Code for training the models

[cguerner]

Hi,

I'm trying to add llama2 as one of the models under the framework. I'm curious re. the fit method here, seems you omitted it for some reason.

CEBaB/code/eval_pipeline/models/gpt2.py

Line 42 in 40e6851

# assume model was already trained

Without this function or a training script, I'm thinking I could just take a LlamaForSequenceClassification model and train it on the exclusive train set for the different number of overall sentiment classes, for example following this script: https://huggingface.co/blog/Lora-for-sequence-classification-with-Roberta-Llama-Mistral#llama-2.

Thanks!

Best,
Clément