

(self, config, prompt_template=None)

src/flashrag/pipeline/pipeline.py:14

↓ 3 callersMethod__init__

(self)

src/verl/single_controller/ray/base.py:482

↓ 3 callersMethod__init__

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

src/verl/models/qwen2/megatron/layers/parallel_attention.py:37

↓ 3 callersMethod_batch_search

(self, query, num, return_score)

src/flashrag/retriever/retriever.py:184

↓ 3 callersFunction_broadcast_tp_shard_tensor

broadcast tensor in tp shards across mp_group

src/verl/models/qwen2/megatron/checkpoint_utils/qwen2_loader_depracated.py:181

↓ 3 callersFunction_broadcast_tp_shard_tensor

broadcast tensor in tp shards across mp_group

src/verl/models/mcore/loader.py:181

↓ 3 callersFunction_broadcast_tp_shard_tensor

broadcast tensor in tp shards across mp_group

src/verl/models/llama/megatron/checkpoint_utils/llama_loader_depracated.py:185

↓ 3 callersMethod_build_param_references

(self, pp_rank, maintain_weight=False)

src/verl/workers/sharding_manager/megatron_vllm.py:107

↓ 3 callersFunction_concat_data_proto_or_future

(output: List)

src/verl/single_controller/base/decorator.py:129

↓ 3 callersFunction_fetch_tp_shard_tensor

fetch tensor in tp shards

src/verl/models/qwen2/megatron/checkpoint_utils/qwen2_loader.py:122

↓ 3 callersFunction_fetch_tp_shard_tensor

fetch tensor in tp shards

src/verl/models/llama/megatron/checkpoint_utils/llama_loader.py:126

↓ 3 callersFunction_get_gpt_model

(model)

src/verl/models/qwen2/megatron/checkpoint_utils/qwen2_saver.py:85

↓ 3 callersFunction_get_gpt_model

(model)

src/verl/models/mcore/saver.py:98

↓ 3 callersFunction_get_gpt_model

(model)

src/verl/models/llama/megatron/checkpoint_utils/llama_saver.py:85

↓ 3 callersMethod_judge_input_modal

(self, query)

src/flashrag/retriever/retriever.py:498

↓ 3 callersMethod_search

r"""Retrieve topk relevant documents in corpus. Return: list: contains information related to the document, including:

src/flashrag/retriever/retriever.py:171

↓ 3 callersMethod_set_cos_sin_cache

(self, seq_len, device, dtype)

src/verl/models/llama/megatron/layers/parallel_attention.py:51

↓ 3 callersFunction_str_is_int

(x: str)

src/verl/utils/reward_score/prime_math/__init__.py:111

↓ 3 callersFunctionall_gather_data_proto

(data: DataProto, process_group)

src/verl/protocol.py:764

↓ 3 callersMethodbackward

(ctx: Any, grad_output: Tensor)

src/verl/utils/ulysses.py:226

↓ 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

src/verl/utils/memory_buffer.py:71

↓ 3 callersFunctioncalc_padded_numel

for cuda memory alignment, make sure alignment by 128-bits

src/verl/utils/memory_buffer.py:54

↓ 3 callersFunctioncompute_transformers_input_shapes

(batches, meta_info)

src/verl/utils/megatron/pipeline_parallel.py:22

↓ 3 callersFunctionconvert_image_to_base64

(image)

src/flashrag/generator/utils.py:70

↓ 3 callersMethodconvert_tokens_to_string

Converts a sequence of tokens in a single string. The most simple way to do it is `" ".join(tokens)` but we often want to remove sub-

src/verl/workers/rollout/tokenizer.py:147

↓ 3 callersFunctioncopy

r"""Works like shutil.copy() for file, and shutil.copytree for dir, and supports hdfs. Copy data and mode bits ("cp src dst"). Return the file's

src/verl/utils/hdfs_io.py:84

↓ 3 callersFunctioncreate_device_mesh

(world_size, fsdp_size)

src/verl/workers/fsdp_workers.py:49

↓ 3 callersMethodexecute_all_sync

(self, method_name: str, *args, **kwargs)

