tiedeman commited on
Commit
a5241c6
1 Parent(s): 7a459c1

Initial commit

Browse files
README.md ADDED
@@ -0,0 +1,72 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ language:
3
+ - fi
4
+ - en
5
+
6
+ tags:
7
+ - translation
8
+
9
+ license: apache-2.0
10
+ ---
11
+ ### fi-en
12
+
13
+ * source group: Finnish
14
+ * target group: English
15
+ * OPUS readme: [fin-eng](https://github.com/Helsinki-NLP/Tatoeba-Challenge/tree/master/models/fin-eng/README.md)
16
+
17
+ * model: transformer-align
18
+ * source language(s): fin
19
+ * target language(s): eng
20
+ * model: transformer-align
21
+ * pre-processing: normalization + SentencePiece (spm32k,spm32k)
22
+ * download original weights: [opusTCv20210807+bt-2021-08-25.zip](https://object.pouta.csc.fi/Tatoeba-MT-models/fin-eng/opusTCv20210807+bt-2021-08-25.zip)
23
+ * test set translations: [opusTCv20210807+bt-2021-08-25.test.txt](https://object.pouta.csc.fi/Tatoeba-MT-models/fin-eng/opusTCv20210807+bt-2021-08-25.test.txt)
24
+ * test set scores: [opusTCv20210807+bt-2021-08-25.eval.txt](https://object.pouta.csc.fi/Tatoeba-MT-models/fin-eng/opusTCv20210807+bt-2021-08-25.eval.txt)
25
+
26
+ ## Benchmarks
27
+
28
+ | testset | BLEU | chr-F | #sent | #words | BP |
29
+ |---------|-------|-------|-------|--------|----|
30
+ | newsdev2015-enfi.fin-eng | 27.1 | 0.550 | 1500 | 32104 | 0.988 |
31
+ | newstest2015-enfi.fin-eng | 28.5 | 0.560 | 1370 | 27356 | 0.980 |
32
+ | newstest2016-enfi.fin-eng | 31.7 | 0.586 | 3000 | 63043 | 1.000 |
33
+ | newstest2017-enfi.fin-eng | 34.6 | 0.610 | 3002 | 61936 | 0.988 |
34
+ | newstest2018-enfi.fin-eng | 25.4 | 0.530 | 3000 | 62325 | 0.981 |
35
+ | newstest2019-fien.fin-eng | 30.6 | 0.577 | 1996 | 36227 | 0.994 |
36
+ | newstestB2016-enfi.fin-eng | 25.8 | 0.538 | 3000 | 63043 | 0.987 |
37
+ | newstestB2017-enfi.fin-eng | 29.6 | 0.572 | 3002 | 61936 | 0.999 |
38
+ | newstestB2017-fien.fin-eng | 29.6 | 0.572 | 3002 | 61936 | 0.999 |
39
+ | Tatoeba-test-v2021-08-07.fin-eng | 54.1 | 0.700 | 10000 | 75212 | 0.988 |
40
+
41
+
42
+ ### System Info:
43
+ - hf_name: fi-en
44
+ - source_languages: fin
45
+ - target_languages: eng
46
+ - opus_readme_url: https://github.com/Helsinki-NLP/Tatoeba-Challenge/tree/master/models/fin-eng/README.md
47
+ - original_repo: Tatoeba-Challenge
48
+ - tags: ['translation']
49
+ - languages: ['fi', 'en']
50
+ - src_constituents: ('Finnish', {'fin'})
51
+ - tgt_constituents: ('English', {'eng'})
52
+ - src_multilingual: False
53
+ - tgt_multilingual: False
54
+ - long_pair: fin-eng
55
+ - prepro: normalization + SentencePiece (spm32k,spm32k)
56
+ - url_model: https://object.pouta.csc.fi/Tatoeba-MT-models/fin-eng/opusTCv20210807+bt-2021-08-25.zip
57
+ - url_test_set: https://object.pouta.csc.fi/Tatoeba-MT-models/fin-eng/opusTCv20210807+bt-2021-08-25.test.txt
58
+ - src_alpha3: fin
59
+ - tgt_alpha3: eng
60
+ - chrF2_score: 0.7
61
+ - bleu: 54.1
62
+ - src_name: Finnish
63
+ - tgt_name: English
64
+ - train_date: 2021-08-25 00:00:00
65
+ - src_alpha2: fi
66
+ - tgt_alpha2: en
67
+ - prefer_old: False
68
+ - short_pair: fi-en
69
+ - helsinki_git_sha: 2ef219d5b67f0afb0c6b732cd07001d84181f002
70
+ - transformers_git_sha: 12b4d66a80419db30a15e7b9d4208ceb9887c03b
71
+ - port_machine: LM0-400-22516.local
72
+ - port_time: 2021-11-04-21:36
config.json ADDED
@@ -0,0 +1,43 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "activation_dropout": 0.0,
3
+ "activation_function": "swish",
4
+ "architectures": [
5
+ "MarianMTModel"
6
+ ],
7
+ "attention_dropout": 0.0,
8
+ "bad_words_ids": [
9
+ [
10
+ 57829
11
+ ]
12
+ ],
13
+ "bos_token_id": 0,
14
+ "classifier_dropout": 0.0,
15
+ "d_model": 512,
16
+ "decoder_attention_heads": 8,
17
+ "decoder_ffn_dim": 2048,
18
+ "decoder_layerdrop": 0.0,
19
+ "decoder_layers": 6,
20
+ "decoder_start_token_id": 57829,
21
+ "dropout": 0.1,
22
+ "encoder_attention_heads": 8,
23
+ "encoder_ffn_dim": 2048,
24
+ "encoder_layerdrop": 0.0,
25
+ "encoder_layers": 6,
26
+ "eos_token_id": 0,
27
+ "forced_eos_token_id": 0,
28
+ "init_std": 0.02,
29
+ "is_encoder_decoder": true,
30
+ "max_length": 512,
31
+ "max_position_embeddings": 512,
32
+ "model_type": "marian",
33
+ "normalize_embedding": false,
34
+ "num_beams": 6,
35
+ "num_hidden_layers": 6,
36
+ "pad_token_id": 57829,
37
+ "scale_embedding": true,
38
+ "static_position_embeddings": true,
39
+ "torch_dtype": "float16",
40
+ "transformers_version": "4.11.3",
41
+ "use_cache": true,
42
+ "vocab_size": 57830
43
+ }
metadata.json ADDED
@@ -0,0 +1 @@
 
