update the readme

README.md

@@ -3,6 +3,8 @@ license: mit
 ---
 # M2M100 418M
 
+***This an ONNX checkpoint exported from [facebook/m2m100_418M](https://huggingface.co/facebook/m2m100_418M) with [🤗 Optimum](https://huggingface.co/docs/optimum/index) v1.14.1***
+
 M2M100 is a multilingual encoder-decoder (seq-to-seq) model trained for Many-to-Many multilingual translation.
 It was introduced in this [paper](https://arxiv.org/abs/2010.11125) and first released in [this](https://github.com/pytorch/fairseq/tree/master/examples/m2m_100) repository.
 
@@ -41,6 +43,17 @@ tokenizer.batch_decode(generated_tokens, skip_special_tokens=True)
 # => "Life is like a box of chocolate."
 ```
 
+If the checkpoint is not working correctly, it might be due to recent update in the `🤗 Optimum` library, you could export the checkpoint from PyTorch to ONNX yourself with the following:
+
+```python
+from optimum.onnxruntime import ORTModelForSeq2SeqLM
+
+model = ORTModelForSeq2SeqLM.from_pretrained("facebook/m2m100_418M", export=True)
+model.save_pretrained("m2m100_418M/")
+```
+
+Feel free to open a pull request and contribute your update, 🤗 thx!
+
 
 See the [model hub](https://huggingface.co/models?filter=m2m_100) to look for more fine-tuned versions.