"ValueError: Callback[train] is not in the specification" in utils_elf.py

[alatyshe]

build by cmake go_game and after that try to run sh train_df.sh and get:

PID: 18580
========== Args ============
Loader: actor_only=False,list_file="/home/yuandong/local/go/go_gogod/train.lst",verbose=False,data_aug=-1,ratio_pre_moves=0,start_ratio_pre_moves=0.5,num_games_per_thread=5,move_cutoff=-1,mode="online",use_mcts=False,gpu=None
ContextArgs: num_games=512,batchsize=128,game_multi=None,T=1,eval=False,wait_per_group=False,num_collectors=0,verbose_comm=False,verbose_collector=False,mcts_threads=0,mcts_rollout_per_thread=1,mcts_verbose=False,mcts_save_tree_filename="",mcts_verbose_time=False,mcts_use_prior=False,mcts_pseudo_games=0,mcts_pick_method="most_visited"
MoreLabels: additional_labels=None
MultiplePrediction: multipred_no_backprop=False
Sampler: sample_policy="epsilon-greedy",greedy=False,epsilon=0.0,sample_nodes="pi,a"
ModelLoader: load=None,onload=None,omit_keys=None,no_bn=False,no_leaky_relu=False,num_layer=39,dim=128
ModelInterface: opt_method="adam",lr=0.001,adam_eps=0.001
Trainer: freq_update=1
Evaluator: keys_in_reply=""
Stats: trainer_stats="rewards"
ModelSaver: record_dir="./record",save_prefix="save",save_dir="./",latest_symlink="latest"
SingleProcessRun: num_minibatch=5000,num_episode=10000,tqdm=True
========== End of Args ============
#Game: 512
#Max_thread: 0
#Collectors: 0
T: 1
Wait per group: False
Maximal #moves (0 = no constraint): 0
#Threads: 0
#Rollout per thread: 1
Verbose: False, Verbose_time: False
Use prior: False
Persistent tree: False
#Pseudo game: 0
Pick method: most_visited

Loading /home/yuandong/local/go/go_gogod/train.lst failed!
Version:  6a769a02dc0ab11e5a7633c337b5d3ce0d0bf511_staged
Num Actions:  361
#recv_thread = 2
Group 0:
  Collector[0] Batchsize: 128 Info: [gid=0][T=1][name="human_actor"]
  Collector[1] Batchsize: 128 Info: [gid=1][T=1][name="human_actor"]
Group 1:
  Collector[2] Batchsize: 128 Info: [gid=2][T=1][name="actor"]
  Collector[3] Batchsize: 128 Info: [gid=3][T=1][name="actor"]

KEY HERE :  train
self IDX:  defaultdict(<class 'list'>, {'human_actor': [0, 1], 'actor': [2, 3]})
cb :  <bound method Trainer.train of <rlpytorch.trainer.trainer.Trainer object at 0x7fe6209eeac8>>
Traceback (most recent call last):
  File "train.py", line 29, in <module>
    GC.reg_callback("train", trainer.train)
  File "/home/ubuntu/arvi_dima/ELF/elf/utils_elf.py", line 325, in reg_callback
    raise ValueError("Callback[%s] is not in the specification" % key)
ValueError: Callback[train] is not in the specification

in key i had "train" argument and in dict i have:
defaultdict(<class 'list'>, {'human_actor': [0, 1], 'actor': [2, 3]})
Wth, and how can i fix it?

[fantianwen]

the same problem...wish someone's help.

[yuandong-tian]

Try adding GC.reg_callback("train", None) before GC.start() in your script.

[yuandong-tian]

Also ELF is updated and the codebase is here.