我的代码是：
config = {
"num_workers" : 19,
#"num_gpus": 1,
"gamma" : tune.grid_search([0, 0.2, 0.4, 0.6, 0.8, 1]),
"lr" : tune.grid_search([1, 0.1, 0.01, 0.001, 0.0001, 0.00001, 0.000001])}
和：
analysis = tune.run(config=config)
当我运行这个时，我有：
Number of trials: 23/42 (22 PENDING, 1 RUNNING)
+----------------------------+----------+------------------------+---------+-------+--------+------------------+--------+----------+----------------------+----------------------+--------------------+
| Trial name | status | loc | gamma | lr | iter | total time (s) | ts | reward | episode_reward_max | episode_reward_min | episode_len_mean |
|----------------------------+----------+------------------------+---------+-------+--------+------------------+--------+----------+----------------------+----------------------+--------------------|
| A2C_TradingEnv_b9572_00000 | RUNNING | 192.168.252.130:361637 | 0 | 1 | 33 | 326.923 | 335920 | nan | nan | nan | nan |
| A2C_TradingEnv_b9572_00001 | PENDING | | 0.2 | 1 | | | | | | | |
| A2C_TradingEnv_b9572_00002 | PENDING | | 0.4 | 1 | | | | | | | |
| A2C_TradingEnv_b9572_00003 | PENDING | | 0.6 | 1 | | | | | | | |
| A2C_TradingEnv_b9572_00004 | PENDING | | 0.8 | 1 | | | | | | | |
| A2C_TradingEnv_b9572_00005 | PENDING | | 1 | 1 | | | | | | | |
| A2C_TradingEnv_b9572_00006 | PENDING | | 0 | 0.1 | | | | | | | |
| A2C_TradingEnv_b9572_00007 | PENDING | | 0.2 | 0.1 | | | | | | | |
| A2C_TradingEnv_b9572_00008 | PENDING | | 0.4 | 0.1 | | | | | | | |
| A2C_TradingEnv_b9572_00009 | PENDING | | 0.6 | 0.1 | | | | | | | |
| A2C_TradingEnv_b9572_00010 | PENDING | | 0.8 | 0.1 | | | | | | | |
| A2C_TradingEnv_b9572_00011 | PENDING | | 1 | 0.1 | | | | | | | |
| A2C_TradingEnv_b9572_00012 | PENDING | | 0 | 0.01 | | | | | | | |
| A2C_TradingEnv_b9572_00013 | PENDING | | 0.2 | 0.01 | | | | | | | |
| A2C_TradingEnv_b9572_00014 | PENDING | | 0.4 | 0.01 | | | | | | | |
| A2C_TradingEnv_b9572_00015 | PENDING | | 0.6 | 0.01 | | | | | | | |
| A2C_TradingEnv_b9572_00016 | PENDING | | 0.8 | 0.01 | | | | | | | |
| A2C_TradingEnv_b9572_00017 | PENDING | | 1 | 0.01 | | | | | | | |
| A2C_TradingEnv_b9572_00018 | PENDING | | 0 | 0.001 | | | | | | | |
| A2C_TradingEnv_b9572_00019 | PENDING | | 0.2 | 0.001 | | | | | | | |
+----------------------------+----------+------------------------+---------+-------+--------+------------------+--------+----------+----------------------+----------------------+--------------------+
... 3 more trials not shown (3 PENDING)
所以只有一个试验在进行。我想同时进行多项试验。当我想用在单个CPU上运行每个试用版时
analysis = tune.run(
config=config,
resources_per_trial = {"cpu": 1, "gpu": 0})
我有错误：
Exception has occurred: ValueError
Resources for <class 'ray.rllib.agents.trainer_template.A2C'> have been automatically set to <ray.tune.utils.placement_groups.PlacementGroupFactory object at 0x7fe119c3f7c0> by its `default_resource_request()` method. Please clear the `resources_per_trial` option.
我应该如何进行多个并行试运行，每个试运行有1个CPU？
我现在面临着同样的问题，但在我更改了{ ..., "num_workers": 1, ... }之后，试验仍然是一个。
config_train = {
"train_batch_size": args.batch_size,
"horizon": args.horizon,
"model": { "fcnet_hiddens": model_structure },
"num_workers": args.num_workers,
"env_config": { "generalize": True,
"run_valid": False,
"env": args.env,
"mid_range_init": args.mid_range_init },
"framework": args.framework,
"episodes_per_batch": args.episodes_per_batch,
"seed" : args.seed,
"lr" : args.lr,
"num_gpus": 4,
"num_workers": 1
}
trials = tune.run_experiments( { args.experiment: {
"run": args.algo,
"checkpoint_freq": 5,
"keep_checkpoints_num": 1,
"local_dir": args.output_path,
"env": env,
"stop": { "episode_reward_mean": -0.02 },
"checkpoint_at_end": True,
"config": config_train
}
}
#progress_reporter=reporter
)
Resources requested: 2.0/40 CPUs, 4.0/4 GPUs, 0.0/73.24 GiB heap, 0.0/35.38 GiB objects (0.0/1.0 accelerator_type:GTX)
我应该怎么做才能进行多次平行试验？
Number of trials: 1/1 (1 RUNNING)