Index

_

__init__() (trinity.algorithm.advantage_fn.grpo_advantage.GRPOAdvantageFn method) (trinity.algorithm.advantage_fn.grpo_advantage.GRPOGroupedAdvantage method) (trinity.algorithm.advantage_fn.GRPOAdvantageFn method) (trinity.algorithm.advantage_fn.GRPOGroupedAdvantage method) (trinity.algorithm.advantage_fn.opmd_advantage.OPMDAdvantageFn method) (trinity.algorithm.advantage_fn.opmd_advantage.OPMDGroupAdvantage method) (trinity.algorithm.advantage_fn.OPMDAdvantageFn method) (trinity.algorithm.advantage_fn.OPMDGroupAdvantage method) (trinity.algorithm.advantage_fn.ppo_advantage.PPOAdvantageFn method) (trinity.algorithm.advantage_fn.PPOAdvantageFn method) (trinity.algorithm.advantage_fn.reinforce_plus_plus_advantage.REINFORCEPLUSPLUSAdvantageFn method) (trinity.algorithm.advantage_fn.REINFORCEPLUSPLUSAdvantageFn method) (trinity.algorithm.advantage_fn.remax_advantage.REMAXAdvantageFn method) (trinity.algorithm.advantage_fn.REMAXAdvantageFn method) (trinity.algorithm.advantage_fn.rloo_advantage.RLOOAdvantageFn method) (trinity.algorithm.advantage_fn.RLOOAdvantageFn method) (trinity.algorithm.advantage_fn.StepWiseGRPOAdvantageFn method) (trinity.algorithm.algorithm_manager.AlgorithmManager method) (trinity.algorithm.entropy_loss_fn.entropy_loss_fn.DefaultEntropyLossFn method) (trinity.algorithm.entropy_loss_fn.entropy_loss_fn.DummyEntropyLossFn method) (trinity.algorithm.entropy_loss_fn.entropy_loss_fn.MixEntropyLossFn method) (trinity.algorithm.key_mapper.KeyMapper method) (trinity.algorithm.kl_fn.kl_fn.KLFn method) (trinity.algorithm.kl_fn.KLFn method) (trinity.algorithm.KLFn method) (trinity.algorithm.policy_loss_fn.dpo_loss.DPOLossFn method) (trinity.algorithm.policy_loss_fn.DPOLossFn method) (trinity.algorithm.policy_loss_fn.GSPOLossFn method) (trinity.algorithm.policy_loss_fn.mix_policy_loss.MIXPolicyLossFn method) (trinity.algorithm.policy_loss_fn.MIXCHORDPolicyLossFn method) (trinity.algorithm.policy_loss_fn.MIXPolicyLossFn method) (trinity.algorithm.policy_loss_fn.opmd_policy_loss.OPMDPolicyLossFn method) (trinity.algorithm.policy_loss_fn.OPMDPolicyLossFn method) (trinity.algorithm.policy_loss_fn.policy_loss_fn.PolicyLossFn method) (trinity.algorithm.policy_loss_fn.PolicyLossFn method) (trinity.algorithm.policy_loss_fn.ppo_policy_loss.PPOPolicyLossFn method) (trinity.algorithm.policy_loss_fn.PPOPolicyLossFn method) (trinity.algorithm.policy_loss_fn.sft_loss.SFTLossFn method) (trinity.algorithm.policy_loss_fn.SFTISLossFn method) (trinity.algorithm.policy_loss_fn.SFTLossFn method) (trinity.algorithm.policy_loss_fn.SFTPhiLossFn method) (trinity.algorithm.PolicyLossFn method) (trinity.algorithm.sample_strategy.DefaultSampleStrategy method) (trinity.algorithm.sample_strategy.mix_sample_strategy.MixSampleStrategy method) (trinity.algorithm.sample_strategy.MixSampleStrategy method) (trinity.algorithm.sample_strategy.sample_strategy.DefaultSampleStrategy method) (trinity.algorithm.sample_strategy.sample_strategy.SampleStrategy method) (trinity.algorithm.sample_strategy.sample_strategy.WarmupSampleStrategy method) (trinity.algorithm.sample_strategy.SampleStrategy method) (trinity.algorithm.sample_strategy.WarmupSampleStrategy method) (trinity.algorithm.SampleStrategy method) (trinity.buffer.queue.AsyncPriorityQueue method) (trinity.buffer.queue.AsyncQueue method) (trinity.buffer.ray_wrapper.DBWrapper method) (trinity.buffer.ray_wrapper.FileWrapper method) (trinity.buffer.ray_wrapper.QueueWrapper method) (trinity.buffer.reader.file_reader.DPODataReader method) (trinity.buffer.reader.file_reader.DummyProgressBar method) (trinity.buffer.reader.file_reader.RawDataReader method) (trinity.buffer.reader.file_reader.RolloutDataReader method) (trinity.buffer.reader.file_reader.SFTDataReader method) (trinity.buffer.reader.queue_reader.QueueReader method) (trinity.buffer.reader.sql_reader.SQLReader method) (trinity.buffer.schema.Base method) (trinity.buffer.writer.file_writer.JSONWriter method) (trinity.buffer.writer.queue_writer.QueueWriter method) (trinity.buffer.writer.sql_writer.SQLWriter method) (trinity.common.config.AlgorithmConfig method) (trinity.common.config.BufferConfig method) (trinity.common.config.ClusterConfig method) (trinity.common.config.Config method) (trinity.common.config.DataJuicerServiceConfig method) (trinity.common.config.DataProcessorConfig method) (trinity.common.config.ExperiencePipelineConfig method) (trinity.common.config.ExplorerConfig method) (trinity.common.config.ExplorerInput method) (trinity.common.config.FormatConfig method) (trinity.common.config.GenerationConfig method) (trinity.common.config.InferenceModelConfig method) (trinity.common.config.ModelConfig method) (trinity.common.config.MonitorConfig method) (trinity.common.config.OperatorConfig method) (trinity.common.config.ServiceConfig method) (trinity.common.config.StorageConfig method) (trinity.common.config.SynchronizerConfig method) (trinity.common.config.TaskPipelineConfig method) (trinity.common.config.TrainerConfig method) (trinity.common.config.TrainerInput method) (trinity.common.experience.CustomField method) (trinity.common.experience.EID method) (trinity.common.experience.Experience method) (trinity.common.experience.Experiences method) (trinity.common.models.model.ModelWrapper method) (trinity.common.models.vllm_model.vLLMRolloutModel method) (trinity.common.rewards.accuracy_reward.AccuracyReward method) (trinity.common.rewards.AccuracyReward method) (trinity.common.rewards.countdown_reward.CountDownRewardFn method) (trinity.common.rewards.CountDownRewardFn method) (trinity.common.rewards.format_reward.FormatReward method) (trinity.common.rewards.FormatReward method) (trinity.common.rewards.math_reward.MathBoxedRewardFn method) (trinity.common.rewards.math_reward.MathRewardFn method) (trinity.common.rewards.MathBoxedRewardFn method) (trinity.common.rewards.MathDAPORewardFn method) (trinity.common.rewards.MathRewardFn method) (trinity.common.rewards.reward_fn.RewardFn method) (trinity.common.rewards.reward_fn.RMGalleryFn method) (trinity.common.rewards.RewardFn method) (trinity.common.rewards.RMGalleryFn method) (trinity.common.verl_config.Actor method) (trinity.common.verl_config.ActorModel method) (trinity.common.verl_config.ActorRolloutRef method) (trinity.common.verl_config.Algorithm method) (trinity.common.verl_config.Checkpoint method) (trinity.common.verl_config.Critic method) (trinity.common.verl_config.CriticModel method) (trinity.common.verl_config.CustomRewardFunction method) (trinity.common.verl_config.Data method) (trinity.common.verl_config.FSDPConfig method) (trinity.common.verl_config.FusedKernelOptions method) (trinity.common.verl_config.KL_Ctrl method) (trinity.common.verl_config.Optim method) (trinity.common.verl_config.Ref method) (trinity.common.verl_config.RewardModel method) (trinity.common.verl_config.Rollout method) (trinity.common.verl_config.Trainer method) (trinity.common.verl_config.veRLConfig method) (trinity.common.verl_config.WrapPolicy method) (trinity.common.workflows.AgentScopeReactV2MathWorkflow method) (trinity.common.workflows.AlfworldWorkflow method) (trinity.common.workflows.math_rm_workflow.MathRMWorkflow method) (trinity.common.workflows.MathEvalWorkflow method) (trinity.common.workflows.MathRMWorkflow method) (trinity.common.workflows.MathWorkflow method) (trinity.common.workflows.RAFTAlfworldWorkflow method) (trinity.common.workflows.RAFTReflectAlfworldWorkflow method) (trinity.common.workflows.SciWorldWorkflow method) (trinity.common.workflows.SimpleWorkflow method) (trinity.common.workflows.StepWiseAlfworldWorkflow method) (trinity.common.workflows.Task method) (trinity.common.workflows.WebShopWorkflow method) (trinity.common.workflows.Workflow method) (trinity.common.workflows.workflow.MathWorkflow method) (trinity.common.workflows.workflow.MultiTurnWorkflow method) (trinity.common.workflows.workflow.SimpleWorkflow method) (trinity.common.workflows.workflow.Task method) (trinity.common.workflows.workflow.Workflow method) (trinity.explorer.Explorer method) (trinity.explorer.explorer.Explorer method) (trinity.explorer.scheduler.RunnerWrapper method) (trinity.explorer.scheduler.Scheduler method) (trinity.explorer.scheduler.TaskWrapper method) (trinity.explorer.workflow_runner.Status method) (trinity.explorer.workflow_runner.WorkflowRunner method) (trinity.manager.CacheManager method) (trinity.manager.config_manager.ConfigManager method) (trinity.manager.config_registry.config_registry.ConfigRegistry method) (trinity.manager.manager.CacheManager method) (trinity.manager.Synchronizer method) (trinity.trainer.Trainer method) (trinity.trainer.trainer.Trainer method) (trinity.trainer.verl.dp_actor.DataParallelPPOActor method) (trinity.trainer.verl.fsdp_workers.ActorRolloutRefWorker method) (trinity.trainer.verl.fsdp_workers.CriticWorker method) (trinity.trainer.verl_trainer.VerlPPOTrainerWrapper method) (trinity.utils.dlc_utils.ClusterStatus method) (trinity.utils.log.NewLineFormatter method) (trinity.utils.monitor.MlflowMonitor method) (trinity.utils.monitor.Monitor method) (trinity.utils.monitor.TensorboardMonitor method) (trinity.utils.monitor.WandbMonitor method) (trinity.utils.registry.Registry method) (trinity.utils.timer.Timer method)

_ready_condition (trinity.manager.Synchronizer attribute)

A

AbsFn (class in trinity.algorithm.kl_fn.kl_fn) AccuracyReward (class in trinity.common.rewards) (class in trinity.common.rewards.accuracy_reward) acquire() (trinity.buffer.buffer_writer.BufferWriter method) (trinity.buffer.BufferWriter method) (trinity.buffer.ray_wrapper.DBWrapper method) (trinity.buffer.ray_wrapper.FileWrapper method) (trinity.buffer.ray_wrapper.QueueWrapper method) (trinity.buffer.writer.file_writer.JSONWriter method) (trinity.buffer.writer.queue_writer.QueueWriter method) (trinity.buffer.writer.sql_writer.SQLWriter method) action_mask (trinity.common.experience.Experience attribute) action_masks (trinity.common.experience.Experiences attribute) Actor (class in trinity.common.verl_config) actor (trinity.common.verl_config.ActorRolloutRef attribute) actor_grad_clip (trinity.common.config.TrainerConfig attribute) actor_rollout_ref (trinity.common.verl_config.veRLConfig attribute) ActorModel (class in trinity.common.verl_config) ActorRolloutRef (class in trinity.common.verl_config) ActorRolloutRefWorker (class in trinity.trainer.verl.fsdp_workers) ADD (trinity.common.constants.OpType attribute) add_module() (trinity.manager.Synchronizer method) adv_estimator (trinity.common.verl_config.Algorithm attribute) advantage_fn (trinity.common.config.AlgorithmConfig attribute) advantage_fn_args (trinity.common.config.AlgorithmConfig attribute) AdvantageFn (class in trinity.algorithm) (class in trinity.algorithm.advantage_fn) (class in trinity.algorithm.advantage_fn.advantage_fn)

advantages (trinity.common.experience.Experience attribute) (trinity.common.experience.Experiences attribute) AgentScopeReactV2MathWorkflow (class in trinity.common.workflows) AlfworldWorkflow (class in trinity.common.workflows) Algorithm (class in trinity.common.verl_config) algorithm (trinity.common.config.Config attribute) (trinity.common.verl_config.veRLConfig attribute) algorithm_type (trinity.common.config.AlgorithmConfig attribute) (trinity.common.config.StorageConfig attribute) AlgorithmConfig (class in trinity.common.config) AlgorithmManager (class in trinity.algorithm.algorithm_manager) AlgorithmType (class in trinity.algorithm) (class in trinity.algorithm.algorithm) api_server_ready() (trinity.common.models.vllm_model.vLLMRolloutModel method) apply_kl_penalty_to_reward() (trinity.algorithm.kl_fn.kl_fn.DummyKLFn method) (trinity.algorithm.kl_fn.kl_fn.KLFn method) (trinity.algorithm.kl_fn.KLFn method) (trinity.algorithm.KLFn method) args (trinity.common.config.OperatorConfig attribute) AsyncPriorityQueue (class in trinity.buffer.queue) AsyncQueue (class in trinity.buffer.queue) attention_masks (trinity.common.experience.Experiences attribute) auto_start (trinity.common.config.DataJuicerServiceConfig attribute) auxiliary_models (trinity.common.config.ExplorerConfig attribute)