src/verl/single_controller/ray/base.py:372

↓ 3 callersMethodformat_reference

(self, retrieval_result)

src/flashrag/pipeline/active_pipeline.py:941

↓ 3 callersMethodgenerate

(self, input_list: list, batch_size=None, **params)

src/flashrag/generator/multimodal_generator.py:75

↓ 3 callersFunctionget_cosine_schedule_with_warmup

Create a schedule with a learning rate that decreases following the values of the cosine function between the initial lr set in the optimizer

src/verl/utils/torch_functional.py:415

↓ 3 callersFunctionget_default_kwargs_for_model_parallel_config

()

src/verl/utils/megatron/tensor_parallel.py:32

↓ 3 callersFunctionget_micro_data_parallel_world_size

()

src/verl/workers/sharding_manager/megatron_vllm.py:481

↓ 3 callersFunctionget_reverse_idx

(idx_map)

src/verl/utils/seqlen_balancing.py:259

↓ 3 callersFunctionget_sharding_strategy

(device_mesh)

src/verl/workers/fsdp_workers.py:59

↓ 3 callersFunctionhf_processor

Create a huggingface processor to process multimodal data. Args: name_or_path (str): The name of the processor. Returns: tra

src/verl/utils/tokenizer.py:62

↓ 3 callersFunctionimport_external_libs

(external_libs=None)

src/verl/utils/import_utils.py:51

↓ 3 callersMethodload_checkpoint

(self, local_path, hdfs_path=None, del_local_after_load=True)

src/verl/workers/fsdp_workers.py:906

↓ 3 callersFunctionload_mcore_dist_weights

(parallel_model, dist_weight_path, is_value_model=False)

src/verl/utils/model.py:409

↓ 3 callersFunctionload_megatron_gptmodel_weights

(config, model_config, parallel_model,

src/verl/utils/model.py:367

↓ 3 callersFunctionload_model

(model_path: str, use_fp16: bool = False)

src/flashrag/retriever/utils.py:54

↓ 3 callersFunctionmake_batch_generator

(batches, vpp_size)

src/verl/utils/megatron/pipeline_parallel.py:43

↓ 3 callersFunctionmcore_model_parallel_config

( sequence_parallel: bool, params_dtype: torch.dtype, )

src/verl/utils/megatron_utils.py:216

↓ 3 callersMethodnaive_run

(self, dataset, do_eval=True, pred_process_fun=None)

src/flashrag/pipeline/pipeline.py:73

↓ 3 callersFunctionnormalize_final_answer

Normalize a final answer to a quantitative reasoning question. Args: final_answer: The answer string to normalize Return

src/verl/utils/reward_score/math_dapo.py:142

↓ 3 callersFunctionparse_query

processing query for different encoders

src/flashrag/retriever/utils.py:100

↓ 3 callersFunctionprocess_sequence_data

(rate, start, end, sequence, is_dict=False)

src/flashrag/refiner/llmlingua_compressor.py:174

↓ 3 callersFunctionremove_images

(data: Any)

src/flashrag/dataset/utils.py:73

↓ 3 callersMethodrerank

r"""Rerank doc_list.

src/flashrag/retriever/reranker.py:35

↓ 3 callersMethodrun

The overall inference process of a RAG framework.

src/flashrag/pipeline/pipeline.py:24

↓ 3 callersMethodset_input_tensor

Set input tensor to be used instead of forward()'s input. When doing pipeline parallelism the input from the previous stage comes fro

src/verl/models/qwen2/megatron/modeling_qwen2_megatron.py:461

↓ 3 callersFunctionset_random_seed

(seed)

src/verl/workers/megatron_workers.py:53

↓ 3 callersFunctionsplit_dict_tensor_into_batches

(tensors: TensorDict, batch_size)

src/verl/utils/torch_functional.py:225

↓ 3 callersFunctiontime_limit

(seconds: float)

src/verl/utils/reward_score/prime_math/grader.py:345

← previous next →101–200 of 1,539, ranked by callers