Exporting RL Neural Net To DofBot

Hi,

I am working through the training example from an Nvidia repo for training a Dofbot to reach.

I am wondering how I will be able to export the saved .pth to an actual dofbot and run it. What kinds of inputs to the loaded model from the .pth will I need to give it, and what will the output be?

I’m aware that I may be able to use PyTorch to load the model, but I don’t know what kinds of inputs I will need to give it.

In the Config for this task the observation type is set to full. What exactly does the “full” observation look like?

So I’ve found how how to export the model for the dofbot using this code:


https://github.com/Denys88/rl_games/blob/master/notebooks/train_and_export_onnx_example_continuous.ipynb

so I included in the training script from the Dofbot repo to train the model in rl_games train:

def run(self):
    class ModelWrapper(torch.nn.Module):
        '''
        Main idea is to ignore outputs which we don't need from model
        '''

        def __init__(self, model):
            torch.nn.Module.__init__(self)
            self._model = model

        def forward(self, input_dict):
            input_dict['obs'] = self._model.norm_obs(input_dict['obs'])
            '''
            just model export doesn't work. Looks like onnx issue with torch distributions
            thats why we are exporting only neural network
            '''
            # print(input_dict)
            # output_dict = self._model.a2c_network(input_dict)
            # input_dict['is_train'] = False
            # return output_dict['logits'], output_dict['values']
            return self._model.a2c_network(input_dict)

    # create runner and set the settings
    runner = Runner(RLGPUAlgoObserver())
    runner.load(self.rlg_config_dict)
    runner.reset()

    # dump config dict
    experiment_dir = os.path.join('runs', self.cfg.train.params.config.name)
    os.makedirs(experiment_dir, exist_ok=True)
    with open(os.path.join(experiment_dir, 'config.yaml'), 'w') as f:
        f.write(OmegaConf.to_yaml(self.cfg))


    agent = runner.create_player()
    agent.restore('/home/dylan/Desktop/repos/OmniIsaacGymEnvs-DofbotReacher/runs/DofbotReacher/nn/DofbotReacher.pth')

    inputs = {
        'obs': torch.zeros((1,) + agent.obs_shape).to(agent.device),
        'rnn_states': agent.states,
    }

    with torch.no_grad():
        adapter = flatten.TracingAdapter(ModelWrapper(agent.model), inputs, allow_non_tensor=True)
        traced = torch.jit.trace(adapter, adapter.flattened_inputs, check_trace=False)
        flattened_outputs = traced(*adapter.flattened_inputs)
        print(flattened_outputs)

    torch.onnx.export(traced, *adapter.flattened_inputs, "dofbotreacherDefault.onnx", verbose=True, input_names=['obs'],
                      output_names=['mu', 'log_std', 'value'])

    runner.run({
        'train': not self.cfg.test,
        'play': self.cfg.test,
        'checkpoint': self.cfg.checkpoint,
        'sigma': None
    })

But when I run this in a notebook to compare with the output of the model in Isaac Sim I am getting different results:


while True:
    input_data = np.concatenate((jointPos, jointVel, goalPos, goalRot, goalRotRel, prevAct), dtype=np.float32)
    input_dict = {input_name: input_data.reshape(1, -1)}
    actions = []
    output = sess.run(None, input_dict)
    # Run the model on the input data
    mus = output[0][0]
    sigmas = output[1][0]
    for mu, sigma in zip(mus, sigmas):
        #print("mu is " + str(mu))
        #print("sigma is " + str(sigma))
        sigma = np.exp(sigma)
        action = np.random.normal(mu, sigma)
        action = clip_actions(action)
        #print(str(action))
        actions.append(action)
        # Print the output
    prevAct = np.array(actions, dtype=np.float32)
    joinPos = prevAct
    mus = clip_actions(mus)
    print("From Mu: " + str(mus))
    print("Sampled Actions: " + str(actions)+"\n")

My results from exported ONNX:


From Mu: [-1.         -1.         -1.         -0.8046344   1.          0.03140484]
Sampled Actions: [-1.0, -1.0, -1.0, -0.9712634764535495, 1.0, 0.13085964262265282]

From Mu: [-1.        -1.        -1.        -0.7526153  0.3962637 -0.4832963]
Sampled Actions: [-1.0, -1.0, -1.0, -0.673332441238304, 0.4277837343190856, -0.8004600149667951]

From Mu: [-1.         -1.         -1.         -0.7781227   1.          0.01266351]
Sampled Actions: [-1.0, -1.0, -1.0, -0.9150055130283129, 1.0, 0.5220340378975656]

From Mu: [-1.         -1.         -1.         -0.75321215  0.4043142  -0.48851803]
Sampled Actions: [-1.0, -1.0, -1.0, -0.740595988656948, 0.6001872415307878, -0.43890275166219206]

From Mu: [-1.         -1.         -1.         -0.79603606  1.          0.02533225]
Sampled Actions: [-1.0, -1.0, -1.0, -0.7589437855821842, 0.9001211409978054, 0.45835817870457995]

From Mu: [-1.         -1.         -1.         -0.7440234   0.44310746 -0.50922453]
Sampled Actions: [-1.0, -1.0, -1.0, -0.7435969780643082, 0.12811670977753048, -0.7244098554326204]

From Mu: [-1.         -1.         -1.         -0.7519526   1.          0.01125173]
Sampled Actions: [-1.0, -1.0, -1.0, -0.7532973499744635, 1.0, -0.40081315228268555]

From Mu: [-1.         -1.         -1.         -0.83088285  0.99210644  0.0389543 ]
Sampled Actions: [-1.0, -1.0, -1.0, -0.8324133591496548, 0.706666445481611, 0.19869922991846953]

From Mu: [-1.         -1.         -1.         -0.7238199   0.46477643 -0.5139754 ]
Sampled Actions: [-1.0, -1.0, -1.0, -0.5591049802680348, 0.4525925046550397, -0.8477365401897607]

From Mu: [-1.0000000e+00 -1.0000000e+00 -1.0000000e+00 -7.7655083e-01
  1.0000000e+00  8.5781515e-04]
Sampled Actions: [-1.0, -1.0, -1.0, -0.9417623079946027, 1.0, 0.19555459127998726]

From Mu: [-1.         -1.         -1.         -0.752935    0.40048864 -0.4860393 ]
Sampled Actions: [-1.0, -1.0, -1.0, -0.7758784773062144, 0.2824237395466518, -0.6451741533484725]

From Mu: [-1.        -1.        -1.        -0.7677086  1.         0.0195474]
Sampled Actions: [-1.0, -1.0, -1.0, -0.7224911522708941, 1.0, 0.013253213240240352]

From Mu: [-1.         -1.         -1.         -0.76368755  0.4935695  -0.4761499 ]
Sampled Actions: [-1.0, -1.0, -1.0, -0.711067045811111, 0.45817385742559213, -0.33301126978630297]

From Mu: [-1.         -1.         -1.         -0.78214306  1.          0.01774313]
Sampled Actions: [-1.0, -1.0, -1.0, -0.8200303164566594, 1.0, 0.06691762403021183]

From Mu: [-1.         -1.         -1.         -0.7540979   0.41784108 -0.49724796]
Sampled Actions: [-1.0, -1.0, -1.0, -0.971417280472688, 0.24631265870297667, -0.420806805659078]

From Mu: [-1.         -1.         -1.         -0.7697392   1.          0.04017575]
Sampled Actions: [-1.0, -1.0, -1.0, -0.8351657458525144, 1.0, -0.3143587497067801]

From Mu: [-1.         -1.         -1.         -0.83230096  0.9839373   0.04801101]
Sampled Actions: [-1.0, -1.0, -1.0, -0.6531447447963699, 1.0, -0.5401700070822363]

From Mu: [-1.         -1.         -1.         -0.82877606  1.          0.02789856]
Sampled Actions: [-1.0, -1.0, -1.0, -0.7325868211515937, 1.0, -0.2056215451291925]

From Mu: [-1.         -1.         -1.         -0.8304811   0.99420536  0.03666638]
Sampled Actions: [-0.9596524949814464, -1.0, -1.0, -0.727783274374398, 1.0, -0.022099674545825593]

From Mu: [-1.         -1.         -1.         -0.83203524  0.98964196  0.02937362]
Sampled Actions: [-1.0, -1.0, -1.0, -0.9240667476074332, 1.0, 0.14339611663589916]

From Mu: [-1.         -1.         -1.         -0.75311977  0.4030195  -0.4876795 ]
Sampled Actions: [-1.0, -1.0, -1.0, -0.6717779185202091, -0.18275692793347031, -0.36308104834416527]

From Mu: [-1.         -1.         -1.         -0.7185393   1.         -0.00630318]
Sampled Actions: [-1.0, -1.0, -1.0, -0.6100275823929414, 1.0, 0.016816065477693667]

From Mu: [-1.        -1.        -1.        -0.7555036  0.4474652 -0.51618  ]
Sampled Actions: [-1.0, -1.0, -1.0, -0.7180995217046809, 1.0, -0.13469952553495773]

From Mu: [-1.         -1.         -1.         -0.8301895   0.99568105  0.03506646]
Sampled Actions: [-1.0, -1.0, -1.0, -0.7588848485509612, 0.9381901139756421, 0.31158853516230894]

From Mu: [-1.         -1.         -1.         -0.74803036  0.4368003  -0.50681484]
Sampled Actions: [-1.0, -1.0, -1.0, -0.7154093862788289, 0.8633711819668368, -1.0]

From Mu: [-1.         -1.         -1.         -0.81857425  1.          0.03061933]
Sampled Actions: [-1.0, -1.0, -1.0, -0.9605380108456176, 1.0, 0.224580063371297]

From Mu: [-1.         -1.         -1.         -0.75273305  0.39780074 -0.48429456]
Sampled Actions: [-1.0, -1.0, -1.0, -0.7599399310258989, 0.683415490179879, 0.10554475773474592]

From Mu: [-1.         -1.         -1.         -0.7216615   0.47854942 -0.5221823 ]
Sampled Actions: [-1.0, -1.0, -1.0, -0.8500937557664596, 0.44775010561508727, -0.062483218958119646]

From Mu: [-1.         -1.         -1.         -0.78516716  1.          0.0327966 ]
Sampled Actions: [-1.0, -1.0, -1.0, -0.6297888878813683, 0.8676291846536264, -0.044572628309073714]

From Mu: [-1.         -1.         -1.         -0.8168054   1.          0.02131866]
Sampled Actions: [-1.0, -1.0, -1.0, -0.9869841126204636, 0.749818234488756, -0.0865036930558632]

From Mu: [-1.         -1.         -1.         -0.81412536  1.          0.0572635 ]
Sampled Actions: [-1.0, -1.0, -1.0, -0.8774458476527681, 0.9262720416298681, -0.3105193372811678]

From Mu: [-1.         -1.         -1.         -0.8269616   0.99148035  0.05044907]
Sampled Actions: [-1.0, -1.0, -1.0, -0.956779717342482, 1.0, -0.09833272962712514]

From Mu: [-1.         -1.         -1.         -0.8339178   0.97218335  0.06150183]
Sampled Actions: [-1.0, -1.0, -1.0, -0.8847926518222826, 0.7811633147768765, 0.6468734889113454]

From Mu: [-1.         -1.         -1.         -0.7310789   0.4452337  -0.50467825]
Sampled Actions: [-1.0, -1.0, -1.0, -0.7219534542825735, 0.6578086633571447, -0.5079010698438633]

From Mu: [-1.         -1.         -1.         -0.80070335  1.          0.02505435]
Sampled Actions: [-1.0, -1.0, -1.0, -0.7067711793896203, 1.0, 0.04010780364819768]

From Mu: [-1.         -1.         -1.         -0.75495094  0.43386662 -0.50752133]
Sampled Actions: [-1.0, -1.0, -1.0, -0.734413969039702, 0.5941167050192434, -0.5086914582627033]

From Mu: [-1.         -1.         -1.         -0.79531425  1.          0.02446295]
Sampled Actions: [-1.0, -1.0, -1.0, -0.8421573316399269, 1.0, 0.30989272641437204]

From Mu: [-1.         -1.         -1.         -0.75390536  0.4146967  -0.49522403]
Sampled Actions: [-1.0, -1.0, -1.0, -0.9163837434115838, 0.31018013332085653, -0.6052221552721155]

From Mu: [-1.         -1.         -1.         -0.77425617  1.          0.03597975]
Sampled Actions: [-1.0, -1.0, -1.0, -0.6653456375410538, 1.0, -0.06353721821121802]

From Mu: [-1.         -1.         -1.         -0.8290542   1.          0.02924436]
Sampled Actions: [-1.0, -1.0, -1.0, -0.6330052804822425, 0.29690473715661647, -0.2526023832207899]

From Mu: [-1.         -1.         -1.         -0.7643934   1.          0.00425477]
Sampled Actions: [-1.0, -1.0, -1.0, -0.644647362984077, 1.0, -0.039213324141717576]

From Mu: [-1.         -1.         -1.         -0.82857746  1.          0.02696133]
Sampled Actions: [-1.0, -1.0, -1.0, -0.6490890462102662, 0.9928362688072035, -0.020496762619332073]

From Mu: [-1.         -1.         -1.         -0.8280749   1.          0.02723466]
Sampled Actions: [-1.0, -1.0, -1.0, -0.7456473961946039, 1.0, -0.19916520520425451]

