Output logits of converted tensflow lite model is different from original google/tapas-small-finetuned-wtq model

[Muhd Baasit]

Hello,

I am implementing table question answering task using google/tapas-small-finetuned-wtq model. for that i have converted the mode into Tf-Lite model but output_logits of the original model and converted TF-Lite model is different. please refer below code. please let me know how i can resolve the issues