trinity.explorer package

Submodules

• trinity.explorer.explorer module
  • Explorer
    • Explorer.__init__()
    • Explorer.setup_weight_sync_group()
    • Explorer.prepare()
    • Explorer.get_weight()
    • Explorer.explore()
    • Explorer.explore_step()
    • Explorer.need_sync()
    • Explorer.need_eval()
    • Explorer.eval()
    • Explorer.benchmark()
    • Explorer.save_checkpoint()
    • Explorer.sync_weight()
    • Explorer.shutdown()
    • Explorer.is_alive()
    • Explorer.get_actor()
• trinity.explorer.scheduler module
  • TaskWrapper
    • TaskWrapper.task
    • TaskWrapper.batch_id
    • TaskWrapper.run_id_base
    • TaskWrapper.repeat_times
    • TaskWrapper.__init__()
  • RunnerWrapper
    • RunnerWrapper.__init__()
    • RunnerWrapper.run_with_retry()
    • RunnerWrapper.restart_runner()
  • sort_batch_id()
  • Scheduler
    • Scheduler.__init__()
    • Scheduler.task_done_callback()
    • Scheduler.start()
    • Scheduler.stop()
    • Scheduler.schedule()
    • Scheduler.get_results()
    • Scheduler.has_step()
    • Scheduler.wait_all()
• trinity.explorer.workflow_runner module
  • Status
    • Status.ok
    • Status.metric
    • Status.message
    • Status.__init__()
  • WorkflowRunner
    • WorkflowRunner.__init__()
    • WorkflowRunner.is_alive()
    • WorkflowRunner.run_task()

Module contents

class trinity.explorer.Explorer(config: Config)

Bases: object

Responsible for exploring the taskset.

__init__(config: Config)

async setup_weight_sync_group(master_address: str, master_port: int, state_dict_meta: List | None = None)

async prepare() → None

Preparation before running.

async get_weight(name: str) → Tensor

Get the weight of the loaded model (For checkpoint weights update).

async explore() → str

The timeline of the exploration process:

<——————————— one period ————————————-> |

explorer | <—————- step_1 ————–> | |

| <—————- step_2 ————–> | |

… |

| <—————- step_n —————> | |

| <———————- eval ——————–> | <– sync –> |

|--------------------------------------------------------------------------------------|

trainer | <– idle –> | <– step_1 –> | <– step_2 –> | … | <– step_n –> | <– sync –> |

async explore_step() → bool

async need_sync() → bool

need_eval() → bool

async eval()

Evaluation on all evaluation data samples.

async benchmark() → bool

Benchmark the model checkpoints.

async save_checkpoint(sync_weight: bool = False) → None

async sync_weight() → None

Synchronize model weights.

async shutdown() → None

async is_alive() → bool

Check if the explorer is alive.

classmethod get_actor(config: Config)

Get a Ray actor for the explorer.