From Mu: [-1.         -1.         -1.         -0.83073646  0.99288005  0.03810912]
Sampled Actions: [-1.0, -1.0, -1.0, -0.6872249670162497, 1.0, -0.1842744928744335]

From Mu: [-1.         -1.         -1.         -0.82953936  0.99884725  0.03165848]
Sampled Actions: [-1.0, -1.0, -1.0, -0.7027429862791732, 1.0, 0.1565127741614349]

From Mu: [-1.         -1.         -1.         -0.75497705  0.43443447 -0.5078842 ]
Sampled Actions: [-1.0, -1.0, -1.0, -0.9300769908160136, 0.5547780652801778, -0.15824015832012317]

From Mu: [-1.         -1.         -1.         -0.7966184   1.          0.04495192]
Sampled Actions: [-1.0, -1.0, -1.0, -0.9351516547556644, 1.0, 0.41995618692489717]

From Mu: [-1.         -1.         -1.         -0.75300497  0.4014346  -0.48665246]
Sampled Actions: [-1.0, -1.0, -1.0, -0.799439295000151, 0.33643433600371603, -0.1238350057988532]

From Mu: [-1.         -1.         -1.         -0.7734776   1.          0.02380122]
Sampled Actions: [-1.0, -1.0, -1.0, -0.7137572110777022, 1.0, 0.03234684937291753]

From Mu: [-1.         -1.         -1.         -0.75490457  0.4328815  -0.50689197]
Sampled Actions: [-1.0, -1.0, -1.0, -0.6739407059979791, 0.3401257993163045, -0.5932283956507928]

From Mu: [-1.         -1.         -1.         -0.76990277  1.          0.01006894]
Sampled Actions: [-1.0, -1.0, -1.0, -0.9218036317157808, 1.0, 0.19921441468520712]

From Mu: [-1.         -1.         -1.         -0.753143    0.4033429  -0.48788902]
Sampled Actions: [-1.0, -1.0, -1.0, -0.8677409427333201, 0.10603287720424603, -1.0]

From Mu: [-1.         -1.         -1.         -0.7537337   1.          0.02431748]
Sampled Actions: [-1.0, -1.0, -1.0, -0.7987103700020154, 0.9748768000782158, -0.19254955620029196]

From Mu: [-1.         -1.         -1.         -0.82964367  0.99151677  0.04321043]
Sampled Actions: [-1.0, -1.0, -1.0, -0.7772603301650468, 1.0, 0.08330701284505027]

From Mu: [-1.         -1.         -1.         -0.7544461   0.42390615 -0.5011452 ]
Sampled Actions: [-1.0, -1.0, -1.0, -0.7215235033994709, 0.7475780753105198, -0.4463962371186046]

From Mu: [-1.         -1.         -1.         -0.8086596   1.          0.02776166]
Sampled Actions: [-1.0, -1.0, -1.0, -0.7416580639653408, 0.7908667095687227, -0.7007411216283626]

From Mu: [-1.         -1.         -1.         -0.8129317   1.          0.03130444]
Sampled Actions: [-1.0, -1.0, -1.0, -0.9641518084761704, 0.7152171875335328, -0.1790646387606805]

From Mu: [-1.        -1.        -1.        -0.8109068  1.         0.0536619]
Sampled Actions: [-1.0, -1.0, -1.0, -0.9449071105724618, 0.9214226188393199, 0.3830525353828998]

From Mu: [-1.         -1.         -1.         -0.7448434   0.41333145 -0.49061635]
Sampled Actions: [-1.0, -1.0, -1.0, -0.6139587415753119, 0.27964950178425085, -0.805540133657932]

From Mu: [-1.         -1.         -1.         -0.7620633   1.          0.00163948]
Sampled Actions: [-1.0, -1.0, -1.0, -0.7513908686503362, 1.0, 0.23165483971241815]

From Mu: [-1.         -1.         -1.         -0.754641    0.42756668 -0.5034921 ]
Sampled Actions: [-1.0, -1.0, -1.0, -0.9274740125963947, 0.5669085915310174, -0.6694073326868109]

From Mu: [-1.         -1.         -1.         -0.79761654  1.          0.04503421]
Sampled Actions: [-1.0, -1.0, -1.0, -0.829177380456725, 0.9792390638350275, 0.7460004057255771]

From Mu: [-1.         -1.         -1.         -0.7518366   0.4200245  -0.49774817]
Sampled Actions: [-1.0, -1.0, -1.0, -0.7531855640293652, 0.3724516385546356, -0.81092287920861]

From Mu: [-1.         -1.         -1.         -0.77546066  1.          0.01978657]
Sampled Actions: [-1.0, -1.0, -1.0, -0.8006624931542228, 1.0, -0.2035726014573656]

From Mu: [-1.         -1.         -1.         -0.831736    0.9873554   0.04419161]
Sampled Actions: [-1.0, -1.0, -1.0, -0.8211956396909531, 1.0, 0.3126758327660015]

From Mu: [-1.         -1.         -1.         -0.75408775  0.41767544 -0.497142  ]
Sampled Actions: [-1.0, -1.0, -1.0, -0.6773146200950056, 0.23898131031657155, -0.00888535581840777]

From Mu: [-1.         -1.         -1.         -0.7500414   1.         -0.12462715]
Sampled Actions: [-1.0, -1.0, -1.0, -0.7044294522807475, 1.0, -0.37893207805010704]

From Mu: [-1.         -1.         -1.         -0.8299064   0.9970793   0.03355742]
Sampled Actions: [-1.0, -1.0, -1.0, -0.8042123596741024, 1.0, 0.02978162273492279]

From Mu: [-1.         -1.         -1.         -0.7542301   0.42008585 -0.49869195]
Sampled Actions: [-1.0, -1.0, -1.0, -0.7994773223681491, 0.35195154559923064, -0.6513128507643774]

From Mu: [-1.         -1.         -1.         -0.77492505  1.          0.02427842]
Sampled Actions: [-1.0, -1.0, -1.0, -0.8146382622800785, 1.0, 0.4632515711692034]

From Mu: [-1.         -1.         -1.         -0.7541432   0.41860655 -0.49774054]
Sampled Actions: [-1.0, -1.0, -1.0, -0.6399486189712743, 0.47991460152377413, -0.6106060048915335]

From Mu: [-1.        -1.        -1.        -0.7818957  1.         0.0105686]
Sampled Actions: [-1.0, -1.0, -1.0, -0.7521512450297217, 1.0, -0.101485989421196]

From Mu: [-1.         -1.         -1.         -0.8308614   0.9922222   0.03882762]
Sampled Actions: [-1.0, -1.0, -1.0, -0.8318709297241236, 0.7954753471454886, 0.18972583585456484]

From Mu: [-1.         -1.         -1.         -0.7328147   0.45023966 -0.50883967]
Sampled Actions: [-1.0, -1.0, -1.0, -0.7645353328437561, 0.8173021942662632, -0.4628886097273362]

From Mu: [-1.         -1.         -1.         -0.8157343   1.          0.03463554]
Sampled Actions: [-1.0, -1.0, -1.0, -0.8481815342708638, 1.0, 0.03233081407351638]

From Mu: [-1.        -1.        -1.        -0.7538517  0.4138401 -0.4946716]
Sampled Actions: [-1.0, -1.0, -1.0, -0.8667782367159919, 0.5177156208302721, -0.3209698993271739]

From Mu: [-1.         -1.         -1.         -0.7919213   1.          0.03677167]
Sampled Actions: [-1.0, -1.0, -1.0, -0.7166605067083957, 0.7848380977631081, -0.903113004857943]

From Mu: [-1.        -1.        -1.        -0.8118021  1.         0.0283642]
Sampled Actions: [-1.0, -1.0, -1.0, -0.7483624175285033, 0.9690469629087698, 0.3157288374427686]

From Mu: [-1.         -1.         -1.         -0.75136656  0.43315125 -0.50578666]
Sampled Actions: [-1.0, -1.0, -1.0, -0.7228498155618657, 0.8739042277649753, -0.8068623520948508]

From Mu: [-1.        -1.        -1.        -0.8196482  1.         0.0317601]
Sampled Actions: [-1.0, -1.0, -1.0, -0.7101558499317228, 1.0, 0.11094376540214405]

From Mu: [-1.         -1.         -1.         -0.7549284   0.43338925 -0.5072165 ]
Sampled Actions: [-1.0, -1.0, -1.0, -0.7380984420750469, 0.19013209276868123, -0.4474742833012317]

From Mu: [-1.         -1.         -1.         -0.7577391   1.          0.01255114]
Sampled Actions: [-1.0, -1.0, -1.0, -0.7153524683163135, 1.0, 0.42582526401823645]

From Mu: [-1.         -1.         -1.         -0.7548941   0.43265638 -0.5067481 ]
Sampled Actions: [-1.0, -1.0, -1.0, -0.8254276575165649, 0.7881192382061699, -0.5197676868750642]

From Mu: [-1.         -1.         -1.         -0.8145303   1.          0.04047535]
Sampled Actions: [-1.0, -1.0, -1.0, -0.8888925976178152, 0.5152770032077211, -1.0]

From Mu: [-1.         -1.         -1.         -0.7922334   1.          0.03915513]
Sampled Actions: [-1.0, -1.0, -1.0, -0.7928917736744503, 1.0, 0.7313546877440096]

From Mu: [-1.         -1.         -1.         -0.75432247  0.4216915  -0.4997233 ]
Sampled Actions: [-1.0, -1.0, -1.0, -0.7441526863909724, 0.38976482033239357, -0.06913269820327372]

From Mu: [-1.         -1.         -1.         -0.77680117  1.          0.0193164 ]
Sampled Actions: [-1.0, -1.0, -1.0, -0.7109446039021524, 1.0, 0.0463337171791479]

From Mu: [-1.         -1.         -1.         -0.7549233   0.43327805 -0.50714546]
Sampled Actions: [-1.0, -1.0, -1.0, -0.9411369457112222, 0.6229244546911966, -0.002945034717867112]

From Mu: [-1.         -1.         -1.         -0.7737288   0.87156916 -0.20136413]
Sampled Actions: [-1.0, -1.0, -1.0, -0.8333429132267878, 0.7866244967082299, -0.40523307862356206]

From Mu: [-1.         -1.         -1.         -0.8145625   1.          0.04130588]
Sampled Actions: [-1.0, -1.0, -1.0, -0.8137097281918882, 0.8703228699255615, -0.11254037521151103]

From Mu: [-1.         -1.         -1.         -0.8212401   1.          0.04168546]
Sampled Actions: [-1.0, -1.0, -1.0, -0.8345495408115013, 0.964417183788387, -0.08616398051944929]

From Mu: [-1.         -1.         -1.         -0.8293955   0.989624    0.04685815]
Sampled Actions: [-1.0, -1.0, -1.0, -0.9619933018895461, 0.9294147044207864, 0.26190701677129324]

From Mu: [-1.         -1.         -1.         -0.7454954   0.40955535 -0.4885219 ]
Sampled Actions: [-1.0, -1.0, -1.0, -0.7298318947599646, 0.7249129003673476, -0.44874197696222995]

From Mu: [-1.         -1.         -1.         -0.80687493  1.          0.0279835 ]
Sampled Actions: [-1.0, -1.0, -1.0, -0.9172371300540336, 1.0, -0.11950898063572937]

From Mu: [-1.         -1.         -1.         -0.83345515  0.97595435  0.05710997]
Sampled Actions: [-1.0, -1.0, -1.0, -0.7520426333517087, 0.7108703606577311, -0.2643116840263049]

From Mu: [-1.         -1.         -1.         -0.80619746  1.          0.03000315]
Sampled Actions: [-1.0, -1.0, -1.0, -0.8700702838087512, 1.0, -0.5868225448990657]

From Mu: [-1.         -1.         -1.         -0.83282405  0.9805167   0.0518783 ]
Sampled Actions: [-1.0, -1.0, -1.0, -0.8459504345277016, 0.852532244547307, 0.21485077773330233]

From Mu: [-1.         -1.         -1.         -0.73856825  0.43881717 -0.5040919 ]
Sampled Actions: [-1.0, -1.0, -1.0, -0.6042425399210104, 0.38998532721734425, -0.6512789363508023]

From Mu: [-1.         -1.         -1.         -0.7722377   1.          0.00392138]
Sampled Actions: [-1.0, -1.0, -1.0, -0.8335485796836798, 1.0, 0.6453908524881675]

From Mu: [-1.         -1.         -1.         -0.7539808   0.41592062 -0.4960123 ]
Sampled Actions: [-1.0, -1.0, -1.0, -0.6598709852418955, 0.3168695489524249, -0.836174569712293]

From Mu: [-1.         -1.         -1.         -0.76722413  1.          0.00781376]
Sampled Actions: [-1.0, -1.0, -1.0, -0.8826941159915369, 0.9815507997243897, 0.09223066650737678]

From Mu: [-1.         -1.         -1.         -0.7516089   0.41203526 -0.49267414]
Sampled Actions: [-1.0, -1.0, -1.0, -0.7262354023240403, 0.611571876077073, -0.3748896004425093]

From Mu: [-1.         -1.         -1.         -0.7966631   1.          0.02409985]
Sampled Actions: [-1.0, -1.0, -1.0, -0.8437298752685033, 1.0, -0.21390895491626874]

