Links for neural machine translations for the March 20th workshop at INALCO
-
Pascale 's session on Compare AI https://comparia.beta.gouv.fr/
-
FIT position paper https://github.com/nballier/INALCO26/blob/main/EUniWell_Inalco_FIT_Extrait_IMP.pdf
-
More on Tokenisation : OpenAI on-line page for Tokenisation https://platform.openai.com/tokenizer
-
Testing SLMs Small Language Models with LM Studio : https://lmstudio.ai/
-
Using CoT (Chain of Thought) Prompting: Briakou, E., Luo, J., Cherry, C., & Freitag, M. (2024, November). Translating step-by-step: Decomposing the translation process for improved translation quality of long-form texts. In Proceedings of the Ninth Conference on Machine Translation (pp. 1301-1317) https://aclanthology.org/2024.wmt-1.123.pdf
-
From the ERASMUSPLUS MUT_NMT Project (Erasmus Plus) https://ntradumatica.uab.cat/about
A website for NMT literacy: you need a Google account to explore models https://ntradumatica.uab.cat/auth/google
The free book as a deliverable for the project: Kenny, D. (Ed.). (2022). Machine translation for everyone. Lang Sci Press. https://langsci-press.org/catalog/book/342
- Using RAG for translation : <Google's NotebookLM> Upload a relevant text for your translation as a source (eg., Pride and Prejudice) https://www.gutenberg.org/files/1342/1342-0.txt , and use NotebookLM to translate your text, adopting the same textual features.
** Using Google Collab notebooks for applications (AI lteracy: using a jupyter notebook) **
Using OpenAI for translation (speech to text): Whisper4translation.ipynb
(ipynb: interactive python notebook, Jupyter notebooks usable in Google's Colab)
September : Translating and improving translations with Fine-tuning in our joint chapter to be published at the end of this year as a result of the LT-lider project, see for example the set of interviews <MT professional interviews>