| ---
|
| library_name: transformers
|
| tags:
|
| - biology
|
| - medical
|
| - biomedical
|
| - natural-language-processing
|
| - relation-extraction
|
| - named-entity-recognition
|
| - question-answering
|
| - synthetic-biology
|
| - synthetic-data
|
| - low-resource
|
| - biomedbert
|
| ---
|
|
|
| # MultiAgent4SyntheticBioNLP experiment models
|
|
|
| This repository archives Hugging Face models from multiple experiments and
|
| datasets in the MultiAgent4SyntheticBioNLP project. Remote paths mirror the
|
| project hierarchy beginning below `experiments/`, while local-only `model` or
|
| `models` container directories are removed.
|
|
|
| For example:
|
|
|
| ```text
|
| local: overall-results/phosphorylation/model/<run>/model/*
|
| remote: overall-results/phosphorylation/<run>/*
|
| ```
|
|
|
| Each run directory is directly loadable as a Transformers subfolder:
|
|
|
| ```python
|
| from transformers import AutoModel, AutoTokenizer
|
|
|
| repo_id = "OWNER/REPOSITORY"
|
| subfolder = "overall-results/phosphorylation/RUN_NAME"
|
| tokenizer = AutoTokenizer.from_pretrained(repo_id, subfolder=subfolder)
|
| model = AutoModel.from_pretrained(repo_id, subfolder=subfolder)
|
| ```
|
|
|
| Each run contains only its final inference-model artifacts. Per-run
|
| `metrics.json`, `prepared-data/`, prepared training CSVs, and checkpoints are
|
| not uploaded.
|
|
|
| ## Intended use and limitations
|
|
|
| These are research artifacts for studying synthetic-data augmentation in
|
| biomedical NLP. They are not intended for clinical diagnosis or treatment
|
| decisions. Review the corresponding local experiment documentation, metrics,
|
| and data-selection manifest before reuse.
|
|
|
