ctheodoris

/

Geneformer

like 200

hello, thanks for this great tool Geneformer !!!

I try to redo the examples from here for cell_classification.

However, I am new to hugging face and I met with a problem:

    # reload pretrained model
    model = BertForSequenceClassification.from_pretrained("/mypath/Geneformer/", 
                        num_labels=len(organ_label_dict.keys()),
                        output_attentions = False,
                        output_hidden_states = False)   ## I do not have any GPU resources, so I remove to("cuda")


    # create the trainer
    trainer = Trainer(
        model=model,
        args=training_args_init,
        data_collator=DataCollatorForCellClassification(),
        train_dataset=organ_trainset,
        eval_dataset=organ_evalset,
        compute_metrics=compute_metrics
    )

the error info:

---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
Cell In[10], line 55
     49 training_args_init = TrainingArguments(**training_args)
     51 # create the trainer
     52 trainer = Trainer(
     53     model=model,
     54     args=training_args_init,
---> 55     data_collator=DataCollatorForCellClassification(),
     56     train_dataset=organ_trainset,
     57     eval_dataset=organ_evalset,
     58     compute_metrics=compute_metrics
     59 )
     60 # train the cell type classifier
     61 trainer.train()

TypeError: __init__() missing 1 required positional argument: 'tokenizer'

Could you help with this problem ? Thanks !!!
Is the above error caused by a lack of GPU resources on my system?

Thanks !!!