Upload README.md with huggingface_hub
Browse files
README.md
CHANGED
|
@@ -1,20 +1,14 @@
|
|
| 1 |
# Translator
|
| 2 |
-
|
|
|
|
|
|
|
| 3 |
|
| 4 |
# How to launch
|
| 5 |
-
-
|
| 6 |
-
|
| 7 |
-
|
| 8 |
-
```
|
| 9 |
-
|
| 10 |
-
#
|
| 11 |
-
|
| 12 |
-
|
| 13 |
-
[Link to get from OPUS](https://opus.nlpl.eu/results/en&ru/corpus-result-table)
|
| 14 |
-
|
| 15 |
-
[Direct download link](https://object.pouta.csc.fi/OPUS-OpenSubtitles/v2024/moses/en-ru.txt.zip)
|
| 16 |
-
|
| 17 |
-
# Tokenizer
|
| 18 |
-
We use Sentencepiece, params for the tokenizer:
|
| 19 |
-
- Vocabulary length = 10000
|
| 20 |
-
- Training pairs = 200000
|
|
|
|
| 1 |
# Translator
|
| 2 |
+
This is a research project to create a translator from an article Attention Is All You Need.
|
| 3 |
+
At current state, I don't have enough resources to train a model for this task,
|
| 4 |
+
so I'm presenting only the decoder, that can generate some text based on the input.
|
| 5 |
|
| 6 |
# How to launch
|
| 7 |
+
- Clone repository
|
| 8 |
+
- Install missing dependencies
|
| 9 |
+
- Run code
|
| 10 |
+
```python
|
| 11 |
+
from Translator import Writer
|
| 12 |
+
writer = Writer.from_pretrained("Sashavav/Translator") # .to("cuda")
|
| 13 |
+
print(writer(input_seq="One day I saw a "))
|
| 14 |
+
```
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|