opus-mt-hi-en-onnx

Export

File layout

./                      fp32 ONNX graphs (encoder_model.onnx, decoder_model.onnx, decoder_with_past_model.onnx) + tokenizer files
./int8/                 int8 dynamic-quantized ONNX graphs

Parity check

• src: नमस्ते, आज आप कैसे हैं?
  • pytorch greedy: Hello, how are you today?
  • onnx fp32 greedy: Hello, how are you today? (match)
  • pytorch beam4: Hello, how are you today?
  • onnx fp32 beam4: Hello, how are you today? (match)
  • onnx int8 greedy: Hello, how are you today?
• src: बहुत बहुत धन्यवाद।
  • pytorch greedy: Thank you very much.
  • onnx fp32 greedy: Thank you very much. (match)
  • pytorch beam4: Thank you very much.
  • onnx fp32 beam4: Thank you very much. (match)
  • onnx int8 greedy: Thank you very much.

Usage