Run "scripts/submit_gym.py" fails about rllab HOT 19 CLOSED

Hi @Alex-zhai, can you paste the full command you ran? You should pass the gym_log directory to the submit_gym.py command. For example this is the full command on my computer:

python scripts/submit_gym.py /Users/dementrock/research/rllab/data/local/experiment/experiment_2016_05_31_10_50_06_0001/gym_log

from rllab.

Yes, I passed the gym_logdirectory to the submit_gym.py command, and the problem is still exits.

python scripts/submit_gym.py /home/alex-zhai/Downloads/rllab-master1/data/local/experiment/experiment_2016_05_31_23_34_50_0001/gym_log

from rllab.

What files do you have under the gym_log folder?

from rllab.

The results produced by running example/trpo_gym.py .

from rllab.

Can you paste the results of running ls?

from rllab.

openaigym.episode_batch.None.10971.10971.stats.json openaigym.manifest.None.10971.manifest.json openaigym.video.None.10971.video000000.meta.json openaigym.video.None.10971.video000000.mp4 openaigym.video.None.10971.video000001.meta.json openaigym.video.None.10971.video000001.mp4 openaigym.video.None.10971.video000008.meta.json openaigym.video.None.10971.video000008.mp4 openaigym.video.None.10971.video000027.meta.json openaigym.video.None.10971.video000027.mp4 openaigym.video.None.10971.video000064.meta.json openaigym.video.None.10971.video000064.mp4 openaigym.video.None.10971.video000125.meta.json openaigym.video.None.10971.video000125.mp4 openaigym.video.None.10971.video000216.meta.json openaigym.video.None.10971.video000216.mp4 openaigym.video.None.10971.video000343.meta.json openaigym.video.None.10971.video000343.mp4 openaigym.video.None.10971.video000512.meta.json openaigym.video.None.10971.video000512.mp4 openaigym.video.None.10971.video000729.meta.json openaigym.video.None.10971.video000729.mp4

from rllab.

I'm sorry but I couldn't reproduce the issue locally. Can you check which commit you are using for the rllab repo, and check the version for openai gym? You can get that by running python -c "import gym; print gym.version.VERSION".

from rllab.

Also, what is the operating system you are using?

from rllab.

The gym version is 0.1.1 in the anaconda2. But in my own python, the version is 0.1.2.
And the operating system is 14.04 ubantu. So how to update the gym in anaconda2?

from rllab.

You can run pip install --upgrade git+https://github.com/openai/gym.git, but I don't think the gym version is the issue. What is the content of the file openaigym.manifest.None.10971.manifest.json?

from rllab.

{"env_info": {"env_id": "Pendulum-v0", "gym_version": "0.1.1"}, "stats": "openaigym.episode_batch.None.10971.10971.stats.json", "videos": []}

from rllab.

Ok I'm out of clues. Can you zip the content of the gym_log folder and send me an email? My email address is [email protected].

from rllab.

Ok , my email is [email protected]. Thank you!!!

from rllab.

Hi @Alex-zhai, I inspected the content you sent. It seems like the openaigym.episode_batch.None.10971.10971.stats.json file failed to record any episode data, which caused the error you saw. Another thing that seems abnormal is the monitor id, which should be an integer following openaigym.episode_batch but right now it is None. Unfortunately I could not reproduce such behavior on a Linux machine.

Did you notice anything wrong when running trpo_gym.py? Can you set n_itr to 1 (only run for 1 iteration), and paste the full log here?

from rllab.

