Upload 10 files
Browse files- added_tokens.json +137 -0
- artifacts.ckpt +3 -0
- config.json +24 -0
- config.yaml +29 -0
- events.out.tfevents.1692227373.ip-172-31-42-230.eu-central-1.compute.internal +3 -0
- hparams.yaml +1 -0
- pytorch_model.bin +3 -0
- sentencepiece.bpe.model +3 -0
- special_tokens_map.json +18 -0
- tokenizer_config.json +22 -0
added_tokens.json
ADDED
@@ -0,0 +1,137 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"</s_cashprice>": 57549,
|
3 |
+
"</s_changeprice>": 57551,
|
4 |
+
"</s_cnt>": 57529,
|
5 |
+
"</s_creditcardprice>": 57563,
|
6 |
+
"</s_discount_price>": 57557,
|
7 |
+
"</s_discountprice>": 57567,
|
8 |
+
"</s_emoneyprice>": 57569,
|
9 |
+
"</s_etc>": 57541,
|
10 |
+
"</s_h_bic>": 57638,
|
11 |
+
"</s_h_contactperson>": 57632,
|
12 |
+
"</s_h_creditorcity>": 57604,
|
13 |
+
"</s_h_creditorcountry>": 57612,
|
14 |
+
"</s_h_creditoriban>": 57606,
|
15 |
+
"</s_h_creditorname>": 57614,
|
16 |
+
"</s_h_creditorpobox>": 57648,
|
17 |
+
"</s_h_creditorpostcode>": 57592,
|
18 |
+
"</s_h_creditorsalestaxid>": 57596,
|
19 |
+
"</s_h_creditorsearch>": 57652,
|
20 |
+
"</s_h_creditorstreet>": 57590,
|
21 |
+
"</s_h_creditortaxid>": 57620,
|
22 |
+
"</s_h_currency1>": 57602,
|
23 |
+
"</s_h_deldate>": 57610,
|
24 |
+
"</s_h_deliverynotenumber>": 57630,
|
25 |
+
"</s_h_discount>": 57642,
|
26 |
+
"</s_h_documentdate>": 57600,
|
27 |
+
"</s_h_documenttype>": 57582,
|
28 |
+
"</s_h_extracharges>": 57588,
|
29 |
+
"</s_h_grossamount>": 57580,
|
30 |
+
"</s_h_invoicenumber>": 57586,
|
31 |
+
"</s_h_invoicerecipient>": 57616,
|
32 |
+
"</s_h_netamount1>": 57594,
|
33 |
+
"</s_h_ordernumber>": 57622,
|
34 |
+
"</s_h_paymentterms>": 57634,
|
35 |
+
"</s_h_receiptdate>": 57608,
|
36 |
+
"</s_h_recipientsalestaxid>": 57636,
|
37 |
+
"</s_h_recipienttaxid>": 57654,
|
38 |
+
"</s_h_taxamount1>": 57618,
|
39 |
+
"</s_h_taxamount2>": 57626,
|
40 |
+
"</s_h_taxamount3>": 57644,
|
41 |
+
"</s_h_taxbasisamount1>": 57598,
|
42 |
+
"</s_h_taxbasisamount2>": 57628,
|
43 |
+
"</s_h_taxbasisamount3>": 57650,
|
44 |
+
"</s_h_taxrate1>": 57584,
|
45 |
+
"</s_h_taxrate2>": 57624,
|
46 |
+
"</s_h_taxrate3>": 57646,
|
47 |
+
"</s_h_web>": 57640,
|
48 |
+
"</s_itemsubtotal>": 57577,
|
49 |
+
"</s_menu>": 57525,
|
50 |
+
"</s_menuqty_cnt>": 57555,
|
51 |
+
"</s_menutype_cnt>": 57553,
|
52 |
+
"</s_nm>": 57527,
|
53 |
+
"</s_num>": 57565,
|
54 |
+
"</s_othersvc_price>": 57573,
|
55 |
+
"</s_price>": 57531,
|
56 |
+
"</s_service_price>": 57537,
|
57 |
+
"</s_sub>": 57547,
|
58 |
+
"</s_sub_total>": 57533,
|
59 |
+
"</s_subtotal_price>": 57535,
|
60 |
+
"</s_tax_price>": 57539,
|
61 |
+
"</s_total>": 57543,
|
62 |
+
"</s_total_etc>": 57561,
|
63 |
+
"</s_total_price>": 57545,
|
64 |
+
"</s_unitprice>": 57559,
|
65 |
+
"</s_vatyn>": 57575,
|
66 |
+
"</s_void_menu>": 57571,
|
67 |
+
"<s_cashprice>": 57550,
|
68 |
+
"<s_changeprice>": 57552,
|
69 |
+
"<s_cnt>": 57530,
|
70 |
+
"<s_cord-v2>": 57579,
|
71 |
+
"<s_creditcardprice>": 57564,
|
72 |
+
