Instructions to use Rostlab/prot_bert with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use Rostlab/prot_bert with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("fill-mask", model="Rostlab/prot_bert")# Load model directly from transformers import AutoModelForMaskedLM model = AutoModelForMaskedLM.from_pretrained("Rostlab/prot_bert", device_map="auto") - Inference
- Notebooks
- Google Colab
- Kaggle
Get more than 5 estimations in output
Hi,
First of all - thank you very much for sharing pretrained ProtBert model for masking.
I started working with the model, and I got from it each time only 5 estimations for a masked station in the sequence.
Is it possible to let the model giving back as output estimations for all types of amino-acids, so getting 25 estimations for a single mask instead 5?
If it's so - how I may do it?
Regards,
Aviv.
Hi Aviv,
thanks for your interest in our models.
That is indeed weird but without any information on the code you are running, I won't be able to help you out.
Usually, for BERT-like models, you should easily be able to derive the logits for all possible tokens in your vocab as described for example here: https://huggingface.co/docs/transformers/main_classes/output