Initial commit

.gitattributes

@@ -25,3 +25,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zstandard filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+*.spm filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,211 @@
+---
+language:
+- be
+- de
+- ru
+- uk
+- zle
+
+tags:
+- translation
+
+license: cc-by-4.0
+model-index:
+- name: opus-mt-tc-big-de-zle
+  results:
+  - task:
+      name: Translation deu-rus
+      type: translation
+      args: deu-rus
+    dataset:
+      name: flores101-devtest
+      type: flores_101
+      args: deu rus devtest
+    metrics:
+       - name: BLEU
+         type: bleu
+         value: 26.3
+  - task:
+      name: Translation deu-ukr
+      type: translation
+      args: deu-ukr
+    dataset:
+      name: flores101-devtest
+      type: flores_101
+      args: deu ukr devtest
+    metrics:
+       - name: BLEU
+         type: bleu
+         value: 24.2
+  - task:
+      name: Translation deu-bel
+      type: translation
+      args: deu-bel
+    dataset:
+      name: tatoeba-test-v2021-08-07
+      type: tatoeba_mt
+      args: deu-bel
+    metrics:
+       - name: BLEU
+         type: bleu
+         value: 29.5
+  - task:
+      name: Translation deu-rus
+      type: translation
+      args: deu-rus
+    dataset:
+      name: tatoeba-test-v2021-08-07
+      type: tatoeba_mt
+      args: deu-rus
+    metrics:
+       - name: BLEU
+         type: bleu
+         value: 46.1
+  - task:
+      name: Translation deu-ukr
+      type: translation
+      args: deu-ukr
+    dataset:
+      name: tatoeba-test-v2021-08-07
+      type: tatoeba_mt
+      args: deu-ukr
+    metrics:
+       - name: BLEU
+         type: bleu
+         value: 40.7
+  - task:
+      name: Translation deu-rus
+      type: translation
+      args: deu-rus
+    dataset:
+      name: newstest2012
+      type: wmt-2012-news
+      args: deu-rus
+    metrics:
+       - name: BLEU
+         type: bleu
+         value: 20.8
+  - task:
+      name: Translation deu-rus
+      type: translation
+      args: deu-rus
+    dataset:
+      name: newstest2013
+      type: wmt-2013-news
+      args: deu-rus
+    metrics:
+       - name: BLEU
+         type: bleu
+         value: 24.9
+---
+# opus-mt-tc-big-de-zle
+
+Neural machine translation model for translating from German (de) to East Slavic languages (zle).
+
+This model is part of the [OPUS-MT project](https://github.com/Helsinki-NLP/Opus-MT), an effort to make neural machine translation models widely available and accessible for many languages in the world. All models are originally trained using the amazing framework of [Marian NMT](https://marian-nmt.github.io/), an efficient NMT implementation written in pure C++. The models have been converted to pyTorch using the transformers library by huggingface. Training data is taken from [OPUS](https://opus.nlpl.eu/) and training pipelines use the procedures of [OPUS-MT-train](https://github.com/Helsinki-NLP/Opus-MT-train).
+
+* Publications: [OPUS-MT – Building open translation services for the World](https://aclanthology.org/2020.eamt-1.61/) and [The Tatoeba Translation Challenge – Realistic Data Sets for Low Resource and Multilingual MT](https://aclanthology.org/2020.wmt-1.139/) (Please, cite if you use this model.)
+
+```
+@inproceedings{tiedemann-thottingal-2020-opus,
+    title = "{OPUS}-{MT} {--} Building open translation services for the World",
+    author = {Tiedemann, J{\"o}rg  and Thottingal, Santhosh},
+    booktitle = "Proceedings of the 22nd Annual Conference of the European Association for Machine Translation",
+    month = nov,
+    year = "2020",
+    address = "Lisboa, Portugal",
+    publisher = "European Association for Machine Translation",
+    url = "https://aclanthology.org/2020.eamt-1.61",
+    pages = "479--480",
+}
+
+@inproceedings{tiedemann-2020-tatoeba,
+    title = "The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}",
+    author = {Tiedemann, J{\"o}rg},
+    booktitle = "Proceedings of the Fifth Conference on Machine Translation",
+    month = nov,
+    year = "2020",
+    address = "Online",
