how to display and continue train?

[Jiang1206]

I'm very interested in this project and try to use it, but I'm not very clear about the code running. I encountered the following problems :

1. I try to run python -m maddpg_o/experiments/train_normal --display to display the render, but it not work,only training and not display. what should i do?

2. When the training process is interrupted, I try to add --restore to train_grassland_epc.sh to continue training, but the designated --load-dir can not continue training, is the designated XX right?

[lzh-awesome]

@Jiang1206 i have the same problems with you.i want to see the gif while training. i also find the param --display doesn't work, only training but not display.this repo use --save-gif-data to save gif-data,--render-gif to render gif,but error occurs as follow:
(DARL) lzh@lzh:~/epciclr2020-master$ bash train_grassland_att.sh
/home/lzh/software/anaconda3/envs/DARL/lib/python3.7/site-packages/tensorflow/python/framework/dtypes.py:526: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.
_np_qint8 = np.dtype([("qint8", np.int8, 1)])
/home/lzh/software/anaconda3/envs/DARL/lib/python3.7/site-packages/tensorflow/python/framework/dtypes.py:527: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.
_np_quint8 = np.dtype([("quint8", np.uint8, 1)])
/home/lzh/software/anaconda3/envs/DARL/lib/python3.7/site-packages/tensorflow/python/framework/dtypes.py:528: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.
_np_qint16 = np.dtype([("qint16", np.int16, 1)])
/home/lzh/software/anaconda3/envs/DARL/lib/python3.7/site-packages/tensorflow/python/framework/dtypes.py:529: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.
_np_quint16 = np.dtype([("quint16", np.uint16, 1)])
/home/lzh/software/anaconda3/envs/DARL/lib/python3.7/site-packages/tensorflow/python/framework/dtypes.py:530: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.
_np_qint32 = np.dtype([("qint32", np.int32, 1)])
/home/lzh/software/anaconda3/envs/DARL/lib/python3.7/site-packages/tensorflow/python/framework/dtypes.py:535: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.
np_resource = np.dtype([("resource", np.ubyte, 1)])
arglist Namespace(adv_num_units=None, adv_policy='maddpg', adv_share_weights=True, alpha=0.0, batch_size=1024, benchmark=False, benchmark_iters=10000, checkpoint_rate=1000, gamma=0.95, good_num_units=None, good_policy='att-maddpg', good_share_weights=True, load_dir='./test/grassland_att_maddpg_vs_maddpg_3-3', lr=0.01, map_size='normal', max_episode_len=25, n_cpu_per_agent=50, n_envs=100, no_wheel=False, num_adversaries=3, num_episodes=1000, num_food=4, num_good=3, num_units=32, render_gif=True, restore=False, save_dir='./result/grassland_att_maddpg_vs_maddpg_3-3', save_gif_data=False, save_rate=1000, save_summary=True, scenario='grassland', show_attention=False, sight=100.0, timeout=0.03, train_rate=100, use_gpu=False)
collaborate!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
step 0
<mpe_local.multiagent.core.Agent object at 0x7fcd7bdd6f90> 0.4684688118032705 0.16335185712704492
<mpe_local.multiagent.core.Agent object at 0x7fcd7bddf2d0> -0.10195838246557276 -0.39889553780498166
<mpe_local.multiagent.core.Agent object at 0x7fcd7bddffd0> -0.19384971553561936 -0.7719600924113763
<mpe_local.multiagent.core.Agent object at 0x7fcda0ac9f90> -0.3486847856606383 -0.058348183584626456
<mpe_local.multiagent.core.Agent object at 0x7fcd7be38e50> 0.4716939570295191 0.7397263145673387
<mpe_local.multiagent.core.Agent object at 0x7fcd7bde6590> -0.7219922968790204 0.04681808024141132
<mpe_local.multiagent.core.Landmark object at 0x7fcd7bde6a10> -0.441491092426334 -0.5367035254244059
<mpe_local.multiagent.core.Landmark object at 0x7fcd7bde6c90> 0.20347805705033506 0.2878178351567865
<mpe_local.multiagent.core.Landmark object at 0x7fcd7bde6ed0> -0.6017387856782823 0.42144077074574693
<mpe_local.multiagent.core.Landmark object at 0x7fcd7bded150> -0.10416289694013015 -0.833392872054225
Traceback (most recent call last):
File "/home/lzh/software/anaconda3/envs/DARL/lib/python3.7/runpy.py", line 193, in _run_module_as_main
"main", mod_spec)
File "/home/lzh/software/anaconda3/envs/DARL/lib/python3.7/runpy.py", line 85, in _run_code
exec(code, run_globals)
File "/home/lzh/epciclr2020-master/maddpg_o/experiments/train_normal.py", line 6, in
train(arglist)
File "/home/lzh/epciclr2020-master/maddpg_o/experiments/train_helper/train_helpers.py", line 798, in train
frames = env.render_from_memory(gif_data, mode='rgb_array')
File "/home/lzh/epciclr2020-master/mpe_local/multiagent/environment.py", line 232, in render_from_memory
ret.append(self.render(mode=mode, world=world)[0])
File "/home/lzh/epciclr2020-master/mpe_local/multiagent/environment.py", line 342, in render
results.append(self.viewers[i].render(return_rgb_array = mode=='rgb_array')) # todo
File "/home/lzh/epciclr2020-master/mpe_local/multiagent/rendering.py", line 104, in render
arr = np.fromstring(image_data.data, dtype=np.uint8, sep='')
AttributeError: 'ImageData' object has no attribute 'data'
please let me know if you solve this problem!!!!thxs

[Aib-star]

Has this problem been solved?