Hi there! I would like to fine-tune the pre-trained model on a custom dataset. If I understand correctly, I have to train a new word2vec model and skip-instructions model on the custom dataset. Using these models, I create the custom HDF5 file. Then, I can fine-tune the pre-trained model by training it on the custom HDF5 file with the parameter -finetune 1.
Is this the correct way to do this? I’m asking because I’m not sure if I should train the word2vec and skip-instruction model on the custom dataset only, or if these should be trained on a concatenation of the Recipe1M and the custom dataset. Additionally, the main.lua file contains opts.finetune = opts.finetune ~= 0, but I have not been able to figure out how this parameter is used during training. Thanks!
Hi there! I would like to fine-tune the pre-trained model on a custom dataset. If I understand correctly, I have to train a new word2vec model and skip-instructions model on the custom dataset. Using these models, I create the custom HDF5 file. Then, I can fine-tune the pre-trained model by training it on the custom HDF5 file with the parameter
-finetune 1.Is this the correct way to do this? I’m asking because I’m not sure if I should train the word2vec and skip-instruction model on the custom dataset only, or if these should be trained on a concatenation of the Recipe1M and the custom dataset. Additionally, the main.lua file contains
opts.finetune = opts.finetune ~= 0, but I have not been able to figure out how this parameter is used during training. Thanks!