Instructions to use multimolecule/bpfold with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- MultiMolecule
How to use multimolecule/bpfold with MultiMolecule:
pip install multimolecule
from multimolecule import AutoModel, AutoTokenizer tokenizer = AutoTokenizer.from_pretrained("multimolecule/bpfold") model = AutoModel.from_pretrained("multimolecule/bpfold") inputs = tokenizer("UAGCUUAUCAGACUGAUGUUGA", return_tensors="pt") outputs = model(**inputs) embeddings = outputs.last_hidden_state - Notebooks
- Google Colab
- Kaggle
| { | |
| "architectures": [ | |
| "BpfoldModel" | |
| ], | |
| "attention_head_size": 32, | |
| "bos_token_id": 1, | |
| "dtype": "float32", | |
| "eos_token_id": 2, | |
| "hidden_dropout": 0.1, | |
| "hidden_size": 256, | |
| "id2label": null, | |
| "intermediate_size": 768, | |
| "label2id": null, | |
| "mask_token_id": 4, | |
| "max_length": 600, | |
| "model_type": "bpfold", | |
| "motif_radius": 3, | |
| "null_token_id": 5, | |
| "num_hidden_layers": 12, | |
| "num_labels": 1, | |
| "num_members": 6, | |
| "num_pairwise_convolutions": 3, | |
| "pad_token_id": 0, | |
| "pairwise_kernel_size": 3, | |
| "pos_weight": 300.0, | |
| "positional_embedding": "dyn", | |
| "postprocess_iterations": 100, | |
| "postprocess_lr_max": 0.1, | |
| "postprocess_lr_min": 0.01, | |
| "postprocess_nc_rho": 0.5, | |
| "postprocess_nc_s": 0.5, | |
| "postprocess_rho": 1.6, | |
| "postprocess_s": 1.5, | |
| "postprocess_with_l1": true, | |
| "separate_outer_inner_energy": true, | |
| "threshold": 0.5, | |
| "tie_word_embeddings": true, | |
| "transformers_version": "5.7.0", | |
| "unk_token_id": 3, | |
| "use_base_pair_energy": true, | |
| "use_base_pair_probability": false, | |
| "use_postprocessing": false, | |
| "use_squeeze_excitation": true, | |
| "vocab_size": 11 | |
| } | |