kiddothe2b
commited on
Commit
•
7a846fa
1
Parent(s):
14f0263
100k steps (512)
Browse files- README.md +58 -1
- config.json +32 -0
- pytorch_model.bin +3 -0
- special_tokens_map.json +1 -0
- tokenizer.json +0 -0
- tokenizer_config.json +1 -0
- vocab.txt +0 -0
README.md
CHANGED
@@ -1,3 +1,60 @@
|
|
1 |
---
|
2 |
-
license: cc-by-nc-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
---
|
2 |
+
license: cc-by-nc-4.0
|
3 |
+
pipeline_tag: fill-mask
|
4 |
+
tags:
|
5 |
+
- legal
|
6 |
+
language:
|
7 |
+
- da
|
8 |
+
datasets:
|
9 |
+
- multi_eurlex
|
10 |
+
- DDSC/partial-danish-gigaword-no-twitter
|
11 |
+
model-index:
|
12 |
+
- name: coastalcph/danish-legal-bert-base
|
13 |
+
results: []
|
14 |
---
|
15 |
+
|
16 |
+
# Danish LegalBERT (derivative of Maltehb/danish-bert-botxo)
|
17 |
+
|
18 |
+
This model is a derivative of [Maltehb/danish-bert-botxo](https://huggingface.co/Maltehb/danish-bert-botxo) adapted to legal text. It has been pre-trained on a combination of the Danish part of the MultiEURLEX (Chalkidis et al., 2021) dataset comprising EU legislation and two subsets (`retsinformationdk`, `retspraksis`) of the Danish Gigaword Corpus (Derczynski et al., 2021) comprising legal proceedings.
|
19 |
+
It achieves the following results on the evaluation set:
|
20 |
+
- Loss: -
|
21 |
+
|
22 |
+
## Model description
|
23 |
+
|
24 |
+
This is a BERT model (Devlin et al., 2018) model pre-trained on Danish legal corpora. It follows a base configuration with 12 Transformer layers, each one with 768 hidden units and 12 attention heads.
|
25 |
+
|
26 |
+
## Intended uses & limitations
|
27 |
+
|
28 |
+
More information needed
|
29 |
+
|
30 |
+
## Training and evaluation data
|
31 |
+
|
32 |
+
This model is pre-training on a combination of the Danish part of the MultiEURLEX dataset and two subsets (`retsinformationdk`, `retspraksis`) of the Danish Gigaword Corpus.
|
33 |
+
|
34 |
+
## Training procedure
|
35 |
+
|
36 |
+
The model was initially pre-trained for 500k steps with sequences up to 128 tokens, and then continued pre-training for additional 100k with sequences up to 512 tokens.
|
37 |
+
|
38 |
+
### Training hyperparameters
|
39 |
+
|
40 |
+
The following hyperparameters were used during training:
|
41 |
+
- learning_rate: 0.00001
|
42 |
+
- train_batch_size: 16
|
43 |
+
- eval_batch_size: 16
|
44 |
+
- seed: 42
|
45 |
+
- distributed_type: tpu
|
46 |
+
- num_devices: 8
|
47 |
+
- gradient_accumulation_steps: 2
|
48 |
+
- total_train_batch_size: 256
|
49 |
+
- total_eval_batch_size: 128
|
50 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
51 |
+
- lr_scheduler_type: cosine
|
52 |
+
- lr_scheduler_warmup_ratio: 0.05
|
53 |
+
- training_steps: 100000
|
54 |
+
|
55 |
+
### Training results
|
56 |
+
|
57 |
+
| Training Loss | Length | Step | Validation Loss |
|
58 |
+
|:-------------:|:------:|:-------:|:---------------:|
|
59 |
+
| 1.0030 | 128 | 50000 | - |
|
60 |
+
| 0.9593 | 128 | 100000 | - |
|
config.json
ADDED
@@ -0,0 +1,32 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_name_or_path": "Maltehb/danish-bert-botxo",
|
3 |
+
"architectures": [
|
4 |
+
"BertForMaskedLM"
|
5 |
+
],
|
6 |
+
"attention_probs_dropout_prob": 0.1,
|
7 |
+
"classifier_dropout": null,
|
8 |
+
"directionality": "bidi",
|
9 |
+
"gradient_checkpointing": false,
|
10 |
+
"hidden_act": "gelu",
|
11 |
+
"hidden_dropout_prob": 0.1,
|
12 |
+
"hidden_size": 768,
|
13 |
+
"initializer_range": 0.02,
|
14 |
+
"intermediate_size": 3072,
|
15 |
+
"layer_norm_eps": 1e-12,
|
16 |
+
"max_position_embeddings": 512,
|
17 |
+
"model_type": "bert",
|
18 |
+
"num_attention_heads": 12,
|
19 |
+
"num_hidden_layers": 12,
|
20 |
+
"pad_token_id": 0,
|
21 |
+
"pooler_fc_size": 768,
|
22 |
+
"pooler_num_attention_heads": 12,
|
23 |
+
"pooler_num_fc_layers": 3,
|
24 |
+
"pooler_size_per_head": 128,
|
25 |
+
"pooler_type": "first_token_transform",
|
26 |
+
"position_embedding_type": "absolute",
|
27 |
+
"torch_dtype": "float32",
|
28 |
+
"transformers_version": "4.18.0",
|
29 |
+
"type_vocab_size": 2,
|
30 |
+
"use_cache": true,
|
31 |
+
"vocab_size": 31748
|
32 |
+
}
|
pytorch_model.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:31c60cc5e92d5dcc0d10fe322985fe8a40619267851f2d54d6b66a02431414d3
|
3 |
+
size 539554343
|
special_tokens_map.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
{"unk_token": "[UNK]", "sep_token": "[SEP]", "pad_token": "[PAD]", "cls_token": "[CLS]", "mask_token": "[MASK]"}
|
tokenizer.json
ADDED
The diff for this file is too large to render.
See raw diff
|
|
tokenizer_config.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
{"do_lower_case": true, "unk_token": "[UNK]", "sep_token": "[SEP]", "pad_token": "[PAD]", "cls_token": "[CLS]", "mask_token": "[MASK]", "tokenize_chinese_chars": true, "strip_accents": false, "model_max_length": 512, "special_tokens_map_file": null, "name_or_path": "Maltehb/danish-bert-botxo", "do_basic_tokenize": true, "never_split": null, "tokenizer_class": "BertTokenizer"}
|
vocab.txt
ADDED
The diff for this file is too large to render.
See raw diff
|
|