matejulcar commited on
Commit
16a2cf1
1 Parent(s): 33b66b7

first release

Browse files
README.md CHANGED
@@ -1,3 +1,34 @@
1
  ---
2
- license: cc-by-sa-4.0
 
 
 
 
 
3
  ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  ---
2
+ tags:
3
+ - generated_from_trainer
4
+ language:
5
+ - sl
6
+ - en
7
+ licence: cc-by-sa-4.0
8
  ---
9
+
10
+ # SloBERTa-SlEng
11
+
12
+ SloBERTa-SlEng is a masked language model, based on the [SloBERTa](https://huggingface.co/EMBEDDIA/sloberta) Slovene model.
13
+
14
+ SloBERTa-SlEng replaces the tokenizer, vocabulary and the embeddings layer of the SloBERTa model.
15
+ The tokenizer and vocabulary used are bilingual, Slovene-English, based on conversational, non-standard, and slang language the model was trained on.
16
+ They are the same as in the [SlEng-bert](https://huggingface.co/cjvt/sleng-bert) model.
17
+ The new embedding weights were initialized from the SloBERTa embeddings.
18
+
19
+ The new SloBERTa-SlEng model is SloBERTa model, which was further pre-trained for two epochs on the conversational English and Slovene corpora,
20
+ the same as the [SlEng-bert](https://huggingface.co/cjvt/sleng-bert) model.
21
+
22
+ ## Training corpora
23
+
24
+ The model was trained on English and Slovene tweets, Slovene corpora [MaCoCu](http://hdl.handle.net/11356/1517) and [Frenk](http://hdl.handle.net/11356/1201),
25
+ and a small subset of English [Oscar](https://huggingface.co/datasets/oscar) corpus. We tried to keep the sizes of English and Slovene corpora as equal as possible.
26
+ Training corpora had in total about 2.7 billion words.
27
+
28
+
29
+ ### Framework versions
30
+
31
+ - Transformers 4.22.0.dev0
32
+ - Pytorch 1.13.0a0+d321be6
33
+ - Datasets 2.4.0
34
+ - Tokenizers 0.12.1
all_results.json ADDED
@@ -0,0 +1,15 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 2.0,
3
+ "eval_accuracy": 0.5992899920812731,
4
+ "eval_loss": 2.087890625,
5
+ "eval_runtime": 4563.1687,
6
+ "eval_samples": 636089,
7
+ "eval_samples_per_second": 139.396,
8
+ "eval_steps_per_second": 2.178,
9
+ "perplexity": 8.067879020505075,
10
+ "train_loss": 0.10199262492233857,
11
+ "train_runtime": 71458.7973,
12
+ "train_samples": 125699346,
13
+ "train_samples_per_second": 3518.093,
14
+ "train_steps_per_second": 27.485
15
+ }
config.json ADDED
@@ -0,0 +1,28 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "./slen-from-sloberta",
3
+ "architectures": [
4
+ "CamembertForMaskedLM"
5
+ ],
6
+ "attention_probs_dropout_prob": 0.1,
7
+ "bos_token_id": 0,
8
+ "classifier_dropout": null,
9
+ "eos_token_id": 2,
10
+ "gradient_checkpointing": false,
11
+ "hidden_act": "gelu",
12
+ "hidden_dropout_prob": 0.1,
13
+ "hidden_size": 768,
14
+ "initializer_range": 0.02,
15
+ "intermediate_size": 3072,
16
+ "layer_norm_eps": 1e-05,
17
+ "max_position_embeddings": 514,
18
+ "model_type": "camembert",
19
+ "num_attention_heads": 12,
20
+ "num_hidden_layers": 12,
21
+ "pad_token_id": 1,
22
+ "position_embedding_type": "absolute",
23
+ "torch_dtype": "float16",
24
+ "transformers_version": "4.22.0.dev0",
25
+ "type_vocab_size": 1,
26
+ "use_cache": true,
27
+ "vocab_size": 40005
28
+ }
eval_results.json ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 2.0,
3
+ "eval_accuracy": 0.5992899920812731,
4
+ "eval_loss": 2.087890625,
5
+ "eval_runtime": 4563.1687,
6
+ "eval_samples": 636089,
7
+ "eval_samples_per_second": 139.396,
8
+ "eval_steps_per_second": 2.178,
9
+ "perplexity": 8.067879020505075
10
+ }
pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c86db419a34f6c0e2d88e8933a800452fc3dfbeadc4862eb7551ae57e150ec8d
3
+ size 233648127
special_tokens_map.json ADDED
@@ -0,0 +1,19 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "additional_special_tokens": [
3
+ "<s>NOTUSED",
4
+ "</s>NOTUSED"
5
+ ],
6
+ "bos_token": "<s>",
7
+ "cls_token": "<s>",
8
+ "eos_token": "</s>",
9
+ "mask_token": {
10
+ "content": "<mask>",
11
+ "lstrip": true,
12
+ "normalized": true,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": "<pad>",
17
+ "sep_token": "</s>",
18
+ "unk_token": "<unk>"
19
+ }
tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
tokenizer_config.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "additional_special_tokens": [
3
+ "<s>NOTUSED",
4
+ "</s>NOTUSED"
5
+ ],
6
+ "bos_token": "<s>",
7
+ "cls_token": "<s>",
8
+ "eos_token": "</s>",
9
+ "mask_token": {
10
+ "__type": "AddedToken",
11
+ "content": "<mask>",
12
+ "lstrip": true,
13
+ "normalized": true,
14
+ "rstrip": false,
15
+ "single_word": false
16
+ },
17
+ "name_or_path": "./slen-from-sloberta",
18
+ "pad_token": "<pad>",
19
+ "sep_token": "</s>",
20
+ "sp_model_kwargs": {},
21
+ "special_tokens_map_file": null,
22
+ "tokenizer_class": "CamembertTokenizer",
23
+ "unk_token": "<unk>"
24
+ }
train_results.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 2.0,
3
+ "train_loss": 0.10199262492233857,
4
+ "train_runtime": 71458.7973,
5
+ "train_samples": 125699346,
6
+ "train_samples_per_second": 3518.093,
7
+ "train_steps_per_second": 27.485
8
+ }
trainer_state.json ADDED
The diff for this file is too large to render. See raw diff
 
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4c8f8e4af5ab3b5506b2586f94a4784233af0c0a55ccfb6fc4f881fc3920bc6d
3
+ size 4347