Support for ONNX models & INT8 quantization

[bkaruman]

Thanks for the excellent toolkit.
Are there plans to support ONNX models? It would be nice to see some speed-up data on quantized(INT8) attention.

[feihugis]

@bkaruman We did not have a plan to support ONNX models yet.

[JiushengChen]

There is tool to convert PyTorch model to ONNX model, it should be applicable to FastSeq as well.