B

balance_batch (trinity.common.verl_config.Trainer attribute) Base (class in trinity.buffer.schema) BaseFileReader (class in trinity.buffer.reader.file_reader) batch (trinity.common.experience.EID attribute) batch_id (trinity.common.workflows.Task attribute) (trinity.common.workflows.workflow.Task attribute) (trinity.explorer.scheduler.TaskWrapper attribute) batch_size (trinity.common.config.BufferConfig attribute) (trinity.common.experience.Experiences property) beginner_mode() (trinity.manager.config_manager.ConfigManager method) bench_on_latest_checkpoint (trinity.common.config.ExplorerConfig attribute)

benchmark() (trinity.explorer.Explorer method) (trinity.explorer.explorer.Explorer method) betas (trinity.common.verl_config.Optim attribute) broadcast_advantages() (trinity.algorithm.advantage_fn.StepWiseGRPOAdvantageFn method) buffer (trinity.common.config.Config attribute) (trinity.common.verl_config.veRLConfig attribute) BufferConfig (class in trinity.common.config) BufferReader (class in trinity.buffer) (class in trinity.buffer.buffer_reader) BufferWriter (class in trinity.buffer) (class in trinity.buffer.buffer_writer) bundle_indices (trinity.common.config.InferenceModelConfig attribute)

C

cache_dir (trinity.common.config.BufferConfig attribute) (trinity.common.config.MonitorConfig attribute) CacheManager (class in trinity.manager) (class in trinity.manager.manager) calculate_group_advantage() (trinity.algorithm.advantage_fn.advantage_fn.GroupAdvantage method) (trinity.algorithm.advantage_fn.GroupAdvantage method) (trinity.algorithm.advantage_fn.grpo_advantage.GRPOGroupedAdvantage method) (trinity.algorithm.advantage_fn.GRPOGroupedAdvantage method) (trinity.algorithm.advantage_fn.opmd_advantage.OPMDGroupAdvantage method) (trinity.algorithm.advantage_fn.OPMDGroupAdvantage method) calculate_kl() (trinity.algorithm.kl_fn.kl_fn.AbsFn method) (trinity.algorithm.kl_fn.kl_fn.DummyKLFn method) (trinity.algorithm.kl_fn.kl_fn.K1Fn method) (trinity.algorithm.kl_fn.kl_fn.K2Fn method) (trinity.algorithm.kl_fn.kl_fn.K3Fn method) (trinity.algorithm.kl_fn.kl_fn.KLFn method) (trinity.algorithm.kl_fn.KLFn method) (trinity.algorithm.KLFn method) calculate_kl_loss() (trinity.algorithm.kl_fn.kl_fn.DummyKLFn method) (trinity.algorithm.kl_fn.kl_fn.KLFn method) (trinity.algorithm.kl_fn.KLFn method) (trinity.algorithm.KLFn method) calculate_last_step_advantage() (trinity.algorithm.advantage_fn.StepWiseGRPOAdvantageFn method) calculate_metrics() (trinity.utils.monitor.Monitor method) can_balance_batch (trinity.algorithm.algorithm.AlgorithmType attribute) (trinity.algorithm.algorithm.DPOAlgorithm attribute) (trinity.algorithm.algorithm.GRPOAlgorithm attribute) (trinity.algorithm.algorithm.MIXAlgorithm attribute) (trinity.algorithm.algorithm.MIXCHORDAlgorithm attribute) (trinity.algorithm.algorithm.OPMDAlgorithm attribute) (trinity.algorithm.algorithm.PPOAlgorithm attribute) (trinity.algorithm.algorithm.RAFTAlgorithm attribute) (trinity.algorithm.algorithm.SFTAlgorithm attribute) (trinity.algorithm.AlgorithmType attribute) capacity (trinity.buffer.queue.AsyncPriorityQueue attribute) (trinity.common.config.StorageConfig attribute) CaseInsensitiveEnum (class in trinity.common.constants) CaseInsensitiveEnumMeta (class in trinity.common.constants) chat() (trinity.common.models.model.InferenceModel method) (trinity.common.models.model.ModelWrapper method) (trinity.common.models.vllm_model.vLLMRolloutModel method) chat_async() (trinity.common.models.model.ModelWrapper method) chat_template (trinity.common.config.FormatConfig attribute) (trinity.common.config.InferenceModelConfig attribute) CHATPAIR (trinity.common.constants.PromptType attribute) check_and_update() (trinity.common.config.Config method) check_auxiliary_models() (in module trinity.manager.config_registry.explorer_config_manager) check_checkpoint_root_dir() (in module trinity.manager.config_registry.model_config_manager) check_config() (trinity.algorithm.algorithm.AlgorithmType class method) (trinity.algorithm.algorithm.DPOAlgorithm class method) (trinity.algorithm.AlgorithmType class method) check_experience_buffer_path() (in module trinity.manager.config_registry.buffer_config_manager) check_model_path() (in module trinity.manager.config_registry.model_config_manager) check_resume_from_path() (in module trinity.manager.config_registry.trainer_config_manager) check_sft_warmup_dataset_path() (in module trinity.manager.config_registry.buffer_config_manager) check_taskset_path() (in module trinity.manager.config_registry.buffer_config_manager) check_tensor_parallel_size() (in module trinity.manager.config_registry.explorer_config_manager) check_train_batch_size() (in module trinity.manager.config_registry.buffer_config_manager) Checkpoint (class in trinity.common.verl_config) CHECKPOINT (trinity.common.constants.SyncMethod attribute) checkpoint (trinity.common.verl_config.Actor attribute) (trinity.common.verl_config.Critic attribute) (trinity.common.verl_config.Ref attribute) checkpoint_job_dir (trinity.common.config.Config attribute) checkpoint_root_dir (trinity.common.config.Config attribute) checkpoint_shard_counter (trinity.manager.Synchronizer attribute) chosen (trinity.buffer.schema.sql_schema.DPODataModel attribute) (trinity.common.experience.Experience attribute) chosen_key (trinity.common.config.FormatConfig attribute)

chosen_text (trinity.common.experience.Experience attribute) clear_optimizer_state() (trinity.trainer.verl.fsdp_workers.ActorRolloutRefWorker method) (trinity.trainer.verl.fsdp_workers.CriticWorker method) clip_ratio (trinity.common.verl_config.Actor attribute) cliprange_value (trinity.common.verl_config.Critic attribute) close() (trinity.buffer.queue.AsyncPriorityQueue method) (trinity.buffer.queue.AsyncQueue method) (trinity.buffer.queue.QueueBuffer method) (trinity.buffer.reader.file_reader.DummyProgressBar method) (trinity.utils.monitor.MlflowMonitor method) (trinity.utils.monitor.Monitor method) (trinity.utils.monitor.TensorboardMonitor method) (trinity.utils.monitor.WandbMonitor method) cluster (trinity.common.config.Config attribute) ClusterConfig (class in trinity.common.config) ClusterStatus (class in trinity.utils.dlc_utils) compute_advantage_in_trainer (trinity.algorithm.algorithm.AlgorithmType attribute) (trinity.algorithm.algorithm.DPOAlgorithm attribute) (trinity.algorithm.algorithm.GRPOAlgorithm attribute) (trinity.algorithm.algorithm.MIXAlgorithm attribute) (trinity.algorithm.algorithm.MIXCHORDAlgorithm attribute) (trinity.algorithm.algorithm.OPMDAlgorithm attribute) (trinity.algorithm.algorithm.PPOAlgorithm attribute) (trinity.algorithm.algorithm.RAFTAlgorithm attribute) (trinity.algorithm.algorithm.SFTAlgorithm attribute) (trinity.algorithm.AlgorithmType attribute) compute_in_trainer() (trinity.algorithm.advantage_fn.advantage_fn.AdvantageFn class method) (trinity.algorithm.advantage_fn.advantage_fn.GroupAdvantage class method) (trinity.algorithm.advantage_fn.AdvantageFn class method) (trinity.algorithm.advantage_fn.GroupAdvantage class method) (trinity.algorithm.advantage_fn.StepWiseGRPOAdvantageFn class method) (trinity.algorithm.AdvantageFn class method) compute_log_prob() (trinity.trainer.verl.fsdp_workers.ActorRolloutRefWorker method) compute_overlong_penalty() (trinity.common.rewards.MathDAPORewardFn method) compute_ref_log_prob() (trinity.trainer.verl.fsdp_workers.ActorRolloutRefWorker method) compute_score() (in module trinity.utils.eval_utils) compute_values() (trinity.trainer.verl.fsdp_workers.CriticWorker method) Config (class in trinity.common.config) config_path (trinity.common.config.TaskPipelineConfig attribute) ConfigManager (class in trinity.manager.config_manager) ConfigRegistry (class in trinity.manager.config_registry.config_registry) ConstantMeta (class in trinity.algorithm.algorithm) construct_sft_data() (trinity.common.workflows.RAFTReflectAlfworldWorkflow method) consumed (trinity.buffer.schema.sql_schema.DPODataModel attribute) (trinity.buffer.schema.sql_schema.ExperienceModel attribute) (trinity.buffer.schema.sql_schema.SFTDataModel attribute) continue_from_checkpoint (trinity.common.config.Config attribute) convert_api_output_to_experience() (in module trinity.common.models.model) convert_messages_to_experience() (trinity.common.models.model.InferenceModel method) (trinity.common.models.model.ModelWrapper method) (trinity.common.models.vllm_model.vLLMRolloutModel method) convert_messages_to_experience_async() (trinity.common.models.model.ModelWrapper method) CountDownRewardFn (class in trinity.common.rewards) (class in trinity.common.rewards.countdown_reward) create_dynamic_table() (in module trinity.buffer.schema) (in module trinity.buffer.schema.sql_schema) create_environment() (trinity.common.workflows.RAFTAlfworldWorkflow method) create_inference_models() (in module trinity.common.models) Critic (class in trinity.common.verl_config) critic (trinity.common.verl_config.veRLConfig attribute) critic_model_path (trinity.common.config.ModelConfig attribute) critic_warmup (trinity.common.verl_config.Trainer attribute) CriticModel (class in trinity.common.verl_config) CriticWorker (class in trinity.trainer.verl.fsdp_workers) custom_chat_template (trinity.common.config.ModelConfig attribute) (trinity.common.verl_config.ActorModel attribute) custom_fields (trinity.common.experience.Experiences attribute) custom_reward_function (trinity.common.verl_config.veRLConfig attribute) CustomField (class in trinity.common.experience) CustomRewardFunction (class in trinity.common.verl_config)

D

Data (class in trinity.common.verl_config) data (trinity.common.verl_config.veRLConfig attribute) data_juicer (trinity.common.config.ServiceConfig attribute) data_processor (trinity.common.config.Config attribute) data_processor_url (trinity.common.config.DataProcessorConfig attribute) data_type (trinity.common.experience.CustomField attribute) DataJuicerServiceConfig (class in trinity.common.config) DataParallelPPOActor (class in trinity.trainer.verl.dp_actor) DataProcessorConfig (class in trinity.common.config) DBWrapper (class in trinity.buffer.ray_wrapper) DEFAULT (trinity.common.constants.ReadStrategy attribute) DEFAULT_ANSWER_PARSER() (trinity.common.rewards.math_reward.MathRewardFn method) (trinity.common.rewards.MathRewardFn method) default_args() (trinity.algorithm.advantage_fn.advantage_fn.AdvantageFn class method) (trinity.algorithm.advantage_fn.AdvantageFn class method) (trinity.algorithm.advantage_fn.grpo_advantage.GRPOAdvantageFn class method) (trinity.algorithm.advantage_fn.grpo_advantage.GRPOGroupedAdvantage class method) (trinity.algorithm.advantage_fn.GRPOAdvantageFn class method) (trinity.algorithm.advantage_fn.GRPOGroupedAdvantage class method) (trinity.algorithm.advantage_fn.opmd_advantage.OPMDAdvantageFn class method) (trinity.algorithm.advantage_fn.opmd_advantage.OPMDGroupAdvantage class method) (trinity.algorithm.advantage_fn.OPMDAdvantageFn class method) (trinity.algorithm.advantage_fn.OPMDGroupAdvantage class method) (trinity.algorithm.advantage_fn.ppo_advantage.PPOAdvantageFn class method) (trinity.algorithm.advantage_fn.PPOAdvantageFn class method) (trinity.algorithm.advantage_fn.reinforce_plus_plus_advantage.REINFORCEPLUSPLUSAdvantageFn class method) (trinity.algorithm.advantage_fn.REINFORCEPLUSPLUSAdvantageFn class method) (trinity.algorithm.advantage_fn.remax_advantage.REMAXAdvantageFn class method) (trinity.algorithm.advantage_fn.REMAXAdvantageFn class method) (trinity.algorithm.advantage_fn.rloo_advantage.RLOOAdvantageFn class method) (trinity.algorithm.advantage_fn.RLOOAdvantageFn class method) (trinity.algorithm.advantage_fn.StepWiseGRPOAdvantageFn class method) (trinity.algorithm.AdvantageFn class method) (trinity.algorithm.entropy_loss_fn.entropy_loss_fn.EntropyLossFn class method) (trinity.algorithm.entropy_loss_fn.EntropyLossFn class method) (trinity.algorithm.EntropyLossFn class method) (trinity.algorithm.kl_fn.kl_fn.KLFn class method) (trinity.algorithm.kl_fn.KLFn class method) (trinity.algorithm.KLFn class method) (trinity.algorithm.policy_loss_fn.dpo_loss.DPOLossFn class method) (trinity.algorithm.policy_loss_fn.DPOLossFn class method) (trinity.algorithm.policy_loss_fn.GSPOLossFn class method) (trinity.algorithm.policy_loss_fn.mix_policy_loss.MIXPolicyLossFn class method) (trinity.algorithm.policy_loss_fn.MIXCHORDPolicyLossFn class method) (trinity.algorithm.policy_loss_fn.MIXPolicyLossFn class method) (trinity.algorithm.policy_loss_fn.opmd_policy_loss.OPMDPolicyLossFn class method) (trinity.algorithm.policy_loss_fn.OPMDPolicyLossFn class method) (trinity.algorithm.policy_loss_fn.policy_loss_fn.PolicyLossFn class method) (trinity.algorithm.policy_loss_fn.PolicyLossFn class method) (trinity.algorithm.policy_loss_fn.ppo_policy_loss.PPOPolicyLossFn class method) (trinity.algorithm.policy_loss_fn.PPOPolicyLossFn class method) (trinity.algorithm.policy_loss_fn.sft_loss.SFTLossFn class method) (trinity.algorithm.policy_loss_fn.SFTISLossFn class method) (trinity.algorithm.policy_loss_fn.SFTLossFn class method) (trinity.algorithm.policy_loss_fn.SFTPhiLossFn class method) (trinity.algorithm.PolicyLossFn class method) (trinity.algorithm.sample_strategy.DefaultSampleStrategy class method) (trinity.algorithm.sample_strategy.mix_sample_strategy.MixSampleStrategy class method) (trinity.algorithm.sample_strategy.MixSampleStrategy class method) (trinity.algorithm.sample_strategy.sample_strategy.DefaultSampleStrategy class method) (trinity.algorithm.sample_strategy.sample_strategy.SampleStrategy class method) (trinity.algorithm.sample_strategy.sample_strategy.WarmupSampleStrategy class method) (trinity.algorithm.sample_strategy.SampleStrategy class method) (trinity.algorithm.sample_strategy.WarmupSampleStrategy class method) (trinity.algorithm.SampleStrategy class method) (trinity.utils.monitor.MlflowMonitor class method) (trinity.utils.monitor.Monitor class method) (trinity.utils.monitor.WandbMonitor class method)

