Christian Geishauser

Thanks for your quick answer! > In the script ConvLab-2/convlab2/policy/vector/vector_multiwoz.py, the method state_vectorize(self, state) is used to create the vector version of the given state. It does not try to...

Hey Nick, regarding your question in https://github.com/thu-coai/ConvLab-2/issues/8#issuecomment-658690011 I used the best_mle model as a starting point and did not train my own one. I managed to reproduce the results once...

Hey Yen-Chen! I am not sure what has been changed but maybe the following happens with the reward: https://github.com/thu-coai/ConvLab-2/blob/master/convlab2/evaluator/multiwoz_eval.py#L417 If you look here, you get a reward of 5 if...

@ruleGreen > These two results are not the same at all. This is reasonable as sentence-level creates a more difficult environment for the policy. On action-level, you basically have the...

Hi @nomawni, this tutorial has not been maintained since years. Nevertheless, your bug should be fixed easily: https://github.com/ConvLab/ConvLab-3/blob/master/tutorials/Train_RL_Policies/example_train.py#L60 Here in line 60, you currently have s_vec = torch.Tensor(policy.vector.state_vectorize(s)) You should...

Hi @JamesCao2048, thanks a lot for all your questions! I hope I can answer them sufficiently for you: 1. For MLE training, this is explained in the README: https://github.com/ConvLab/ConvLab-3/tree/master/convlab/policy/mle So...

Hi @silviatti! First of all, thank you very much for using ConvLab-3 and helping us improve with your bug issues and solution suggestions, that is great! :) I will now...

@silviatti > Just a quick update. I have realized that I was loading a model for the UserPolicy that was trained on multiwoz (it was automatically downloaded from here: https://zenodo.org/record/7369429/files/multiwoz_0.zip)...

Hi @silviatti ! I will investigate it more on Monday but could you just check whether it prints "Loaded policy checkpoint from file: convlab/policy/vtrace_DPT/supervised.pol.mdl" before the dialogue collection starts? If...

Hi @silviatti, sorry for the late response! I think you missed to change the user NLU as I explained above: **You should have:** from convlab.nlu.jointBERT.unified_datasets.nlu import BERTNLU user_nlu = BERTNLU(mode='sys',...