[Chatllama] it seems you don't support flan_t5_xl to generate rewards training data

[lonelydancer]

you claim"If you prefer avoiding external paid APIs, we suggest using HuggingFace’s models (e.g. flan_t5_xl) as described in more detail in the Supported models section."
howerver,
python artifacts/generate_rewards.py ./artifacts/datasets/reward_training_data.json --model flan_t5_xl

self.llm = LLMChain(llm=openai_llm, prompt=prompt_template)

[PierpaoloSorbellini]

Hi @lonelydancer thank you for the feedback.
Yep you are right, we actually added the support to T5 only in extend_rlhf_dataset.py . The error in the readme is due to the fact that we wanted to update the documentation and code as soon as possible to fix as many issues as possible in the shortest amount of time. Unfortunately, we didn't make to add all the things that we want to, as a result that sentence wasn't properly removed from the readme.md. Thank you for spotting the mistake. We will adjust this inconsistency as soon as possible and possibly add the feature.

[avacaondata]

I've faced the same issue

[pengwei-iie]

I've faced the same issue, have you fixed it?