default_config (trinity.manager.config_registry.config_registry.ConfigRegistry property) default_config() (trinity.algorithm.algorithm.AlgorithmType class method) (trinity.algorithm.algorithm.DPOAlgorithm class method) (trinity.algorithm.algorithm.GRPOAlgorithm class method) (trinity.algorithm.algorithm.MIXAlgorithm class method) (trinity.algorithm.algorithm.MIXCHORDAlgorithm class method) (trinity.algorithm.algorithm.OPMDAlgorithm class method) (trinity.algorithm.algorithm.PPOAlgorithm class method) (trinity.algorithm.algorithm.RAFTAlgorithm class method) (trinity.algorithm.algorithm.SFTAlgorithm class method) (trinity.algorithm.AlgorithmType class method) default_eval_workflow_type (trinity.common.config.ExplorerInput attribute) (trinity.common.config.StorageConfig attribute) DEFAULT_FORMAT_PATTERN (trinity.common.rewards.math_reward.MathRewardFn attribute) (trinity.common.rewards.MathRewardFn attribute) default_hdfs_dir (trinity.common.verl_config.Trainer attribute) default_local_dir (trinity.common.verl_config.Trainer attribute) default_reward_fn_type (trinity.common.config.ExplorerInput attribute) (trinity.common.config.StorageConfig attribute) default_storage_path() (in module trinity.buffer.utils) default_workflow_type (trinity.common.config.ExplorerInput attribute) (trinity.common.config.StorageConfig attribute) DefaultEntropyLossFn (class in trinity.algorithm.entropy_loss_fn.entropy_loss_fn) DefaultSampleStrategy (class in trinity.algorithm.sample_strategy) (class in trinity.algorithm.sample_strategy.sample_strategy) del_local_ckpt_after_load (trinity.common.verl_config.Trainer attribute) deserialize() (trinity.common.experience.Experience class method) destination_field (trinity.common.experience.CustomField attribute) device (trinity.common.verl_config.Trainer attribute) DIV (trinity.common.constants.OpType attribute) DPOAlgorithm (class in trinity.algorithm.algorithm) DPODataModel (class in trinity.buffer.schema.sql_schema) DPODataReader (class in trinity.buffer.reader.file_reader) DPOLossFn (class in trinity.algorithm.policy_loss_fn) (class in trinity.algorithm.policy_loss_fn.dpo_loss) dtype (trinity.common.config.InferenceModelConfig attribute) DummyEntropyLossFn (class in trinity.algorithm.entropy_loss_fn.entropy_loss_fn) DummyKLFn (class in trinity.algorithm.kl_fn.kl_fn) DummyProgressBar (class in trinity.buffer.reader.file_reader) DYNAMIC_BY_EXPLORER (trinity.common.constants.SyncStyle attribute) DYNAMIC_BY_TRAINER (trinity.common.constants.SyncStyle attribute)

E

EID (class in trinity.common.experience) eid (trinity.common.experience.Experience attribute) eids (trinity.common.experience.Experiences attribute) empty_experiences() (in module trinity.common.experience) enable (trinity.common.verl_config.RewardModel attribute) enable_auto_tool_choice (trinity.common.config.InferenceModelConfig attribute) enable_chunked_prefill (trinity.common.config.InferenceModelConfig attribute) enable_exp_buffer (trinity.common.verl_config.Trainer attribute) enable_gradient_checkpointing (trinity.common.verl_config.ActorModel attribute) (trinity.common.verl_config.CriticModel attribute) enable_history (trinity.common.config.InferenceModelConfig attribute) enable_openai_api (trinity.common.config.InferenceModelConfig attribute) enable_prefix_caching (trinity.common.config.InferenceModelConfig attribute) enable_preview (trinity.common.config.TrainerConfig attribute) (trinity.common.verl_config.veRLConfig attribute) enable_progress_bar (trinity.common.config.StorageConfig attribute) enable_ray_timeline (trinity.common.config.MonitorConfig attribute) enable_thinking (trinity.common.config.InferenceModelConfig attribute) enforce_eager (trinity.common.config.InferenceModelConfig attribute) engine_num (trinity.common.config.InferenceModelConfig attribute) engine_type (trinity.common.config.InferenceModelConfig attribute) entropy_checkpointing (trinity.common.verl_config.Actor attribute) (trinity.common.verl_config.Ref attribute) entropy_coeff (trinity.common.verl_config.Actor attribute) entropy_from_logits_with_chunking (trinity.common.verl_config.Actor attribute) (trinity.common.verl_config.Ref attribute) entropy_loss_fn (trinity.common.config.AlgorithmConfig attribute) entropy_loss_fn_args (trinity.common.config.AlgorithmConfig attribute) EntropyLossFn (class in trinity.algorithm) (class in trinity.algorithm.entropy_loss_fn) (class in trinity.algorithm.entropy_loss_fn.entropy_loss_fn) env_vars (trinity.common.config.ExplorerConfig attribute) EVAL (trinity.common.constants.TaskType attribute) eval() (trinity.explorer.Explorer method) (trinity.explorer.explorer.Explorer method)

eval_alfworld() (trinity.common.workflows.RAFTAlfworldWorkflow method) eval_interval (trinity.common.config.ExplorerConfig attribute) eval_on_startup (trinity.common.config.ExplorerConfig attribute) eval_tasksets (trinity.common.config.ExplorerInput attribute) evaluate_equation() (in module trinity.utils.eval_utils) Experience (class in trinity.common.experience) experience_buffer (trinity.common.config.TrainerInput attribute) experience_pipeline (trinity.common.config.DataProcessorConfig attribute) ExperienceModel (class in trinity.buffer.schema.sql_schema) ExperiencePipelineConfig (class in trinity.common.config) Experiences (class in trinity.common.experience) experiment_name (trinity.common.verl_config.Trainer attribute) expert_mode() (trinity.manager.config_manager.ConfigManager method) EXPLORE (trinity.common.constants.TaskType attribute) explore() (trinity.explorer.Explorer method) (trinity.explorer.explorer.Explorer method) explore_step() (trinity.explorer.Explorer method) (trinity.explorer.explorer.Explorer method) Explorer (class in trinity.explorer) (class in trinity.explorer.explorer) explorer (trinity.common.config.Config attribute) explorer_input (trinity.common.config.BufferConfig attribute) explorer_name (trinity.common.verl_config.ActorRolloutRef attribute) explorer_output (trinity.common.config.BufferConfig attribute) explorer_status_counts (trinity.manager.Synchronizer attribute) explorer_visible() (in module trinity.manager.config_registry.explorer_config_manager) explorer_world_size (trinity.common.config.SynchronizerConfig attribute) ExplorerConfig (class in trinity.common.config) ExplorerInput (class in trinity.common.config) external_lib (trinity.common.verl_config.ActorModel attribute) (trinity.common.verl_config.CriticModel attribute) extract_experience_from_history() (trinity.common.models.model.ModelWrapper method) extract_logprobs() (in module trinity.common.models.model) extract_solution() (in module trinity.utils.eval_utils)

F

FIFO (trinity.common.constants.ReadStrategy attribute) FILE (trinity.common.constants.StorageType attribute) FileWrapper (class in trinity.buffer.ray_wrapper) find_boxed_answer() (in module trinity.utils.eval_utils) finish() (trinity.utils.dlc_utils.ClusterStatus method) FIXED (trinity.common.constants.SyncStyle attribute) flatten() (trinity.common.config.Config method) format (trinity.common.config.StorageConfig attribute) format() (trinity.utils.log.NewLineFormatter method) format_args (trinity.common.workflows.Task attribute) (trinity.common.workflows.workflow.Task attribute) format_messages() (trinity.common.workflows.MathEvalWorkflow method) (trinity.common.workflows.SimpleWorkflow method) (trinity.common.workflows.workflow.SimpleWorkflow method) format_prompt() (trinity.common.workflows.customized_math_workflows.MathBoxedWorkflow method) (trinity.common.workflows.MathBoxedWorkflow method) (trinity.common.workflows.ToolCallWorkflow method) FormatConfig (class in trinity.common.config)

FormatReward (class in trinity.common.rewards) (class in trinity.common.rewards.format_reward) forward_max_token_len_per_gpu (trinity.common.verl_config.Critic attribute) (trinity.common.verl_config.RewardModel attribute) forward_micro_batch_size (trinity.common.verl_config.Critic attribute) forward_micro_batch_size_per_gpu (trinity.common.verl_config.Critic attribute) forward_prefetch (trinity.common.verl_config.FSDPConfig attribute) from_experience() (trinity.buffer.schema.sql_schema.ExperienceModel class method) from_hf_datasets() (in module trinity.common.experience) from_messages() (trinity.buffer.schema.sql_schema.SFTDataModel class method) from_trinity() (trinity.algorithm.key_mapper.KeyMapper method) fsdp_config (trinity.common.verl_config.Actor attribute) (trinity.common.verl_config.CriticModel attribute) (trinity.common.verl_config.Ref attribute) fsdp_size (trinity.common.verl_config.FSDPConfig attribute) FSDPConfig (class in trinity.common.verl_config) fused_kernel_options (trinity.common.verl_config.ActorModel attribute) FusedKernelOptions (class in trinity.common.verl_config)

G

gamma (trinity.common.verl_config.Algorithm attribute) gather() (trinity.common.experience.Experience class method) gather_action_masks() (in module trinity.common.experience) gather_advantages() (in module trinity.common.experience) gather_attention_masks() (in module trinity.common.experience) gather_experiences() (trinity.common.experience.Experiences class method) gather_logprobs() (in module trinity.common.experience) gather_metrics() (in module trinity.utils.monitor) gather_returns() (in module trinity.common.experience) gather_token_ids() (in module trinity.common.experience) generate() (trinity.common.models.model.InferenceModel method) (trinity.common.models.model.ModelWrapper method) (trinity.common.models.vllm_model.vLLMRolloutModel method) generate_async() (trinity.common.models.model.ModelWrapper method) generate_config() (trinity.manager.config_manager.ConfigManager method) generate_env_inference_samples() (trinity.common.workflows.AlfworldWorkflow method) (trinity.common.workflows.SciWorldWorkflow method) (trinity.common.workflows.WebShopWorkflow method) GenerationConfig (class in trinity.common.config) get() (trinity.buffer.queue.AsyncPriorityQueue method) (trinity.buffer.queue.QueueBuffer method) (trinity.manager.config_registry.config_registry.ConfigRegistry method) (trinity.utils.registry.Registry method) get_actor() (trinity.explorer.Explorer class method) (trinity.explorer.explorer.Explorer class method) (trinity.manager.Synchronizer class method) (trinity.trainer.Trainer class method) (trinity.trainer.trainer.Trainer class method) get_available_address() (trinity.common.models.model.InferenceModel method) get_available_port() (in module trinity.utils.distributed) get_batch() (trinity.buffer.ray_wrapper.QueueWrapper method) get_buffer_reader() (in module trinity.buffer) (in module trinity.buffer.buffer) get_buffer_writer() (in module trinity.buffer) (in module trinity.buffer.buffer) get_check_func() (trinity.manager.config_registry.config_registry.ConfigRegistry method) get_checkpoint_dir_with_step_num() (in module trinity.common.models.utils) get_configs() (trinity.manager.config_manager.ConfigManager method) (trinity.manager.config_registry.config_registry.ConfigRegistry method) get_current_algorithm_config() (trinity.algorithm.algorithm_manager.AlgorithmManager method) get_dlc_env_vars() (in module trinity.utils.dlc_utils) get_explorer_status_counts() (trinity.manager.Synchronizer method) get_logger() (in module trinity.utils.log)

get_model_response() (trinity.common.workflows.AlfworldWorkflow method) (trinity.common.workflows.SciWorldWorkflow method) (trinity.common.workflows.WebShopWorkflow method) get_model_response_text() (trinity.common.workflows.AlfworldWorkflow method) (trinity.common.workflows.SciWorldWorkflow method) (trinity.common.workflows.WebShopWorkflow method) get_model_state_dict() (trinity.manager.Synchronizer method) get_model_version() (trinity.common.models.model.InferenceModel method) (trinity.common.models.vllm_model.vLLMRolloutModel method) get_openai_client() (trinity.common.models.model.ModelWrapper method) get_queue() (trinity.buffer.queue.QueueBuffer class method) get_results() (trinity.explorer.scheduler.Scheduler method) get_state_dict_meta() (trinity.manager.Synchronizer method) get_trainer_status() (trinity.manager.Synchronizer method) get_trainer_wrapper() (in module trinity.trainer) (in module trinity.trainer.trainer) get_verl_checkpoint_info() (in module trinity.common.models.utils) get_weight() (trinity.explorer.Explorer method) (trinity.explorer.explorer.Explorer method) get_wrapper() (trinity.buffer.ray_wrapper.DBWrapper class method) (trinity.buffer.ray_wrapper.FileWrapper class method) (trinity.buffer.ray_wrapper.QueueWrapper class method) gpu_memory_utilization (trinity.common.config.InferenceModelConfig attribute) gpu_per_node (trinity.common.config.ClusterConfig attribute) grad_clip (trinity.common.verl_config.Actor attribute) (trinity.common.verl_config.Critic attribute) group (trinity.common.config.Config attribute) group_by() (in module trinity.common.experience) group_experiences() (trinity.algorithm.advantage_fn.advantage_fn.GroupAdvantage method) (trinity.algorithm.advantage_fn.GroupAdvantage method) (trinity.algorithm.advantage_fn.grpo_advantage.GRPOGroupedAdvantage method) (trinity.algorithm.advantage_fn.GRPOGroupedAdvantage method) (trinity.algorithm.advantage_fn.opmd_advantage.OPMDGroupAdvantage method) (trinity.algorithm.advantage_fn.OPMDGroupAdvantage method) group_name (trinity.common.verl_config.Trainer attribute) GroupAdvantage (class in trinity.algorithm.advantage_fn) (class in trinity.algorithm.advantage_fn.advantage_fn) GRPOAdvantageFn (class in trinity.algorithm.advantage_fn) (class in trinity.algorithm.advantage_fn.grpo_advantage) GRPOAlgorithm (class in trinity.algorithm.algorithm) GRPOGroupedAdvantage (class in trinity.algorithm.advantage_fn) (class in trinity.algorithm.advantage_fn.grpo_advantage) GSPOLossFn (class in trinity.algorithm.policy_loss_fn)

