Hi folks! I’m currently working on my master thesis and need to implement a text-classification model. Therefore I’m fine-tuning a base BERT model on my data. So far so good. To evaluate my fine-tuned model, I’m passing my test data to it for predictions. For this purpose I’m using the “pipeline” A…

Different outputs when using pipeline

mtyrrell July 20, 2023, 10:08pm 3

I had the same issue. I forgot to include problem_type in the model config. Added “multi_label_classification” and it worked fine.

Topic		Replies	Views
Sentiment Analysis Pipeline on single label function_to_apply not working 🤗Transformers	1	1031	March 17, 2022
Huggingface classification struggling with prediction 🤗Transformers	0	835	April 5, 2022
Is there a way to get the logits as output in zero shot classification Pipeline? 🤗Transformers	0	480	July 30, 2021
Different results between pipeline and model() with multiple inputs 🤗Transformers	0	549	April 20, 2022
Get logits from Inference API Classification Model (for Regression) 🤗Hub	2	1088	January 24, 2024