Add TF weights
Model converted by the transformers' pt_to_tf CLI. All converted model outputs and hidden layers were validated against its Pytorch counterpart.
Maximum crossload output difference=1.482e-05; Maximum crossload hidden layer difference=1.122e-03;
Maximum conversion output difference=1.482e-05; Maximum conversion hidden layer difference=1.122e-03;
List of maximum output differences above the threshold (1e-19):
logits: 1.216e-05
cls_logits: 1.192e-05
distillation_logits: 1.482e-05
List of maximum hidden layer differences above the threshold (1e-19):
hidden_states[0]: 7.391e-06
hidden_states[1]: 1.657e-05
hidden_states[2]: 2.766e-05
hidden_states[3]: 2.813e-05
hidden_states[4]: 2.557e-05
hidden_states[5]: 2.247e-05
hidden_states[6]: 2.891e-05
hidden_states[7]: 4.929e-05
hidden_states[8]: 1.543e-04
hidden_states[9]: 4.997e-04
hidden_states[10]: 5.622e-04
hidden_states[11]: 9.489e-04
hidden_states[12]: 1.122e-03