File size: 2,889 Bytes
d9776fb |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 |
---
license: mit
datasets:
- uvci/Koumankan_mt_dyu_fr
language:
- fr
metrics:
- bleu
- chrf
library_name: adapter-transformers
pipeline_tag: translation
---
# Adeptschneider/dyu-fr-joeynmt
A machine translation model that translates Dyula to French using the [JoeyNMT framework](https://github.com/joeynmt/joeynmt).
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Usage
### Load and use for inference
```python
import torch
from joeynmt.config import load_config, parse_global_args
from joeynmt.prediction import predict, prepare
from huggingface_hub import snapshot_download
# Download model
snapshot_download(
repo_id="Adeptschneider/joeynmt-dyu-fr-v19.0",
local_dir="/path/to/save/locally"
)
# Define model interface
class JoeyNMTModel:
'''
JoeyNMTModel which load JoeyNMT model for inference.
:param config_path: Path to YAML config file
:param n_best: return this many hypotheses, <= beam (currently only 1)
'''
def __init__(self, config_path: str, n_best: int = 1):
seed = 42
torch.manual_seed(seed)
cfg = load_config(config_path)
args = parse_global_args(cfg, rank=0, mode="translate")
self.args = args._replace(test=args.test._replace(n_best=n_best))
# build model
self.model, _, _, self.test_data = prepare(self.args, rank=0, mode="translate")
def _translate_data(self):
_, _, hypotheses, trg_tokens, trg_scores, _ = predict(
model=self.model,
data=self.test_data,
compute_loss=False,
device=self.args.device,
rank=0,
n_gpu=self.args.n_gpu,
normalization="none",
num_workers=self.args.num_workers,
args=self.args.test,
autocast=self.args.autocast,
)
return hypotheses, trg_tokens, trg_scores
def translate(self, sentence) -> list:
'''
Translate the given sentence.
:param sentence: Sentence to be translated
:return:
- translations: (list of str) possible translations of the sentence.
'''
self.test_data.set_item(sentence.strip())
translations, _, _ = self._translate_data()
assert len(translations) == len(self.test_data) * self.args.test.n_best
self.test_data.reset_cache()
return translations
# Load model
config_path = "/path/to/lean_model/config_local.yaml" # Change this to the path to your model congig file
model = JoeyNMTModel(config_path=config_path, n_best=1)
# Translate
model.translate(sentence="i tɔgɔ bi cogodɔ")
```
## Training procedure
### Training hyperparameters
More information needed
### Training results
More information needed
### Framework versions
- JoeyNMT 2.3.0
- Torch 2.0.1+cu118 |