Yarongef

Rank:

Average Model Cost: $0.0000

Number of Runs: 8,334

Models by this creator

DistilProtBert

DistilProtBert

yarongef

DistilProtBert is a distilled version of the ProtBert-UniRef100 model, pre-trained on a large dataset of protein sequences. It is designed for protein feature extraction and can be fine-tuned on downstream tasks. The model has been trained using masked language modeling (MLM) and uses a combination of cross-entropy and cosine teacher-student loss functions. It only works with capital letter amino acids. The model achieves good results in distinguishing between real proteins and their randomly shuffled counterparts. It was trained on the Uniref50 dataset consisting of approximately 43 million protein sequences. The model can be used in the same way as ProtBert, along with ProtBert's tokenizer. Please cite the relevant paper when using this model.

Read more

$-/run

8.3K

Huggingface

Similar creators