From Mu: [-1.         -1.         -1.         -0.83243376  0.9830944   0.04895948]
Sampled Actions: [-1.0, -1.0, -1.0, -1.0, 0.5754243191292447, 0.23529930719799402]

From Mu: [-1.         -1.         -1.         -0.71029884  0.463244   -0.50490886]
Sampled Actions: [-1.0, -1.0, -1.0, -0.73022477601822, 0.022821628964901475, -0.5316323107425646]

From Mu: [-1.         -1.         -1.         -0.7412426   1.          0.00652762]
Sampled Actions: [-1.0, -1.0, -1.0, -0.7626166359492124, 1.0, -0.06862099196679089]

From Mu: [-1.         -1.         -1.         -0.83105797  0.99116623  0.03998423]
Sampled Actions: [-1.0, -1.0, -1.0, -0.9733124724446618, 0.37299390582550784, 0.44632602453678194]

From Mu: [-1.         -1.         -1.         -0.691082    0.49967584 -0.51800704]
Sampled Actions: [-1.0, -1.0, -1.0, -0.6466158853593572, 0.10894997588414462, -1.0]

From Mu: [-1.0000000e+00 -1.0000000e+00 -1.0000000e+00 -7.4669188e-01
  1.0000000e+00 -2.7045608e-06]
Sampled Actions: [-1.0, -1.0, -1.0, -0.8063693254883623, 1.0, 0.5692572123634391]

From Mu: [-1.         -1.         -1.         -0.7542121   0.41978022 -0.49849543]
Sampled Actions: [-1.0, -1.0, -1.0, -0.7238280431509154, 0.38645082434177647, -0.9589964281916483]

From Mu: [-1.         -1.         -1.         -0.77586526  1.          0.01697315]
Sampled Actions: [-1.0, -1.0, -1.0, -0.831368703825368, 1.0, -0.5139026712857898]

From Mu: [-1.         -1.         -1.         -0.8322415   0.9843118   0.04759049]
Sampled Actions: [-1.0, -1.0, -1.0, -0.7360280364090481, 0.9232017585720846, 0.5881593626744092]

From Mu: [-1.        -1.        -1.        -0.746584   0.4424964 -0.5098611]
Sampled Actions: [-1.0, -1.0, -1.0, -0.6488643998651968, 0.2686077691383384, -0.3111219097689356]

From Mu: [-1.         -1.         -1.         -0.76224035  1.          0.00513411]
Sampled Actions: [-1.0, -1.0, -1.0, -0.8595738078202146, 1.0, -0.512847792326814]

From Mu: [-1.         -1.         -1.         -0.8326718   0.98154134  0.05071499]
Sampled Actions: [-1.0, -1.0, -1.0, -0.8643575460178847, 0.862350834502499, -0.38874976606175027]

From Mu: [-1.         -1.         -1.         -0.8214912   1.          0.04705023]
Sampled Actions: [-1.0, -1.0, -1.0, -0.6588649259331928, 1.0, -0.5046605473797209]

From Mu: [-1.         -1.         -1.         -0.8289071   1.          0.02852955]
Sampled Actions: [-1.0, -1.0, -1.0, -0.8517433537320316, 1.0, 0.05951976909559868]

From Mu: [-1.         -1.         -1.         -0.75381917  0.41333315 -0.4943451 ]
Sampled Actions: [-1.0, -1.0, -1.0, -0.6605816309924912, 0.5240264577972836, -0.36094379048087666]

From Mu: [-1.         -1.         -1.         -0.7866599   1.          0.01418263]
Sampled Actions: [-1.0, -1.0, -1.0, -0.8125747178432652, 1.0, -0.6301311988732121]

From Mu: [-1.         -1.         -1.         -0.8319364   0.98617125  0.04550968]
Sampled Actions: [-1.0, -1.0, -1.0, -0.9180389179816161, 1.0, 0.28216419973977785]

From Mu: [-1.        -1.        -1.        -0.7531814  0.4038807 -0.4882373]
Sampled Actions: [-1.0, -1.0, -1.0, -0.6380547114189727, 0.03953814606399447, -0.2504484085064785]

From Mu: [-1.        -1.        -1.        -0.7395368  1.        -0.0030805]
Sampled Actions: [-1.0, -1.0, -1.0, -0.6973475279789967, 0.8115279369858396, 0.07558269950511493]

From Mu: [-1.         -1.         -1.         -0.735009    0.46627623 -0.520403  ]
Sampled Actions: [-1.0, -1.0, -1.0, -0.622644426546135, 0.4373438596924615, -0.08400922156111618]

From Mu: [-1.         -1.         -1.         -0.77733594  1.          0.00737545]
Sampled Actions: [-1.0, -1.0, -1.0, -0.7586348374902103, 0.7959879168119375, 0.17666705980918532]

From Mu: [-1.         -1.         -1.         -0.7331888   0.46033216 -0.51566833]
Sampled Actions: [-1.0, -1.0, -1.0, -0.5334251034131978, 0.3052061611582506, -0.5437402490291854]

From Mu: [-1.         -1.         -1.         -0.7615332   1.         -0.00640586]
Sampled Actions: [-1.0, -1.0, -1.0, -0.7560654529868349, 1.0, 0.539021922370829]

From Mu: [-1.        -1.        -1.        -0.7546068  0.4269057 -0.5030684]
Sampled Actions: [-1.0, -1.0, -1.0, -0.8423875899476196, 0.4653336632067847, -1.0]

From Mu: [-1.         -1.         -1.         -0.7865687   1.          0.03247565]
Sampled Actions: [-1.0, -1.0, -1.0, -0.7663988395721106, 1.0, -0.17325005915778452]

From Mu: [-1.         -1.         -1.         -0.83112794  0.99078536  0.04040235]
Sampled Actions: [-1.0, -1.0, -1.0, -0.9676105318614661, 0.8365514947440982, 0.5571944694536217]

From Mu: [-1.         -1.         -1.         -0.7360707   0.4243949  -0.49354514]
Sampled Actions: [-1.0, -1.0, -1.0, -0.7366622475797552, 0.390995301230382, -0.03810309075937229]

From Mu: [-1.         -1.         -1.         -0.77668625  1.          0.01852713]
Sampled Actions: [-1.0, -1.0, -1.0, -0.8130309624714379, 1.0, 0.44082986876843167]

From Mu: [-1.         -1.         -1.         -0.75415725  0.41883454 -0.49788752]
Sampled Actions: [-1.0, -1.0, -1.0, -0.6394138721748022, 0.5443130386856704, -0.4965374936603323]

From Mu: [-1.         -1.         -1.         -0.78786045  1.          0.01247414]
Sampled Actions: [-1.0, -1.0, -1.0, -0.8427138518191807, 1.0, 0.16638944786908302]

From Mu: [-1.         -1.         -1.         -0.75390023  0.41461757 -0.49517307]
Sampled Actions: [-1.0, -1.0, -1.0, -0.6390485261141253, 0.5640379344859947, -0.5100361164760574]

From Mu: [-1.         -1.         -1.         -0.78967327  1.          0.01303856]
Sampled Actions: [-1.0, -1.0, -1.0, -0.7034092600547694, 0.7546028299960432, 0.05844635706625497]

From Mu: [-1.         -1.         -1.         -0.72906953  0.47473878 -0.5232934 ]
Sampled Actions: [-1.0, -1.0, -1.0, -0.5826544421244291, 0.5793368480820541, -0.4019668633045341]

From Mu: [-1.         -1.         -1.         -0.78923744  1.          0.00731073]
Sampled Actions: [-1.0, -1.0, -1.0, -0.6381532763226888, 1.0, 0.3110264941862362]

From Mu: [-1.         -1.         -1.         -0.7553595   0.4435204  -0.51367366]
Sampled Actions: [-1.0, -1.0, -1.0, -0.7857085545448248, 0.494469254950458, -0.6016415142932836]

From Mu: [-1.         -1.         -1.         -0.78768235  1.          0.02708494]
Sampled Actions: [-1.0, -1.0, -1.0, -0.8083727740193192, 1.0, -0.24852696109124245]

From Mu: [-1.         -1.         -1.         -0.83186644  0.98658824  0.04504448]
Sampled Actions: [-1.0, -1.0, -1.0, -0.776346378269931, 1.0, -0.28950093764337215]

From Mu: [-1.         -1.         -1.         -0.8313095   0.98978597  0.04150212]
Sampled Actions: [-1.0, -1.0, -1.0, -0.6960893478644005, 0.9014041015396248, -0.22198072487502907]

From Mu: [-1.         -1.         -1.         -0.82138604  1.          0.02964587]
Sampled Actions: [-0.9768299492942232, -1.0, -1.0, -0.8759363600846799, 1.0, 0.12790216985630348]

From Mu: [-1.         -1.         -1.         -0.75762206  0.40363485 -0.49327156]
Sampled Actions: [-1.0, -1.0, -1.0, -0.7063905887776372, 0.20395591216468664, -0.486436042214649]

From Mu: [-1.         -1.         -1.         -0.75801116  1.          0.00948197]
Sampled Actions: [-1.0, -1.0, -1.0, -0.9542299599565807, 1.0, 0.11332216743685222]

From Mu: [-1.         -1.         -1.         -0.75280184  0.39870414 -0.4848815 ]
Sampled Actions: [-1.0, -1.0, -1.0, -0.6944751518937373, 0.20959305591299868, 0.04993119163857018]

From Mu: [-1.         -1.         -1.         -0.67443234  0.5622158  -0.55371433]
Sampled Actions: [-1.0, -1.0, -1.0, -0.665550245298805, 0.38887023213162863, -0.1318197420317252]

From Mu: [-1.         -1.         -1.         -0.7742209   1.          0.01062669]
Sampled Actions: [-1.0, -1.0, -1.0, -0.745085411438627, 0.9742014617465792, -0.4184609329492054]

From Mu: [-1.         -1.         -1.         -0.8285844   0.9970036   0.03726296]
Sampled Actions: [-1.0, -1.0, -1.0, -0.7205167559993333, 0.8334237806172715, -0.26492267600600583]

From Mu: [-1.         -1.         -1.         -0.81611484  1.          0.03027082]
Sampled Actions: [-1.0, -1.0, -1.0, -0.8613384701109654, 0.9797665377255256, 0.24349944851927657]

From Mu: [-1.         -1.         -1.         -0.7516162   0.41537166 -0.49475768]
Sampled Actions: [-1.0, -1.0, -1.0, -0.8702310431423083, 0.22695640724913071, -0.7334373026744053]

From Mu: [-1.         -1.         -1.         -0.765258    1.          0.02830486]
Sampled Actions: [-1.0, -1.0, -1.0, -0.7504270744637468, 1.0, -0.3195996561245299]

From Mu: [-1.         -1.         -1.         -0.83082813  0.9923965   0.03863757]
Sampled Actions: [-1.0, -1.0, -1.0, -0.8663857049559778, 0.5733008315297083, 0.7173743299851452]

From Mu: [-1.         -1.         -1.         -0.71049124  0.48186427 -0.51813126]
Sampled Actions: [-1.0, -1.0, -1.0, -0.8041422951433205, 0.5554909080842381, 0.20613152230278142]

From Mu: [-1.         -1.         -1.         -0.70880026  0.49321756 -0.5251183 ]
Sampled Actions: [-1.0, -1.0, -1.0, -0.614003760194867, 0.7653976137533172, -0.9266509000795795]

From Mu: [-1.         -1.         -1.         -0.80729204  1.          0.01647381]
Sampled Actions: [-1.0, -1.0, -1.0, -0.6881985439225536, 0.9390496989627011, -0.2663311783688894]

From Mu: [-1.         -1.         -1.         -0.8244148   1.          0.02991825]
Sampled Actions: [-1.0, -1.0, -1.0, -0.8310000061950585, 1.0, -0.31772355246662787]

From Mu: [-1.         -1.         -1.         -0.8322353   0.98434806  0.04754952]
Sampled Actions: [-1.0, -1.0, -1.0, -0.9037621099534194, 0.8333114817000511, -0.6050761211733053]

From Mu: [-1.         -1.         -1.         -0.81974214  1.          0.05053739]
Sampled Actions: [-1.0, -1.0, -1.0, -0.8060724344365541, 1.0, -0.17339006375675628]

From Mu: [-1.         -1.         -1.         -0.8318277   0.98681706  0.04479015]
Sampled Actions: [-1.0, -1.0, -1.0, -1.0, 0.8159709382943136, 0.024050554393333336]

From Mu: [-1.        -1.        -1.        -0.7337516  0.4232876 -0.4916418]
Sampled Actions: [-1.0, -1.0, -1.0, -0.7702245885232373, 0.5973745937048496, -0.26677264893490593]

