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

↓ 1 callersFunctionare_equal_under_sympy

(ground_truth_normalized: str, given_normalized: str)

src/verl/utils/reward_score/prime_math/__init__.py:240

↓ 1 callersMethodbatch_execute

(self, env_list: List[str], tool_calls_list: List[List[str]])

src/verl/workers/rollout/vllm_rollout/vllm_rollout_spmd.py:356

↓ 1 callersMethodbeautify_context

(self, context: str)

src/flashrag/refiner/selective_context_compressor.py:239

↓ 1 callersFunctionbroadcast_params

(module)

src/verl/models/qwen2/megatron/checkpoint_utils/qwen2_loader_depracated.py:69

↓ 1 callersFunctionbroadcast_params

(module)

src/verl/models/mcore/loader.py:67

↓ 1 callersFunctionbroadcast_params

(module)

src/verl/models/llama/megatron/checkpoint_utils/llama_loader_depracated.py:73

↓ 1 callersMethodbuild_bm25_index_bm25s

Building BM25 index based on bm25s library.

src/flashrag/retriever/index_builder.py:190

↓ 1 callersMethodbuild_bm25_index_pyserini

Building BM25 index based on Pyserini library. Reference: https://github.com/castorini/pyserini/blob/master/docs/usage-index.md#building-a-bm

src/flashrag/retriever/index_builder.py:133

↓ 1 callersMethodbuild_dense_index

Obtain the representation of documents based on the embedding model(BERT-based) and construct a faiss index.

src/flashrag/retriever/index_builder.py:262

↓ 1 callersMethodbuild_index

r"""Constructing different indexes based on selective retrieval method.

src/flashrag/retriever/index_builder.py:121

↓ 1 callersFunctionbuild_memory_reference

Build the memory references. The memory buffers are built using the build_memory_buffer API. This API will allocate a weight buffer pointer to the

src/verl/utils/memory_buffer.py:116

↓ 1 callersMethodbuild_prompt_for_reasoning_chain

( self, hop, question, existing_triples, candidate_triples, ge

src/flashrag/refiner/kg_refiner.py:295

↓ 1 callersMethodbuild_single_doc_prompt

(self, question: str, doc_list: List[str])

src/flashrag/pipeline/branching_pipeline.py:25

↓ 1 callersMethodbuild_transform

(self, input_size)

src/flashrag/generator/multimodal_generator.py:161

↓ 1 callersFunctioncal_llm_judge_metric

(args)

scripts/evaluation/llm_judge.py:39

↓ 1 callersMethodcalculate_em

(self, prediction: str, golden_answers: list)

src/flashrag/evaluator/metrics.py:146

↓ 1 callersMethodcalculate_metric

(self, data)

src/flashrag/evaluator/metrics.py:473

↓ 1 callersMethodcalculate_sub_em

(self, prediction: str, golden_answers: list)

src/flashrag/evaluator/metrics.py:187

↓ 1 callersFunctioncall_method

(method, inputs)

src/verl/utils/reward_score/prime_code/testing_util.py:617

↓ 1 callersMethodcat_assistant_response

(self, curr_prompt, assistant_response)

src/re_call/inference/re_call.py:53

↓ 1 callersMethodcat_tool_results

(self, curr_prompt, tool_calls, results)

src/re_call/inference/re_call.py:56

↓ 1 callersFunctionceildiv

(a, b)

src/verl/utils/seqlen_balancing.py:220

↓ 1 callersMethodcheck_consistency

Check the consistency of the DataProto. Mainly for batch and non_tensor_batch We expose this function as a public one so that user can call th

src/verl/protocol.py:280

↓ 1 callersFunctionclean_prompt_image

(input)

src/flashrag/dataset/utils.py:96

↓ 1 callersFunctioncollect_dp_compute

(worker_group, output)

src/verl/single_controller/base/decorator.py:270

↓ 1 callersFunctioncollect_megatron_compute

Only collect the data from the tp=0 and pp=last and every dp ranks

src/verl/single_controller/base/decorator.py:103

↓ 1 callersFunctioncollect_megatron_pp_as_dp

treat pp as dp. Only collect data on tp=0

src/verl/single_controller/base/decorator.py:213

↓ 1 callersMethodcompress_prompt_llmlingua2

Compresses the given context, instruction and question. Args: context (List[str]): List of context strings that form the

src/flashrag/refiner/llmlingua_compressor.py:825

↓ 1 callersFunctioncompute_advantage

(data: DataProto, adv_estimator, gamma=1.0, lam=1.0, num_repeat=1)

src/verl/trainer/ppo/ray_trainer.py:177

↓ 1 callersFunctioncompute_data_metrics

(batch: DataProto, use_critic: bool = True)

src/verl/trainer/ppo/metric_utils.py:48

↓ 1 callersMethodcompute_log_prob

(self, data: DataProto)

src/verl/workers/megatron_workers.py:455

↓ 1 callersMethodcompute_log_prob

(self, data: DataProto)

