Update README.md

README.md

@@ -7,22 +7,36 @@ tags: []
 
 All credit for the original work goes to Tymor Hamamsy and the following authors of this paper https://www.biorxiv.org/content/10.1101/2023.11.26.568742v1
 
-We have added a Huggingface compatible wrapper for the model in models.protein_vec.src_run.huggingface_protein_vec.py
+We have added a Huggingface compatible wrapper for the model in protvec.py
 
+Clone and install
 ```
-git clone https://github.com/lhallee/Mixture-of-Experts-Sentence-Similarity
+git clone https://github.com/lhallee/ProteinVecHuggingface.git
+pip install torch pytorch_lightning transformers
 ```
 
 To use from hugggingface
-
 ```
 from transformers import T5Tokenizer
-from huggingface_protein_vec import ProteinVec, ProteinVecConfig
+from protvec import ProteinVec, ProteinVecConfig
 
 tokenizer = T5Tokenizer.from_pretrained('lhallee/ProteinVec')
 model = ProteinVec.from_pretrained('lhallee/ProteinVec', config=ProteinVecConfig())
 ```
 
+Embed a single sequence with ```embed```
+```
+model.to_eval()
+model = model.cuda() # remove if cpu inference
+embedding = model.embed('SEQWENCE').detach().cpu() # torch.tensor(1, 512)
+```
+
+To half precision weights
+```
+model.to_half()
+```
+
+
 ## The license for the protein vec code
 
 ### BSD 3-Clause License
@@ -54,4 +68,3 @@ CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
 OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
 OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 
-