config_runtimes.txt

Validation enabled + batch size 50 + no jit

[epoch 49] avg train loss: 40.239465771484376: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████| 50/50 [04:19<00:00,  5.29s/it]

Validation disabled + batch size 50 + no jit

[epoch 49] avg train loss: 41.71224888916016: 100%|███████████████████████████████████████████████████████████████████████████████████████████████████| 50/50 [04:09<00:00,  4.94s/it]

Validation disabled + batch size 250 + no jit

[epoch 49] avg train loss: 46.55022587890625: 100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 50/50 [02:49<00:00,  3.28s/it]

Validation disabled + batch size 1000 + no jit

[epoch 49] avg train loss: 54.4259671875: 100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 50/50 [02:36<00:00,  3.12s/it]


JIT warnings:
/home/asal/src/project/pyro/distributions/util.py:96: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!
  elif reversed_shape[i] == 1 and not strict:
/home/asal/src/project/pyro/distributions/util.py:98: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!
  elif reversed_shape[i] != size and (size != 1 or strict):
/home/asal/src/project/pyro/ops/sumproduct.py:121: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!
  any(s < t for s, t in zip_align_right(shape, target_shape)):
/home/asal/src/project/pyro/ops/sumproduct.py:137: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!
  if size != 1]
/home/asal/src/project/pyro/ops/sumproduct.py:144: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!
  if size != 1])
/home/asal/src/project/pyro/ops/einsum/paths.py:172: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!
  heapq.heapify(queue)
/home/asal/anaconda2/envs/pyro-env/lib/python3.7/site-packages/numpy/lib/stride_tricks.py:120: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!
  if any(size < 0 for size in shape):
/home/asal/anaconda2/envs/pyro-env/lib/python3.7/site-packages/numpy/lib/stride_tricks.py:128: TracerWarning: Converting a tensor to a Python index might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!
  op_flags=[op_flag], itershape=shape, order='C')
/home/asal/src/project/pyro/ops/sumproduct.py:124: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!
  if i >= len(shape) or shape[-1-i] < target_shape[-1-i]:
/home/asal/src/project/pyro/ops/sumproduct.py:126: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!
  while smaller_shape and smaller_shape[0] == 1:

Validation disabled + batch size 250 + jit 
[epoch 49] avg train loss: 46.57264775390625: 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 50/50 [07:37<00:00,  9.25s/it]

Validation disabled + batch size 1000 + jit
[epoch 49] avg train loss: 52.344290625: 100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 50/50 [07:14<00:00,  8.63s/it]