+    publisher = "Association for Computational Linguistics",
+    url = "https://aclanthology.org/2020.wmt-1.139",
+    pages = "1174--1182",
+}
+```
+
+## Model info
+
+* Release: 2022-03-23
+* source language(s): deu
+* target language(s): bel rus ukr
+* valid target language labels: >>bel<< >>rus<< >>ukr<<
+* model: transformer-big
+* data: opusTCv20210807 ([source](https://github.com/Helsinki-NLP/Tatoeba-Challenge))
+* tokenization: SentencePiece (spm32k,spm32k)
+* original model: [opusTCv20210807_transformer-big_2022-03-23.zip](https://object.pouta.csc.fi/Tatoeba-MT-models/deu-zle/opusTCv20210807_transformer-big_2022-03-23.zip)
+* more information released models: [OPUS-MT deu-zle README](https://github.com/Helsinki-NLP/Tatoeba-Challenge/tree/master/models/deu-zle/README.md)
+* more information about the model: [MarianMT](https://huggingface.co/docs/transformers/model_doc/marian)
+
+This is a multilingual translation model with multiple target languages. A sentence initial language token is required in the form of `>>id<<` (id = valid target language ID), e.g. `>>bel<<`
+
+## Usage
+
+A short example code:
+
+```python
+from transformers import MarianMTModel, MarianTokenizer
+
+src_text = [
+    ">>ukr<< Der Soldat hat mir Wasser gegeben.",
+    ">>ukr<< Ich will hier nicht essen."
+]
+
+model_name = "pytorch-models/opus-mt-tc-big-de-zle"
+tokenizer = MarianTokenizer.from_pretrained(model_name)
+model = MarianMTModel.from_pretrained(model_name)
+translated = model.generate(**tokenizer(src_text, return_tensors="pt", padding=True))
+
+for t in translated:
+    print( tokenizer.decode(t, skip_special_tokens=True) )
+
+# expected output:
+#     Солдат дав мені воду.
+#     Я не хочу тут їсти.
+```
+
+You can also use OPUS-MT models with the transformers pipelines, for example:
+
+```python
+from transformers import pipeline
+pipe = pipeline("translation", model="Helsinki-NLP/opus-mt-tc-big-de-zle")
+print(pipe(">>ukr<< Der Soldat hat mir Wasser gegeben."))
+
+# expected output: Солдат дав мені воду.
+```
+
+## Benchmarks
+
+* test set translations: [opusTCv20210807_transformer-big_2022-03-23.test.txt](https://object.pouta.csc.fi/Tatoeba-MT-models/deu-zle/opusTCv20210807_transformer-big_2022-03-23.test.txt)
+* test set scores: [opusTCv20210807_transformer-big_2022-03-23.eval.txt](https://object.pouta.csc.fi/Tatoeba-MT-models/deu-zle/opusTCv20210807_transformer-big_2022-03-23.eval.txt)
+* benchmark results: [benchmark_results.txt](benchmark_results.txt)
+* benchmark output: [benchmark_translations.zip](benchmark_translations.zip)
+
+| langpair | testset | chr-F | BLEU  | #sent | #words |
+|----------|---------|-------|-------|-------|--------|
+| deu-bel | tatoeba-test-v2021-08-07 | 0.53128 | 29.5 | 551 | 3601 |
+| deu-rus | tatoeba-test-v2021-08-07 | 0.67143 | 46.1 | 12800 | 87296 |
+| deu-ukr | tatoeba-test-v2021-08-07 | 0.62737 | 40.7 | 10319 | 56287 |
+| deu-rus | flores101-devtest | 0.54152 | 26.3 | 1012 | 23295 |
+| deu-ukr | flores101-devtest | 0.53286 | 24.2 | 1012 | 22810 |
+| deu-rus | newstest2012 | 0.49409 | 20.8 | 3003 | 64790 |
+| deu-rus | newstest2013 | 0.52631 | 24.9 | 3000 | 58560 |
+
+## Acknowledgements
+
+The work is supported by the [European Language Grid](https://www.european-language-grid.eu/) as [pilot project 2866](https://live.european-language-grid.eu/catalogue/#/resource/projects/2866), by the [FoTran project](https://www.helsinki.fi/en/researchgroups/natural-language-understanding-with-cross-lingual-grounding), funded by the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (grant agreement No 771113), and the [MeMAD project](https://memad.eu/), funded by the European Union’s Horizon 2020 Research and Innovation Programme under grant agreement No 780069. We are also grateful for the generous computational resources and IT infrastructure provided by [CSC -- IT Center for Science](https://www.csc.fi/), Finland.
+
+## Model conversion info
+
+* transformers version: 4.16.2
+* OPUS-MT git hash: 1bdabf7
+* port time: Thu Mar 24 01:29:09 EET 2022
+* port machine: LM0-400-22516.local

