Fine-tune broken for QRNNs?

[daemon]

I made some modifications to the codebase, so this might be a problem on my end... But does finetune.py require SplitCrossEntropyLoss to be used for the criterion instead? The decoder is called in SplitCrossEntropyLoss only. I added the appropriate SplitCrossEntropyLoss in finetune.py, and it works as expected.

[Smerity]

Finetune is currently broken - I forgot to update that code when updating main.py.

See https://github.com/salesforce/awd-lstm-lm/issues/26

I'll close this when I've fixed the issue.

[soyoung97]

Hi @daemon, I'm trying to work this finetune.py out. Can you tell me what you changed to make this work? I imported SplitCrossEntropyLoss and changed criterion on https://github.com/salesforce/awd-lstm-lm/blob/master/finetune.py#L115 and https://github.com/salesforce/awd-lstm-lm/blob/master/finetune.py#L146 to SplitCrossEntropyLoss(...) and changed the torch.load(f) on https://github.com/salesforce/awd-lstm-lm/blob/master/finetune.py#L177 to model, _, _ = torch.load(f) , but this still gives me an error. Any help would be greatly appreciated!

--------> found your github repo, https://github.com/daemon/awd-lstm-lm/commit/91b8271b368d4c5caa800c956b2817925e4f7b3d I think this will do!