KeyError: 'training' when using dpgen

[ghost]

Greetings everyone,
I plan to use dpgen to develop nueral network potentials for KCl and MgCl2 mixtures. However, after making the input files, I keep getting the following error:

Traceback (most recent call last):
File "/home/abdul/sfw/anaconda3/2019/envs/deepmd_gpu/bin/dpgen", line 8, in
sys.exit(main())
File "/home/abdul/sfw/anaconda3/2019/envs/deepmd_gpu/lib/python3.9/site-packages/dpgen/main.py", line 175, in main
args.func(args)
File "/home/abdul/sfw/anaconda3/2019/envs/deepmd_gpu/lib/python3.9/site-packages/dpgen/generator/run.py", line 3001, in gen_run
run_iter (args.PARAM, args.MACHINE)
File "/home/abdul/sfw/anaconda3/2019/envs/deepmd_gpu/lib/python3.9/site-packages/dpgen/generator/run.py", line 2963, in run_iter
make_train (ii, jdata, mdata)
File "/home/abdul/sfw/anaconda3/2019/envs/deepmd_gpu/lib/python3.9/site-packages/dpgen/generator/run.py", line 324, in make_train
jinput['training']['systems'] = init_data_sys
KeyError: 'training'

From my understanding, this implies that the initial data for training is not found by dpgen.

I checked the directories specified in the .json files. However, the error status does not change. My initial training data consists of 40310 (divided into 10 sets) snapshots obtained from AIMD calculations and I am using ver. 2.0.3 of the deepmd package. I am also using machine.json inputs for job submission with slurm. Using deepmd normally without dpgen is not an issue. Could this perhaps be because of a compatibility issue with deepmd versions > 2.0.0 or perhaps my input files are incorrect?

Below are my input files, any help would be greatly appreciated.

param.json

input_run.txt

machine.json

machine.txt

Thank you for your time.

Abdullah

[wanghan-iapcm]

You are using a deepmd-kit input compatible to v0.12, which is too old. Please upgrade it to v2.x compatibility.

[ghost]

I suspected as much. Thank you for the help!