Error log with Baseball example on GPU

[springcoil]

I think this is probably not surprising.

But I get

    transformed_name, transform.apply(distribution), total_size=total_size)
  File "/home/ubuntu/miniconda3/envs/bunnies/lib/python3.5/site-packages/pymc3/model.py", line 508, in Var
    total_size=total_size, model=self)
  File "/home/ubuntu/miniconda3/envs/bunnies/lib/python3.5/site-packages/pymc3/model.py", line 800, in __init__
    self.logp_elemwiset = distribution.logp(self)
  File "/home/ubuntu/miniconda3/envs/bunnies/lib/python3.5/site-packages/pymc3/distributions/transforms.py", line 74, in logp
    return (self.dist.logp(self.transform_used.backward(x)) +
  File "/home/ubuntu/miniconda3/envs/bunnies/lib/python3.5/site-packages/pymc3/distributions/transforms.py", line 145, in backward
    r = tt.exp(x) + a
  File "/home/ubuntu/miniconda3/envs/bunnies/lib/python3.5/site-packages/theano/gof/op.py", line 625, in __call__
    storage_map[ins] = [self._get_test_value(ins)]
  File "/home/ubuntu/miniconda3/envs/bunnies/lib/python3.5/site-packages/theano/gof/op.py", line 562, in _get_test_value
    ret = v.type.filter(v.tag.test_value)
  File "/home/ubuntu/miniconda3/envs/bunnies/lib/python3.5/site-packages/theano/tensor/type.py", line 173, in filter
    raise TypeError(err_msg, data)
TypeError: For compute_test_value, one input test value does not have the requested type.

The error when converting the test value to that variable type:
TensorType(float32, scalar) cannot store accurately value 0.693043224656, it would be represented as 0.6930432319641113. If you do not mind this precision loss, you can: 1) explicitly convert your data to a numpy array of dtype float32, or 2) set "allow_input_downcast=True" when calling "function".```
When running the baseball example on a gpu. I'll run a few more in the examples folders just to see what they say. 

[springcoil]

cc #2033

[twiecki]

Make sure to cast your input data to float32.

[springcoil]

I'm pretty sure that's the solution. I'll try it out soon.

[springcoil]

cc #2035 -- still getting errors.

[twiecki]

Same float64 errors?

[springcoil]

Seems to be caused by some tt.log applications - I'll run pdb :)