My results from the IsaacSim:

new position tensor([[0.1949, 0.1820, 0.2187]], device='cuda:0')
tensor([[-0.7803, -1.0000, -1.0000, -1.0000,  1.0000, -0.5157]],
       device='cuda:0')
tensor([[-0.5791, -1.0000, -1.0000, -1.0000,  0.3410, -0.5124]],
       device='cuda:0')
tensor([[-0.4997, -1.0000, -1.0000, -1.0000,  0.2800, -0.5737]],
       device='cuda:0')
tensor([[-0.4843, -1.0000, -1.0000, -1.0000,  0.2260, -0.5793]],
       device='cuda:0')
tensor([[-0.5000, -1.0000, -1.0000, -1.0000,  0.1828, -0.5856]],
       device='cuda:0')
tensor([[-0.5323, -1.0000, -1.0000, -1.0000,  0.1400, -0.5825]],
       device='cuda:0')
tensor([[-0.5508, -1.0000, -1.0000, -1.0000,  0.1113, -0.6239]],
       device='cuda:0')
tensor([[-0.6025, -1.0000, -1.0000, -1.0000,  0.0443, -0.5531]],
       device='cuda:0')
tensor([[-0.5848, -1.0000, -1.0000, -1.0000,  0.0372, -0.6465]],
       device='cuda:0')
tensor([[-0.5442, -1.0000, -1.0000, -1.0000,  0.0082, -0.6643]],
       device='cuda:0')
tensor([[-0.4736, -1.0000, -1.0000, -1.0000, -0.0145, -0.7116]],
       device='cuda:0')
tensor([[-0.5291, -1.0000, -1.0000, -1.0000, -0.0703, -0.7217]],
       device='cuda:0')
tensor([[-0.4950, -1.0000, -1.0000, -1.0000, -0.0821, -0.7635]],
       device='cuda:0')
tensor([[-0.5306, -1.0000, -1.0000, -1.0000, -0.1220, -0.8100]],
       device='cuda:0')
tensor([[-0.5168, -1.0000, -1.0000, -1.0000, -0.1207, -0.8285]],
       device='cuda:0')
tensor([[-0.5442, -1.0000, -1.0000, -1.0000, -0.0972, -0.8447]],
       device='cuda:0')
tensor([[-0.5368, -1.0000, -1.0000, -1.0000, -0.0495, -0.7873]],
       device='cuda:0')
tensor([[-0.5217, -1.0000, -1.0000, -1.0000,  0.0289, -0.7510]],
       device='cuda:0')
tensor([[-0.4257, -0.7880, -1.0000, -1.0000,  0.1470, -0.6592]],
       device='cuda:0')
tensor([[-0.2987, -0.0884, -1.0000, -1.0000,  0.2608, -0.5015]],
       device='cuda:0')
tensor([[-0.4287,  0.1850, -1.0000, -1.0000, -0.2809, -0.4096]],
       device='cuda:0')
tensor([[-0.5971,  0.0153, -1.0000, -1.0000, -0.1758, -0.5567]],
       device='cuda:0')
tensor([[-0.6339, -0.6090, -1.0000, -1.0000, -0.1512, -0.5023]],
       device='cuda:0')
tensor([[-0.3895, -0.3582, -1.0000, -1.0000,  0.3381, -0.3637]],
       device='cuda:0')
tensor([[-0.2472,  0.5666, -1.0000, -0.9056,  0.4192, -0.1518]],
       device='cuda:0')
tensor([[-0.5179,  0.4278, -0.8854, -0.8633, -0.1279, -0.2017]],
       device='cuda:0')
tensor([[-0.7538, -0.2676, -1.0000, -1.0000, -0.1480, -0.3828]],
       device='cuda:0')
tensor([[-0.5746, -0.3179, -1.0000, -1.0000,  0.1659, -0.4121]],
       device='cuda:0')
tensor([[-0.3536,  0.1061, -1.0000, -1.0000,  0.2730, -0.0825]],
       device='cuda:0')
tensor([[-0.4643,  0.4024, -1.0000, -0.9495, -0.2056, -0.2248]],
       device='cuda:0')
tensor([[-0.6804, -0.0439, -1.0000, -1.0000, -0.2589, -0.5425]],
       device='cuda:0')
tensor([[-0.6613, -0.5530, -1.0000, -1.0000, -0.1297, -0.4616]],
       device='cuda:0')
tensor([[-0.4080, -0.3370, -1.0000, -1.0000,  0.2396, -0.3226]],
       device='cuda:0')
tensor([[-0.3005,  0.5082, -1.0000, -0.9445,  0.3195, -0.1575]],
       device='cuda:0')
tensor([[-0.5592,  0.3544, -0.8949, -0.9353, -0.2282, -0.2062]],
       device='cuda:0')
tensor([[-0.7875, -0.3733, -1.0000, -1.0000, -0.2678, -0.4217]],
       device='cuda:0')
tensor([[-0.5984, -0.5766, -1.0000, -1.0000,  0.0464, -0.5609]],
       device='cuda:0')
tensor([[-0.2753, -0.0271, -1.0000, -1.0000,  0.3200, -0.1748]],
       device='cuda:0')
tensor([[-0.3842,  0.4044, -1.0000, -0.9492, -0.2539, -0.2206]],
       device='cuda:0')
tensor([[-6.4951e-01, -4.9989e-04, -1.0000e+00, -1.0000e+00, -2.5767e-01,
         -5.5067e-01]], device='cuda:0')
tensor([[-0.6404, -0.5183, -1.0000, -1.0000, -0.1491, -0.4507]],
       device='cuda:0')
tensor([[-0.3876, -0.3133, -1.0000, -1.0000,  0.2379, -0.3083]],
       device='cuda:0')
tensor([[-0.2889,  0.4871, -1.0000, -0.9488,  0.2882, -0.1684]],
       device='cuda:0')
tensor([[-0.5556,  0.3315, -0.8842, -0.9432, -0.2478, -0.2145]],
       device='cuda:0')
tensor([[-0.7738, -0.3490, -1.0000, -1.0000, -0.2870, -0.4362]],
       device='cuda:0')
tensor([[-0.5808, -0.5886, -1.0000, -1.0000,  0.0482, -0.5692]],
       device='cuda:0')
tensor([[-0.2938, -0.0618, -1.0000, -1.0000,  0.2222, -0.1649]],
       device='cuda:0')
tensor([[-0.3892,  0.3719, -1.0000, -0.9886, -0.2349, -0.2306]],
       device='cuda:0')
tensor([[-0.6363, -0.0031, -1.0000, -1.0000, -0.2966, -0.5841]],
       device='cuda:0')
tensor([[-0.6284, -0.5140, -1.0000, -1.0000, -0.1419, -0.4399]],
       device='cuda:0')
tensor([[-0.3731, -0.3099, -1.0000, -1.0000,  0.2275, -0.3025]],
       device='cuda:0')
tensor([[-0.2805,  0.4742, -1.0000, -0.9569,  0.2677, -0.1801]],
       device='cuda:0')
tensor([[-0.5527,  0.3227, -0.8953, -0.9522, -0.2743, -0.2294]],
       device='cuda:0')
tensor([[-0.7651, -0.3246, -1.0000, -1.0000, -0.3078, -0.4567]],
       device='cuda:0')
tensor([[-0.5705, -0.4992, -1.0000, -1.0000,  0.0953, -0.5551]],
       device='cuda:0')
tensor([[-0.2890,  0.0339, -1.0000, -1.0000,  0.2248, -0.1274]],
       device='cuda:0')
tensor([[-0.4380,  0.2835, -1.0000, -1.0000, -0.3538, -0.2844]],
       device='cuda:0')
tensor([[-0.6462, -0.0538, -1.0000, -1.0000, -0.2957, -0.5578]],
       device='cuda:0')
tensor([[-0.6251, -0.5323, -1.0000, -1.0000, -0.1346, -0.4648]],
       device='cuda:0')
tensor([[-0.3709, -0.3283, -1.0000, -1.0000,  0.2052, -0.3104]],
       device='cuda:0')
tensor([[-0.2769,  0.4703, -1.0000, -0.9636,  0.2715, -0.1852]],
       device='cuda:0')
tensor([[-0.5517,  0.3236, -0.9033, -0.9585, -0.2962, -0.2394]],
       device='cuda:0')
tensor([[-0.7642, -0.3140, -1.0000, -1.0000, -0.3197, -0.4654]],
       device='cuda:0')
tensor([[-0.5859, -0.4512, -1.0000, -1.0000,  0.0852, -0.5649]],
       device='cuda:0')
tensor([[-0.3117, -0.0497, -1.0000, -1.0000,  0.1923, -0.1549]],
       device='cuda:0')
tensor([[-0.4155,  0.2877, -1.0000, -1.0000, -0.3070, -0.3069]],
       device='cuda:0')
tensor([[-0.6284, -0.0307, -1.0000, -1.0000, -0.2997, -0.5704]],
       device='cuda:0')
tensor([[-0.6128, -0.5206, -1.0000, -1.0000, -0.1389, -0.4519]],
       device='cuda:0')
tensor([[-0.3663, -0.3786, -1.0000, -1.0000,  0.2125, -0.3009]],
       device='cuda:0')
tensor([[-0.2375,  0.4121, -1.0000, -1.0000,  0.2636, -0.1947]],
       device='cuda:0')
tensor([[-0.5190,  0.3540, -0.9795, -0.9512, -0.2981, -0.2786]],
       device='cuda:0')
tensor([[-0.7273, -0.2057, -1.0000, -1.0000, -0.3108, -0.4971]],
       device='cuda:0')
tensor([[-0.6037, -0.4561, -1.0000, -1.0000,  0.0101, -0.5149]],
       device='cuda:0')
tensor([[-0.3503, -0.1681, -1.0000, -1.0000,  0.1722, -0.2132]],
       device='cuda:0')
tensor([[-0.3533,  0.4006, -1.0000, -1.0000, -0.0638, -0.2369]],
       device='cuda:0')
tensor([[-0.6016,  0.0863, -1.0000, -1.0000, -0.2942, -0.5542]],
       device='cuda:0')
tensor([[-0.6093, -0.4735, -1.0000, -1.0000, -0.1238, -0.4090]],
       device='cuda:0')
tensor([[-0.3702, -0.3148, -1.0000, -1.0000,  0.2179, -0.2760]],
       device='cuda:0')
tensor([[-0.2747,  0.4814, -1.0000, -0.9528,  0.2588, -0.1864]],
       device='cuda:0')
tensor([[-0.5562,  0.3113, -0.8848, -0.9549, -0.3052, -0.2550]],
       device='cuda:0')
tensor([[-0.7650, -0.3159, -1.0000, -1.0000, -0.3159, -0.4580]],
       device='cuda:0')
tensor([[-0.5799, -0.4475, -1.0000, -1.0000,  0.0854, -0.5715]],
       device='cuda:0')
tensor([[-0.3075, -0.0206, -1.0000, -1.0000,  0.1872, -0.1780]],
       device='cuda:0')
tensor([[-0.4277,  0.2658, -1.0000, -1.0000, -0.3383, -0.3298]],
       device='cuda:0')
tensor([[-0.6292, -0.0382, -1.0000, -1.0000, -0.2885, -0.5607]],
       device='cuda:0')
tensor([[-0.6091, -0.5214, -1.0000, -1.0000, -0.1365, -0.4597]],
       device='cuda:0')
tensor([[-0.3623, -0.3609, -1.0000, -1.0000,  0.2145, -0.2960]],
       device='cuda:0')
tensor([[-0.2515,  0.4707, -1.0000, -0.9550,  0.3045, -0.2080]],
       device='cuda:0')
tensor([[-0.5397,  0.3307, -0.8848, -0.9445, -0.3026, -0.2513]],
       device='cuda:0')
tensor([[-0.7513, -0.2513, -1.0000, -1.0000, -0.3006, -0.4622]],
       device='cuda:0')
tensor([[-0.5949, -0.4879, -1.0000, -1.0000,  0.0299, -0.5475]],
       device='cuda:0')
tensor([[-0.3309, -0.1638, -1.0000, -1.0000,  0.1689, -0.2289]],
       device='cuda:0')
tensor([[-0.3504,  0.3755, -1.0000, -1.0000, -0.1056, -0.2730]],
       device='cuda:0')
tensor([[-0.6006,  0.0498, -1.0000, -1.0000, -0.3149, -0.6093]],
       device='cuda:0')
tensor([[-0.6184, -0.5330, -1.0000, -1.0000, -0.1380, -0.4106]],
       device='cuda:0')
tensor([[-0.3039, -0.2306, -1.0000, -1.0000,  0.3539, -0.2617]],
       device='cuda:0')
tensor([[-0.2620,  0.4741, -1.0000, -0.9444,  0.1596, -0.2112]],
       device='cuda:0')
tensor([[-0.5721,  0.2255, -0.9462, -0.9829, -0.3101, -0.3593]],
       device='cuda:0')
tensor([[-0.7366, -0.3789, -1.0000, -1.0000, -0.2878, -0.4670]],
       device='cuda:0')
tensor([[-0.5170, -0.5863, -1.0000, -1.0000,  0.0614, -0.5741]],
       device='cuda:0')
tensor([[-0.2426,  0.3216, -1.0000, -1.0000,  0.3061, -0.1198]],
       device='cuda:0')
tensor([[-0.4919,  0.3570, -1.0000, -0.9442, -0.2668, -0.2665]],
       device='cuda:0')
tensor([[-0.6887, -0.1136, -1.0000, -1.0000, -0.2764, -0.5098]],
