Matej Martinc
commited on
Commit
•
88a2027
1
Parent(s):
25d64e7
adding model files
Browse files- README.md +44 -3
- config.json +25 -0
- pytorch_model.bin +3 -0
- special_tokens_map.json +7 -0
- tokenizer.json +0 -0
- tokenizer_config.json +3 -0
README.md
CHANGED
@@ -1,3 +1,44 @@
|
|
1 |
-
|
2 |
-
|
3 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# bert-small-buddhist-nonbuddhist-sanskrit
|
2 |
+
|
3 |
+
BERT model trained on a lemmatized corpus containing Buddhist and non-Buddhist Sanskrit texts.
|
4 |
+
|
5 |
+
## Model description
|
6 |
+
|
7 |
+
The model has the bert architecture and was pretrained from scratch as a masked language model
|
8 |
+
on the lemmatized Sanskrit corpus. Due to lack of resources and to prevent overfitting, the model is smaller than bert-base,
|
9 |
+
i.e. the number of attention heads and hidden layers have been reduced to 8 and the context has been reduced to 128 tokens. Vocabulary size is 10000 tokens.
|
10 |
+
|
11 |
+
## How to use it
|
12 |
+
|
13 |
+
```
|
14 |
+
model = AutoModelForMaskedLM.from_pretrained("Matej/bert-small-buddhist-nonbuddhist-sanskrit")
|
15 |
+
tokenizer = AutoTokenizer.from_pretrained("Matej/bert-small-buddhist-nonbuddhist-sanskrit", use_fast=True)
|
16 |
+
```
|
17 |
+
|
18 |
+
## Intended uses & limitations
|
19 |
+
|
20 |
+
MIT license, no limitations
|
21 |
+
|
22 |
+
## Training and evaluation data
|
23 |
+
|
24 |
+
See the paper 'Embeddings models for Buddhist Sanskrit' for details on the corpora and the evaluation procedure.
|
25 |
+
|
26 |
+
### Training hyperparameters
|
27 |
+
|
28 |
+
The following hyperparameters were used during training:
|
29 |
+
- learning_rate: 5e-05
|
30 |
+
- train_batch_size: 32
|
31 |
+
- eval_batch_size: 4
|
32 |
+
- seed: 42
|
33 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
34 |
+
- lr_scheduler_type: linear
|
35 |
+
- num_epochs: 200
|
36 |
+
|
37 |
+
### Framework versions
|
38 |
+
|
39 |
+
- Transformers 4.20.0
|
40 |
+
- Pytorch 1.9.0
|
41 |
+
- Datasets 2.3.2
|
42 |
+
- Tokenizers 0.12.1
|
43 |
+
|
44 |
+
|
config.json
ADDED
@@ -0,0 +1,25 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_name_or_path": "configs/models_reference_tokens_all_short_small.json",
|
3 |
+
"architectures": [
|
4 |
+
"BertForMaskedLM"
|
5 |
+
],
|
6 |
+
"attention_probs_dropout_prob": 0.1,
|
7 |
+
"classifier_dropout": null,
|
8 |
+
"hidden_act": "gelu",
|
9 |
+
"hidden_dropout_prob": 0.1,
|
10 |
+
"hidden_size": 768,
|
11 |
+
"initializer_range": 0.02,
|
12 |
+
"intermediate_size": 3072,
|
13 |
+
"layer_norm_eps": 1e-12,
|
14 |
+
"max_position_embeddings": 128,
|
15 |
+
"model_type": "bert",
|
16 |
+
"num_attention_heads": 8,
|
17 |
+
"num_hidden_layers": 8,
|
18 |
+
"pad_token_id": 0,
|
19 |
+
"position_embedding_type": "absolute",
|
20 |
+
"torch_dtype": "float32",
|
21 |
+
"transformers_version": "4.20.1",
|
22 |
+
"type_vocab_size": 2,
|
23 |
+
"use_cache": true,
|
24 |
+
"vocab_size": 10000
|
25 |
+
}
|
pytorch_model.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:868f218809024aa909b83949a12ec39dacbf173141e27fd5d3890e40427c559d
|
3 |
+
size 260392875
|
special_tokens_map.json
ADDED
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"cls_token": "[CLS]",
|
3 |
+
"mask_token": "[MASK]",
|
4 |
+
"pad_token": "[PAD]",
|
5 |
+
"sep_token": "[SEP]",
|
6 |
+
"unk_token": "[UNK]"
|
7 |
+
}
|
tokenizer.json
ADDED
The diff for this file is too large to render.
See raw diff
|
|
tokenizer_config.json
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"tokenizer_class": "PreTrainedTokenizerFast"
|
3 |
+
}
|