xiangrongzeng

The model weights are not in my hand row. I have reached such perfromance most of the time (but not every time, sometimes the performance dropped a lot suddenly). Please...

There is no relation embedding. We recognize the relation by "classification".

Exactly

Using the same tokenizer for actor and critic in step3 is beneficial. Considering that RM model is easier to train, in step2, I try to use the actor tokenizer during...

@guijuzhejiang Yes, the pre-trained parameters may not helpful in this case. By doing this, my assumption is that RM is easier to train, which not heavily rely on the pre-trained...

@Pattaro This project is currently not support llama yet. I used the llama tokenizer with OPT model. According to the introduction, the llama model will soon be supported officially.

@guijuzhejiang Hi, check this paper https://arxiv.org/pdf/2304.08177.pdf > We initialize the Chinese-LLaMA model with the original LLaMA weights and pre-train the model on general Chinese corpora