       device='cuda:0')
tensor([[-0.6197, -0.5364, -1.0000, -1.0000, -0.0683, -0.4892]],
       device='cuda:0')
tensor([[-0.3600, -0.2609, -1.0000, -1.0000,  0.1965, -0.2614]],
       device='cuda:0')
tensor([[-0.3056,  0.4651, -1.0000, -0.9931,  0.1579, -0.1807]],
       device='cuda:0')
tensor([[-0.5930,  0.2190, -1.0000, -1.0000, -0.3760, -0.4224]],
       device='cuda:0')
tensor([[-0.7439, -0.4696, -1.0000, -1.0000, -0.3038, -0.4682]],
       device='cuda:0')
tensor([[-0.5041, -0.4962, -1.0000, -1.0000,  0.0699, -0.5173]],
       device='cuda:0')
tensor([[-0.2610,  0.4201, -1.0000, -1.0000,  0.2998, -0.1075]],
       device='cuda:0')
tensor([[-0.5212,  0.3667, -0.9522, -0.9346, -0.2736, -0.2663]],
       device='cuda:0')
tensor([[-0.7315, -0.1952, -1.0000, -1.0000, -0.2742, -0.4625]],
       device='cuda:0')
tensor([[-0.6028, -0.4618, -1.0000, -1.0000,  0.0102, -0.5052]],
       device='cuda:0')
tensor([[-0.3440, -0.1487, -1.0000, -1.0000,  0.1982, -0.1760]],
       device='cuda:0')
tensor([[-0.3500,  0.4253, -1.0000, -1.0000, -0.0682, -0.1813]],
       device='cuda:0')
tensor([[-0.6112,  0.0680, -1.0000, -1.0000, -0.3068, -0.5655]],
       device='cuda:0')
tensor([[-0.6324, -0.5011, -1.0000, -1.0000, -0.1624, -0.4363]],
       device='cuda:0')
tensor([[-0.3712, -0.2944, -1.0000, -1.0000,  0.2400, -0.2900]],
       device='cuda:0')
tensor([[-0.2769,  0.4727, -1.0000, -0.9504,  0.2645, -0.1758]],
       device='cuda:0')
tensor([[-0.5501,  0.3142, -0.8718, -0.9462, -0.2640, -0.2227]],
       device='cuda:0')
tensor([[-0.7571, -0.2981, -1.0000, -1.0000, -0.2960, -0.4488]],
       device='cuda:0')
tensor([[-0.5788, -0.4767, -1.0000, -1.0000,  0.0852, -0.5659]],
       device='cuda:0')
tensor([[-0.2948, -0.0130, -1.0000, -1.0000,  0.2310, -0.1274]],
       device='cuda:0')
tensor([[-0.4120,  0.3153, -1.0000, -1.0000, -0.2923, -0.2375]],
       device='cuda:0')
tensor([[-0.6391, -0.0498, -1.0000, -1.0000, -0.3179, -0.5817]],
       device='cuda:0')
tensor([[-0.6172, -0.5367, -1.0000, -1.0000, -0.1271, -0.4614]],
       device='cuda:0')
tensor([[-0.3289, -0.3178, -1.0000, -1.0000,  0.2880, -0.2920]],
       device='cuda:0')
tensor([[-0.2594,  0.4598, -1.0000, -0.9648,  0.2270, -0.2140]],
       device='cuda:0')
tensor([[-0.5611,  0.2765, -0.9529, -0.9875, -0.3387, -0.3135]],
       device='cuda:0')
tensor([[-0.7513, -0.3564, -1.0000, -1.0000, -0.3331, -0.4927]],
       device='cuda:0')
tensor([[-0.5533, -0.5903, -1.0000, -1.0000,  0.0234, -0.5746]],
       device='cuda:0')
tensor([[-0.2820,  0.0669, -1.0000, -1.0000,  0.2080, -0.2167]],
       device='cuda:0')
tensor([[-0.4586,  0.3136, -1.0000, -0.9769, -0.3332, -0.2762]],
       device='cuda:0')
tensor([[-0.6570, -0.0157, -1.0000, -1.0000, -0.3026, -0.5532]],
       device='cuda:0')
tensor([[-0.6438, -0.5656, -1.0000, -1.0000, -0.1471, -0.4519]],
       device='cuda:0')
tensor([[-0.3821, -0.3863, -1.0000, -1.0000,  0.1983, -0.3089]],
       device='cuda:0')
tensor([[-0.2656,  0.3550, -1.0000, -1.0000,  0.2119, -0.1957]],
       device='cuda:0')
tensor([[-0.5420,  0.2903, -1.0000, -0.9970, -0.3192, -0.3038]],
       device='cuda:0')
tensor([[-0.7308, -0.3409, -1.0000, -1.0000, -0.3388, -0.5208]],
       device='cuda:0')
tensor([[-0.5581, -0.5369, -1.0000, -1.0000,  0.0437, -0.5418]],
       device='cuda:0')
tensor([[-0.2932,  0.0973, -1.0000, -1.0000,  0.2266, -0.1316]],
       device='cuda:0')
tensor([[-0.4664,  0.3543, -1.0000, -0.9457, -0.2565, -0.2119]],
       device='cuda:0')
tensor([[-0.6658, -0.0367, -1.0000, -1.0000, -0.2832, -0.5260]],
       device='cuda:0')
tensor([[-0.6488, -0.5717, -1.0000, -1.0000, -0.1510, -0.4685]],
       device='cuda:0')
tensor([[-0.3904, -0.3878, -1.0000, -1.0000,  0.1926, -0.3095]],
       device='cuda:0')
tensor([[-0.2743,  0.3604, -1.0000, -1.0000,  0.2141, -0.1859]],
       device='cuda:0')
tensor([[-0.5458,  0.2983, -1.0000, -0.9954, -0.3098, -0.2879]],
       device='cuda:0')
tensor([[-0.7392, -0.3606, -1.0000, -1.0000, -0.3406, -0.5152]],
       device='cuda:0')
tensor([[-0.5596, -0.5389, -1.0000, -1.0000,  0.0377, -0.5348]],
       device='cuda:0')
tensor([[-0.2958,  0.0905, -1.0000, -1.0000,  0.2264, -0.1330]],
       device='cuda:0')
tensor([[-0.4637,  0.3599, -1.0000, -0.9418, -0.2539, -0.2074]],
       device='cuda:0')
tensor([[-0.6640, -0.0266, -1.0000, -1.0000, -0.2773, -0.5223]],
       device='cuda:0')
tensor([[-0.6519, -0.5701, -1.0000, -1.0000, -0.1571, -0.4642]],
       device='cuda:0')
tensor([[-0.3946, -0.3933, -1.0000, -1.0000,  0.1940, -0.3094]],
       device='cuda:0')
tensor([[-0.2767,  0.3558, -1.0000, -1.0000,  0.2127, -0.1834]],
       device='cuda:0')
tensor([[-0.5450,  0.3001, -1.0000, -0.9940, -0.2994, -0.2769]],
       device='cuda:0')
tensor([[-0.7398, -0.3625, -1.0000, -1.0000, -0.3396, -0.5144]],
       device='cuda:0')
tensor([[-0.5616, -0.5463, -1.0000, -1.0000,  0.0372, -0.5347]],
       device='cuda:0')
tensor([[-0.2937,  0.0803, -1.0000, -1.0000,  0.2323, -0.1242]],
       device='cuda:0')
tensor([[-0.4606,  0.3558, -1.0000, -0.9441, -0.2692, -0.2118]],
       device='cuda:0')
tensor([[-0.6626, -0.0223, -1.0000, -1.0000, -0.2699, -0.5206]],
       device='cuda:0')
tensor([[-0.6534, -0.5786, -1.0000, -1.0000, -0.1597, -0.4666]],
       device='cuda:0')
tensor([[-0.3960, -0.4064, -1.0000, -1.0000,  0.1931, -0.3110]],
       device='cuda:0')
tensor([[-0.2769,  0.3459, -1.0000, -1.0000,  0.2079, -0.1825]],
       device='cuda:0')
tensor([[-0.5451,  0.2930, -1.0000, -0.9983, -0.2989, -0.2777]],
       device='cuda:0')
tensor([[-0.7378, -0.3694, -1.0000, -1.0000, -0.3389, -0.5180]],
       device='cuda:0')
tensor([[-0.5600, -0.5524, -1.0000, -1.0000,  0.0367, -0.5340]],
       device='cuda:0')
tensor([[-0.2921,  0.0833, -1.0000, -1.0000,  0.2365, -0.1189]],
       device='cuda:0')
tensor([[-0.4625,  0.3526, -1.0000, -0.9450, -0.2664, -0.2104]],
       device='cuda:0')
tensor([[-0.6639, -0.0311, -1.0000, -1.0000, -0.2720, -0.5216]],
       device='cuda:0')
tensor([[-0.6510, -0.5727, -1.0000, -1.0000, -0.1522, -0.4663]],
       device='cuda:0')
tensor([[-0.3952, -0.3910, -1.0000, -1.0000,  0.1923, -0.3038]],
       device='cuda:0')
tensor([[-0.2799,  0.3555, -1.0000, -1.0000,  0.2075, -0.1828]],
       device='cuda:0')
tensor([[-0.5500,  0.2884, -1.0000, -1.0000, -0.3110, -0.2906]],
       device='cuda:0')
tensor([[-0.7410, -0.3850, -1.0000, -1.0000, -0.3411, -0.5151]],
       device='cuda:0')
tensor([[-0.5556, -0.5451, -1.0000, -1.0000,  0.0359, -0.5305]],
       device='cuda:0')
tensor([[-0.2935,  0.1061, -1.0000, -1.0000,  0.2300, -0.1316]],
       device='cuda:0')
tensor([[-0.4671,  0.3619, -1.0000, -0.9397, -0.2398, -0.1984]],
       device='cuda:0')
tensor([[-0.6684, -0.0395, -1.0000, -1.0000, -0.2807, -0.5203]],
       device='cuda:0')
tensor([[-0.6503, -0.5712, -1.0000, -1.0000, -0.1466, -0.4675]],
       device='cuda:0')
tensor([[-0.3930, -0.3828, -1.0000, -1.0000,  0.1917, -0.3033]],
       device='cuda:0')
tensor([[-0.2787,  0.3659, -1.0000, -1.0000,  0.2145, -0.1820]],
       device='cuda:0')
tensor([[-0.5502,  0.2953, -1.0000, -0.9991, -0.3162, -0.2921]],
       device='cuda:0')
tensor([[-0.7430, -0.3796, -1.0000, -1.0000, -0.3441, -0.5145]],
       device='cuda:0')
tensor([[-0.5566, -0.5356, -1.0000, -1.0000,  0.0355, -0.5298]],
       device='cuda:0')
tensor([[-0.2972,  0.1030, -1.0000, -1.0000,  0.2212, -0.1427]],
       device='cuda:0')
tensor([[-0.4690,  0.3586, -1.0000, -0.9432, -0.2481, -0.2100]],
       device='cuda:0')
tensor([[-0.6678, -0.0381, -1.0000, -1.0000, -0.2809, -0.5212]],
       device='cuda:0')
tensor([[-0.6487, -0.5649, -1.0000, -1.0000, -0.1484, -0.4652]],
       device='cuda:0')
tensor([[-0.3909, -0.3769, -1.0000, -1.0000,  0.1939, -0.3051]],
       device='cuda:0')
tensor([[-0.2749,  0.3757, -1.0000, -1.0000,  0.2268, -0.1811]],
       device='cuda:0')
tensor([[-0.5449,  0.3117, -1.0000, -0.9883, -0.3067, -0.2754]],
       device='cuda:0')
tensor([[-0.7440, -0.3547, -1.0000, -1.0000, -0.3436, -0.5119]],
       device='cuda:0')
tensor([[-0.5660, -0.5379, -1.0000, -1.0000,  0.0352, -0.5353]],
       device='cuda:0')
tensor([[-0.3007,  0.0550, -1.0000, -1.0000,  0.2162, -0.1443]],
       device='cuda:0')
tensor([[-0.4445,  0.3094, -1.0000, -0.9723, -0.3125, -0.2306]],
       device='cuda:0')
tensor([[-0.6594, -0.0551, -1.0000, -1.0000, -0.2756, -0.5282]],
       device='cuda:0')
tensor([[-0.6346, -0.5281, -1.0000, -1.0000, -0.1268, -0.4572]],
       device='cuda:0')
tensor([[-0.3833, -0.3197, -1.0000, -1.0000,  0.1984, -0.2862]],
       device='cuda:0')
tensor([[-0.2968,  0.4602, -1.0000, -0.9756,  0.2503, -0.1755]],
       device='cuda:0')
tensor([[-0.5684,  0.2985, -0.9363, -0.9852, -0.3234, -0.2635]],
       device='cuda:0')
tensor([[-0.7731, -0.3885, -1.0000, -1.0000, -0.3421, -0.4787]],
       device='cuda:0')
tensor([[-0.5725, -0.6008, -1.0000, -1.0000,  0.0023, -0.5603]],
       device='cuda:0')
tensor([[-0.2975, -0.0499, -1.0000, -1.0000,  0.1893, -0.2351]],
       device='cuda:0')
tensor([[-0.4025,  0.3389, -1.0000, -0.9975, -0.2409, -0.2402]],
       device='cuda:0')
tensor([[-0.6324, -0.0222, -1.0000, -1.0000, -0.2976, -0.5784]],
       device='cuda:0')
tensor([[-0.6186, -0.5112, -1.0000, -1.0000, -0.1244, -0.4340]],