H

has_api_server() (trinity.common.models.vllm_model.vLLMRolloutModel method) has_step() (trinity.explorer.scheduler.Scheduler method)

horizon (trinity.common.verl_config.KL_Ctrl attribute) hybrid_engine (trinity.common.verl_config.ActorRolloutRef attribute)

I

id (trinity.buffer.schema.sql_schema.DPODataModel attribute) (trinity.buffer.schema.sql_schema.ExperienceModel attribute) (trinity.buffer.schema.sql_schema.SFTDataModel attribute) (trinity.buffer.schema.sql_schema.TaskModel attribute) ignore_eos (trinity.common.config.InferenceModelConfig attribute) ignore_keys (trinity.algorithm.policy_loss_fn.policy_loss_fn.PolicyLossFnMeta attribute) impl_backend (trinity.common.verl_config.FusedKernelOptions attribute) index (trinity.common.config.StorageConfig attribute) InferenceModel (class in trinity.common.models.model) InferenceModelConfig (class in trinity.common.config) info (trinity.common.experience.Experience attribute) init_model() (trinity.trainer.verl.fsdp_workers.ActorRolloutRefWorker method) (trinity.trainer.verl.fsdp_workers.CriticWorker method) init_process_group() (in module trinity.utils.distributed) (trinity.common.models.vllm_model.vLLMRolloutModel method) (trinity.common.models.vllm_worker.WorkerExtension method)

init_workers() (trinity.trainer.verl_trainer.VerlPPOTrainerWrapper method) input_save_path (trinity.common.config.ExperiencePipelineConfig attribute) inputs (trinity.common.config.ExperiencePipelineConfig attribute) (trinity.common.config.TaskPipelineConfig attribute) is_alive() (trinity.explorer.Explorer method) (trinity.explorer.explorer.Explorer method) (trinity.explorer.workflow_runner.WorkflowRunner method) (trinity.trainer.Trainer method) (trinity.trainer.trainer.Trainer method) is_database_url() (in module trinity.buffer.ray_wrapper) is_equiv() (in module trinity.utils.eval_utils) is_eval (trinity.common.workflows.Task attribute) (trinity.common.workflows.workflow.Task attribute) is_ipv6_address() (in module trinity.utils.distributed) is_json_file() (in module trinity.buffer.ray_wrapper) is_port_available() (in module trinity.utils.distributed) is_running() (in module trinity.utils.dlc_utils)

J

JSONWriter (class in trinity.buffer.writer.file_writer)

K

K1Fn (class in trinity.algorithm.kl_fn.kl_fn) K2Fn (class in trinity.algorithm.kl_fn.kl_fn) K3Fn (class in trinity.algorithm.kl_fn.kl_fn) KeyMapper (class in trinity.algorithm.key_mapper) kl_coef (trinity.common.verl_config.KL_Ctrl attribute) KL_Ctrl (class in trinity.common.verl_config) kl_ctrl (trinity.common.verl_config.Algorithm attribute) kl_loss_coef (trinity.common.verl_config.Actor attribute)

kl_loss_fn (trinity.common.config.AlgorithmConfig attribute) kl_loss_fn_args (trinity.common.config.AlgorithmConfig attribute) kl_loss_type (trinity.common.verl_config.Actor attribute) kl_penalty (trinity.common.verl_config.Algorithm attribute) kl_penalty_fn (trinity.common.config.AlgorithmConfig attribute) kl_penalty_fn_args (trinity.common.config.AlgorithmConfig attribute) KLFn (class in trinity.algorithm) (class in trinity.algorithm.kl_fn) (class in trinity.algorithm.kl_fn.kl_fn)

L

label_key (trinity.common.config.FormatConfig attribute) lam (trinity.common.verl_config.Algorithm attribute) last_boxed_only_string() (in module trinity.utils.eval_utils) length() (trinity.buffer.ray_wrapper.QueueWrapper method) LFU (trinity.common.constants.ReadStrategy attribute) linear_decay_priority() (in module trinity.buffer.queue) list() (trinity.utils.registry.Registry method) load_checkpoint() (trinity.trainer.verl.fsdp_workers.ActorRolloutRefWorker method) (trinity.trainer.verl.fsdp_workers.CriticWorker method) load_config() (in module trinity.common.config) (in module trinity.common.verl_config) load_contents (trinity.common.verl_config.Checkpoint attribute) load_explorer() (trinity.manager.CacheManager method) (trinity.manager.manager.CacheManager method) load_from_file() (in module trinity.utils.plugin_loader) load_plugin_from_dirs() (in module trinity.utils.plugin_loader) load_plugins() (in module trinity.utils.plugin_loader) load_state_dict() (in module trinity.common.models.utils) load_state_dict_from_verl_checkpoint() (in module trinity.common.models.utils) load_trainer() (trinity.manager.CacheManager method) (trinity.manager.manager.CacheManager method) log() (trinity.utils.monitor.MlflowMonitor method) (trinity.utils.monitor.Monitor method) (trinity.utils.monitor.TensorboardMonitor method) (trinity.utils.monitor.WandbMonitor method)

log_prob_max_token_len_per_gpu (trinity.common.verl_config.Ref attribute) log_prob_micro_batch_size (trinity.common.verl_config.Ref attribute) (trinity.common.verl_config.Rollout attribute) log_prob_micro_batch_size_per_gpu (trinity.common.verl_config.Ref attribute) (trinity.common.verl_config.Rollout attribute) log_prob_use_dynamic_bsz (trinity.common.verl_config.Ref attribute) log_table() (trinity.utils.monitor.MlflowMonitor method) (trinity.utils.monitor.Monitor method) (trinity.utils.monitor.TensorboardMonitor method) (trinity.utils.monitor.WandbMonitor method) logger (trinity.common.verl_config.Trainer attribute) logprobs (trinity.common.config.GenerationConfig attribute) (trinity.common.experience.Experience attribute) (trinity.common.experience.Experiences attribute) logprobs() (trinity.common.models.model.InferenceModel method) (trinity.common.models.model.ModelWrapper method) (trinity.common.models.vllm_model.vLLMRolloutModel method) logprobs_async() (trinity.common.models.model.ModelWrapper method) loss_agg_mode (trinity.common.verl_config.Actor attribute) (trinity.common.verl_config.Critic attribute) lr (trinity.common.verl_config.Optim attribute) lr_warmup_steps (trinity.common.verl_config.Optim attribute) lr_warmup_steps_ratio (trinity.common.verl_config.Optim attribute) LRU (trinity.common.constants.ReadStrategy attribute)

M

maintain_session_state() (trinity.manager.config_manager.ConfigManager method) masked_mean() (in module trinity.algorithm.utils) masked_sum() (in module trinity.algorithm.utils) masked_var() (in module trinity.algorithm.utils) masked_whiten() (in module trinity.algorithm.utils) MathBoxedRewardFn (class in trinity.common.rewards) (class in trinity.common.rewards.math_reward) MathBoxedWorkflow (class in trinity.common.workflows) (class in trinity.common.workflows.customized_math_workflows) MathDAPORewardFn (class in trinity.common.rewards) MathEvalWorkflow (class in trinity.common.workflows) MathRewardFn (class in trinity.common.rewards) (class in trinity.common.rewards.math_reward) MathRMWorkflow (class in trinity.common.workflows) (class in trinity.common.workflows.math_rm_workflow) MathWorkflow (class in trinity.common.workflows) (class in trinity.common.workflows.workflow) max_actor_ckpt_to_keep (trinity.common.verl_config.Trainer attribute) max_critic_ckpt_to_keep (trinity.common.verl_config.Trainer attribute) max_length (trinity.common.verl_config.RewardModel attribute) max_model_len (trinity.common.config.InferenceModelConfig attribute) (trinity.common.config.ModelConfig attribute) max_prompt_tokens (trinity.common.config.InferenceModelConfig attribute) (trinity.common.config.ModelConfig attribute) max_read_timeout (trinity.common.config.StorageConfig attribute) max_repeat_times_per_runner (trinity.common.config.ExplorerConfig attribute) max_response_tokens (trinity.common.config.InferenceModelConfig attribute) (trinity.common.config.ModelConfig attribute) max_retry_interval (trinity.common.config.BufferConfig attribute) max_retry_times (trinity.common.config.BufferConfig attribute) (trinity.common.config.ExplorerConfig attribute) max_step_num (trinity.common.workflows.StepWiseAlfworldWorkflow property) max_timeout (trinity.common.config.ExplorerConfig attribute) MEMORY (trinity.common.constants.SyncMethod attribute) merge_by_placement() (in module trinity.common.models.utils) message (trinity.explorer.workflow_runner.Status attribute) messages (trinity.buffer.schema.sql_schema.SFTDataModel attribute) MESSAGES (trinity.common.constants.PromptType attribute) messages (trinity.common.experience.Experience attribute) messages_key (trinity.common.config.FormatConfig attribute) metadata (trinity.buffer.schema.Base attribute) metric (trinity.explorer.workflow_runner.Status attribute) metrics (trinity.common.experience.Experience attribute) micro_batch_size_per_gpu (trinity.common.verl_config.RewardModel attribute) min_lr_ratio (trinity.common.verl_config.Optim attribute) min_num_params (trinity.common.verl_config.WrapPolicy attribute) min_response_tokens (trinity.common.config.InferenceModelConfig attribute) (trinity.common.config.ModelConfig attribute) MIXAlgorithm (class in trinity.algorithm.algorithm) MIXCHORDAlgorithm (class in trinity.algorithm.algorithm) MIXCHORDPolicyLossFn (class in trinity.algorithm.policy_loss_fn) MixEntropyLossFn (class in trinity.algorithm.entropy_loss_fn.entropy_loss_fn) MIXPolicyLossFn (class in trinity.algorithm.policy_loss_fn) (class in trinity.algorithm.policy_loss_fn.mix_policy_loss) MixSampleStrategy (class in trinity.algorithm.sample_strategy) (class in trinity.algorithm.sample_strategy.mix_sample_strategy) MlflowMonitor (class in trinity.utils.monitor) mode (trinity.common.config.Config attribute) model (trinity.common.config.Config attribute) (trinity.common.verl_config.ActorRolloutRef attribute) (trinity.common.verl_config.Critic attribute) (trinity.common.verl_config.RewardModel attribute) model_path (trinity.common.config.InferenceModelConfig attribute) (trinity.common.config.ModelConfig attribute) model_state_dict (trinity.manager.Synchronizer attribute) model_version (trinity.common.models.model.ModelWrapper property) (trinity.manager.Synchronizer attribute) ModelConfig (class in trinity.common.config) ModelWrapper (class in trinity.common.models.model) module trinity trinity.algorithm trinity.algorithm.advantage_fn trinity.algorithm.advantage_fn.advantage_fn trinity.algorithm.advantage_fn.grpo_advantage trinity.algorithm.advantage_fn.opmd_advantage trinity.algorithm.advantage_fn.ppo_advantage trinity.algorithm.advantage_fn.reinforce_plus_plus_advantage trinity.algorithm.advantage_fn.remax_advantage trinity.algorithm.advantage_fn.rloo_advantage trinity.algorithm.algorithm trinity.algorithm.algorithm_manager trinity.algorithm.entropy_loss_fn trinity.algorithm.entropy_loss_fn.entropy_loss_fn trinity.algorithm.key_mapper trinity.algorithm.kl_fn trinity.algorithm.kl_fn.kl_fn trinity.algorithm.policy_loss_fn trinity.algorithm.policy_loss_fn.dpo_loss trinity.algorithm.policy_loss_fn.mix_policy_loss trinity.algorithm.policy_loss_fn.opmd_policy_loss trinity.algorithm.policy_loss_fn.policy_loss_fn trinity.algorithm.policy_loss_fn.ppo_policy_loss trinity.algorithm.policy_loss_fn.sft_loss trinity.algorithm.sample_strategy trinity.algorithm.sample_strategy.mix_sample_strategy trinity.algorithm.sample_strategy.sample_strategy trinity.algorithm.sample_strategy.utils trinity.algorithm.utils trinity.buffer trinity.buffer.buffer trinity.buffer.buffer_reader trinity.buffer.buffer_writer trinity.buffer.queue trinity.buffer.ray_wrapper trinity.buffer.reader trinity.buffer.reader.file_reader trinity.buffer.reader.queue_reader trinity.buffer.reader.sql_reader trinity.buffer.schema trinity.buffer.schema.sql_schema trinity.buffer.utils trinity.buffer.writer trinity.buffer.writer.file_writer trinity.buffer.writer.queue_writer trinity.buffer.writer.sql_writer trinity.common trinity.common.config trinity.common.constants trinity.common.experience trinity.common.models trinity.common.models.model trinity.common.models.utils trinity.common.models.vllm_model trinity.common.models.vllm_worker trinity.common.rewards trinity.common.rewards.accuracy_reward trinity.common.rewards.agents_reward trinity.common.rewards.countdown_reward trinity.common.rewards.format_reward trinity.common.rewards.human_reward trinity.common.rewards.math_reward trinity.common.rewards.reward_fn trinity.common.rewards.tool_reward trinity.common.rewards.utils trinity.common.verl_config trinity.common.workflows trinity.common.workflows.customized_math_workflows trinity.common.workflows.math_rm_workflow trinity.common.workflows.workflow trinity.explorer trinity.explorer.explorer trinity.explorer.scheduler trinity.explorer.workflow_runner trinity.manager trinity.manager.config_manager trinity.manager.config_registry trinity.manager.config_registry.algorithm_config_manager trinity.manager.config_registry.buffer_config_manager trinity.manager.config_registry.config_registry trinity.manager.config_registry.explorer_config_manager trinity.manager.config_registry.model_config_manager trinity.manager.config_registry.trainer_config_manager trinity.manager.manager trinity.plugins trinity.trainer trinity.trainer.trainer trinity.trainer.verl trinity.trainer.verl.dp_actor trinity.trainer.verl.fsdp_workers trinity.trainer.verl_trainer trinity.utils trinity.utils.distributed trinity.utils.dlc_utils trinity.utils.eval_utils trinity.utils.log trinity.utils.monitor trinity.utils.plugin_loader trinity.utils.registry trinity.utils.timer

