Spaces:
Sleeping
Request for code
Hello! Great work with the model! I really like the way it works and the results given by it.
I'm new to machine learning overall and I'm trying to train the model myself using SimpleTOD open source code from github. (just to learn and add it to my portofolio)
I've started to preprocess the SGD data myself, but I got stuck. Right now for me the code from that repo is very complex and I saw that it also has some issues.
Is there any possibilty that you share the code of preprocessing and training of the model wirh me so I can reverse engineer it? It would mean the world to me.
Thank you for your time reading this.
Hi there, I based my implementation on code from this repo => https://github.com/facebookresearch/accentor. Their paper focuses on adding chitchat to task-oriented dialogue, and simpletod is part of their pipeline.
It's overall much cleaner and training is based on the transformers library language modeling script => https://github.com/huggingface/transformers/blob/main/examples/pytorch/language-modeling/run_clm.py. They basically just added special tokens (<|context|>, <|endofcontext|>, <|belief|> etc...)
Hope this helps !
Thanks a lot man! It definetly helps. I will take this approach.
Really appreciate your help ❤️
Just as an update, finally everything is more clear to me and I was able to generate the model myself, following the code from accentor repo. I just want to thank you again for the fast answer and for the guidance.