       device='cuda:0')
tensor([[-0.3425, -0.2851, -1.0000, -1.0000,  0.2705, -0.2789]],
       device='cuda:0')
tensor([[-0.2801,  0.4456, -1.0000, -0.9735,  0.2027, -0.2032]],
       device='cuda:0')
tensor([[-0.5727,  0.2421, -0.9875, -1.0000, -0.3456, -0.3397]],
       device='cuda:0')
tensor([[-0.7417, -0.4398, -1.0000, -1.0000, -0.3372, -0.5035]],
       device='cuda:0')
tensor([[-0.5249, -0.5705, -1.0000, -1.0000,  0.0377, -0.5532]],
       device='cuda:0')
tensor([[-0.2761,  0.2282, -1.0000, -1.0000,  0.2312, -0.1817]],
       device='cuda:0')
tensor([[-0.4915,  0.3450, -1.0000, -0.9554, -0.2565, -0.2331]],
       device='cuda:0')
tensor([[-0.6886, -0.1143, -1.0000, -1.0000, -0.3012, -0.5236]],
       device='cuda:0')
tensor([[-0.6273, -0.5337, -1.0000, -1.0000, -0.0746, -0.4898]],
       device='cuda:0')
tensor([[-0.3690, -0.2700, -1.0000, -1.0000,  0.1821, -0.2686]],
       device='cuda:0')
tensor([[-0.3092,  0.4642, -1.0000, -0.9923,  0.1708, -0.1753]],
       device='cuda:0')
tensor([[-0.5942,  0.2197, -1.0000, -1.0000, -0.3850, -0.4139]],
       device='cuda:0')
tensor([[-0.7496, -0.4757, -1.0000, -1.0000, -0.3146, -0.4704]],
       device='cuda:0')
tensor([[-0.5107, -0.5037, -1.0000, -1.0000,  0.0571, -0.5195]],
       device='cuda:0')
tensor([[-0.2735,  0.3697, -1.0000, -1.0000,  0.2654, -0.1314]],
       device='cuda:0')
tensor([[-0.5175,  0.3619, -0.9812, -0.9413, -0.2688, -0.2516]],
       device='cuda:0')
tensor([[-0.7245, -0.1942, -1.0000, -1.0000, -0.2958, -0.4839]],
       device='cuda:0')
tensor([[-0.6035, -0.4493, -1.0000, -1.0000,  0.0116, -0.5023]],
       device='cuda:0')
tensor([[-0.3480, -0.1405, -1.0000, -1.0000,  0.1921, -0.1755]],
       device='cuda:0')
tensor([[-0.3610,  0.4093, -1.0000, -1.0000, -0.0885, -0.1934]],
       device='cuda:0')
tensor([[-0.6171,  0.0422, -1.0000, -1.0000, -0.3134, -0.5867]],
       device='cuda:0')
tensor([[-0.6270, -0.4968, -1.0000, -1.0000, -0.1434, -0.4275]],
       device='cuda:0')
tensor([[-0.3687, -0.2836, -1.0000, -1.0000,  0.2332, -0.2812]],
       device='cuda:0')
tensor([[-0.2839,  0.4661, -1.0000, -0.9599,  0.2380, -0.1787]],
       device='cuda:0')
tensor([[-0.5601,  0.2964, -0.9055, -0.9665, -0.2953, -0.2546]],
       device='cuda:0')
tensor([[-0.7620, -0.3437, -1.0000, -1.0000, -0.3194, -0.4655]],
       device='cuda:0')
tensor([[-0.5668, -0.6080, -1.0000, -1.0000,  0.0233, -0.5792]],
       device='cuda:0')
tensor([[-0.2823, -0.0082, -1.0000, -1.0000,  0.2097, -0.2072]],
       device='cuda:0')
tensor([[-0.4125,  0.3243, -1.0000, -0.9950, -0.2805, -0.2418]],
       device='cuda:0')
tensor([[-0.6387, -0.0382, -1.0000, -1.0000, -0.2991, -0.5709]],
       device='cuda:0')
tensor([[-0.6202, -0.5244, -1.0000, -1.0000, -0.1260, -0.4491]],
       device='cuda:0')
tensor([[-0.3534, -0.3054, -1.0000, -1.0000,  0.2377, -0.2896]],
       device='cuda:0')
tensor([[-0.2779,  0.4545, -1.0000, -0.9713,  0.2296, -0.1969]],
       device='cuda:0')
tensor([[-0.5664,  0.2728, -0.9579, -0.9930, -0.3407, -0.3069]],
       device='cuda:0')
tensor([[-0.7567, -0.3899, -1.0000, -1.0000, -0.3412, -0.4927]],
       device='cuda:0')
tensor([[-0.5520, -0.5938, -1.0000, -1.0000,  0.0150, -0.5662]],
       device='cuda:0')
tensor([[-0.2852,  0.0616, -1.0000, -1.0000,  0.2043, -0.2251]],
       device='cuda:0')
tensor([[-0.4429,  0.2867, -1.0000, -0.9821, -0.3252, -0.2566]],
       device='cuda:0')
tensor([[-0.6506, -0.0616, -1.0000, -1.0000, -0.2743, -0.5350]],
       device='cuda:0')
tensor([[-0.6218, -0.5319, -1.0000, -1.0000, -0.1154, -0.4649]],
       device='cuda:0')
tensor([[-0.3687, -0.3097, -1.0000, -1.0000,  0.2049, -0.2828]],
       device='cuda:0')
tensor([[-0.2902,  0.4575, -1.0000, -0.9771,  0.2327, -0.1853]],
       device='cuda:0')
tensor([[-0.5735,  0.2711, -0.9651, -0.9999, -0.3530, -0.3132]],
       device='cuda:0')
tensor([[-0.7612, -0.4135, -1.0000, -1.0000, -0.3484, -0.4963]],
       device='cuda:0')
tensor([[-0.5513, -0.5688, -1.0000, -1.0000,  0.0157, -0.5500]],
       device='cuda:0')
tensor([[-0.2928,  0.0781, -1.0000, -1.0000,  0.1962, -0.2245]],
       device='cuda:0')
tensor([[-0.4668,  0.3245, -1.0000, -0.9637, -0.2954, -0.2663]],
       device='cuda:0')
tensor([[-0.6535, -0.0235, -1.0000, -1.0000, -0.2711, -0.5323]],
       device='cuda:0')
tensor([[-0.6397, -0.5590, -1.0000, -1.0000, -0.1520, -0.4603]],
       device='cuda:0')
tensor([[-0.3828, -0.3794, -1.0000, -1.0000,  0.2000, -0.3075]],
       device='cuda:0')
tensor([[-0.2671,  0.3634, -1.0000, -1.0000,  0.2212, -0.1915]],
       device='cuda:0')
tensor([[-0.5403,  0.3045, -1.0000, -0.9888, -0.3071, -0.2816]],
       device='cuda:0')
tensor([[-0.7354, -0.3340, -1.0000, -1.0000, -0.3396, -0.5161]],
       device='cuda:0')
tensor([[-0.5632, -0.5288, -1.0000, -1.0000,  0.0488, -0.5384]],
       device='cuda:0')
tensor([[-0.2956,  0.0771, -1.0000, -1.0000,  0.2260, -0.1219]],
       device='cuda:0')
tensor([[-0.4649,  0.3393, -1.0000, -0.9580, -0.2975, -0.2316]],
       device='cuda:0')
tensor([[-0.6617, -0.0296, -1.0000, -1.0000, -0.2781, -0.5297]],
       device='cuda:0')
tensor([[-0.6490, -0.5648, -1.0000, -1.0000, -0.1574, -0.4645]],
       device='cuda:0')
tensor([[-0.3916, -0.3865, -1.0000, -1.0000,  0.1944, -0.3089]],
       device='cuda:0')
tensor([[-0.2736,  0.3598, -1.0000, -1.0000,  0.2178, -0.1838]],
       device='cuda:0')
tensor([[-0.5422,  0.3061, -1.0000, -0.9891, -0.2977, -0.2717]],
       device='cuda:0')
tensor([[-0.7390, -0.3460, -1.0000, -1.0000, -0.3380, -0.5128]],
       device='cuda:0')
tensor([[-0.5650, -0.5434, -1.0000, -1.0000,  0.0395, -0.5373]],
       device='cuda:0')
tensor([[-0.2955,  0.0631, -1.0000, -1.0000,  0.2286, -0.1241]],
       device='cuda:0')
tensor([[-0.4432,  0.3116, -1.0000, -0.9676, -0.3068, -0.2153]],
       device='cuda:0')
tensor([[-0.6620, -0.0624, -1.0000, -1.0000, -0.2770, -0.5266]],
       device='cuda:0')
tensor([[-0.6356, -0.5370, -1.0000, -1.0000, -0.1215, -0.4635]],
       device='cuda:0')
tensor([[-0.3831, -0.3208, -1.0000, -1.0000,  0.1958, -0.2832]],
       device='cuda:0')
tensor([[-0.2987,  0.4597, -1.0000, -0.9780,  0.2456, -0.1769]],
       device='cuda:0')
tensor([[-0.5725,  0.2902, -0.9486, -0.9927, -0.3359, -0.2793]],
       device='cuda:0')
tensor([[-0.7722, -0.4037, -1.0000, -1.0000, -0.3474, -0.4860]],
       device='cuda:0')
tensor([[-0.5678, -0.5846, -1.0000, -1.0000,  0.0061, -0.5510]],
       device='cuda:0')
tensor([[-0.2985, -0.0166, -1.0000, -1.0000,  0.1880, -0.2318]],
       device='cuda:0')
tensor([[-0.4202,  0.3187, -1.0000, -0.9969, -0.2733, -0.2526]],
       device='cuda:0')
tensor([[-0.6377, -0.0311, -1.0000, -1.0000, -0.2880, -0.5636]],
       device='cuda:0')
tensor([[-0.6216, -0.5131, -1.0000, -1.0000, -0.1241, -0.4385]],
       device='cuda:0')
tensor([[-0.3661, -0.3043, -1.0000, -1.0000,  0.2182, -0.2817]],
       device='cuda:0')
tensor([[-0.2868,  0.4543, -1.0000, -0.9735,  0.2333, -0.1871]],
       device='cuda:0')
tensor([[-0.5685,  0.2741, -0.9537, -0.9930, -0.3401, -0.2991]],
       device='cuda:0')
tensor([[-0.7611, -0.3956, -1.0000, -1.0000, -0.3423, -0.4906]],
       device='cuda:0')
tensor([[-0.5562, -0.5984, -1.0000, -1.0000,  0.0098, -0.5641]],
       device='cuda:0')
tensor([[-0.2873,  0.0339, -1.0000, -1.0000,  0.1985, -0.2318]],
       device='cuda:0')
tensor([[-0.4360,  0.2954, -1.0000, -0.9887, -0.3062, -0.2536]],
       device='cuda:0')
tensor([[-0.6461, -0.0558, -1.0000, -1.0000, -0.2811, -0.5465]],
       device='cuda:0')
tensor([[-0.6211, -0.5295, -1.0000, -1.0000, -0.1152, -0.4591]],
       device='cuda:0')
tensor([[-0.3654, -0.3063, -1.0000, -1.0000,  0.2103, -0.2800]],
       device='cuda:0')
tensor([[-0.2900,  0.4546, -1.0000, -0.9775,  0.2273, -0.1877]],
       device='cuda:0')
tensor([[-0.5742,  0.2648, -0.9708, -1.0000, -0.3558, -0.3214]],
       device='cuda:0')
tensor([[-0.7586, -0.4212, -1.0000, -1.0000, -0.3472, -0.4968]],
       device='cuda:0')
tensor([[-0.5457, -0.5672, -1.0000, -1.0000,  0.0183, -0.5494]],
       device='cuda:0')
tensor([[-0.2905,  0.1064, -1.0000, -1.0000,  0.1999, -0.2211]],
       device='cuda:0')
tensor([[-0.4722,  0.3323, -1.0000, -0.9586, -0.2664, -0.2508]],
       device='cuda:0')
tensor([[-0.6605, -0.0421, -1.0000, -1.0000, -0.2817, -0.5315]],
       device='cuda:0')
tensor([[-0.6381, -0.5622, -1.0000, -1.0000, -0.1392, -0.4667]],
       device='cuda:0')
tensor([[-0.3794, -0.3645, -1.0000, -1.0000,  0.1975, -0.3016]],
       device='cuda:0')
tensor([[-0.2668,  0.3997, -1.0000, -1.0000,  0.2547, -0.1815]],
       device='cuda:0')
tensor([[-0.5397,  0.3284, -0.9867, -0.9767, -0.3101, -0.2639]],
       device='cuda:0')
tensor([[-0.7473, -0.3165, -1.0000, -1.0000, -0.3421, -0.5055]],
       device='cuda:0')
tensor([[-0.5759, -0.4503, -1.0000, -1.0000,  0.0931, -0.5285]],
       device='cuda:0')
tensor([[-0.3063,  0.0434, -1.0000, -1.0000,  0.2139, -0.1019]],
       device='cuda:0')
tensor([[-0.4515,  0.2717, -1.0000, -1.0000, -0.3592, -0.2826]],
       device='cuda:0')
tensor([[-0.6523, -0.0604, -1.0000, -1.0000, -0.3011, -0.5513]],
       device='cuda:0')
tensor([[-0.6299, -0.5288, -1.0000, -1.0000, -0.1293, -0.4625]],
       device='cuda:0')
tensor([[-0.3766, -0.3207, -1.0000, -1.0000,  0.1954, -0.2984]],
