arXiv:1912.02047v2 [cs.CL] 29 Sep 2020
PDF) Refining the state-of-the-art in Machine Translation, optimizing NMT for the JA <-> EN language pair by leveraging personal domain expertise
ConvTranspose1d extremely slow on GPU (T4), even slower than CPU · Issue #41382 · pytorch/pytorch · GitHub
Marian :: Features & Benchmarks
arXiv:1912.02047v2 [cs.CL] 29 Sep 2020
Efficient machine translation
Information | Free Full-Text | Knowledge Distillation: A Method for Making Neural Machine Translation More Efficient
Don't load entire corpus into memory on start up (enhancement request) · Issue #148 · marian-nmt/marian-dev · GitHub
how can i decode faster without adding more gpus · Issue #280 · marian-nmt/ marian-dev · GitHub
PDF) Marian: Fast Neural Machine Translation in C++
Neural Machine Translation – Explained - tcworld magazine
Marian :: Features & Benchmarks
Batched translation speed is very slow compared to Fairseq · Issue #266 · marian-nmt/marian-dev · GitHub
Slow decoding on Marian v1.10.0 · Issue #375 · marian-nmt/marian · GitHub
![Tuning] Results are GPU-number and batch-size dependent · Issue #444 · tensorflow/tensor2tensor · GitHub](https://user-images.githubusercontent.com/15141326/33256270-a3795912-d351-11e7-83e4-ea941ba95dd5.png "Tuning] Results are GPU-number and batch-size dependent · Issue #444 · tensorflow/tensor2tensor · GitHub")
Tuning] Results are GPU-number and batch-size dependent · Issue #444 · tensorflow/tensor2tensor · GitHub
beam search · Issue #186 · marian-nmt/marian · GitHub
Conversation with the creator of OPUS-CAT | La Linterna del Traductor
Efficient machine translation
Information | Free Full-Text | Knowledge Distillation: A Method for Making Neural Machine Translation More Efficient
PDF) SGNMT – A Flexible NMT Decoding Platform for Quick Prototyping of New Models and Search Strategies
problem with workspace > 26000 · Issue #515 · marian-nmt/marian-dev · GitHub