modules (trinity.utils.registry.Registry property) Monitor (class in trinity.utils.monitor) monitor (trinity.common.config.Config attribute) monitor_args (trinity.common.config.MonitorConfig attribute) monitor_type (trinity.common.config.MonitorConfig attribute) MonitorConfig (class in trinity.common.config) MonitorType (class in trinity.common.constants) MUL (trinity.common.constants.OpType attribute) multi_turn (trinity.common.verl_config.Rollout attribute) MultiTurnWorkflow (class in trinity.common.workflows.workflow)

N

n (trinity.common.config.GenerationConfig attribute) (trinity.common.verl_config.Rollout attribute) n_gpus_per_node (trinity.common.verl_config.Trainer attribute) name (trinity.common.config.Config attribute) (trinity.common.config.ExplorerConfig attribute) (trinity.common.config.OperatorConfig attribute) (trinity.common.config.StorageConfig attribute) (trinity.common.config.TrainerConfig attribute) (trinity.common.verl_config.CustomRewardFunction attribute) (trinity.utils.registry.Registry property) name() (trinity.algorithm.algorithm.AlgorithmType class method) (trinity.algorithm.AlgorithmType class method)

NCCL (trinity.common.constants.SyncMethod attribute) need_eval() (trinity.explorer.Explorer method) (trinity.explorer.explorer.Explorer method) need_save() (trinity.algorithm.algorithm_manager.AlgorithmManager method) need_sync() (trinity.explorer.Explorer method) (trinity.explorer.explorer.Explorer method) (trinity.trainer.Trainer method) (trinity.trainer.trainer.Trainer method) NewLineFormatter (class in trinity.utils.log) nnodes (trinity.common.verl_config.Trainer attribute) node_num (trinity.common.config.ClusterConfig attribute) norm_adv_by_std_in_grpo (trinity.common.verl_config.Algorithm attribute) num_process (trinity.common.config.TaskPipelineConfig attribute)

O

offload_policy (trinity.common.verl_config.FSDPConfig attribute) ok (trinity.explorer.workflow_runner.Status attribute) OperatorConfig (class in trinity.common.config) operators (trinity.common.config.ExperiencePipelineConfig attribute) (trinity.common.config.TaskPipelineConfig attribute) OPMDAdvantageFn (class in trinity.algorithm.advantage_fn) (class in trinity.algorithm.advantage_fn.opmd_advantage) OPMDAlgorithm (class in trinity.algorithm.algorithm) OPMDGroupAdvantage (class in trinity.algorithm.advantage_fn) (class in trinity.algorithm.advantage_fn.opmd_advantage)

OPMDPolicyLossFn (class in trinity.algorithm.policy_loss_fn) (class in trinity.algorithm.policy_loss_fn.opmd_policy_loss) Optim (class in trinity.common.verl_config) optim (trinity.common.verl_config.Actor attribute) (trinity.common.verl_config.Critic attribute) optimizer_offload (trinity.common.verl_config.FSDPConfig attribute) OpType (class in trinity.common.constants) output (trinity.common.config.ExperiencePipelineConfig attribute) (trinity.common.config.TaskPipelineConfig attribute) override_config (trinity.common.verl_config.ActorModel attribute) (trinity.common.verl_config.CriticModel attribute)

P

pad_token_id (trinity.common.config.BufferConfig attribute) param_offload (trinity.common.verl_config.FSDPConfig attribute) path (trinity.common.config.StorageConfig attribute) (trinity.common.verl_config.ActorModel attribute) (trinity.common.verl_config.CriticModel attribute) (trinity.common.verl_config.CustomRewardFunction attribute) PLAINTEXT (trinity.common.constants.PromptType attribute) policy_loss_fn (trinity.common.config.AlgorithmConfig attribute) policy_loss_fn_args (trinity.common.config.AlgorithmConfig attribute) PolicyLossFn (class in trinity.algorithm) (class in trinity.algorithm.policy_loss_fn) (class in trinity.algorithm.policy_loss_fn.policy_loss_fn) PolicyLossFnMeta (class in trinity.algorithm.policy_loss_fn.policy_loss_fn) port (trinity.common.config.DataJuicerServiceConfig attribute) ppo_epochs (trinity.common.verl_config.Actor attribute) (trinity.common.verl_config.Critic attribute) ppo_max_token_len_per_gpu (trinity.common.verl_config.Actor attribute) (trinity.common.verl_config.Critic attribute) ppo_micro_batch_size (trinity.common.verl_config.Actor attribute) (trinity.common.verl_config.Critic attribute) ppo_micro_batch_size_per_gpu (trinity.common.verl_config.Actor attribute) (trinity.common.verl_config.Critic attribute) ppo_mini_batch_size (trinity.common.verl_config.Actor attribute) (trinity.common.verl_config.Critic attribute) PPOAdvantageFn (class in trinity.algorithm.advantage_fn) (class in trinity.algorithm.advantage_fn.ppo_advantage) PPOAlgorithm (class in trinity.algorithm.algorithm) PPOPolicyLossFn (class in trinity.algorithm.policy_loss_fn) (class in trinity.algorithm.policy_loss_fn.ppo_policy_loss) prefix_metrics() (in module trinity.algorithm.utils)

prepare() (trinity.explorer.Explorer method) (trinity.explorer.explorer.Explorer method) (trinity.trainer.TrainEngineWrapper method) (trinity.trainer.Trainer method) (trinity.trainer.trainer.TrainEngineWrapper method) (trinity.trainer.trainer.Trainer method) (trinity.trainer.verl_trainer.VerlPPOTrainerWrapper method) priority (trinity.buffer.schema.sql_schema.ExperienceModel attribute) PRIORITY (trinity.common.constants.ReadStrategy attribute) priority_fn (trinity.buffer.queue.AsyncPriorityQueue attribute) priority_groups (trinity.buffer.queue.AsyncPriorityQueue attribute) priority_weights (trinity.common.config.TaskPipelineConfig attribute) process() (trinity.algorithm.advantage_fn.advantage_fn.GroupAdvantage method) (trinity.algorithm.advantage_fn.GroupAdvantage method) (trinity.algorithm.advantage_fn.grpo_advantage.GRPOGroupedAdvantage method) (trinity.algorithm.advantage_fn.GRPOGroupedAdvantage method) (trinity.algorithm.advantage_fn.StepWiseGRPOAdvantageFn method) process_messages_to_experience() (trinity.common.workflows.workflow.MultiTurnWorkflow method) project (trinity.common.config.Config attribute) project_name (trinity.common.verl_config.Trainer attribute) prompt (trinity.buffer.schema.sql_schema.ExperienceModel attribute) prompt_key (trinity.common.config.FormatConfig attribute) prompt_length (trinity.common.experience.Experience attribute) (trinity.common.experience.Experiences attribute) prompt_text (trinity.common.experience.Experience attribute) prompt_type (trinity.common.config.FormatConfig attribute) PromptType (class in trinity.common.constants) put() (trinity.buffer.queue.AsyncPriorityQueue method) (trinity.buffer.queue.QueueBuffer method) put_batch() (trinity.buffer.ray_wrapper.QueueWrapper method)

Q

qsize() (trinity.buffer.queue.AsyncPriorityQueue method) (trinity.buffer.queue.QueueBuffer method) QUEUE (trinity.common.constants.StorageType attribute)

QueueBuffer (class in trinity.buffer.queue) QueueReader (class in trinity.buffer.reader.queue_reader) QueueWrapper (class in trinity.buffer.ray_wrapper) QueueWriter (class in trinity.buffer.writer.queue_writer)

R

RAFTAlfworldWorkflow (class in trinity.common.workflows) RAFTAlgorithm (class in trinity.algorithm.algorithm) RAFTReflectAlfworldWorkflow (class in trinity.common.workflows) RANDOM (trinity.common.constants.ReadStrategy attribute) raw (trinity.common.config.StorageConfig attribute) raw_task (trinity.common.workflows.Task attribute) (trinity.common.workflows.workflow.Task attribute) RawDataReader (class in trinity.buffer.reader.file_reader) ray_namespace (trinity.common.config.Config attribute) (trinity.common.config.StorageConfig attribute) (trinity.common.config.SynchronizerConfig attribute) re_explore_with_context() (trinity.common.workflows.RAFTReflectAlfworldWorkflow method) read() (trinity.buffer.buffer_reader.BufferReader method) (trinity.buffer.BufferReader method) (trinity.buffer.ray_wrapper.DBWrapper method) (trinity.buffer.ray_wrapper.FileWrapper method) (trinity.buffer.reader.file_reader.DPODataReader method) (trinity.buffer.reader.file_reader.RawDataReader method) (trinity.buffer.reader.file_reader.RolloutDataReader method) (trinity.buffer.reader.file_reader.SFTDataReader method) (trinity.buffer.reader.queue_reader.QueueReader method) (trinity.buffer.reader.sql_reader.SQLReader method) read_async() (trinity.buffer.buffer_reader.BufferReader method) (trinity.buffer.BufferReader method) (trinity.buffer.reader.file_reader.BaseFileReader method) (trinity.buffer.reader.queue_reader.QueueReader method) (trinity.buffer.reader.sql_reader.SQLReader method) ReadStrategy (class in trinity.common.constants) ready_to_nccl_sync() (trinity.manager.Synchronizer method) reasoning_parser (trinity.common.config.InferenceModelConfig attribute) Ref (class in trinity.common.verl_config) ref (trinity.common.verl_config.ActorRolloutRef attribute) register_check() (trinity.manager.config_registry.config_registry.ConfigRegistry method) register_config() (trinity.manager.config_registry.config_registry.ConfigRegistry method) register_module() (trinity.utils.registry.Registry method) Registry (class in trinity.utils.registry) registry (trinity.buffer.schema.Base attribute) REINFORCEPLUSPLUSAdvantageFn (class in trinity.algorithm.advantage_fn) (class in trinity.algorithm.advantage_fn.reinforce_plus_plus_advantage) rejected (trinity.buffer.schema.sql_schema.DPODataModel attribute) (trinity.common.experience.Experience attribute) rejected_key (trinity.common.config.FormatConfig attribute) rejected_text (trinity.common.experience.Experience attribute) release() (trinity.buffer.buffer_writer.BufferWriter method) (trinity.buffer.BufferWriter method) (trinity.buffer.ray_wrapper.DBWrapper method) (trinity.buffer.ray_wrapper.FileWrapper method) (trinity.buffer.ray_wrapper.QueueWrapper method) (trinity.buffer.writer.file_writer.JSONWriter method) (trinity.buffer.writer.queue_writer.QueueWriter method) (trinity.buffer.writer.sql_writer.SQLWriter method) REMAXAdvantageFn (class in trinity.algorithm.advantage_fn) (class in trinity.algorithm.advantage_fn.remax_advantage) remove_boxed() (in module trinity.utils.eval_utils) remove_previous_ckpt_in_save (trinity.common.verl_config.Trainer attribute) remove_right_units() (in module trinity.utils.eval_utils) repeat_times (trinity.common.config.AlgorithmConfig attribute) (trinity.common.config.StorageConfig attribute) (trinity.common.workflows.Task attribute) (trinity.common.workflows.workflow.Task attribute) (trinity.explorer.scheduler.TaskWrapper attribute) repeatable (trinity.common.workflows.AgentScopeReactV2MathWorkflow property) (trinity.common.workflows.MathEvalWorkflow property) (trinity.common.workflows.Workflow property) (trinity.common.workflows.workflow.Workflow property) replay_buffer_kwargs (trinity.common.config.StorageConfig attribute) reply_prefix (trinity.common.config.ExplorerInput attribute) (trinity.common.config.FormatConfig attribute) representative_sample() (in module trinity.algorithm.sample_strategy.utils) REQUIRE_SYNC (trinity.common.constants.RunningStatus attribute) reset() (trinity.common.workflows.AgentScopeReactV2MathWorkflow method) (trinity.common.workflows.customized_math_workflows.MathBoxedWorkflow method) (trinity.common.workflows.MathBoxedWorkflow method) (trinity.common.workflows.MathWorkflow method) (trinity.common.workflows.RAFTAlfworldWorkflow method) (trinity.common.workflows.SimpleWorkflow method) (trinity.common.workflows.ToolCallWorkflow method) (trinity.common.workflows.WebShopWorkflow method) (trinity.common.workflows.Workflow method) (trinity.common.workflows.workflow.MathWorkflow method) (trinity.common.workflows.workflow.SimpleWorkflow method) (trinity.common.workflows.workflow.Workflow method) reset_prefix_cache() (trinity.common.models.vllm_model.vLLMRolloutModel method)

