Update README.md
Browse files
README.md
CHANGED
|
@@ -1 +1,68 @@
|
|
| 1 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: cc-by-sa-4.0
|
| 3 |
+
library_name: pytorch
|
| 4 |
+
language:
|
| 5 |
+
- ru
|
| 6 |
+
- vep
|
| 7 |
+
datasets:
|
| 8 |
+
- Lynxpda/back-translated-veps-russian
|
| 9 |
+
---
|
| 10 |
+
|
| 11 |
+
# Model Card for Veps - Russian version 1.0
|
| 12 |
+
|
| 13 |
+
A model of translation from Vepsian into Russian.
|
| 14 |
+
In archive initial weights of the model trained with OpenNMT-py (Locomotive).
|
| 15 |
+
The model has 457M parameters and is trained from scratch.
|
| 16 |
+
Also presented are model weights converted for Ctranslate2 and a package for installation and use with Argostranslate/Libretranslate.
|
| 17 |
+
|
| 18 |
+
## Model Architecture and Objective
|
| 19 |
+
|
| 20 |
+
```
|
| 21 |
+
dec_layers: 20
|
| 22 |
+
decoder_type: transformer
|
| 23 |
+
enc_layers: 20
|
| 24 |
+
encoder_type: transformer
|
| 25 |
+
heads: 8
|
| 26 |
+
hidden_size: 512
|
| 27 |
+
max_relative_positions: 20
|
| 28 |
+
model_dtype: fp16
|
| 29 |
+
pos_ffn_activation_fn: gated-gelu
|
| 30 |
+
position_encoding: false
|
| 31 |
+
share_decoder_embeddings: true
|
| 32 |
+
share_embeddings: true
|
| 33 |
+
share_vocab: true
|
| 34 |
+
src_vocab_size: 32000
|
| 35 |
+
tgt_vocab_size: 32000
|
| 36 |
+
transformer_ff: 6144
|
| 37 |
+
word_vec_size: 512
|
| 38 |
+
```
|
| 39 |
+
|
| 40 |
+
# Citing & Authors
|
| 41 |
+
|
| 42 |
+
Authors: Maksim Migukin, Maksim Kuznetsov, Alexey Kutashov.
|
| 43 |
+
|
| 44 |
+
## Credits
|
| 45 |
+
|
| 46 |
+
Data compiled by [Opus](https://opus.nlpl.eu/).
|
| 47 |
+
|
| 48 |
+
Includes pretrained models from [Stanza](https://github.com/stanfordnlp/stanza/).
|
| 49 |
+
|
| 50 |
+
Data from Vepsian [WiKi](https://vep.wikipedia.org/wiki/)
|
| 51 |
+
|
| 52 |
+
Data from [Lehme No 2051 // Open corpus of Vepsian and Karelian languages VepKar.](http://dictorpus.krc.karelia.ru/)
|
| 53 |
+
|
| 54 |
+
Data from [OMAMEDIA](https://omamedia.ru/)
|
| 55 |
+
|
| 56 |
+
CCMatrix
|
| 57 |
+
|
| 58 |
+
http://opus.nlpl.eu/CCMatrix-v1.php
|
| 59 |
+
|
| 60 |
+
If you use the dataset or code, please cite (pdf) and, please, acknowledge OPUS (bib, pdf) as well for this release.
|
| 61 |
+
|
| 62 |
+
This corpus has been extracted from web crawls using the margin-based bitext mining techniques described here. The original distribution is available from http://data.statmt.org/cc-matrix/
|
| 63 |
+
|
| 64 |
+
OpenSubtitles
|
| 65 |
+
|
| 66 |
+
http://opus.nlpl.eu/OpenSubtitles-v2018.php
|
| 67 |
+
|
| 68 |
+
Please cite the following article if you use any part of the corpus in your own work: P. Lison and J. Tiedemann, 2016, OpenSubtitles2016: Extracting Large Parallel Corpora from Movie and TV Subtitles. In Proceedings of the 10th International Conference on Language Resources and Evaluation (LREC 2016)
|