How can we run it on our own data-set using our Infersent model?

[ashutoshsingh25]

I am unable to run it on my own data-set using Infersent model.
Please share some example, where we can integrate our own Infersent model.

[JaSchuste]

I got the same question right now. Did you figure it out and maybe tell me how this is possible?

[jjkiljanski]

To use your custom dataset you need to add your own task, as the implemented task classes refer to dataset paths with specific names. If it is a probing task, you can simply edit the engine.py file by adding the custom task name ‘MyTask’ to self.list_tasks in the SE init and adding elif instance in the eval function:
elif name == 'MyTask':
self.evaluation = MyTaskEval(‘path/to/my/task’, seed=self.params.seed).
Then you need to edit probing.py by adding your new task path. E.g.
class MyTaskEval(PROBINGEval):
def init(self, task_path, seed=1111): # don’t replace task_path
task_path = os.path.join(task_path, 'my_task_name.txt') # don’t replace task_path
PROBINGEval.init(self, 'MyTask', task_path, seed)