src/verl/workers/fsdp_workers.py:540

↓ 1 callersMethodcompute_log_prob

Compute logits given a batch of data. Args: data (DataProto): a batch of data represented by DataProto. It must contain key ```in

src/verl/workers/actor/base.py:39

↓ 1 callersMethodcompute_ref_log_prob

(self, data: DataProto)

src/verl/workers/fsdp_workers.py:574

↓ 1 callersMethodcompute_reward

Computing reward given input_ids. The transformers should output a tensor with shape [batch_size, sequence_length], and the value at [EOS]

src/verl/workers/reward_model/base.py:29

↓ 1 callersMethodcompute_rm_score

(self, data: DataProto)

src/verl/workers/fsdp_workers.py:1156

↓ 1 callersFunctioncompute_score

(solution_str, ground_truth)

src/verl/utils/reward_score/math.py:17

↓ 1 callersFunctioncompute_throughout_metrics

(batch: DataProto, timing_raw: Dict[str, float], n_gpus: int)

src/verl/trainer/ppo/metric_utils.py:160

↓ 1 callersFunctioncompute_timing_metrics

(batch: DataProto, timing_raw: Dict[str, float])

src/verl/trainer/ppo/metric_utils.py:136

↓ 1 callersMethodcompute_values

(self, data: DataProto)

src/verl/workers/fsdp_workers.py:831

↓ 1 callersFunctionconcat_dict_to_str

(dict: Dict, step)

src/verl/utils/logger/aggregate_logger.py:21

↓ 1 callersMethodconcate_segment_info

( self, segment_info: List[List[tuple]], )

src/flashrag/refiner/llmlingua_compressor.py:2078

↓ 1 callersMethodcontrol_context_budget

( self, context: List[str], context_tokens_length: List[int], target_token: fl

src/flashrag/refiner/llmlingua_compressor.py:1223

↓ 1 callersMethodcontrol_sentence_budget

( self, context: List[str], target_token: float, keep_first_sentence: int = 0,

src/flashrag/refiner/llmlingua_compressor.py:1288

↓ 1 callersMethodconvert_candidate_triples_to_choices

(self, candidates)

src/flashrag/refiner/kg_refiner.py:289

↓ 1 callersFunctionconvert_gate_up_shard

(full_tensor, gate_name, up_name)

src/verl/utils/megatron_utils.py:364

↓ 1 callersFunctionconvert_megatron_model_to_transformers_model

Convert megatron model to transformers model.

src/verl/utils/megatron_utils.py:307

↓ 1 callersFunctionconvert_numpy

(data: Any)

src/flashrag/dataset/utils.py:6

↓ 1 callersFunctionconvert_numpy

Recursively convert numpy objects in nested dictionaries or lists to native Python types.

src/flashrag/retriever/utils.py:11

↓ 1 callersFunctionconvert_qkv_shard

(full_tensor, q_name, k_name, v_name)

src/verl/utils/megatron_utils.py:316

↓ 1 callersFunctionconvert_to_regular_types

Convert Hydra configs and other special types to regular Python types.

src/verl/trainer/fsdp_sft_trainer.py:67

↓ 1 callersFunctioncount_unknown_letters_in_expr

(expr: str)

src/verl/utils/reward_score/prime_math/__init__.py:216

↓ 1 callersFunctioncreate_colocated_worker_cls

This function should return a class instance that delegates the calls to every cls in cls_dict

src/verl/single_controller/ray/base.py:460

↓ 1 callersFunctioncreate_huggingface_actor

Args: model_name: override_config_kwargs: Returns:

src/verl/utils/model.py:75

↓ 1 callersMethodcreate_resource_pool

(self)

src/verl/trainer/ppo/ray_trainer.py:88

↓ 1 callersFunctioncreate_worker_group_register_center

(name, info)

src/verl/single_controller/base/register_center/ray.py:28

↓ 1 callersFunctioncustom_compare_

(output, ground_truth)

src/verl/utils/reward_score/prime_code/testing_util.py:595

↓ 1 callersMethoddefault_tp_concat_fn

name: name of the parameter param: training parameters infer_params (Iterable[torch.Tensor]): a iterator towards list of para

src/verl/workers/sharding_manager/megatron_vllm.py:309

↓ 1 callersMethoddispatch_fn

(x, i, chunks)

src/verl/protocol.py:741

↓ 1 callersFunctiondispatch_megatron_compute

User passes in dp data. The data is dispatched to all tp/pp ranks with the same dp

src/verl/single_controller/base/decorator.py:74

↓ 1 callersFunctiondispatch_megatron_pp_as_dp

treat pp as dp.

