MCPcopy Create free account
hub / github.com/YeWR/EfficientZero / __init__

Method __init__

core/reanalyze_worker.py:268–296  ·  view source on GitHub ↗

GPU Batch Worker for reanalyzing targets, see Appendix. receive the context from CPU maker and deal with GPU overheads Parameters ---------- worker_id: int id of the worker replay_buffer: Any Replay buffer storage: Any

(self, worker_id, replay_buffer, storage, batch_storage, mcts_storage, config)

Source from the content-addressed store, hash-verified

266@ray.remote(num_gpus=0.125)
267class BatchWorker_GPU(object):
268 def __init__(self, worker_id, replay_buffer, storage, batch_storage, mcts_storage, config):
269 """GPU Batch Worker for reanalyzing targets, see Appendix.
270 receive the context from CPU maker and deal with GPU overheads
271 Parameters
272 ----------
273 worker_id: int
274 id of the worker
275 replay_buffer: Any
276 Replay buffer
277 storage: Any
278 The model storage
279 batch_storage: Any
280 The batch storage (batch queue)
281 mcts_storage: Ant
282 The mcts-related contexts storage
283 """
284 self.replay_buffer = replay_buffer
285 self.config = config
286 self.worker_id = worker_id
287
288 self.model = config.get_uniform_network()
289 self.model.to(config.device)
290 self.model.eval()
291
292 self.mcts_storage = mcts_storage
293 self.storage = storage
294 self.batch_storage = batch_storage
295
296 self.last_model_index = 0
297
298 def _prepare_reward_value(self, reward_value_context):
299 """prepare reward and value targets from the context of rewards and values

Callers

nothing calls this directly

Calls 1

get_uniform_networkMethod · 0.45

Tested by

no test coverage detected