"<s_discount_price>": 57558,
|
73 |
+
"<s_discountprice>": 57568,
|
74 |
+
"<s_emoneyprice>": 57570,
|
75 |
+
"<s_etc>": 57542,
|
76 |
+
"<s_h_bic>": 57639,
|
77 |
+
"<s_h_contactperson>": 57633,
|
78 |
+
"<s_h_creditorcity>": 57605,
|
79 |
+
"<s_h_creditorcountry>": 57613,
|
80 |
+
"<s_h_creditoriban>": 57607,
|
81 |
+
"<s_h_creditorname>": 57615,
|
82 |
+
"<s_h_creditorpobox>": 57649,
|
83 |
+
"<s_h_creditorpostcode>": 57593,
|
84 |
+
"<s_h_creditorsalestaxid>": 57597,
|
85 |
+
"<s_h_creditorsearch>": 57653,
|
86 |
+
"<s_h_creditorstreet>": 57591,
|
87 |
+
"<s_h_creditortaxid>": 57621,
|
88 |
+
"<s_h_currency1>": 57603,
|
89 |
+
"<s_h_deldate>": 57611,
|
90 |
+
"<s_h_deliverynotenumber>": 57631,
|
91 |
+
"<s_h_discount>": 57643,
|
92 |
+
"<s_h_documentdate>": 57601,
|
93 |
+
"<s_h_documenttype>": 57583,
|
94 |
+
"<s_h_extracharges>": 57589,
|
95 |
+
"<s_h_grossamount>": 57581,
|
96 |
+
"<s_h_invoicenumber>": 57587,
|
97 |
+
"<s_h_invoicerecipient>": 57617,
|
98 |
+
"<s_h_netamount1>": 57595,
|
99 |
+
"<s_h_ordernumber>": 57623,
|
100 |
+
"<s_h_paymentterms>": 57635,
|
101 |
+
"<s_h_receiptdate>": 57609,
|
102 |
+
"<s_h_recipientsalestaxid>": 57637,
|
103 |
+
"<s_h_recipienttaxid>": 57655,
|
104 |
+
"<s_h_taxamount1>": 57619,
|
105 |
+
"<s_h_taxamount2>": 57627,
|
106 |
+
"<s_h_taxamount3>": 57645,
|
107 |
+
"<s_h_taxbasisamount1>": 57599,
|
108 |
+
"<s_h_taxbasisamount2>": 57629,
|
109 |
+
"<s_h_taxbasisamount3>": 57651,
|
110 |
+
"<s_h_taxrate1>": 57585,
|
111 |
+
"<s_h_taxrate2>": 57625,
|
112 |
+
"<s_h_taxrate3>": 57647,
|
113 |
+
"<s_h_web>": 57641,
|
114 |
+
"<s_iitcdip>": 57523,
|
115 |
+
"<s_invoice_dataset_large_cleaned_2>": 57656,
|
116 |
+
"<s_itemsubtotal>": 57578,
|
117 |
+
"<s_menu>": 57526,
|
118 |
+
"<s_menuqty_cnt>": 57556,
|
119 |
+
"<s_menutype_cnt>": 57554,
|
120 |
+
"<s_nm>": 57528,
|
121 |
+
"<s_num>": 57566,
|
122 |
+
"<s_othersvc_price>": 57574,
|
123 |
+
"<s_price>": 57532,
|
124 |
+
"<s_service_price>": 57538,
|
125 |
+
"<s_sub>": 57548,
|
126 |
+
"<s_sub_total>": 57534,
|
127 |
+
"<s_subtotal_price>": 57536,
|
128 |
+
"<s_synthdog>": 57524,
|
129 |
+
"<s_tax_price>": 57540,
|
130 |
+
"<s_total>": 57544,
|
131 |
+
"<s_total_etc>": 57562,
|
132 |
+
"<s_total_price>": 57546,
|
133 |
+
"<s_unitprice>": 57560,
|
134 |
+
"<s_vatyn>": 57576,
|
135 |
+
"<s_void_menu>": 57572,
|
136 |
+
"<sep/>": 57522
|
137 |
+
}
|
artifacts.ckpt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:002af47ed9323d85605a541017a3714fd7fa4a8220d14e35132dde57f8e23631
|
3 |
+
size 1609929307
|
config.json
ADDED
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_name_or_path": "naver-clova-ix/donut-base-finetuned-cord-v2",
|
3 |
+
"align_long_axis": false,
|
4 |
+
"architectures": [
|
5 |
+
"DonutModel"
|
6 |
+
],
|
7 |
+
"decoder_layer": 4,
|
8 |
+
"encoder_layer": [
|
9 |
+
2,
|
10 |
+
2,
|
11 |
+
14,
|
12 |
+
2
|
13 |
+
],
|
14 |
+
"input_size": [
|
15 |
+
960,
|
16 |
+
1280
|
17 |
+
],
|
18 |
+
"max_length": 768,
|
19 |
+
"max_position_embeddings": 768,
|
