how to write a training json file? which parameters should be changed?

[tonystarkiss]

I have got the POSCAR and OUTCAR file, you can get it from the link below. and I also convert the OUTCAR to RAW file by dpdata. the elements include Pt, H, O, F
链接：https://pan.baidu.com/s/1dq6716iqAohCS6MDjsm89Q
提取码：eynf

when I ran the training with the example json file water_se_a.json, it run into errors, but it does not output what error.

off course, I do not modify much parameters, but the type_map, I know it is not enough, but I do not know which parameters I shoud change? and how to change? can anybody can write an json for me? that will very help for me to understand it and continue the work.

{
"_comment": " model parameters",
"model": {
"type_map": ["Pt","O", "H","F"],
"descriptor" :{
"type": "se_a",
"sel": [46, 92],
"rcut_smth": 5.80,
"rcut": 6.00,
"neuron": [25, 50, 100],
"resnet_dt": false,
"axis_neuron": 16,
"seed": 1,
"_comment": " that's all"
},
"fitting_net" : {
"neuron": [240, 240, 240],
"resnet_dt": true,
"seed": 1,
"_comment": " that's all"
},
"_comment": " that's all"
},
$ dp train

2021-07-31 16:03:14.118738: I tensorflow/compiler/xla/service/service.cc:175] StreamExecutor device (0): NVIDIA Tesla V100-SXM2-16GB, Compute Capability 7.0
2021-07-31 16:03:14.118765: I tensorflow/compiler/xla/service/service.cc:175] StreamExecutor device (1): NVIDIA Tesla V100-SXM2-16GB, Compute Capability 7.0
2021-07-31 16:03:14.118772: I tensorflow/compiler/xla/service/service.cc:175] StreamExecutor device (2): NVIDIA Tesla V100-SXM2-16GB, Compute Capability 7.0
2021-07-31 16:03:14.118780: I tensorflow/compiler/xla/service/service.cc:175] StreamExecutor device (3): NVIDIA Tesla V100-SXM2-16GB, Compute Capability 7.0
2021-07-31 16:03:14.118787: I tensorflow/compiler/xla/service/service.cc:175] StreamExecutor device (4): NVIDIA Tesla V100-SXM2-16GB, Compute Capability 7.0
2021-07-31 16:03:14.118794: I tensorflow/compiler/xla/service/service.cc:175] StreamExecutor device (5): NVIDIA Tesla V100-SXM2-16GB, Compute Capability 7.0
2021-07-31 16:03:14.118801: I tensorflow/compiler/xla/service/service.cc:175] StreamExecutor device (6): NVIDIA Tesla V100-SXM2-16GB, Compute Capability 7.0
2021-07-31 16:03:14.118808: I tensorflow/compiler/xla/service/service.cc:175] StreamExecutor device (7): NVIDIA Tesla V100-SXM2-16GB, Compute Capability 7.0
Traceback (most recent call last):
File "/opt/deepmd-kit/bin/dp", line 10, in
sys.exit(main())
File "/opt/deepmd-kit/lib/python3.7/site-packages/deepmd/main.py", line 73, in main
train(args)
File "/opt/deepmd-kit/lib/python3.7/site-packages/deepmd/train.py", line 87, in train
_do_work(jdata, run_opt)
File "/opt/deepmd-kit/lib/python3.7/site-packages/deepmd/train.py", line 134, in _do_work
modifier = modifier)
File "/opt/deepmd-kit/lib/python3.7/site-packages/deepmd/DataSystem.py", line 31, in init
modifier = modifier))
File "/opt/deepmd-kit/lib/python3.7/site-packages/deepmd/Data.py", line 25, in init
assert(len(self.type_map) >= max(self.atom_type)+1)
AssertionError
(base) root@2687b2d74806:/data/examples/water/train#

[njzjz]

The length of the list of "sel": [46, 92], should be equal to the element length. Check the document for details. https://github.com/deepmodeling/deepmd-kit/blob/v2.0.0.b3/doc/train-input-auto.rst

[tonystarkiss]

I Have reset "sel" parameters, but encountered the same error:

Traceback (most recent call last):
File "/usr/local/bin/dp", line 8, in
sys.exit(main())
File "/usr/local/lib/python3.8/dist-packages/deepmd/main.py", line 73, in main
train(args)
File "/usr/local/lib/python3.8/dist-packages/deepmd/train.py", line 87, in train
_do_work(jdata, run_opt)
File "/usr/local/lib/python3.8/dist-packages/deepmd/train.py", line 128, in _do_work
data = DeepmdDataSystem(systems,
File "/usr/local/lib/python3.8/dist-packages/deepmd/DataSystem.py", line 27, in init
self.data_systems.append(DeepmdData(ii,
File "/usr/local/lib/python3.8/dist-packages/deepmd/Data.py", line 25, in init
assert(len(self.type_map) >= max(self.atom_type)+1)
AssertionError

the Jason file:

{
"_comment": " model parameters",
"model": {
"type_map": ["F","H", "O","Pt"],
"descriptor" :{
"type": "se_a",
"sel": [120, 164, 82, 10],
"rcut_smth": 5.80,
"rcut": 6.00,
"neuron": [25, 50, 100],
"resnet_dt": false,
"axis_neuron": 16,
"seed": 1,
"_comment": " that's all"
},
"fitting_net" : {
"neuron": [240, 240, 240],
"resnet_dt": true,
"seed": 1,
"_comment": " that's all"
},
"_comment": " that's all"
},

[tonystarkiss]

I have solved the issue, so easy, my type_map.raw file is empty and fill the F O H Pt , it starts to running.

[njzjz]

You can try our new tools DP-GUI to write the training json file: https://deepmodeling.org/dpgui/input/deepmd-kit-2.0

[tonystarkiss]

there are too many optional parameters and other parameters, which parameters are essential? can you point out with "*" in the DP-GUI?