'Unsupported HDF5 version: 1.10.1' when running train.lua

[ajakk]

I cannot seem to start training due to an unsupported HDF5 version error. I installed the torch-hdf5 library using the tutorial in the README. My H5 and JSON training data files are both in the directory data/. The command I'm using along with the error it throws is as follows.

> th train.lua -input_h5 data/trainingData.h5 -input_json data/trainingData.json
/home/jake/Downloads/torch/install/bin/luajit: ...ake/Downloads/torch/install/share/lua/5.1/trep
/init.lua:389: ...ake/Downloads/torch/install/share/lua/5.1/trepl/init.lua:389: .../jake/Downloads/torch/install/share/lua/5.1/hdf5/ffi.lua:71: Unsupported HDF5 version: 1.10.1
stack traceback:
    [C]: in function 'error'
        ...ake/Downloads/torch/install/share/lua/5.1/trepl/init.lua:389: in function 'require'
    train.lua:6: in main chunk
    [C]: in function 'dofile'
    ...oads/torch/install/lib/luarocks/rocks/trepl/scm-1/bin/th:150: in main chunk
    [C]: at 0x7a76420520

[ajakk]

Derp. I'm running Arch Linux, and it seems this program requires an HDF5 library that predates the current one on the Arch repos. I've now gotten past this problem by using hdf5-1.8.14-1.

[martin-g-lee]

I fixed the problem by updating the required version number in order to work with version 1.10.

Edit: /Users//torch/install/share/lua/5.2/hdf5/ffi.lua
change
"if maj[0] ~= 1 or min[0] ~= 8 then"
to
"if maj[0] ~= 1 or min[0] ~= 10 then"

[Sisim]

My Ubuntu Version was 18 LTS.
I downloaded the right version 1.8.12 from here
https://support.hdfgroup.org/ftp/HDF5/releases/hdf5-1.8/hdf5-1.8.12/obtain51812.html
and i just reconfigured the
~torch/install/share/lua/5.1/hdf5/config.lua
to the new paths of
hdf5.h and libhdf5.so
the paths are found in downloaded directory if you search for it.

[mememimis]

@Sisim You are a saviour! your solution works!!!!

[Zunaick]

My Ubuntu Version was 18 LTS.
I downloaded the right version 1.8.12 from here
https://support.hdfgroup.org/ftp/HDF5/releases/hdf5-1.8/hdf5-1.8.12/obtain51812.html
and i just reconfigured the
~torch/install/share/lua/5.1/hdf5/config.lua
to the new paths of
hdf5.h and libhdf5.so
the paths are found in downloaded directory if you search for it.

I did the same but now I am encountering new error. I did search for it but I am not able to fix it. I am pasting my command along with the error encountered.

/data/tools/Basset-master/src/basset_train.lua -job pretrained_params.txt -save cd4_cnn out.h5
{
  conv_filter_sizes : 
    {
      1 : 19
      2 : 11
      3 : 7
    }
  weight_norm : 7
  momentum : 0.98
  learning_rate : 0.002
  hidden_units : 
    {
      1 : 1000
      2 : 1000
    }
  conv_filters : 
    {
      1 : 300
      2 : 200
      3 : 200
    }
  hidden_dropouts : 
    {
      1 : 0.3
      2 : 0.3
    }
  pool_width : 
    {
      1 : 3
      2 : 4
      3 : 4
    }
}
seq_len: 600, filter_size: 19, pad_width: 18	
seq_len: 200, filter_size: 11, pad_width: 10	
seq_len: 50, filter_size: 7, pad_width: 6	
nn.Sequential {
  [input -> (1) -> (2) -> (3) -> (4) -> (5) -> (6) -> (7) -> (8) -> (9) -> (10) -> (11) -> (12) -> (13) -> (14) -> (15) -> (16) -> (17) -> (18) -> (19) -> (20) -> (21) -> (22) -> (23) -> output]
  (1): nn.SpatialConvolution(4 -> 300, 19x1, 1,1, 9,0)
  (2): nn.SpatialBatchNormalization (4D) (300)
  (3): nn.ReLU
  (4): nn.SpatialMaxPooling(3x1, 3,1)
  (5): nn.SpatialConvolution(300 -> 200, 11x1, 1,1, 5,0)
  (6): nn.SpatialBatchNormalization (4D) (200)
  (7): nn.ReLU
  (8): nn.SpatialMaxPooling(4x1, 4,1)
  (9): nn.SpatialConvolution(200 -> 200, 7x1, 1,1, 3,0)
  (10): nn.SpatialBatchNormalization (4D) (200)
  (11): nn.ReLU
  (12): nn.SpatialMaxPooling(4x1, 4,1)
  (13): nn.Reshape(2600)
  (14): nn.Linear(2600 -> 1000)
  (15): nn.BatchNormalization (2D) (1000)
  (16): nn.ReLU
  (17): nn.Dropout(0.300000)
  (18): nn.Linear(1000 -> 1000)
  (19): nn.BatchNormalization (2D) (1000)
  (20): nn.ReLU
  (21): nn.Dropout(0.300000)
  (22): nn.Linear(1000 -> 1)
  (23): nn.Sigmoid
}
/data/torch/install/bin/luajit: /data/torch/install/share/lua/5.1/hdf5/ffi.lua:335: Reading data of class ENUM(50331749) is unsupported
stack traceback:
	[C]: in function 'error'
	/data/juhi/torch/install/share/lua/5.1/hdf5/ffi.lua:335: in function '_getTorchType'
	/data/juhi/torch/install/share/lua/5.1/hdf5/dataset.lua:88: in function 'getTensorFactory'
	/data/juhi/torch/install/share/lua/5.1/hdf5/dataset.lua:138: in function 'partial'
	/data/juhi/tools/Basset-master/src/batcher.lua:39: in function 'next'
	/data/juhi/tools/Basset-master/src/convnet.lua:1009: in function 'train_epoch'
	/data/juhi/tools/Basset-master/src/basset_train.lua:156: in main chunk
	[C]: in function 'dofile'
	...juhi/torch/install/lib/luarocks/rocks/trepl/scm-1/bin/th:150: in main chunk
	[C]: at 0x004064f