How to fine tune AI

Hi,

I want to fine tune an AI to better understand task for creating UMLs.
I found this dataset coai/plantuml_generation · Datasets at Hugging Face
And I have trouble to get Phi3 to use this data for fine tuning.

The database only has a “text” column. In VSC I get this error:

ValueError: The batch received was empty, your model won’t be able to train on it. Double-check that your training dataset contains keys expected by the model: input_ids,attention_mask,position_ids,past_key_values,inputs_embeds,labels,use_cache,output_attentions,output_hidden_states,return_dict,label_ids,labels,label.

I tried to rename the text column to labels but it didn’t sovle the proeblem.
Can some please help me to get this working?

Thanks a lot

oh crazy, I had a bad typo :smiley:

now a different question. Can I force it to run on CPU with only 32gb of ram? I don’t care if it takes 2 weeks :smiley:

I am getting
“The model did not return a loss from the inputs, only the following keys: logits. For reference, the inputs it received are input_ids,attention_mask.”

Do i need to transform the dataset? Because it has all data in just one column