diff options
| author | ZJaume <jzaragoza@prompsit.com> | 2022-07-05 12:41:50 +0300 |
|---|---|---|
| committer | ZJaume <jzaragoza@prompsit.com> | 2022-07-05 12:41:50 +0300 |
| commit | 4afb00e2330581b814977ddcb884aa0f937f24f0 (patch) | |
| tree | 2affb2766c17a8361e8890b411c8da616c2f3fd3 | |
| parent | 3d103e36a525f2b6dda8f82d960b6e954fdf83fe (diff) | |
Update TensorFlow minimum version to 2.6.5
| -rw-r--r-- | README.md | 8 | ||||
| -rw-r--r-- | requirements.txt | 2 |
2 files changed, 7 insertions, 3 deletions
@@ -20,6 +20,9 @@ Full models use fine-tuned XLMRoberta ([Unsupervised Cross-lingual Representatio The use of XLMRoberta and 1:10 positive to negative ratio were inspired in the winner of WMT20 Parallel Corpus Filtering Task paper ([Filtering noisy parallel corpus using transformers with proxy task learning](https://www.statmt.org/wmt20/pdf/2020.wmt-1.105.pdf)). ## Installation & Requirements +- Python >= 3.7 +- TensorFlow >= 2.6.5 +- CUDA 11.2 (for training and inference with full models) Bicleaner AI is written in Python and can be installed using `pip`: @@ -57,11 +60,12 @@ After installation, three binary files (`bicleaner-ai-train`, `bicleaner-ai-clas ### TensorFlow TensorFlow 2 will be installed as a dependency and [GPU support](https://www.tensorflow.org/install/gpu) is required for training. -`pip` will install latest TensorFlow but older versions `>=2.3.2` are supported and can be installed if your machine does not meet TensorFlow CUDA requirements. +`pip` will install latest TensorFlow but older versions `>=2.6.5` are supported and can be installed if your machine does not meet TensorFlow CUDA requirements. See [this](https://www.tensorflow.org/install/source#gpu) table for the CUDA and TensorFlow versions compatibility. +But current allowed versions only supoport **CUDA 11.2**. In case you want a different TensorFlow version, you can downgrade using: ```bash -pip install tensorflow==2.3.2 +pip install tensorflow==2.6.5 ``` TensorFlow logging messages are suppressed by default, in case you want to see them you have to explicitly set `TF_CPP_MIN_LOG_LEVEL` environment variable. diff --git a/requirements.txt b/requirements.txt index 8c773e6..d8bac35 100644 --- a/requirements.txt +++ b/requirements.txt @@ -7,7 +7,7 @@ joblib sacremoses bicleaner-hardrules>=2.1 sentencepiece -tensorflow>=2.3.2 +tensorflow>=2.6.5 glove-python-binary==0.2.0 fuzzywuzzy python-Levenshtein |