Hopefully this helps someone else.
In the case when there is a error that something is not json serializable the answers above will not work. If you only want to save it so that is human readable then you need to recursively call string on all the non dictionary elements of your dictionary. If you want to load it later then save it as a pickle file then load it (e.g. torch.save(obj, f)
works fine).
This is what worked for me:
#%%
def _to_json_dict_with_strings(dictionary):
"""
Convert dict to dict with leafs only being strings. So it recursively makes keys to strings
if they are not dictionaries.
Use case:
- saving dictionary of tensors (convert the tensors to strins!)
- saving arguments from script (e.g. argparse) for it to be pretty
e.g.
"""
if type(dictionary) != dict:
return str(dictionary)
d = {k: _to_json_dict_with_strings(v) for k, v in dictionary.items()}
return d
def to_json(dic):
import types
import argparse
if type(dic) is dict:
dic = dict(dic)
else:
dic = dic.__dict__
return _to_json_dict_with_strings(dic)
def save_to_json_pretty(dic, path, mode='w', indent=4, sort_keys=True):
import json
with open(path, mode) as f:
json.dump(to_json(dic), f, indent=indent, sort_keys=sort_keys)
def my_pprint(dic):
"""
@param dic:
@return:
Note: this is not the same as pprint.
"""
import json
# make all keys strings recursively with their naitve str function
dic = to_json(dic)
# pretty print
pretty_dic = json.dumps(dic, indent=4, sort_keys=True)
print(pretty_dic)
# print(json.dumps(dic, indent=4, sort_keys=True))
# return pretty_dic
import torch
# import json # results in non serializabe errors for torch.Tensors
from pprint import pprint
dic = {'x': torch.randn(1, 3), 'rec': {'y': torch.randn(1, 3)}}
my_pprint(dic)
pprint(dic)
output:
{
"rec": {
"y": "tensor([[-0.3137, 0.3138, 1.2894]])"
},
"x": "tensor([[-1.5909, 0.0516, -1.5445]])"
}
{'rec': {'y': tensor([[-0.3137, 0.3138, 1.2894]])},
'x': tensor([[-1.5909, 0.0516, -1.5445]])}
I don't know why returning the string then printing it doesn't work but it seems you have to put the dumps directly in the print statement. Note pprint
as it has been suggested already works too. Note not all objects can be converted to a dict with dict(dic)
which is why some of my code has checks on this condition.
Context:
I wanted to save pytorch strings but I kept getting the error:
TypeError: tensor is not JSON serializable
so I coded the above. Note that yes, in pytorch you use torch.save
but pickle files aren't readable. Check this related post: https://discuss.pytorch.org/t/typeerror-tensor-is-not-json-serializable/36065/3
PPrint also has indent arguments but I didn't like how it looks:
pprint(stats, indent=4, sort_dicts=True)
output:
{ 'cca': { 'all': {'avg': tensor(0.5132), 'std': tensor(0.1532)},
'avg': tensor([0.5993, 0.5571, 0.4910, 0.4053]),
'rep': {'avg': tensor(0.5491), 'std': tensor(0.0743)},
'std': tensor([0.0316, 0.0368, 0.0910, 0.2490])},
'cka': { 'all': {'avg': tensor(0.7885), 'std': tensor(0.3449)},
'avg': tensor([1.0000, 0.9840, 0.9442, 0.2260]),
'rep': {'avg': tensor(0.9761), 'std': tensor(0.0468)},
'std': tensor([5.9043e-07, 2.9688e-02, 6.3634e-02, 2.1686e-01])},
'cosine': { 'all': {'avg': tensor(0.5931), 'std': tensor(0.7158)},
'avg': tensor([ 0.9825, 0.9001, 0.7909, -0.3012]),
'rep': {'avg': tensor(0.8912), 'std': tensor(0.1571)},
'std': tensor([0.0371, 0.1232, 0.1976, 0.9536])},
'nes': { 'all': {'avg': tensor(0.6771), 'std': tensor(0.2891)},
'avg': tensor([0.9326, 0.8038, 0.6852, 0.2867]),
'rep': {'avg': tensor(0.8072), 'std': tensor(0.1596)},
'std': tensor([0.0695, 0.1266, 0.1578, 0.2339])},
'nes_output': { 'all': {'avg': None, 'std': None},
'avg': tensor(0.2975),
'rep': {'avg': None, 'std': None},
'std': tensor(0.0945)},
'query_loss': { 'all': {'avg': None, 'std': None},
'avg': tensor(12.3746),
'rep': {'avg': None, 'std': None},
'std': tensor(13.7910)}}
compare to:
{
"cca": {
"all": {
"avg": "tensor(0.5144)",
"std": "tensor(0.1553)"
},
"avg": "tensor([0.6023, 0.5612, 0.4874, 0.4066])",
"rep": {
"avg": "tensor(0.5503)",
"std": "tensor(0.0796)"
},
"std": "tensor([0.0285, 0.0367, 0.1004, 0.2493])"
},
"cka": {
"all": {
"avg": "tensor(0.7888)",
"std": "tensor(0.3444)"
},
"avg": "tensor([1.0000, 0.9840, 0.9439, 0.2271])",
"rep": {
"avg": "tensor(0.9760)",
"std": "tensor(0.0468)"
},
"std": "tensor([5.7627e-07, 2.9689e-02, 6.3541e-02, 2.1684e-01])"
},
"cosine": {
"all": {
"avg": "tensor(0.5945)",
"std": "tensor(0.7146)"
},
"avg": "tensor([ 0.9825, 0.9001, 0.7907, -0.2953])",
"rep": {
"avg": "tensor(0.8911)",
"std": "tensor(0.1571)"
},
"std": "tensor([0.0371, 0.1231, 0.1975, 0.9554])"
},
"nes": {
"all": {
"avg": "tensor(0.6773)",
"std": "tensor(0.2886)"
},
"avg": "tensor([0.9326, 0.8037, 0.6849, 0.2881])",
"rep": {
"avg": "tensor(0.8070)",
"std": "tensor(0.1595)"
},
"std": "tensor([0.0695, 0.1265, 0.1576, 0.2341])"
},
"nes_output": {
"all": {
"avg": "None",
"std": "None"
},
"avg": "tensor(0.2976)",
"rep": {
"avg": "None",
"std": "None"
},
"std": "tensor(0.0945)"
},
"query_loss": {
"all": {
"avg": "None",
"std": "None"
},
"avg": "tensor(12.3616)",
"rep": {
"avg": "None",
"std": "None"
},
"std": "tensor(13.7976)"
}
}