Multi-Modal Foundation · biohub
ESMC 300M (Dec 2024)
A 300-million-parameter protein language model for generating protein embeddings in biological workflows.
What ESMC 300M (Dec 2024) is
ESMC 300M (Dec 2024) is a 300-million-parameter protein language model developed by Biohub. Released in December 2024, it provides a compact checkpoint designed for protein-embedding pipelines across biology, medicine, and genomics.
This specific distribution provides model weights formatted for use with the GitHub-compatible esm package. It is retained to maintain backwards compatibility for existing computational pipelines that rely on that original packaging structure.
Key facts
- Parameter count
- 300 million
- Model type
- Protein language model
- Release date
- December 2024
- Software compatibility
- esm package format (GitHub)
What you can use it for
- Protein sequence embedding
- Generate compact vector representations of amino acid sequences for use in downstream machine learning workflows.
- Genomic sequence analysis
- Apply learned protein sequence representations to support analytical pipelines in genomics research.
- Medical and biological informatics
- Serve as a feature extraction step for computational biology and medical predictive tasks.
Run it here
We list this model but cannot run it yet: superseded by biohub/ESMC-300M, already indexed.
Everything on this page still applies. Running it needs a loader for this repository, not a bigger machine.
Strengths and limits
Good at
- Offers a compact 300-million-parameter size suitable for efficient feature extraction pipelines.
- Provides direct compatibility for legacy pipelines relying on the GitHub-compatible esm package format.
Watch out for
- Maintained primarily for backwards compatibility rather than ongoing active development.
- Superseded by alternative Hugging Face-compatible checkpoints such as biohub/ESMC-300M, biohub/ESMC-600M, and biohub/ESMC-6B.
Inputs and outputs
You provide
Sequence or text
A protein or nucleotide sequence, or free text — whichever this model was trained on.
You get back
Mean-pooled representation, one vector per input.
Licence
Check the licence before commercial use · mit
The mit licence carries terms worth reading before you depend on it.
Questions
Can I use ESMC 300M (Dec 2024) commercially?
Check the licence before commercial use. The mit licence carries terms worth reading before you depend on it.
What is the primary purpose of this checkpoint release?
This repository provides weights formatted for the GitHub-compatible esm package, preserved primarily for backwards compatibility with existing pipelines.
Which alternative versions exist for newer projects?
Researchers starting new projects are directed to the Hugging Face-compatible releases, including biohub/ESMC-300M, biohub/ESMC-600M, and biohub/ESMC-6B.
How large is the ESMC 300M model?
The model contains 300 million parameters, offering a compact footprint for protein embedding tasks.
Written from ESMC 300M (Dec 2024)'s own documentation (model card). Statements citing a figure that does not appear there are removed before publication. Licence and running details come from our registry.