       device='cuda:0')
tensor([[-0.2870,  0.4643, -1.0000, -0.9698,  0.2557, -0.1799]],
       device='cuda:0')
tensor([[-0.5615,  0.3053, -0.9184, -0.9746, -0.3173, -0.2565]],
       device='cuda:0')
tensor([[-0.7693, -0.3567, -1.0000, -1.0000, -0.3350, -0.4718]],
       device='cuda:0')
tensor([[-0.5754, -0.6086, -1.0000, -1.0000,  0.0031, -0.5731]],
       device='cuda:0')
tensor([[-0.2957, -0.0666, -1.0000, -1.0000,  0.1879, -0.2403]],
       device='cuda:0')
tensor([[-0.3926,  0.3446, -1.0000, -1.0000, -0.2304, -0.2428]],
       device='cuda:0')
tensor([[-0.6270, -0.0151, -1.0000, -1.0000, -0.3029, -0.5866]],
       device='cuda:0')
tensor([[-0.6143, -0.5114, -1.0000, -1.0000, -0.1282, -0.4333]],
       device='cuda:0')
tensor([[-0.3684, -0.3649, -1.0000, -1.0000,  0.2131, -0.2769]],
       device='cuda:0')
tensor([[-0.2462,  0.4413, -1.0000, -0.9724,  0.2909, -0.1873]],
       device='cuda:0')
tensor([[-0.5350,  0.3368, -0.9183, -0.9514, -0.3124, -0.2609]],
       device='cuda:0')
tensor([[-0.7492, -0.2449, -1.0000, -1.0000, -0.3118, -0.4746]],
       device='cuda:0')
tensor([[-0.6020, -0.4743, -1.0000, -1.0000,  0.0221, -0.5375]],
       device='cuda:0')
tensor([[-0.3425, -0.1788, -1.0000, -1.0000,  0.1572, -0.2315]],
       device='cuda:0')
tensor([[-0.3483,  0.3797, -1.0000, -1.0000, -0.0717, -0.2763]],
       device='cuda:0')
tensor([[-0.5911,  0.0945, -1.0000, -1.0000, -0.2887, -0.5555]],
       device='cuda:0')
tensor([[-0.6273, -0.5177, -1.0000, -1.0000, -0.1635, -0.4249]],
       device='cuda:0')
tensor([[-0.3509, -0.2668, -1.0000, -1.0000,  0.2636, -0.2836]],
       device='cuda:0')
tensor([[-0.2662,  0.4800, -1.0000, -0.9399,  0.2440, -0.1831]],
       device='cuda:0')
tensor([[-0.5456,  0.3145, -0.8532, -0.9345, -0.2498, -0.2192]],
       device='cuda:0')
tensor([[-0.7486, -0.2685, -1.0000, -1.0000, -0.2857, -0.4443]],
       device='cuda:0')
tensor([[-0.5790, -0.4935, -1.0000, -1.0000,  0.0609, -0.5598]],
       device='cuda:0')
tensor([[-0.2991, -0.0533, -1.0000, -1.0000,  0.2202, -0.1620]],
       device='cuda:0')
tensor([[-0.3970,  0.3243, -1.0000, -1.0000, -0.2596, -0.2376]],
       device='cuda:0')
tensor([[-0.6314, -0.0386, -1.0000, -1.0000, -0.3215, -0.5919]],
       device='cuda:0')
tensor([[-0.6143, -0.5364, -1.0000, -1.0000, -0.1279, -0.4557]],
       device='cuda:0')
tensor([[-0.3653, -0.3880, -1.0000, -1.0000,  0.2042, -0.2973]],
       device='cuda:0')
tensor([[-0.2412,  0.3953, -1.0000, -1.0000,  0.2324, -0.2021]],
       device='cuda:0')
tensor([[-0.5306,  0.3223, -1.0000, -0.9756, -0.3312, -0.3241]],
       device='cuda:0')
tensor([[-0.7258, -0.2591, -1.0000, -1.0000, -0.3280, -0.5111]],
       device='cuda:0')
tensor([[-0.5822, -0.3976, -1.0000, -1.0000,  0.0628, -0.5175]],
       device='cuda:0')
tensor([[-0.3265, -0.0089, -1.0000, -1.0000,  0.2026, -0.1570]],
       device='cuda:0')
tensor([[-0.4340,  0.2944, -1.0000, -1.0000, -0.3125, -0.2755]],
       device='cuda:0')
tensor([[-0.6408, -0.0402, -1.0000, -1.0000, -0.2957, -0.5622]],
       device='cuda:0')
tensor([[-0.6219, -0.5171, -1.0000, -1.0000, -0.1333, -0.4510]],
       device='cuda:0')
tensor([[-0.3646, -0.3139, -1.0000, -1.0000,  0.2173, -0.3010]],
       device='cuda:0')
tensor([[-0.2777,  0.4610, -1.0000, -0.9653,  0.2523, -0.1904]],
       device='cuda:0')
tensor([[-0.5564,  0.3037, -0.9157, -0.9693, -0.3101, -0.2576]],
       device='cuda:0')
tensor([[-0.7624, -0.3340, -1.0000, -1.0000, -0.3283, -0.4749]],
       device='cuda:0')
tensor([[-0.5711, -0.5894, -1.0000, -1.0000,  0.0281, -0.5744]],
       device='cuda:0')
tensor([[-0.2914, -0.0353, -1.0000, -1.0000,  0.1924, -0.2171]],
       device='cuda:0')
tensor([[-0.4122,  0.3104, -1.0000, -1.0000, -0.2820, -0.2712]],
       device='cuda:0')
tensor([[-0.6324, -0.0314, -1.0000, -1.0000, -0.2977, -0.5723]],
       device='cuda:0')
tensor([[-0.6174, -0.5197, -1.0000, -1.0000, -0.1295, -0.4449]],
       device='cuda:0')
tensor([[-0.3236, -0.2897, -1.0000, -1.0000,  0.3138, -0.2872]],
       device='cuda:0')
tensor([[-0.2678,  0.4483, -1.0000, -0.9677,  0.1976, -0.2151]],
       device='cuda:0')
tensor([[-0.5694,  0.2433, -0.9802, -1.0000, -0.3393, -0.3408]],
       device='cuda:0')
tensor([[-0.7398, -0.4217, -1.0000, -1.0000, -0.3312, -0.5010]],
       device='cuda:0')
tensor([[-0.5261, -0.5753, -1.0000, -1.0000,  0.0401, -0.5579]],
       device='cuda:0')
tensor([[-0.2711,  0.2308, -1.0000, -1.0000,  0.2407, -0.1709]],
       device='cuda:0')
tensor([[-0.4868,  0.3516, -1.0000, -0.9514, -0.2532, -0.2309]],
       device='cuda:0')
tensor([[-0.6835, -0.0922, -1.0000, -1.0000, -0.2960, -0.5236]],
       device='cuda:0')
tensor([[-0.6337, -0.5523, -1.0000, -1.0000, -0.0995, -0.4834]],
       device='cuda:0')
tensor([[-0.3742, -0.3118, -1.0000, -1.0000,  0.1880, -0.2842]],
       device='cuda:0')
tensor([[-0.2965,  0.4643, -1.0000, -0.9845,  0.2236, -0.1802]],
       device='cuda:0')
tensor([[-0.5807,  0.2652, -0.9948, -1.0000, -0.3720, -0.3464]],
       device='cuda:0')
tensor([[-0.7577, -0.4520, -1.0000, -1.0000, -0.3513, -0.4998]],
       device='cuda:0')
tensor([[-0.5356, -0.5175, -1.0000, -1.0000,  0.0334, -0.5281]],
       device='cuda:0')
tensor([[-0.2951,  0.1973, -1.0000, -1.0000,  0.2050, -0.2014]],
       device='cuda:0')
tensor([[-0.4943,  0.3368, -1.0000, -0.9605, -0.2728, -0.2688]],
       device='cuda:0')
tensor([[-0.6838, -0.0971, -1.0000, -1.0000, -0.2930, -0.5195]],
       device='cuda:0')
tensor([[-0.6246, -0.5082, -1.0000, -1.0000, -0.0903, -0.4670]],
       device='cuda:0')
tensor([[-0.3690, -0.2578, -1.0000, -1.0000,  0.1930, -0.2722]],
       device='cuda:0')
tensor([[-0.3052,  0.4689, -1.0000, -0.9833,  0.1729, -0.1713]],
       device='cuda:0')
tensor([[-0.5918,  0.2232, -1.0000, -1.0000, -0.3742, -0.4007]],
       device='cuda:0')
tensor([[-0.7454, -0.4662, -1.0000, -1.0000, -0.3185, -0.4767]],
       device='cuda:0')
tensor([[-0.5093, -0.5043, -1.0000, -1.0000,  0.0585, -0.5228]],
       device='cuda:0')
tensor([[-0.2698,  0.3824, -1.0000, -1.0000,  0.2699, -0.1312]],
       device='cuda:0')
tensor([[-0.5184,  0.3617, -0.9741, -0.9404, -0.2728, -0.2572]],
       device='cuda:0')
tensor([[-0.7253, -0.1915, -1.0000, -1.0000, -0.2929, -0.4808]],
       device='cuda:0')
tensor([[-0.6030, -0.4550, -1.0000, -1.0000,  0.0075, -0.5042]],
       device='cuda:0')
tensor([[-0.3465, -0.1470, -1.0000, -1.0000,  0.1917, -0.1829]],
       device='cuda:0')
tensor([[-0.3552,  0.4152, -1.0000, -1.0000, -0.0746, -0.1908]],
       device='cuda:0')
tensor([[-0.6127,  0.0601, -1.0000, -1.0000, -0.3068, -0.5679]],
       device='cuda:0')
tensor([[-0.6303, -0.4999, -1.0000, -1.0000, -0.1562, -0.4335]],
       device='cuda:0')
tensor([[-0.3697, -0.2924, -1.0000, -1.0000,  0.2367, -0.2882]],
       device='cuda:0')
tensor([[-0.2783,  0.4694, -1.0000, -0.9534,  0.2566, -0.1777]],
       device='cuda:0')
tensor([[-0.5512,  0.3123, -0.8782, -0.9500, -0.2705, -0.2251]],
       device='cuda:0')
tensor([[-0.7583, -0.3034, -1.0000, -1.0000, -0.3043, -0.4544]],
       device='cuda:0')
tensor([[-0.5802, -0.4745, -1.0000, -1.0000,  0.0860, -0.5687]],
       device='cuda:0')
tensor([[-0.2951, -0.0132, -1.0000, -1.0000,  0.2255, -0.1281]],
       device='cuda:0')
tensor([[-0.4153,  0.3044, -1.0000, -1.0000, -0.3068, -0.2550]],
       device='cuda:0')
tensor([[-0.6379, -0.0512, -1.0000, -1.0000, -0.3172, -0.5788]],
       device='cuda:0')
tensor([[-0.6159, -0.5358, -1.0000, -1.0000, -0.1294, -0.4644]],
       device='cuda:0')
tensor([[-0.3679, -0.3852, -1.0000, -1.0000,  0.1982, -0.2976]],
       device='cuda:0')
tensor([[-0.2418,  0.4039, -1.0000, -1.0000,  0.2379, -0.1959]],
       device='cuda:0')
tensor([[-0.5310,  0.3279, -1.0000, -0.9712, -0.3298, -0.3158]],
       device='cuda:0')
tensor([[-0.7280, -0.2581, -1.0000, -1.0000, -0.3303, -0.5088]],
       device='cuda:0')
tensor([[-0.5844, -0.3979, -1.0000, -1.0000,  0.0591, -0.5166]],
       device='cuda:0')
tensor([[-0.3301, -0.0242, -1.0000, -1.0000,  0.1953, -0.1650]],
       device='cuda:0')
tensor([[-0.4295,  0.2980, -1.0000, -1.0000, -0.3013, -0.2779]],
       device='cuda:0')
tensor([[-0.6378, -0.0355, -1.0000, -1.0000, -0.2954, -0.5645]],
       device='cuda:0')
tensor([[-0.6195, -0.5145, -1.0000, -1.0000, -0.1327, -0.4477]],
       device='cuda:0')
tensor([[-0.3500, -0.2986, -1.0000, -1.0000,  0.2506, -0.2956]],
       device='cuda:0')
tensor([[-0.2765,  0.4547, -1.0000, -0.9668,  0.2283, -0.1980]],
       device='cuda:0')
tensor([[-0.5642,  0.2735, -0.9462, -0.9869, -0.3291, -0.2977]],
       device='cuda:0')
tensor([[-0.7564, -0.3769, -1.0000, -1.0000, -0.3330, -0.4868]],
       device='cuda:0')
tensor([[-0.5547, -0.6106, -1.0000, -1.0000,  0.0137, -0.5759]],
       device='cuda:0')
tensor([[-0.2818,  0.0383, -1.0000, -1.0000,  0.2077, -0.2259]],
       device='cuda:0')
tensor([[-0.4329,  0.3001, -1.0000, -0.9853, -0.3030, -0.2440]],
       device='cuda:0')
tensor([[-0.6476, -0.0591, -1.0000, -1.0000, -0.2828, -0.5476]],
       device='cuda:0')
tensor([[-0.6206, -0.5341, -1.0000, -1.0000, -0.1130, -0.4634]],
       device='cuda:0')
tensor([[-0.3631, -0.3081, -1.0000, -1.0000,  0.2107, -0.2826]],
       device='cuda:0')
tensor([[-0.2881,  0.4559, -1.0000, -0.9776,  0.2280, -0.1894]],