(rllab)alex-zhai@Alex-zhai:~/Downloads/rllab-master1$ python examples/trpo_gym.py
Using gpu device 0: GeForce GTX 970 (CNMeM is disabled, CuDNN 4007)
/home/alex-zhai/anaconda2/envs/rllab/lib/python2.7/site-packages/theano/tensor/signal/downsample.py:6: UserWarning: downsample module has been moved to the theano.tensor.signal.pool module.
"downsample module has been moved to the theano.tensor.signal.pool module.")
python /home/alex-zhai/Downloads/rllab-master1/scripts/run_experiment_lite.py --n_parallel '1' --seed '1' --log_dir '/home/alex-zhai/Downloads/rllab-master1/data/local/experiment/experiment_2016_06_02_08_57_12_0001' --snapshot_mode 'last' --exp_name 'experiment_2016_06_02_08_57_12_0001' --args_data 'Y2NvcHlfcmVnCl9yZWNvbnN0cnVjdG9yCnAxCihjcmxsYWIubWlzYy5pbnN0cnVtZW50ClN0dWJNZXRob2RDYWxsCnAyCmNfX2J1aWx0aW5fXwpvYmplY3QKcDMKTnRScDQKKGRwNQpTJ19fYXJncycKcDYKKGcxCihjcmxsYWIubWlzYy5pbnN0cnVtZW50ClN0dWJPYmplY3QKcDcKZzMKTnRScDgKKGRwOQpTJ2FyZ3MnCnAxMAoodHNTJ3Byb3h5X2NsYXNzJwpwMTEKY3JsbGFiLmFsZ29zLnRycG8KVFJQTwpwMTIKc1Mna3dhcmdzJwpwMTMKKGRwMTQKUydiYXNlbGluZScKcDE1CmcxCihnNwpnMwpOdFJwMTYKKGRwMTcKZzEwCih0c2cxMQpjcmxsYWIuYmFzZWxpbmVzLmxpbmVhcl9mZWF0dXJlX2Jhc2VsaW5lCkxpbmVhckZlYXR1cmVCYXNlbGluZQpwMTgKc2cxMwooZHAxOQpTJ2Vudl9zcGVjJwpwMjAKZzEKKGNybGxhYi5taXNjLmluc3RydW1lbnQKU3R1YkF0dHIKcDIxCmczCk50UnAyMgooZHAyMwpTJ19vYmonCnAyNApnMQooZzcKZzMKTnRScDI1CihkcDI2CmcxMAoodHNnMTEKY3JsbGFiLmVudnMubm9ybWFsaXplZF9lbnYKTm9ybWFsaXplZEVudgpwMjcKc2cxMwooZHAyOApTJ2VudicKcDI5CmcxCihnNwpnMwpOdFJwMzAKKGRwMzEKZzEwCih0c2cxMQpjcmxsYWIuZW52cy5neW1fZW52Ckd5bUVudgpwMzIKc2cxMwooZHAzMwpTJ2Vudl9uYW1lJwpwMzQKUydQZW5kdWx1bS12MCcKcDM1CnNzYnNzYnNTJ19hdHRyX25hbWUnCnAzNgpTJ3NwZWMnCnAzNwpzYnNzYnNTJ2JhdGNoX3NpemUnCnAzOApJNDAwMApzUydkaXNjb3VudCcKcDM5CkYwLjk4OTk5OTk5OTk5OTk5OTk5CnNTJ3N0ZXBfc2l6ZScKcDQwCkYwLjAxCnNTJ25faXRyJwpwNDEKSTEKc2cyOQpnMjUKc1MncG9saWN5JwpwNDIKZzEKKGc3CmczCk50UnA0MwooZHA0NApnMTAKKHRzZzExCmNybGxhYi5wb2xpY2llcy5nYXVzc2lhbl9tbHBfcG9saWN5CkdhdXNzaWFuTUxQUG9saWN5CnA0NQpzZzEzCihkcDQ2CmcyMApnMQooZzIxCmczCk50UnA0NwooZHA0OApnMjQKZzI1CnNnMzYKZzM3CnNic1MnaGlkZGVuX3NpemVzJwpwNDkKKEk4Ckk4CnRwNTAKc3Nic1MnbWF4X3BhdGhfbGVuZ3RoJwpwNTEKZzEKKGcyMQpnMwpOdFJwNTIKKGRwNTMKZzI0CmcyNQpzZzM2ClMnaG9yaXpvbicKcDU0CnNic3NiUyd0cmFpbicKcDU1Cih0KGRwNTYKdHA1NwpzUydfX2t3YXJncycKcDU4CihkcDU5CnNiLg=='
/home/alex-zhai/anaconda2/envs/rllab/lib/python2.7/site-packages/theano/tensor/signal/downsample.py:6: UserWarning: downsample module has been moved to the theano.tensor.signal.pool module.
"downsample module has been moved to the theano.tensor.signal.pool module.")
using seed 1
using seed 1
[2016-06-02 08:57:13,261] Making new env: Pendulum-v0
2016-06-02 08:57:13.968611 CST | [experiment_2016_06_02_08_57_12_0001] Populating workers...
[2016-06-02 08:57:13,980] Making new env: Pendulum-v0
2016-06-02 08:57:14.099682 CST | [experiment_2016_06_02_08_57_12_0001] Populated
0% 100%
[ ][2016-06-02 08:57:14,296] Starting new video recorder writing to /home/alex-zhai/Downloads/rllab-master1/data/local/experiment/experiment_2016_06_02_08_57_12_0001/gym_log/openaigym.video.1.2808.video000000.mp4
avconv version 9.18-6:9.18-0ubuntu0.14.04.1, Copyright (c) 2000-2014 the Libav developers
built on Mar 16 2015 13:19:10 with gcc 4.8 (Ubuntu 4.8.2-19ubuntu1)
[# ] | ETA: 00:04:37[2016-06-02 08:57:28,966] Starting new video recorder writing to /home/alex-zhai/Downloads/rllab-master1/data/local/experiment/experiment_2016_06_02_08_57_12_0001/gym_log/openaigym.video.1.2808.video000001.mp4
avconv version 9.18-6:9.18-0ubuntu0.14.04.1, Copyright (c) 2000-2014 the Libav developers
built on Mar 16 2015 13:19:10 with gcc 4.8 (Ubuntu 4.8.2-19ubuntu1)