1
+ {"hf_name":"fi-en","source_languages":"fin","target_languages":"eng","opus_readme_url":"https:\/\/github.com\/Helsinki-NLP\/Tatoeba-Challenge\/tree\/master\/models\/fin-eng\/README.md","original_repo":"Tatoeba-Challenge","tags":["translation"],"languages":["fi","en"],"src_constituents":["Finnish",["fin"]],"tgt_constituents":["English",["eng"]],"src_multilingual":false,"tgt_multilingual":false,"long_pair":"fin-eng","prepro":" normalization + SentencePiece (spm32k,spm32k)","url_model":"https:\/\/object.pouta.csc.fi\/Tatoeba-MT-models\/fin-eng\/opusTCv20210807+bt-2021-08-25.zip","url_test_set":"https:\/\/object.pouta.csc.fi\/Tatoeba-MT-models\/fin-eng\/opusTCv20210807+bt-2021-08-25.test.txt","src_alpha3":"fin","tgt_alpha3":"eng","chrF2_score":0.7,"bleu":54.1,"src_name":"Finnish","tgt_name":"English","train_date":1629849600000,"src_alpha2":"fi","tgt_alpha2":"en","prefer_old":false,"short_pair":"fi-en","helsinki_git_sha":"2ef219d5b67f0afb0c6b732cd07001d84181f002","transformers_git_sha":"12b4d66a80419db30a15e7b9d4208ceb9887c03b","port_machine":"LM0-400-22516.local","port_time":"2021-11-04-21:36"}
pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:94536c750d1fd46694fc2b9fbc76e4ff7cad00853f48f67ac5c5b6fad588af38
3
+ size 206909571
source.spm ADDED
Binary file (832 kB). View file
special_tokens_map.json ADDED
@@ -0,0 +1 @@
 
1
+ {"eos_token": "</s>", "unk_token": "<unk>", "pad_token": "<pad>"}
target.spm ADDED
Binary file (790 kB). View file
tokenizer_config.json ADDED
@@ -0,0 +1 @@
 
1
+ {"source_lang": "fin", "target_lang": "eng", "unk_token": "<unk>", "eos_token": "</s>", "pad_token": "<pad>", "model_max_length": 512, "sp_model_kwargs": {}, "special_tokens_map_file": null, "tokenizer_file": null, "name_or_path": "marian_ckpt/fin-eng", "tokenizer_class": "MarianTokenizer"}
vocab.json ADDED
The diff for this file is too large to render. See raw diff