20 |
+
"model_type": "donut",
|
21 |
+
"torch_dtype": "float32",
|
22 |
+
"transformers_version": "4.30.1",
|
23 |
+
"window_size": 10
|
24 |
+
}
|
config.yaml
ADDED
@@ -0,0 +1,29 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
resume_from_checkpoint_path: None
|
2 |
+
result_path: './result'
|
3 |
+
pretrained_model_name_or_path: 'naver-clova-ix/donut-base-finetuned-cord-v2'
|
4 |
+
dataset_name_or_paths:
|
5 |
+
- 'verbrannter/invoice_dataset_large_cleaned_2'
|
6 |
+
sort_json_key: False
|
7 |
+
train_batch_sizes:
|
8 |
+
- 2
|
9 |
+
val_batch_sizes:
|
10 |
+
- 1
|
11 |
+
input_size:
|
12 |
+
- 960
|
13 |
+
- 1280
|
14 |
+
max_length: 768
|
15 |
+
align_long_axis: False
|
16 |
+
num_nodes: 1
|
17 |
+
seed: 2022
|
18 |
+
lr: 3e-05
|
19 |
+
warmup_steps: 150
|
20 |
+
num_training_samples_per_epoch: 500
|
21 |
+
max_epochs: 15
|
22 |
+
max_steps: -1
|
23 |
+
num_workers: 256
|
24 |
+
val_check_interval: 1.0
|
25 |
+
check_val_every_n_epoch: 3
|
26 |
+
gradient_clip_val: 1.0
|
27 |
+
verbose: True
|
28 |
+
exp_name: 'train_cord'
|
29 |
+
exp_version: '20230816_221233'
|
events.out.tfevents.1692227373.ip-172-31-42-230.eu-central-1.compute.internal
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:5f1363a6ef38f820978d0a17300cda7a4bf58fa506b520a906c9e5261acbb2b8
|
3 |
+
size 451377
|
hparams.yaml
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
{}
|
pytorch_model.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:37492bde687b424bf705794c0e7901a895aa573a504a8ec7e3fb403613a620d0
|
3 |
+
size 858865985
|
sentencepiece.bpe.model
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:cb9e3dce4c326195d08fc3dd0f7e2eee1da8595c847bf4c1a9c78b7a82d47e2d
|
3 |
+
size 1296245
|
special_tokens_map.json
ADDED
@@ -0,0 +1,18 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"additional_special_tokens": [
|
3 |
+
"<s_invoice_dataset_large_cleaned_2>"
|
4 |
+
],
|
5 |
+
"bos_token": "<s>",
|
6 |
+
"cls_token": "<s>",
|
7 |
+
"eos_token": "</s>",
|
8 |
+
"mask_token": {
|
9 |
+
"content": "<mask>",
|
10 |
+
"lstrip": true,
|
11 |
+
"normalized": true,
|
12 |
+
"rstrip": false,
|
13 |
+
"single_word": false
|
14 |
+
},
|
15 |
+
"pad_token": "<pad>",
|
16 |
+
"sep_token": "</s>",
|
17 |
+
"unk_token": "<unk>"
|
18 |
+
}
|
tokenizer_config.json
ADDED
@@ -0,0 +1,22 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"bos_token": "<s>",
|
3 |
+
"clean_up_tokenization_spaces": true,
|
4 |
+
"cls_token": "<s>",
|
5 |
+
"eos_token": "</s>",
|
6 |
+
"from_slow": true,
|
7 |
+
"mask_token": {
|
8 |
+
"__type": "AddedToken",
|
9 |
+
"content": "<mask>",
|
10 |
+
"lstrip": true,
|
11 |
+
"normalized": true,
|
12 |
+
"rstrip": false,
|
13 |
+
"single_word": false
|
14 |
+
},
|
15 |
+
"model_max_length": 1000000000000000019884624838656,
|
16 |
+
"pad_token": "<pad>",
|
17 |
+
"processor_class": "DonutProcessor",
|
18 |
+
"sep_token": "</s>",
|
19 |
+
"sp_model_kwargs": {},
|
20 |
+
"tokenizer_class": "XLMRobertaTokenizer",
|
21 |
+
"unk_token": "<unk>"
|
22 |
+
}
|