Instructions to use tencent/Hy-MT2-30B-A3B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use tencent/Hy-MT2-30B-A3B with Transformers:
# Use a pipeline as a high-level helper # Warning: Pipeline type "translation" is no longer supported in transformers v5. # You must load the model directly (see below) or downgrade to v4.x with: # 'pip install "transformers<5.0.0' from transformers import pipeline pipe = pipeline("translation", model="tencent/Hy-MT2-30B-A3B")# Load model directly from transformers import AutoTokenizer, AutoModelForMultimodalLM tokenizer = AutoTokenizer.from_pretrained("tencent/Hy-MT2-30B-A3B") model = AutoModelForMultimodalLM.from_pretrained("tencent/Hy-MT2-30B-A3B") - Notebooks
- Google Colab
- Kaggle
Purpose of this model?
#4
by anujchopra - opened
I don't understand what is the purpose of this model. Normal multilingual text generation model like gemma3 27b can translate into 100 languages. That too very good with proper prompting.
Also this model fails in practical translation. I tried translating english to hinglish ( spoken hindi + english in india ). The results are worse than llama 3.1 8b.
China model is my guess.
Hi, thank you for your feedback, and hinglish is not in our supported language list. We evaluate that Hy-MT2-30B-A3B outperforms Gemma4-31B across all 33 languages we support. Detailed metrics can be found in our report: https://arxiv.org/pdf/2605.22064