[############ ] | ETA: 00:00:42[2016-06-02 08:57:42,714] Starting new video recorder writing to /home/alex-zhai/Downloads/rllab-master1/data/local/experiment/experiment_2016_06_02_08_57_12_0001/gym_log/openaigym.video.1.2808.video000008.mp4
avconv version 9.18-6:9.18-0ubuntu0.14.04.1, Copyright (c) 2000-2014 the Libav developers
built on Mar 16 2015 13:19:10 with gcc 4.8 (Ubuntu 4.8.2-19ubuntu1)
[##############################] | ETA: 00:00:00
Total time elapsed: 00:00:42
2016-06-02 08:57:56.653851 CST | [experiment_2016_06_02_08_57_12_0001] itr #0 | fitting baseline...
2016-06-02 08:57:56.719935 CST | [experiment_2016_06_02_08_57_12_0001] itr #0 | fitted
=: Compiling function f_loss
done in 1.068 seconds
/home/alex-zhai/Downloads/rllab-master1/rllab/optimizers/conjugate_gradient_optimizer.py:146: VisibleDeprecationWarning: using a non-integer number instead of an integer will result in an error in the future
n_samples, n_samples * self._subsample_factor, replace=False)
2016-06-02 08:57:57.794056 CST | [experiment_2016_06_02_08_57_12_0001] itr #0 | computing loss before
2016-06-02 08:57:57.798731 CST | [experiment_2016_06_02_08_57_12_0001] itr #0 | performing update
2016-06-02 08:57:57.798881 CST | [experiment_2016_06_02_08_57_12_0001] itr #0 | computing descent direction
=: Compiling function f_grad
done in 1.266 seconds
=: Compiling function f_Hx_plain
done in 3.820 seconds
2016-06-02 08:58:02.926421 CST | [experiment_2016_06_02_08_57_12_0001] itr #0 | descent direction computed
=: Compiling function f_loss_constraint
done in 0.297 seconds
2016-06-02 08:58:03.230280 CST | [experiment_2016_06_02_08_57_12_0001] itr #0 | backtrack iters: 2
2016-06-02 08:58:03.230412 CST | [experiment_2016_06_02_08_57_12_0001] itr #0 | computing loss after
2016-06-02 08:58:03.230478 CST | [experiment_2016_06_02_08_57_12_0001] itr #0 | optimization finished
=: Compiling function constraint
done in 0.174 seconds
2016-06-02 08:58:03.409038 CST | [experiment_2016_06_02_08_57_12_0001] itr #0 | saving snapshot...
2016-06-02 08:58:03.410641 CST | [experiment_2016_06_02_08_57_12_0001] itr #0 | saved
2016-06-02 08:58:03.411401 CST | ----------------------- ---------------
2016-06-02 08:58:03.411506 CST | Iteration 0
2016-06-02 08:58:03.411598 CST | AverageDiscountedReturn -736.944
2016-06-02 08:58:03.411682 CST | AverageReturn -1763.87
2016-06-02 08:58:03.411740 CST | ExplainedVariance 2.16271e-13
2016-06-02 08:58:03.411821 CST | NumTrajs 20
2016-06-02 08:58:03.411900 CST | Entropy 1.41894
2016-06-02 08:58:03.411956 CST | Perplexity 4.13273
2016-06-02 08:58:03.412011 CST | StdReturn 67.0666
2016-06-02 08:58:03.412067 CST | MaxReturn -1613.78
2016-06-02 08:58:03.412123 CST | MinReturn -1855.37
2016-06-02 08:58:03.412179 CST | AveragePolicyStd 1
2016-06-02 08:58:03.412233 CST | LossBefore -6.67484e-10
2016-06-02 08:58:03.412289 CST | LossAfter -0.00232857
2016-06-02 08:58:03.412344 CST | MeanKL 0.00684979
2016-06-02 08:58:03.412398 CST | dLoss 0.00232857
2016-06-02 08:58:03.412467 CST | ----------------------- ---------------

***************************

Training finished! You can upload results to OpenAI Gym by running the following command:

python scripts/submit_gym.py /home/alex-zhai/Downloads/rllab-master1/data/local/experiment/experiment_2016_06_02_08_57_12_0001/gym_log

***************************

from rllab.

Yes, after I set n_iter equals 1. I submited the results successfully. So strange！！

from rllab.

What if you now try a larger n_itr?

from rllab.

Yes, i tried n_itr=10 ,n_itr = 50, everything is OK now! Thank you very much!!!

from rllab.

Ok awesome! Might be just a very rare error then. Please let us know if the error occurs again.

from rllab.