Inference on a new dataset

[prateek-pt]

Is there a way by which I can test the given model on my dataset which has different tags with respect to the conll, etc? I want to use the pre-trained 'eng_conll03' model and fine tune it with my data. Like removing the final dense layer having 5 nodes with 10 nodes (as per my dataset tags).

[juntaoy]

Basically what you need to do is do not read the biaffine part of the system, it is not a dense layer but a biaffine tensor.
You can add a method:

def restore_conll03(self, session):
    # Don't try to restore unused variables from the TF-Hub ELMo module.
    vars_to_restore = [v for v in tf.global_variables() if "module/" not in v.name and "Bilinear/" not in v.name]
    saver = tf.train.Saver(vars_to_restore)
    checkpoint_path = os.path.join(self.config["log_dir"], "model.max.ckpt")
    print("Restoring from {}".format(checkpoint_path))
    session.run(tf.global_variables_initializer())
    saver.restore(session, checkpoint_path)

Then you could do your fine-tuning with different number of NER types