benchmark_results.txt

@@ -0,0 +1,17 @@
+deu-bel	flores101-dev	0.38483	9.3	997	23996
+deu-rus	flores101-dev	0.54342	26.4	997	22657
+deu-ukr	flores101-dev	0.52143	23.3	997	21841
+deu-bel	flores101-devtest	0.39045	9.8	1012	24829
+deu-rus	flores101-devtest	0.54152	26.3	1012	23295
+deu-ukr	flores101-devtest	0.53286	24.2	1012	22810
+deu-rus	newstest2012	0.49409	20.8	3003	64790
+deu-rus	newstest2013	0.52631	24.9	3000	58560
+deu-bel	tatoeba-test-v2020-07-28	0.53161	29.5	550	3595
+deu-rus	tatoeba-test-v2020-07-28	0.67044	46.1	10000	67433
+deu-ukr	tatoeba-test-v2020-07-28	0.62698	40.8	10000	54213
+deu-bel	tatoeba-test-v2021-03-30	0.53247	29.5	554	3617
+deu-rus	tatoeba-test-v2021-03-30	0.66964	45.8	11265	76204
+deu-ukr	tatoeba-test-v2021-03-30	0.62795	40.8	10174	55335
+deu-bel	tatoeba-test-v2021-08-07	0.53128	29.5	551	3601
+deu-rus	tatoeba-test-v2021-08-07	0.67143	46.1	12800	87296
+deu-ukr	tatoeba-test-v2021-08-07	0.62737	40.7	10319	56287

benchmark_translations.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d199b5c82be3eb2488358ce9d0fd2638d5d05978f4130074ec37486b668f24dd
+size 4940129

config.json

@@ -0,0 +1,45 @@
+{
+  "activation_dropout": 0.0,
+  "activation_function": "relu",
+  "architectures": [
+    "MarianMTModel"
+  ],
+  "attention_dropout": 0.0,
+  "bad_words_ids": [
+    [
+      61130
+    ]
+  ],
+  "bos_token_id": 0,
+  "classifier_dropout": 0.0,
+  "d_model": 1024,
+  "decoder_attention_heads": 16,
+  "decoder_ffn_dim": 4096,
+  "decoder_layerdrop": 0.0,
+  "decoder_layers": 6,
+  "decoder_start_token_id": 61130,
+  "decoder_vocab_size": 61131,
+  "dropout": 0.1,
+  "encoder_attention_heads": 16,
+  "encoder_ffn_dim": 4096,
+  "encoder_layerdrop": 0.0,
+  "encoder_layers": 6,
+  "eos_token_id": 23541,
+  "forced_eos_token_id": 23541,
+  "init_std": 0.02,
+  "is_encoder_decoder": true,
+  "max_length": 512,
+  "max_position_embeddings": 1024,
+  "model_type": "marian",
+  "normalize_embedding": false,
+  "num_beams": 4,
+  "num_hidden_layers": 6,
+  "pad_token_id": 61130,
+  "scale_embedding": true,
+  "share_encoder_decoder_embeddings": true,
+  "static_position_embeddings": true,
+  "torch_dtype": "float16",
+  "transformers_version": "4.18.0.dev0",
+  "use_cache": true,
+  "vocab_size": 61131
+}

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6e6822077cba0dff5e91a5a8b5a7a2aed7996635769325cce198e4f6ec8fc7b9
+size 603321283

source.spm

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dde526685600b958138d2b3863bf9b8be10774bf8a99573aede7db619c9b5eb0
+size 822915

special_tokens_map.json

@@ -0,0 +1 @@
+{"eos_token": "</s>", "unk_token": "<unk>", "pad_token": "<pad>"}

target.spm

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3341e4c42304e509ef8068c1b0726be116a8ad90b513c62638db210bf47b4fb6
+size 1009491

tokenizer_config.json

@@ -0,0 +1 @@
+{"source_lang": "de", "target_lang": "zle", "unk_token": "<unk>", "eos_token": "</s>", "pad_token": "<pad>", "model_max_length": 512, "sp_model_kwargs": {}, "separate_vocabs": false, "special_tokens_map_file": null, "name_or_path": "marian-models/opusTCv20210807_transformer-big_2022-03-23/de-zle", "tokenizer_class": "MarianTokenizer"}