ALIAS MA

This is the ALIAS model labeled MA in the manuscript. It is a SentenceTransformer encoder for embedding cell descriptions and biomedical text into a shared space. It is intended for research use.

Field Value
Manuscript label MA
Internal model ID MB_HIAI_Tcells_S2_N1_200k_lr5e5_e15
Training datasets S2_heldout_donor_semantic_200k, N1_ncbi_literature
Base model NeuML/pubmedbert-base-embeddings
Code and dataset provenance ALIAS manuscript reproducibility branch

License and provenance

The ALIAS model release is MIT licensed (see LICENSE.txt). It is fine-tuned from NeuML/pubmedbert-base-embeddings, which is Apache 2.0 licensed; its license text is included as UPSTREAM_APACHE_2_0_LICENSE.txt. The model weights were modified by ALIAS training from the upstream base model.

Use

from sentence_transformers import SentenceTransformer
model = SentenceTransformer("mariakrissmer/ALIAS-MA")
embeddings = model.encode(["CD8 T cell", "cytotoxic T lymphocyte"])

This model is for exploratory research. Its training data and intended comparisons are documented in the linked repository. It has not been validated for clinical use.

Downloads last month
11
Safetensors
Model size
0.1B params
Tensor type
F32
·
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for mariakrissmer/ALIAS-MA