Translate Thai AI style

Today we will show you for translate Thai language on AI style, you can using tools for create script such as https://colab.research.google.com/ or https://www.kaggle.com/

Step 1 : Install python program transformers and torch

!pip install transformers
!pip install torch

Step 2 : Call transformers and torch then import pipeline

from transformers import pipeline
import torch

Step 3 :

You’re creating a translation pipeline using the 🤗 Hugging Face transformers library.

pipeline("translation", ...)

• This creates a pipeline for machine translation — i.e., it translates text from one language to another.
• The pipeline function abstracts away the complexity of tokenization, model loading, and inference. Super convenient!

model="facebook/nllb-200-distilled-600M"

• This specifies the model to use.
• facebook/nllb-200-distilled-600M is a multilingual translation model by Meta AI.
  • NLLB stands for No Language Left Behind.
  • It supports 200+ languages.
  • The “distilled” version is smaller and faster, good for efficiency.
  • “600M” means it has 600 million parameters.

torch_dtype=torch.bfloat16

• This sets the model to use the bfloat16 data type.
  • bfloat16 (Brain Float 16) is a reduced-precision format that’s faster and uses less memory — common in modern GPUs like TPUs or newer NVIDIA cards.
  • Only works well if your hardware supports it.

translator = pipeline("translation",
                      model="facebook/nllb-200-distilled-600M",
                      torch_dtype=torch.bfloat16)

Step 4 : Create text for translate to Thai language

text = """
The little boy walked to the park with his dog on a sunny afternoon."""

Step 5 : create text_translated for translate English to Thai

text_translated = translator(text,
                             src_lang="eng_Latn",
                             tgt_lang="tha_Thai")

You can visit link for other language : https://github.com/facebookresearch/flores/blob/main/flores200/README.md

Step 6 : Call text_translated

Can you copy coding on link below : Thank you.

https://colab.research.google.com/drive/1-000q5TWkLyaQwmWxTcgO3O71jjOomtR?usp=sharing