Hugging Face / Autotrain - LLM Finetuning option disappears after upgrades

I am trying to fine-tune a model like Vicuna 7B on some sample data using a CPU Basic (Free) account. I notice the model starts training in the dashboard but fails after about 3 minutes. It enters into the training status and then stops.

I cannot see where to find out the reason for why it stopped but I am thinking it may be to do with the Free version?? I tried upgrading to CPU Upgrade tier and even NVidia A10G Large. However as soon as I upgrade, the task option of LLM Finetuning Disappears in the Autotrain UI. Only the Text Classification model remains. Weirdly, when I downgrade to CPU Basic, the LLM Finetuning option is no longer there again.

Does anyone know why this happens and how to get back LLM Finetuning after changing tiers? Also if anyone has suggestions why the training failed in the first place, please let me know!