-
## Summary
Train, test, evaluate OpenNMT / Moses (traditional) statistical Machine translation model and compare the quality of both outputs.
Training should be performed on a *standardized* train…
-
Among the backends that we already have, we should consider also ctransformers2 which is high performance: https://opennmt.net/CTranslate2/generation.html besides #796
-
I get this error when `max_relative_positions: -1` or `max_relative_positions: -2`
```
Traceback (most recent call last):
File "/opt/conda/bin/onmt_release_model", line 33, in
sys.exit(load…
-
Hello, I am very happy to finally wait for the demo of UnitySentis. This surprised me. I am a speech recognition algorithm engineer. We often encounter such problems, the inference speed of the model,…
YLQY updated
4 months ago
-
-
OpenNMT has the option to replace an unknown word in a target text with the most likely word from the source text. e.g.:
http://opennmt.net/OpenNMT/translation/unknowns/
Often times symbols wil…
-
@mcarilli
Opening a new issue even though it is similar to #124
However we seem to see this error only since the upgrade to the new API.
here is more detail posted by a user https://github.com…
-
Having some generation issues with NMT models trained with OpenNMT-py, which include OpenNMT-py versions before flash attention existed and one I'm currently training with the most recent which includ…
-
2.6.2 では bi-RNN の前向き隠れ状態と後向き隠れ状態を concat して使っています。
一方 5.1.2 の機械翻訳の実装例では element-wise addition を行っています。
いま OpenNMT のソースを見ると、オプションで concat と sum (element-wise addition) を切り替えられるようになっていました。
https://gi…
-
The following test cases are failing with TensorFlow 2.7 and are currently disabled:
```text
FAILED opennmt/tests/tflite_test.py::TFLiteTest::testTFLiteInterpreter_0
FAILED opennmt/tests/tflite_t…