reset_session_state() (trinity.manager.config_manager.ConfigManager method) resettable (trinity.common.workflows.AgentScopeReactV2MathWorkflow property) (trinity.common.workflows.MathEvalWorkflow property) (trinity.common.workflows.SimpleWorkflow property) (trinity.common.workflows.WebShopWorkflow property) (trinity.common.workflows.Workflow property) (trinity.common.workflows.workflow.SimpleWorkflow property) (trinity.common.workflows.workflow.Workflow property) resettable() (trinity.common.workflows.RAFTAlfworldWorkflow method) reshard_after_forward (trinity.common.verl_config.FSDPConfig attribute) response (trinity.buffer.schema.sql_schema.ExperienceModel attribute) response_key (trinity.common.config.FormatConfig attribute) response_text (trinity.common.experience.Experience attribute) restart_runner() (trinity.explorer.scheduler.RunnerWrapper method) resume_from_path (trinity.common.verl_config.Trainer attribute) resume_mode (trinity.common.verl_config.Trainer attribute) retry_session() (in module trinity.buffer.utils) returns (trinity.common.experience.Experience attribute) (trinity.common.experience.Experiences attribute) reuse_cooldown_time (trinity.buffer.queue.AsyncPriorityQueue attribute) (trinity.common.config.StorageConfig attribute) reward (trinity.buffer.schema.sql_schema.ExperienceModel attribute) (trinity.common.experience.Experience attribute) reward() (trinity.common.workflows.StepWiseAlfworldWorkflow method) reward_fn (trinity.common.workflows.Task attribute) (trinity.common.workflows.workflow.Task attribute) reward_fn_args (trinity.common.config.StorageConfig attribute) (trinity.common.workflows.Task attribute) (trinity.common.workflows.workflow.Task attribute) reward_fn_key (trinity.common.config.FormatConfig attribute) reward_key (trinity.common.config.FormatConfig attribute) reward_manager (trinity.common.verl_config.RewardModel attribute) reward_model (trinity.common.verl_config.veRLConfig attribute) reward_type (trinity.buffer.schema.sql_schema.TaskModel attribute) RewardFn (class in trinity.common.rewards) (class in trinity.common.rewards.reward_fn) RewardModel (class in trinity.common.verl_config) rewards (trinity.common.experience.Experiences attribute) rid (trinity.common.experience.EID property) RLOOAdvantageFn (class in trinity.algorithm.advantage_fn) (class in trinity.algorithm.advantage_fn.rloo_advantage) RMGalleryFn (class in trinity.common.rewards) (class in trinity.common.rewards.reward_fn) Rollout (class in trinity.common.verl_config) rollout (trinity.common.verl_config.ActorRolloutRef attribute) rollout_args (trinity.common.config.StorageConfig attribute) (trinity.common.workflows.Task attribute) (trinity.common.workflows.Workflow property) (trinity.common.workflows.workflow.Task attribute) (trinity.common.workflows.workflow.Workflow property) rollout_model (trinity.common.config.ExplorerConfig attribute) rollout_n (trinity.common.verl_config.Critic attribute) RolloutDataReader (class in trinity.buffer.reader.file_reader) run (trinity.common.experience.EID attribute) run() (trinity.common.workflows.AgentScopeReactV2MathWorkflow method) (trinity.common.workflows.AlfworldWorkflow method) (trinity.common.workflows.customized_math_workflows.MathBoxedWorkflow method) (trinity.common.workflows.math_rm_workflow.MathRMWorkflow method) (trinity.common.workflows.MathBoxedWorkflow method) (trinity.common.workflows.MathEvalWorkflow method) (trinity.common.workflows.MathRMWorkflow method) (trinity.common.workflows.RAFTAlfworldWorkflow method) (trinity.common.workflows.RAFTReflectAlfworldWorkflow method) (trinity.common.workflows.SciWorldWorkflow method) (trinity.common.workflows.SimpleWorkflow method) (trinity.common.workflows.StepWiseAlfworldWorkflow method) (trinity.common.workflows.ToolCallWorkflow method) (trinity.common.workflows.WebShopWorkflow method) (trinity.common.workflows.Workflow method) (trinity.common.workflows.workflow.MultiTurnWorkflow method) (trinity.common.workflows.workflow.SimpleWorkflow method) (trinity.common.workflows.workflow.Workflow method) run_api_server() (trinity.common.models.vllm_model.vLLMRolloutModel method) run_config() (trinity.manager.config_manager.ConfigManager method) run_id_base (trinity.explorer.scheduler.TaskWrapper attribute) run_single_rollout() (trinity.common.workflows.RAFTAlfworldWorkflow method) run_task() (trinity.explorer.workflow_runner.WorkflowRunner method) run_with_retry() (trinity.explorer.scheduler.RunnerWrapper method) runner_num (trinity.common.config.ExplorerConfig attribute) runner_per_model (trinity.common.config.ExplorerConfig attribute) RunnerWrapper (class in trinity.explorer.scheduler) RUNNING (trinity.common.constants.RunningStatus attribute) running() (trinity.utils.dlc_utils.ClusterStatus method) RunningStatus (class in trinity.common.constants)

S

sample() (trinity.algorithm.sample_strategy.DefaultSampleStrategy method) (trinity.algorithm.sample_strategy.mix_sample_strategy.MixSampleStrategy method) (trinity.algorithm.sample_strategy.MixSampleStrategy method) (trinity.algorithm.sample_strategy.sample_strategy.DefaultSampleStrategy method) (trinity.algorithm.sample_strategy.sample_strategy.SampleStrategy method) (trinity.algorithm.sample_strategy.sample_strategy.WarmupSampleStrategy method) (trinity.algorithm.sample_strategy.SampleStrategy method) (trinity.algorithm.sample_strategy.WarmupSampleStrategy method) (trinity.algorithm.SampleStrategy method) sample_strategy (trinity.common.config.AlgorithmConfig attribute) sample_strategy_args (trinity.common.config.AlgorithmConfig attribute) SampleStrategy (class in trinity.algorithm) (class in trinity.algorithm.sample_strategy) (class in trinity.algorithm.sample_strategy.sample_strategy) save() (trinity.common.config.Config method) save_checkpoint() (trinity.explorer.Explorer method) (trinity.explorer.explorer.Explorer method) (trinity.trainer.TrainEngineWrapper method) (trinity.trainer.trainer.TrainEngineWrapper method) (trinity.trainer.verl.fsdp_workers.ActorRolloutRefWorker method) (trinity.trainer.verl.fsdp_workers.CriticWorker method) (trinity.trainer.verl_trainer.VerlPPOTrainerWrapper method) save_contents (trinity.common.verl_config.Checkpoint attribute) save_explorer() (trinity.manager.CacheManager method) (trinity.manager.manager.CacheManager method) save_freq (trinity.common.verl_config.Trainer attribute) save_input (trinity.common.config.ExperiencePipelineConfig attribute) save_interval (trinity.common.config.TrainerConfig attribute) save_state_dict() (trinity.trainer.TrainEngineWrapper method) (trinity.trainer.trainer.TrainEngineWrapper method) (trinity.trainer.verl_trainer.VerlPPOTrainerWrapper method) save_trainer() (trinity.manager.CacheManager method) (trinity.manager.manager.CacheManager method) schedule() (trinity.explorer.scheduler.Scheduler method) Scheduler (class in trinity.explorer.scheduler) schema (trinity.algorithm.algorithm.AlgorithmType attribute) (trinity.algorithm.algorithm.DPOAlgorithm attribute) (trinity.algorithm.algorithm.GRPOAlgorithm attribute) (trinity.algorithm.algorithm.MIXAlgorithm attribute) (trinity.algorithm.algorithm.MIXCHORDAlgorithm attribute) (trinity.algorithm.algorithm.OPMDAlgorithm attribute) (trinity.algorithm.algorithm.PPOAlgorithm attribute) (trinity.algorithm.algorithm.RAFTAlgorithm attribute) (trinity.algorithm.algorithm.SFTAlgorithm attribute) (trinity.algorithm.AlgorithmType attribute) SciWorldWorkflow (class in trinity.common.workflows) seed (trinity.common.config.InferenceModelConfig attribute) select_keys (trinity.algorithm.policy_loss_fn.dpo_loss.DPOLossFn property) (trinity.algorithm.policy_loss_fn.DPOLossFn property) (trinity.algorithm.policy_loss_fn.GSPOLossFn property) (trinity.algorithm.policy_loss_fn.mix_policy_loss.MIXPolicyLossFn property) (trinity.algorithm.policy_loss_fn.MIXCHORDPolicyLossFn property) (trinity.algorithm.policy_loss_fn.MIXPolicyLossFn property) (trinity.algorithm.policy_loss_fn.opmd_policy_loss.OPMDPolicyLossFn property) (trinity.algorithm.policy_loss_fn.OPMDPolicyLossFn property) (trinity.algorithm.policy_loss_fn.policy_loss_fn.PolicyLossFn property) (trinity.algorithm.policy_loss_fn.PolicyLossFn property) (trinity.algorithm.policy_loss_fn.ppo_policy_loss.PPOPolicyLossFn property) (trinity.algorithm.policy_loss_fn.PPOPolicyLossFn property) (trinity.algorithm.policy_loss_fn.sft_loss.SFTLossFn property) (trinity.algorithm.policy_loss_fn.SFTISLossFn property) (trinity.algorithm.policy_loss_fn.SFTLossFn property) (trinity.algorithm.policy_loss_fn.SFTPhiLossFn property) (trinity.algorithm.PolicyLossFn property) serialize() (trinity.common.experience.Experience method) serialized_exp (trinity.buffer.schema.sql_schema.DPODataModel attribute) (trinity.buffer.schema.sql_schema.ExperienceModel attribute) (trinity.buffer.schema.sql_schema.SFTDataModel attribute) server_url (trinity.common.config.DataJuicerServiceConfig attribute) service (trinity.common.config.Config attribute) ServiceConfig (class in trinity.common.config) set_actor_checkpoint() (in module trinity.manager.config_registry.trainer_config_manager) set_actor_entropy_checkpointing() (in module trinity.manager.config_registry.trainer_config_manager) set_actor_entropy_from_logits_with_chunking() (in module trinity.manager.config_registry.trainer_config_manager) set_actor_grad_clip() (in module trinity.manager.config_registry.trainer_config_manager) set_actor_lr() (in module trinity.manager.config_registry.trainer_config_manager) set_actor_lr_warmup_steps_ratio() (in module trinity.manager.config_registry.trainer_config_manager) set_actor_ppo_micro_batch_size_per_gpu() (in module trinity.manager.config_registry.trainer_config_manager) set_actor_ulysses_sequence_parallel_size() (in module trinity.manager.config_registry.trainer_config_manager) set_actor_warmup_style() (in module trinity.manager.config_registry.trainer_config_manager) set_adaptive_in_kl_penalty_fn() (in module trinity.manager.config_registry.algorithm_config_manager) set_advantage_fn() (in module trinity.manager.config_registry.algorithm_config_manager) set_algorithm() (trinity.trainer.verl.dp_actor.DataParallelPPOActor method) (trinity.trainer.verl.fsdp_workers.ActorRolloutRefWorker method) set_algorithm_type() (in module trinity.manager.config_registry.algorithm_config_manager) set_auxiliary_models() (in module trinity.manager.config_registry.explorer_config_manager) set_bench_on_latest_checkpoint() (in module trinity.manager.config_registry.explorer_config_manager) set_beta_in_policy_loss_fn() (in module trinity.manager.config_registry.algorithm_config_manager) set_buffer_max_retry_times() (in module trinity.manager.config_registry.buffer_config_manager) set_checkpoint_root_dir() (in module trinity.manager.config_registry.model_config_manager) set_clip_range_in_policy_loss_fn() (in module trinity.manager.config_registry.algorithm_config_manager) set_critic_checkpoint() (in module trinity.manager.config_registry.trainer_config_manager) set_critic_cliprange_value() (in module trinity.manager.config_registry.trainer_config_manager) set_critic_grad_clip() (in module trinity.manager.config_registry.trainer_config_manager) set_critic_lr() (in module trinity.manager.config_registry.trainer_config_manager) set_critic_lr_warmup_steps_ratio() (in module trinity.manager.config_registry.trainer_config_manager) set_critic_model_path() (in module trinity.manager.config_registry.model_config_manager) set_critic_ppo_micro_batch_size_per_gpu() (in module trinity.manager.config_registry.trainer_config_manager) set_critic_ulysses_sequence_parallel_size() (in module trinity.manager.config_registry.trainer_config_manager) set_critic_warmup() (in module trinity.manager.config_registry.trainer_config_manager) set_critic_warmup_style() (in module trinity.manager.config_registry.trainer_config_manager) set_default_eval_workflow_type() (in module trinity.manager.config_registry.buffer_config_manager) set_default_hdfs_dir() (in module trinity.manager.config_registry.trainer_config_manager) set_default_reward_fn_type() (in module trinity.manager.config_registry.buffer_config_manager) set_default_workflow_type() (in module trinity.manager.config_registry.buffer_config_manager) set_del_local_ckpt_after_load() (in module trinity.manager.config_registry.trainer_config_manager) set_dpo_dataset_kwargs() (in module trinity.manager.config_registry.buffer_config_manager) set_dtype() (in module trinity.manager.config_registry.explorer_config_manager) set_enable_auto_tool_choice() (in module trinity.manager.config_registry.explorer_config_manager) set_enable_chunked_prefill() (in module trinity.manager.config_registry.explorer_config_manager) set_enable_openai_api() (in module trinity.manager.config_registry.explorer_config_manager) set_enable_prefix_caching() (in module trinity.manager.config_registry.explorer_config_manager) set_enable_preview() (in module trinity.manager.config_registry.trainer_config_manager) set_enable_thinking() (in module trinity.manager.config_registry.explorer_config_manager) set_enforce_eager() (in module trinity.manager.config_registry.explorer_config_manager) set_engine_num() (in module trinity.manager.config_registry.explorer_config_manager) set_engine_type() (in module trinity.manager.config_registry.explorer_config_manager) set_entropy_coef_in_entropy_loss_fn() (in module trinity.manager.config_registry.algorithm_config_manager) set_entropy_loss_fn() (in module trinity.manager.config_registry.algorithm_config_manager) set_epsilon_in_advantage_fn() (in module trinity.manager.config_registry.algorithm_config_manager) set_eval_interval() (in module trinity.manager.config_registry.explorer_config_manager) set_eval_tasksets() (in module trinity.manager.config_registry.buffer_config_manager) set_exp_name() (in module trinity.manager.config_registry.model_config_manager) set_experience_buffer_path() (in module trinity.manager.config_registry.buffer_config_manager) set_expert_data_ratio_in_sample_strategy() (in module trinity.manager.config_registry.algorithm_config_manager) set_explore_batch_size() (in module trinity.manager.config_registry.buffer_config_manager) set_explorer_max_retry_times() (in module trinity.manager.config_registry.explorer_config_manager) set_explorer_status() (trinity.manager.Synchronizer method) set_forward_prefetch() (in module trinity.manager.config_registry.trainer_config_manager) set_gamma_in_advantage_fn() (in module trinity.manager.config_registry.algorithm_config_manager) set_gpu_memory_utilization() (in module trinity.manager.config_registry.explorer_config_manager) set_gpu_per_node() (in module trinity.manager.config_registry.model_config_manager) set_horizon() (in module trinity.manager.config_registry.trainer_config_manager) set_impl_backend() (in module trinity.manager.config_registry.trainer_config_manager) set_kl_coef_in_kl_loss_fn() (in module trinity.manager.config_registry.algorithm_config_manager) set_kl_coef_in_kl_penalty_fn() (in module trinity.manager.config_registry.algorithm_config_manager) set_kl_ctrl_coef() (in module trinity.manager.config_registry.trainer_config_manager) set_kl_ctrl_type() (in module trinity.manager.config_registry.trainer_config_manager) set_kl_loss_fn() (in module trinity.manager.config_registry.algorithm_config_manager) set_kl_penalty() (in module trinity.manager.config_registry.trainer_config_manager) set_kl_penalty_fn() (in module trinity.manager.config_registry.algorithm_config_manager) set_label_smoothing_in_policy_loss_fn() (in module trinity.manager.config_registry.algorithm_config_manager) set_lam_in_advantage_fn() (in module trinity.manager.config_registry.algorithm_config_manager) set_max_actor_ckpt_to_keep() (in module trinity.manager.config_registry.trainer_config_manager) set_max_critic_ckpt_to_keep() (in module trinity.manager.config_registry.trainer_config_manager) set_max_model_len() (in module trinity.manager.config_registry.model_config_manager) set_max_response_tokens() (in module trinity.manager.config_registry.model_config_manager)

