Does batch_norm0_gamm have float16 support? I like to run from_darknet with float16 instead of float32 to speed up, but it encounters an error below:
Traceback (most recent call last):
File “./from_darknet_opencl_fp16.py”, line 157, in
graph, lib, params = nnvm.compiler.build(sym, tvm.target.mali(), shape=shape, params=params, dtype=dtype, target_host=‘llvm -mattr=+neon’)
File “/usr/local/lib/python2.7/dist-packages/nnvm-0.8.0-py2.7.egg/nnvm/compiler/build_module.py”, line 232, in build
shape, dtype = _update_shape_dtype(shape, dtype, params)
File “/usr/local/lib/python2.7/dist-packages/nnvm-0.8.0-py2.7.egg/nnvm/compiler/build_module.py”, line 135, in _update_shape_dtype
“%s: dtype not expected %s vs %s” % (k, dtype, v.dtype))
ValueError: batch_norm0_gamma: dtype not expected float16 vs float32
Thanks,