src/verl/single_controller/base/decorator.py:161

↓ 1 callersFunctiondownload_files_distributed

(download_fn)

src/verl/utils/dataset/rm_dataset.py:27

↓ 1 callersFunctiondump_data

(data, name)

src/verl/utils/debug/trajectory_tracker.py:69

↓ 1 callersMethoddynamic_preprocess

(self, image, min_num=1, max_num=12, image_size=448, use_thumbnail=False)

src/flashrag/generator/multimodal_generator.py:190

↓ 1 callersMethodencode

(self, query_list: List[str], batch_size=64, modal="image")

src/flashrag/retriever/encoder.py:193

↓ 1 callersMethodencode_all

(self)

src/flashrag/retriever/index_builder.py:232

↓ 1 callersMethodencode_all_clip

(self)

src/flashrag/retriever/index_builder.py:243

↓ 1 callersMethodencode_passages

(self, batch_text_passages: List[List[str]])

src/flashrag/generator/generator.py:86

↓ 1 callersFunctionexecute

(func, input_list_or_num_samples: Union[list, int], output_path: str, max_workers: int, logger: logging.Logger

scripts/evaluation/utils.py:10

↓ 1 callersMethodexecute_tool_calls

(self, env: str, tool_calls: List[str])

src/re_call/inference/re_call.py:76

↓ 1 callersMethodextract_document_triples

Extract triples from documents associated with each query, handling duplicates and generating prompts for LLM processing.

src/flashrag/refiner/kg_refiner.py:219

↓ 1 callersMethodextract_judge_score

(answer: str, split_str: str = "Total rating:")

src/flashrag/evaluator/metrics.py:546

↓ 1 callersMethodextract_last_assistant_response

(self, text: str)

src/flashrag/pipeline/active_pipeline.py:79

↓ 1 callersFunctionextract_solution

(solution_str, method='strict')

src/verl/utils/reward_score/gsm8k.py:18

↓ 1 callersMethodextract_tool_calls

(self, output_str)

src/re_call/inference/re_call.py:128

↓ 1 callersMethodextract_tool_calls

(self, output_str)

src/verl/workers/rollout/vllm_rollout/vllm_rollout_spmd.py:344

↓ 1 callersFunctionfactory

(name_or_path)

src/flashrag/pipeline/replug_utils.py:232

↓ 1 callersMethodfind_closest_aspect_ratio

(self, aspect_ratio, target_ratios, width, height, image_size)

src/flashrag/generator/multimodal_generator.py:175

↓ 1 callersFunctionfind_latest_ckpt_path

(path, directory_format="global_step_{}")

src/verl/utils/checkpoint/checkpoint_manager.py:124

↓ 1 callersMethodfit

(self)

src/verl/trainer/fsdp_sft_trainer.py:452

↓ 1 callersMethodfit

The training loop of PPO. The driver process only need to call the compute functions of the worker group through RPC to construct the

src/verl/trainer/ppo/ray_trainer.py:799

↓ 1 callersFunctionfix_a_slash_b

(string)

src/verl/utils/reward_score/math.py:125

↓ 1 callersFunctionfix_fracs

(string)

src/verl/utils/reward_score/math.py:93

↓ 1 callersFunctionfix_sqrt

(string)

src/verl/utils/reward_score/math.py:150

↓ 1 callersFunctionflash_attention_forward

Patches flash attention forward to handle 3D position ids in mrope. (3, batch_size, seq_length)

src/verl/models/transformers/qwen2_vl.py:150

↓ 1 callersMethodflush

(self)

src/verl/utils/logger/aggregate_logger.py:37

↓ 1 callersMethodformat_evidences

(self, evidences: List[str])

src/flashrag/pipeline/active_pipeline.py:1234

↓ 1 callersFunctionformat_intervals

(prediction)

src/verl/utils/reward_score/prime_math/grader.py:358

↓ 1 callersMethodformat_ref

(titles, texts)

src/flashrag/pipeline/branching_pipeline.py:130

↓ 1 callersMethodformat_reference

(self, retrieval_result)

src/flashrag/prompt/base_prompt.py:192

↓ 1 callersMethodformat_reference

(self, retrieval_result)

src/flashrag/refiner/refiner.py:62

↓ 1 callersMethodformat_reference

(self, retrieval_result)

src/flashrag/refiner/refiner.py:115

↓ 1 callersMethodformat_reference

(self, retrieval_result)

src/flashrag/refiner/kg_refiner.py:543

↓ 1 callersMethodformat_reference

(self, doc_item)

src/flashrag/pipeline/branching_pipeline.py:28

↓ 1 callersFunctionformat_reward

(predict_str: str)

src/verl/utils/reward_score/geo3k.py:19

↓ 1 callersMethodformat_tool_call

Convert JSON function call description to Python executable code string.

src/re_call/inference/re_call.py:64

↓ 1 callersMethodformat_tool_call

Convert JSON function call description to Python executable code string.

src/verl/workers/rollout/vllm_rollout/vllm_rollout_spmd.py:316

↓ 1 callersMethodforward

(self, *inputs, **kwargs)

src/verl/workers/sharding_manager/megatron_vllm.py:149

↓ 1 callersMethodforward_batch

We assume: - The model takes input: (input_ids, attention_mask, position_ids). No rmpad for the input - The communication sha

src/verl/workers/reward_model/megatron/reward_model.py:180

↓ 1 callersMethodfrom_detached

(cls, worker_names=None, ray_cls_with_init=None)

src/verl/single_controller/ray/base.py:325

← previous next →501–600 of 1,539, ranked by callers