set_max_retry_interval() (in module trinity.manager.config_registry.buffer_config_manager) set_max_timeout() (in module trinity.manager.config_registry.explorer_config_manager) set_model_path() (in module trinity.manager.config_registry.model_config_manager) set_model_state_dict() (trinity.manager.Synchronizer method) set_model_state_dict_with_step_num() (trinity.manager.Synchronizer method) set_monitor_type() (in module trinity.manager.config_registry.model_config_manager) set_mu_in_policy_loss_fn() (in module trinity.manager.config_registry.algorithm_config_manager) set_node_num() (in module trinity.manager.config_registry.model_config_manager) set_norm_adv_by_std_in_grpo() (in module trinity.manager.config_registry.trainer_config_manager) set_opmd_baseline_in_advantage_fn() (in module trinity.manager.config_registry.algorithm_config_manager) set_optimizer_offload() (in module trinity.manager.config_registry.trainer_config_manager) set_param_offload() (in module trinity.manager.config_registry.trainer_config_manager) set_policy_loss_fn() (in module trinity.manager.config_registry.algorithm_config_manager) set_ppo_epochs() (in module trinity.manager.config_registry.trainer_config_manager) set_priority_decay() (in module trinity.manager.config_registry.buffer_config_manager) set_priority_fn() (in module trinity.manager.config_registry.buffer_config_manager) set_project() (in module trinity.manager.config_registry.model_config_manager) set_reasoning_parser() (in module trinity.manager.config_registry.explorer_config_manager) set_ref_log_prob_micro_batch_size_per_gpu() (in module trinity.manager.config_registry.trainer_config_manager) set_remove_previous_ckpt_in_save() (in module trinity.manager.config_registry.trainer_config_manager) set_repeat_times() (in module trinity.manager.config_registry.algorithm_config_manager) (trinity.common.workflows.RAFTAlfworldWorkflow method) (trinity.common.workflows.SimpleWorkflow method) (trinity.common.workflows.Workflow method) (trinity.common.workflows.workflow.MultiTurnWorkflow method) (trinity.common.workflows.workflow.SimpleWorkflow method) (trinity.common.workflows.workflow.Workflow method) set_reply_prefix() (in module trinity.manager.config_registry.buffer_config_manager) set_resume_from_path() (in module trinity.manager.config_registry.trainer_config_manager) set_resume_mode() (in module trinity.manager.config_registry.trainer_config_manager) set_reuse_cooldown_time() (in module trinity.manager.config_registry.buffer_config_manager) set_runner_per_model() (in module trinity.manager.config_registry.explorer_config_manager) set_sample_strategy() (in module trinity.manager.config_registry.algorithm_config_manager) set_save_interval() (in module trinity.manager.config_registry.trainer_config_manager) set_seed() (in module trinity.manager.config_registry.explorer_config_manager) set_sft_warmup_dataset_args() (in module trinity.manager.config_registry.buffer_config_manager) set_sft_warmup_dataset_path() (in module trinity.manager.config_registry.buffer_config_manager) set_sft_warmup_steps() (in module trinity.manager.config_registry.buffer_config_manager) set_storage_type() (in module trinity.manager.config_registry.buffer_config_manager) set_sync_interval() (in module trinity.manager.config_registry.explorer_config_manager) set_sync_method() (in module trinity.manager.config_registry.explorer_config_manager) set_sync_timeout() (in module trinity.manager.config_registry.explorer_config_manager) set_system_prompt() (in module trinity.manager.config_registry.buffer_config_manager) set_target_kl() (in module trinity.manager.config_registry.trainer_config_manager) set_taskset_args() (in module trinity.manager.config_registry.buffer_config_manager) set_taskset_path() (in module trinity.manager.config_registry.buffer_config_manager) set_tau_in_advantage_fn() (in module trinity.manager.config_registry.algorithm_config_manager) set_tau_in_policy_loss_fn() (in module trinity.manager.config_registry.algorithm_config_manager) set_tensor_parallel_size() (in module trinity.manager.config_registry.explorer_config_manager) set_tool_call_parser() (in module trinity.manager.config_registry.explorer_config_manager) set_total_epochs() (in module trinity.manager.config_registry.buffer_config_manager) set_total_gpu_num() (in module trinity.manager.config_registry.model_config_manager) set_total_training_steps() (in module trinity.manager.config_registry.trainer_config_manager) set_train_batch_size() (in module trinity.manager.config_registry.buffer_config_manager) set_trainer_gpu_num() (in module trinity.manager.config_registry.model_config_manager) set_trainer_status() (trinity.manager.Synchronizer method) set_trainer_type() (in module trinity.manager.config_registry.trainer_config_manager) set_training_args() (in module trinity.manager.config_registry.trainer_config_manager) set_training_strategy() (in module trinity.manager.config_registry.trainer_config_manager) set_unfinished_fields() (trinity.manager.config_registry.config_registry.ConfigRegistry method) set_use_kl_in_reward() (in module trinity.manager.config_registry.trainer_config_manager) set_use_priority_queue() (in module trinity.manager.config_registry.buffer_config_manager) set_use_token_level_loss_in_policy_loss_fn() (in module trinity.manager.config_registry.algorithm_config_manager) set_use_v1() (in module trinity.manager.config_registry.explorer_config_manager) setup_data_processor (trinity.common.config.DataProcessorConfig attribute) setup_ray_cluster() (in module trinity.utils.dlc_utils) setup_weight_sync_group() (trinity.explorer.Explorer method) (trinity.explorer.explorer.Explorer method) (trinity.manager.Synchronizer method) (trinity.trainer.verl.fsdp_workers.ActorRolloutRefWorker method) sft_to_rft() (trinity.trainer.verl_trainer.VerlPPOTrainerWrapper method) sft_warmup_dataset (trinity.common.config.TrainerInput attribute) sft_warmup_steps (trinity.common.config.TrainerInput attribute) (trinity.common.verl_config.Trainer attribute) SFTAlgorithm (class in trinity.algorithm.algorithm) SFTDataModel (class in trinity.buffer.schema.sql_schema) SFTDataReader (class in trinity.buffer.reader.file_reader) SFTISLossFn (class in trinity.algorithm.policy_loss_fn) SFTLossFn (class in trinity.algorithm.policy_loss_fn) (class in trinity.algorithm.policy_loss_fn.sft_loss) SFTPhiLossFn (class in trinity.algorithm.policy_loss_fn) shuffle (trinity.common.verl_config.Actor attribute) (trinity.common.verl_config.Critic attribute) shutdown() (trinity.common.models.vllm_model.vLLMRolloutModel method) (trinity.explorer.Explorer method) (trinity.explorer.explorer.Explorer method) (trinity.trainer.Trainer method) (trinity.trainer.trainer.Trainer method) sid (trinity.common.experience.EID property) simple_answer_parser() (in module trinity.utils.eval_utils) SimpleWorkflow (class in trinity.common.workflows) (class in trinity.common.workflows.workflow) sleep() (trinity.common.models.vllm_model.vLLMRolloutModel method) solution_key (trinity.common.config.FormatConfig attribute) sort_batch_id() (in module trinity.explorer.scheduler) source_field (trinity.common.experience.CustomField attribute) split (trinity.common.config.StorageConfig attribute) split_dpo_experience_to_single_turn() (in module trinity.common.experience) SQL (trinity.common.constants.StorageType attribute) SQLReader (class in trinity.buffer.reader.sql_reader) SQLWriter (class in trinity.buffer.writer.sql_writer) start() (trinity.explorer.scheduler.Scheduler method) Status (class in trinity.explorer.workflow_runner) step (trinity.common.experience.EID attribute) step() (trinity.common.workflows.StepWiseAlfworldWorkflow method) StepWiseAlfworldWorkflow (class in trinity.common.workflows) StepWiseGRPOAdvantageFn (class in trinity.algorithm.advantage_fn) stop() (trinity.explorer.scheduler.Scheduler method) stop_ray_cluster() (in module trinity.utils.dlc_utils) STOPPED (trinity.common.constants.RunningStatus attribute) stopped() (trinity.buffer.queue.AsyncPriorityQueue method) (trinity.buffer.queue.AsyncQueue method) (trinity.buffer.queue.QueueBuffer method) storage_type (trinity.common.config.StorageConfig attribute) StorageConfig (class in trinity.common.config) StorageType (class in trinity.common.constants) strategy (trinity.common.verl_config.Actor attribute) (trinity.common.verl_config.Critic attribute) (trinity.common.verl_config.RewardModel attribute) SUB (trinity.common.constants.OpType attribute) subset_name (trinity.common.config.StorageConfig attribute) suffix (trinity.common.experience.EID attribute) sync_freq (trinity.common.verl_config.Trainer attribute) sync_interval (trinity.common.config.SynchronizerConfig attribute) sync_method (trinity.common.config.SynchronizerConfig attribute) sync_model() (trinity.common.models.vllm_model.vLLMRolloutModel method) sync_offset (trinity.common.config.SynchronizerConfig attribute) sync_style (trinity.common.config.SynchronizerConfig attribute) sync_timeout (trinity.common.config.SynchronizerConfig attribute) sync_weight() (trinity.explorer.Explorer method) (trinity.explorer.explorer.Explorer method) (trinity.trainer.TrainEngineWrapper method) (trinity.trainer.Trainer method) (trinity.trainer.trainer.TrainEngineWrapper method) (trinity.trainer.trainer.Trainer method) (trinity.trainer.verl.fsdp_workers.ActorRolloutRefWorker method) (trinity.trainer.verl_trainer.VerlPPOTrainerWrapper method) synchronize_config() (trinity.common.verl_config.veRLConfig method) Synchronizer (class in trinity.manager) synchronizer (trinity.common.config.Config attribute) (trinity.common.verl_config.ActorRolloutRef attribute) (trinity.common.verl_config.veRLConfig attribute) SynchronizerConfig (class in trinity.common.config) SyncMethod (class in trinity.common.constants) SyncMethodEnumMeta (class in trinity.common.constants) SyncStyle (class in trinity.common.constants) system_prompt (trinity.common.config.ExplorerInput attribute) (trinity.common.config.FormatConfig attribute)

