Sentence Similarity
sentence-transformers
Safetensors
roberta
molecular-similarity
feature-extraction
dense
Generated from Trainer
loss:Matryoshka2dLoss
loss:MatryoshkaLoss
loss:TanimotoSentLoss
Eval Results (legacy)
text-embeddings-inference
Instructions to use Derify/ChemMRL-beta with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- sentence-transformers
How to use Derify/ChemMRL-beta with sentence-transformers:
from sentence_transformers import SentenceTransformer model = SentenceTransformer("Derify/ChemMRL-beta") sentences = [ "CC1CCc2c(N)nc(C3CCCC3)n2C1", "CC1CCc2c(N)nc(OC3CC3)n2C1", "CN1CC[NH+](C[C@H](O)C2CC2)C2(CCCCC2)C1", "Cc1c(F)cc(CNCC2CCC(C3CCC(C)CO3)CO2)cc1F" ] embeddings = model.encode(sentences) similarities = model.similarity(embeddings, embeddings) print(similarities.shape) # [4, 4] - Notebooks
- Google Colab
- Kaggle
| { | |
| "__version__": "0.7.2", | |
| "embedding_pooling": "mean", | |
| "eval_metric": "spearman", | |
| "eval_similarity_fct": "tanimoto", | |
| "kl_div_weight": 0.5, | |
| "kl_temperature": 0.3, | |
| "last_layer_weight": 2.0, | |
| "loss_func": "tanimotosentloss", | |
| "model_name": "Derify/ChemBERTa-druglike", | |
| "mrl_dimension_weights": [ | |
| 1, | |
| 1, | |
| 1, | |
| 1, | |
| 1, | |
| 1, | |
| 1, | |
| 1 | |
| ], | |
| "mrl_dimensions": [ | |
| 1024, | |
| 512, | |
| 256, | |
| 128, | |
| 64, | |
| 32, | |
| 16, | |
| 8 | |
| ], | |
| "n_dims_per_step": -1, | |
| "n_layers_per_step": -1, | |
| "prior_layers_weight": 1.0, | |
| "tanimoto_similarity_loss_func": null, | |
| "use_2d_matryoshka": true, | |
| "use_query_tokenizer": false | |
| } |