       device='cuda:0')
tensor([[-0.5741,  0.2655, -0.9735, -1.0000, -0.3572, -0.3237]],
       device='cuda:0')
tensor([[-0.7576, -0.4227, -1.0000, -1.0000, -0.3485, -0.4989]],
       device='cuda:0')
tensor([[-0.5441, -0.5635, -1.0000, -1.0000,  0.0204, -0.5487]],
       device='cuda:0')
tensor([[-0.2902,  0.1190, -1.0000, -1.0000,  0.2009, -0.2196]],
       device='cuda:0')
tensor([[-0.4749,  0.3333, -1.0000, -0.9587, -0.2645, -0.2527]],
       device='cuda:0')
tensor([[-0.6631, -0.0469, -1.0000, -1.0000, -0.2829, -0.5303]],
       device='cuda:0')
tensor([[-0.6367, -0.5581, -1.0000, -1.0000, -0.1357, -0.4671]],
       device='cuda:0')
tensor([[-0.3783, -0.3572, -1.0000, -1.0000,  0.1973, -0.3010]],
       device='cuda:0')
tensor([[-0.2696,  0.4408, -1.0000, -0.9750,  0.2921, -0.1835]],
       device='cuda:0')
tensor([[-0.5492,  0.3224, -0.9253, -0.9684, -0.3154, -0.2500]],
       device='cuda:0')
tensor([[-0.7639, -0.3171, -1.0000, -1.0000, -0.3284, -0.4768]],
       device='cuda:0')
tensor([[-0.5821, -0.4768, -1.0000, -1.0000,  0.0843, -0.5470]],
       device='cuda:0')
tensor([[-0.3083, -0.0279, -1.0000, -1.0000,  0.1926, -0.1390]],
       device='cuda:0')
tensor([[-0.4291,  0.2767, -1.0000, -1.0000, -0.3259, -0.3049]],
       device='cuda:0')
tensor([[-0.6361, -0.0413, -1.0000, -1.0000, -0.2966, -0.5591]],
       device='cuda:0')
tensor([[-0.6179, -0.5236, -1.0000, -1.0000, -0.1365, -0.4554]],
       device='cuda:0')
tensor([[-0.3275, -0.2948, -1.0000, -1.0000,  0.3054, -0.2967]],
       device='cuda:0')
tensor([[-0.2655,  0.4535, -1.0000, -0.9648,  0.2096, -0.2116]],
       device='cuda:0')
tensor([[-0.5652,  0.2602, -0.9596, -0.9924, -0.3316, -0.3198]],
       device='cuda:0')
tensor([[-0.7476, -0.3834, -1.0000, -1.0000, -0.3279, -0.4912]],
       device='cuda:0')
tensor([[-0.5431, -0.5963, -1.0000, -1.0000,  0.0269, -0.5696]],
       device='cuda:0')
tensor([[-0.2748,  0.1234, -1.0000, -1.0000,  0.2256, -0.1959]],
       device='cuda:0')
tensor([[-0.4611,  0.3572, -1.0000, -0.9473, -0.2449, -0.2187]],
       device='cuda:0')
tensor([[-0.6591, -0.0246, -1.0000, -1.0000, -0.2828, -0.5342]],
       device='cuda:0')
tensor([[-0.6472, -0.5962, -1.0000, -1.0000, -0.1562, -0.4751]],
       device='cuda:0')
tensor([[-0.3859, -0.4203, -1.0000, -1.0000,  0.1950, -0.3213]],
       device='cuda:0')
tensor([[-0.2657,  0.3407, -1.0000, -1.0000,  0.2066, -0.1920]],
       device='cuda:0')
tensor([[-0.5398,  0.2922, -1.0000, -0.9978, -0.3045, -0.2912]],
       device='cuda:0')
tensor([[-0.7306, -0.3424, -1.0000, -1.0000, -0.3360, -0.5220]],
       device='cuda:0')
tensor([[-0.5604, -0.5428, -1.0000, -1.0000,  0.0462, -0.5391]],
       device='cuda:0')
tensor([[-0.2906,  0.0907, -1.0000, -1.0000,  0.2410, -0.1080]],
       device='cuda:0')
tensor([[-0.4621,  0.3584, -1.0000, -0.9427, -0.2560, -0.2059]],
       device='cuda:0')
tensor([[-0.6644, -0.0311, -1.0000, -1.0000, -0.2771, -0.5240]],
       device='cuda:0')
tensor([[-0.6517, -0.5789, -1.0000, -1.0000, -0.1538, -0.4693]],
       device='cuda:0')
tensor([[-0.3946, -0.3991, -1.0000, -1.0000,  0.1922, -0.3092]],
       device='cuda:0')
tensor([[-0.2782,  0.3520, -1.0000, -1.0000,  0.2070, -0.1833]],
       device='cuda:0')
tensor([[-0.5481,  0.2900, -1.0000, -1.0000, -0.3077, -0.2891]],
       device='cuda:0')
tensor([[-0.7395, -0.3774, -1.0000, -1.0000, -0.3397, -0.5157]],
       device='cuda:0')
tensor([[-0.5567, -0.5436, -1.0000, -1.0000,  0.0379, -0.5314]],
       device='cuda:0')
tensor([[-0.2926,  0.1049, -1.0000, -1.0000,  0.2339, -0.1244]],
       device='cuda:0')
tensor([[-0.4654,  0.3634, -1.0000, -0.9386, -0.2398, -0.1986]],
       device='cuda:0')
tensor([[-0.6675, -0.0369, -1.0000, -1.0000, -0.2794, -0.5204]],
       device='cuda:0')
tensor([[-0.6504, -0.5719, -1.0000, -1.0000, -0.1483, -0.4675]],
       device='cuda:0')
tensor([[-0.3938, -0.3855, -1.0000, -1.0000,  0.1916, -0.3042]],
       device='cuda:0')
tensor([[-0.2793,  0.3625, -1.0000, -1.0000,  0.2130, -0.1822]],
       device='cuda:0')
tensor([[-0.5499,  0.2943, -1.0000, -0.9994, -0.3139, -0.2910]],
       device='cuda:0')
tensor([[-0.7423, -0.3805, -1.0000, -1.0000, -0.3432, -0.5154]],
       device='cuda:0')
tensor([[-0.5568, -0.5373, -1.0000, -1.0000,  0.0367, -0.5292]],
       device='cuda:0')
tensor([[-0.2953,  0.1041, -1.0000, -1.0000,  0.2254, -0.1373]],
       device='cuda:0')
tensor([[-0.4672,  0.3618, -1.0000, -0.9402, -0.2445, -0.2054]],
       device='cuda:0')
tensor([[-0.6678, -0.0358, -1.0000, -1.0000, -0.2803, -0.5203]],
       device='cuda:0')
tensor([[-0.6505, -0.5678, -1.0000, -1.0000, -0.1505, -0.4656]],
       device='cuda:0')
tensor([[-0.3930, -0.3827, -1.0000, -1.0000,  0.1930, -0.3063]],
       device='cuda:0')
tensor([[-0.2771,  0.3661, -1.0000, -1.0000,  0.2189, -0.1827]],
       device='cuda:0')
tensor([[-0.5463,  0.3044, -1.0000, -0.9927, -0.3070, -0.2801]],
       device='cuda:0')
tensor([[-0.7426, -0.3626, -1.0000, -1.0000, -0.3415, -0.5130]],
       device='cuda:0')
tensor([[-0.5619, -0.5399, -1.0000, -1.0000,  0.0359, -0.5342]],
       device='cuda:0')
tensor([[-0.2976,  0.0771, -1.0000, -1.0000,  0.2219, -0.1377]],
       device='cuda:0')
tensor([[-0.4629,  0.3506, -1.0000, -0.9491, -0.2800, -0.2246]],
       device='cuda:0')
tensor([[-0.6612, -0.0211, -1.0000, -1.0000, -0.2711, -0.5232]],
       device='cuda:0')
tensor([[-0.6513, -0.5664, -1.0000, -1.0000, -0.1612, -0.4615]],
       device='cuda:0')
tensor([[-0.3951, -0.3951, -1.0000, -1.0000,  0.1956, -0.3103]],
       device='cuda:0')
tensor([[-0.2754,  0.3532, -1.0000, -1.0000,  0.2141, -0.1829]],
       device='cuda:0')
tensor([[-0.5417,  0.3040, -1.0000, -0.9904, -0.2919, -0.2684]],
       device='cuda:0')
tensor([[-0.7387, -0.3506, -1.0000, -1.0000, -0.3369, -0.5136]],
       device='cuda:0')
tensor([[-0.5653, -0.5507, -1.0000, -1.0000,  0.0385, -0.5376]],
       device='cuda:0')
tensor([[-0.2936,  0.0613, -1.0000, -1.0000,  0.2346, -0.1166]],
       device='cuda:0')
tensor([[-0.4418,  0.3133, -1.0000, -0.9669, -0.3021, -0.2126]],
       device='cuda:0')
tensor([[-0.6621, -0.0634, -1.0000, -1.0000, -0.2752, -0.5259]],
       device='cuda:0')
tensor([[-0.6350, -0.5378, -1.0000, -1.0000, -0.1184, -0.4642]],
       device='cuda:0')
tensor([[-0.3833, -0.3193, -1.0000, -1.0000,  0.1952, -0.2797]],
       device='cuda:0')
tensor([[-0.3014,  0.4589, -1.0000, -0.9797,  0.2424, -0.1756]],
       device='cuda:0')
tensor([[-0.5755,  0.2839, -0.9570, -0.9976, -0.3422, -0.2881]],
       device='cuda:0')
tensor([[-0.7715, -0.4176, -1.0000, -1.0000, -0.3496, -0.4903]],
       device='cuda:0')
tensor([[-0.5632, -0.5735, -1.0000, -1.0000,  0.0093, -0.5447]],
       device='cuda:0')
tensor([[-0.3001,  0.0128, -1.0000, -1.0000,  0.1886, -0.2270]],
       device='cuda:0')
tensor([[-0.4339,  0.3023, -1.0000, -0.9946, -0.2983, -0.2612]],
       device='cuda:0')
tensor([[-0.6425, -0.0399, -1.0000, -1.0000, -0.2798, -0.5502]],
       device='cuda:0')
tensor([[-0.6240, -0.5151, -1.0000, -1.0000, -0.1230, -0.4439]],
       device='cuda:0')
tensor([[-0.3725, -0.3055, -1.0000, -1.0000,  0.2079, -0.2797]],
       device='cuda:0')
tensor([[-0.2917,  0.4563, -1.0000, -0.9742,  0.2352, -0.1819]],
       device='cuda:0')
tensor([[-0.5709,  0.2754, -0.9506, -0.9937, -0.3405, -0.2957]],
       device='cuda:0')
tensor([[-0.7645, -0.3998, -1.0000, -1.0000, -0.3430, -0.4879]],
       device='cuda:0')
tensor([[-0.5576, -0.5941, -1.0000, -1.0000,  0.0084, -0.5602]],
       device='cuda:0')
tensor([[-0.2913,  0.0260, -1.0000, -1.0000,  0.1954, -0.2322]],
       device='cuda:0')
tensor([[-0.4357,  0.2971, -1.0000, -0.9922, -0.3020, -0.2528]],
       device='cuda:0')
tensor([[-0.6450, -0.0518, -1.0000, -1.0000, -0.2816, -0.5486]],
       device='cuda:0')
tensor([[-0.6214, -0.5245, -1.0000, -1.0000, -0.1165, -0.4540]],
       device='cuda:0')
^Ctensor([[-0.3669, -0.3043, -1.0000, -1.0000,  0.2095, -0.2786]],
       device='cuda:0')
tensor([[-0.2912,  0.4540, -1.0000, -0.9777,  0.2268, -0.1859]],
       device='cuda:0')
tensor([[-0.5744,  0.2642, -0.9697, -1.0000, -0.3542, -0.3188]],
       device='cuda:0')
tensor([[-0.7582, -0.4209, -1.0000, -1.0000, -0.3471, -0.4963]],
       device='cuda:0')
tensor([[-0.5463, -0.5699, -1.0000, -1.0000,  0.0181, -0.5497]],
       device='cuda:0')
tensor([[-0.2900,  0.1025, -1.0000, -1.0000,  0.1999, -0.2202]],
       device='cuda:0')
tensor([[-0.4717,  0.3320, -1.0000, -0.9585, -0.2678, -0.2500]],
       device='cuda:0')
tensor([[-0.6606, -0.0421, -1.0000, -1.0000, -0.2811, -0.5312]],
       device='cuda:0')
tensor([[-0.6372, -0.5624, -1.0000, -1.0000, -0.1388, -0.4672]],
       device='cuda:0')
tensor([[-0.3797, -0.3655, -1.0000, -1.0000,  0.1964, -0.3021]],
       device='cuda:0')
tensor([[-0.2674,  0.3957, -1.0000, -1.0000,  0.2492, -0.1826]],
       device='cuda:0')
tensor([[-0.5407,  0.3236, -0.9925, -0.9799, -0.3135, -0.2689]],
       device='cuda:0')

Can anyone help me figure out how to export a policy and get it running externally from Isaaac Sim? I think this would be a great problem to solve and would help a lot of people.

I’ve tried contacting the repo owner here:

Issue resolved.

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.