T

target_fields (trinity.common.config.TaskPipelineConfig attribute) target_kl (trinity.common.verl_config.KL_Ctrl attribute) Task (class in trinity.common.workflows) (class in trinity.common.workflows.workflow) task (trinity.common.experience.EID attribute) (trinity.explorer.scheduler.TaskWrapper attribute) task_desc (trinity.buffer.schema.sql_schema.TaskModel attribute) (trinity.common.workflows.Task property) (trinity.common.workflows.workflow.Task property) task_done_callback() (trinity.explorer.scheduler.Scheduler method) task_id (trinity.common.workflows.Task attribute) (trinity.common.workflows.workflow.Task attribute) task_pipeline (trinity.common.config.DataProcessorConfig attribute) task_type (trinity.common.config.StorageConfig attribute) TaskModel (class in trinity.buffer.schema.sql_schema) TaskPipelineConfig (class in trinity.common.config) taskset (trinity.common.config.ExplorerInput attribute) TaskType (class in trinity.common.constants) TaskWrapper (class in trinity.explorer.scheduler) temperature (trinity.common.config.GenerationConfig attribute) (trinity.common.verl_config.Rollout attribute) tensor_parallel_size (trinity.common.config.InferenceModelConfig attribute) TENSORBOARD (trinity.common.constants.MonitorType attribute) TensorboardMonitor (class in trinity.utils.monitor) test_freq (trinity.common.verl_config.Trainer attribute) tid (trinity.common.experience.EID property) Timer (class in trinity.utils.timer) to_dict() (trinity.common.experience.EID method) (trinity.common.experience.Experience method) (trinity.common.workflows.Task method) (trinity.common.workflows.workflow.Task method) to_experience() (trinity.buffer.schema.sql_schema.DPODataModel method) (trinity.buffer.schema.sql_schema.ExperienceModel method) (trinity.buffer.schema.sql_schema.SFTDataModel method) to_hf_datasets() (in module trinity.common.experience) to_rm_gallery_messages() (in module trinity.common.rewards.utils) to_trinity() (trinity.algorithm.key_mapper.KeyMapper method) to_workflow() (trinity.common.workflows.Task method) (trinity.common.workflows.workflow.Task method) tokenize_and_mask_messages_default() (in module trinity.common.models.utils) tokenize_and_mask_messages_hf() (in module trinity.common.models.utils) tokenizer_path (trinity.common.config.BufferConfig attribute) (trinity.common.verl_config.CriticModel attribute) tokens (trinity.common.experience.Experience attribute) (trinity.common.experience.Experiences attribute) tool_call_parser (trinity.common.config.InferenceModelConfig attribute) ToolCallWorkflow (class in trinity.common.workflows) tools (trinity.common.experience.Experience attribute) tools_key (trinity.common.config.FormatConfig attribute) top_k (trinity.common.config.GenerationConfig attribute) (trinity.common.config.TaskPipelineConfig attribute) top_p (trinity.common.config.GenerationConfig attribute) total_epochs (trinity.common.config.BufferConfig attribute) (trinity.common.config.StorageConfig attribute) (trinity.common.verl_config.Trainer attribute) total_steps (trinity.common.config.BufferConfig attribute) (trinity.common.config.StorageConfig attribute) total_training_steps (trinity.common.verl_config.Optim attribute) (trinity.common.verl_config.Trainer attribute) train() (trinity.trainer.Trainer method) (trinity.trainer.trainer.Trainer method) train_batch_size (trinity.common.config.BufferConfig attribute) (trinity.common.verl_config.Data attribute) train_step() (trinity.trainer.TrainEngineWrapper method) (trinity.trainer.Trainer method) (trinity.trainer.trainer.TrainEngineWrapper method) (trinity.trainer.trainer.Trainer method) (trinity.trainer.verl_trainer.VerlPPOTrainerWrapper method) train_step_num (trinity.trainer.TrainEngineWrapper property) (trinity.trainer.Trainer property) (trinity.trainer.trainer.TrainEngineWrapper property) (trinity.trainer.trainer.Trainer property) (trinity.trainer.verl_trainer.VerlPPOTrainerWrapper property) TrainEngineWrapper (class in trinity.trainer) (class in trinity.trainer.trainer) Trainer (class in trinity.common.verl_config) (class in trinity.trainer) (class in trinity.trainer.trainer) trainer (trinity.common.config.Config attribute) (trinity.common.verl_config.veRLConfig attribute) trainer_config (trinity.common.config.TrainerConfig attribute) trainer_config_path (trinity.common.config.TrainerConfig attribute) trainer_input (trinity.common.config.BufferConfig attribute) trainer_status (trinity.manager.Synchronizer attribute) trainer_type (trinity.common.config.TrainerConfig attribute) TrainerConfig (class in trinity.common.config) TrainerInput (class in trinity.common.config) training_rollout_mode (trinity.common.verl_config.Trainer attribute) trinity module trinity.algorithm module trinity.algorithm.advantage_fn module trinity.algorithm.advantage_fn.advantage_fn module trinity.algorithm.advantage_fn.grpo_advantage module trinity.algorithm.advantage_fn.opmd_advantage module trinity.algorithm.advantage_fn.ppo_advantage module trinity.algorithm.advantage_fn.reinforce_plus_plus_advantage module trinity.algorithm.advantage_fn.remax_advantage module trinity.algorithm.advantage_fn.rloo_advantage module trinity.algorithm.algorithm module trinity.algorithm.algorithm_manager module trinity.algorithm.entropy_loss_fn module trinity.algorithm.entropy_loss_fn.entropy_loss_fn module trinity.algorithm.key_mapper module trinity.algorithm.kl_fn module trinity.algorithm.kl_fn.kl_fn module trinity.algorithm.policy_loss_fn module trinity.algorithm.policy_loss_fn.dpo_loss module trinity.algorithm.policy_loss_fn.mix_policy_loss module trinity.algorithm.policy_loss_fn.opmd_policy_loss module trinity.algorithm.policy_loss_fn.policy_loss_fn module trinity.algorithm.policy_loss_fn.ppo_policy_loss module trinity.algorithm.policy_loss_fn.sft_loss module trinity.algorithm.sample_strategy module trinity.algorithm.sample_strategy.mix_sample_strategy module trinity.algorithm.sample_strategy.sample_strategy module trinity.algorithm.sample_strategy.utils module trinity.algorithm.utils module

trinity.buffer module trinity.buffer.buffer module trinity.buffer.buffer_reader module trinity.buffer.buffer_writer module trinity.buffer.queue module trinity.buffer.ray_wrapper module trinity.buffer.reader module trinity.buffer.reader.file_reader module trinity.buffer.reader.queue_reader module trinity.buffer.reader.sql_reader module trinity.buffer.schema module trinity.buffer.schema.sql_schema module trinity.buffer.utils module trinity.buffer.writer module trinity.buffer.writer.file_writer module trinity.buffer.writer.queue_writer module trinity.buffer.writer.sql_writer module trinity.common module trinity.common.config module trinity.common.constants module trinity.common.experience module trinity.common.models module trinity.common.models.model module trinity.common.models.utils module trinity.common.models.vllm_model module trinity.common.models.vllm_worker module trinity.common.rewards module trinity.common.rewards.accuracy_reward module trinity.common.rewards.agents_reward module trinity.common.rewards.countdown_reward module trinity.common.rewards.format_reward module trinity.common.rewards.human_reward module trinity.common.rewards.math_reward module trinity.common.rewards.reward_fn module trinity.common.rewards.tool_reward module trinity.common.rewards.utils module trinity.common.verl_config module trinity.common.workflows module trinity.common.workflows.customized_math_workflows module trinity.common.workflows.math_rm_workflow module trinity.common.workflows.workflow module trinity.explorer module trinity.explorer.explorer module trinity.explorer.scheduler module trinity.explorer.workflow_runner module trinity.manager module trinity.manager.config_manager module trinity.manager.config_registry module trinity.manager.config_registry.algorithm_config_manager module trinity.manager.config_registry.buffer_config_manager module trinity.manager.config_registry.config_registry module trinity.manager.config_registry.explorer_config_manager module trinity.manager.config_registry.model_config_manager module trinity.manager.config_registry.trainer_config_manager module trinity.manager.manager module trinity.plugins module trinity.trainer module trinity.trainer.trainer module trinity.trainer.verl module trinity.trainer.verl.dp_actor module trinity.trainer.verl.fsdp_workers module trinity.trainer.verl_trainer module trinity.utils module trinity.utils.distributed module trinity.utils.dlc_utils module trinity.utils.eval_utils module trinity.utils.log module trinity.utils.monitor module trinity.utils.plugin_loader module trinity.utils.registry module trinity.utils.timer module truth (trinity.common.workflows.Task property) (trinity.common.workflows.workflow.Task property) type (trinity.common.verl_config.KL_Ctrl attribute)

U

uid (trinity.common.experience.EID property) ulysses_sequence_parallel_size (trinity.common.verl_config.Actor attribute) (trinity.common.verl_config.Critic attribute) (trinity.common.verl_config.Ref attribute) (trinity.common.verl_config.RewardModel attribute) update() (trinity.buffer.reader.file_reader.DummyProgressBar method) update_actor() (trinity.trainer.verl.fsdp_workers.ActorRolloutRefWorker method) update_critic() (trinity.trainer.verl.fsdp_workers.CriticWorker method) update_kl_coef() (trinity.algorithm.kl_fn.kl_fn.KLFn method) (trinity.algorithm.kl_fn.KLFn method) (trinity.algorithm.KLFn method) update_policy() (trinity.trainer.verl.dp_actor.DataParallelPPOActor method) update_weight() (trinity.common.models.vllm_worker.WorkerExtension method) upload_state_dict() (trinity.trainer.TrainEngineWrapper method) (trinity.trainer.trainer.TrainEngineWrapper method) (trinity.trainer.verl.fsdp_workers.ActorRolloutRefWorker method) (trinity.trainer.verl_trainer.VerlPPOTrainerWrapper method) use_critic (trinity.algorithm.algorithm.AlgorithmType attribute) (trinity.algorithm.algorithm.DPOAlgorithm attribute) (trinity.algorithm.algorithm.GRPOAlgorithm attribute) (trinity.algorithm.algorithm.MIXAlgorithm attribute) (trinity.algorithm.algorithm.MIXCHORDAlgorithm attribute) (trinity.algorithm.algorithm.OPMDAlgorithm attribute) (trinity.algorithm.algorithm.PPOAlgorithm attribute) (trinity.algorithm.algorithm.RAFTAlgorithm attribute) (trinity.algorithm.algorithm.SFTAlgorithm attribute) (trinity.algorithm.AlgorithmType attribute)

use_critic() (in module trinity.manager.config_registry.trainer_config_manager) use_dynamic_bsz (trinity.common.verl_config.Actor attribute) (trinity.common.verl_config.Critic attribute) (trinity.common.verl_config.RewardModel attribute) use_fsdp() (in module trinity.manager.config_registry.trainer_config_manager) use_fused_kernels (trinity.common.verl_config.ActorModel attribute) use_kl_in_reward (trinity.common.verl_config.Algorithm attribute) use_kl_loss (trinity.common.verl_config.Actor attribute) use_priority_queue (trinity.common.config.StorageConfig attribute) use_reference (trinity.algorithm.algorithm.AlgorithmType attribute) (trinity.algorithm.algorithm.DPOAlgorithm attribute) (trinity.algorithm.algorithm.GRPOAlgorithm attribute) (trinity.algorithm.algorithm.MIXAlgorithm attribute) (trinity.algorithm.algorithm.MIXCHORDAlgorithm attribute) (trinity.algorithm.algorithm.OPMDAlgorithm attribute) (trinity.algorithm.algorithm.PPOAlgorithm attribute) (trinity.algorithm.algorithm.RAFTAlgorithm attribute) (trinity.algorithm.algorithm.SFTAlgorithm attribute) (trinity.algorithm.AlgorithmType attribute) use_remove_padding (trinity.common.verl_config.ActorModel attribute) (trinity.common.verl_config.CriticModel attribute) use_rollout (trinity.algorithm.algorithm.MIXAlgorithm attribute) (trinity.algorithm.algorithm.MIXCHORDAlgorithm attribute) use_token_level_loss (trinity.common.config.AlgorithmConfig attribute) use_v1 (trinity.common.config.InferenceModelConfig attribute)

V

val_before_train (trinity.common.verl_config.Trainer attribute) val_generations_to_log_to_wandb (trinity.common.verl_config.Trainer attribute) val_kwargs (trinity.common.verl_config.Rollout attribute) validate_equation() (in module trinity.utils.eval_utils)

validate_think_pattern() (in module trinity.utils.eval_utils) veRLConfig (class in trinity.common.verl_config) VerlPPOTrainerWrapper (class in trinity.trainer.verl_trainer) vLLMRolloutModel (class in trinity.common.models.vllm_model)

W

wait_all() (trinity.explorer.scheduler.Scheduler method) wait_for_checkpoint (trinity.common.config.SynchronizerConfig attribute) wait_for_ray_setup() (in module trinity.utils.dlc_utils) wait_for_ray_worker_nodes() (in module trinity.utils.dlc_utils) wait_new_model_state_dict() (trinity.manager.Synchronizer method) wait_on_save_thread() (trinity.trainer.verl.fsdp_workers.ActorRolloutRefWorker method) (trinity.trainer.verl.fsdp_workers.CriticWorker method) WAITING_SYNC (trinity.common.constants.RunningStatus attribute) wake_up() (trinity.common.models.vllm_model.vLLMRolloutModel method) WANDB (trinity.common.constants.MonitorType attribute) WandbMonitor (class in trinity.utils.monitor) warmup_style (trinity.common.verl_config.Optim attribute) WarmupSampleStrategy (class in trinity.algorithm.sample_strategy) (class in trinity.algorithm.sample_strategy.sample_strategy) WebShopWorkflow (class in trinity.common.workflows) WorkerExtension (class in trinity.common.models.vllm_worker) Workflow (class in trinity.common.workflows) (class in trinity.common.workflows.workflow) workflow (trinity.common.workflows.Task attribute) (trinity.common.workflows.workflow.Task attribute)

workflow_args (trinity.common.config.StorageConfig attribute) (trinity.common.workflows.Task attribute) (trinity.common.workflows.workflow.Task attribute) workflow_key (trinity.common.config.FormatConfig attribute) workflow_type (trinity.buffer.schema.sql_schema.TaskModel attribute) WorkflowRunner (class in trinity.explorer.workflow_runner) wrap_in_ray (trinity.common.config.StorageConfig attribute) wrap_policy (trinity.common.verl_config.FSDPConfig attribute) WrapPolicy (class in trinity.common.verl_config) write() (trinity.buffer.buffer_writer.BufferWriter method) (trinity.buffer.BufferWriter method) (trinity.buffer.ray_wrapper.DBWrapper method) (trinity.buffer.ray_wrapper.FileWrapper method) (trinity.buffer.writer.file_writer.JSONWriter method) (trinity.buffer.writer.queue_writer.QueueWriter method) (trinity.buffer.writer.sql_writer.SQLWriter method) write_async() (trinity.buffer.buffer_writer.BufferWriter method) (trinity.buffer.BufferWriter method) (trinity.buffer.writer.file_writer.JSONWriter method) (trinity.buffer.writer.queue_writer.QueueWriter method) (trinity.buffer.writer.sql_writer.SQLWriter method)