Multi-Modal Foundation · living-models
Botanic1-L
Plant genomic foundation model for sequence embedding, zero-shot variant effect scoring, and downstream fine-tuning.
What Botanic1-L is
Botanic1-L is a plant genomic foundation model developed by living-models. With 2.1 billion parameters, it serves as the largest public release in the Botanic1 model family.
The model addresses computational biology challenges in plant genomics, providing representations for sequence embedding, zero-shot variant effect scoring, and fine-tuning across distinct plant genomic prediction tasks.
How it works
Botanic1-L is constructed using 72 bidirectional Mamba-2 (BiMamba2) blocks, totalling 2.1 billion parameters. It processes plant genomic sequences to produce embeddings and score sequence variants.
Key facts
- Model architecture
- 72 BiMamba2 blocks
- Parameter count
- 2.1 billion
What you can use it for
- Plant sequence embedding
- Generates numerical representations of plant genomic sequences for downstream analytical pipelines.
- Zero-shot variant effect scoring
- Evaluates the functional impacts of plant genetic variants directly without task-specific training.
- Genomic prediction fine-tuning
- Adapts pre-trained parameters to specialised plant genomic prediction tasks through supervised fine-tuning.
Run it here
We list this model but cannot run it yet: gated on the Hub — never run past a gate the user did not accept.
Everything on this page still applies. Running it needs a loader for this repository, not a bigger machine.
Strengths and limits
Good at
- Represents the largest public Botanic1 model, offering 2.1 billion parameters across 72 BiMamba2 blocks.
- Capable of performing zero-shot variant effect scoring without requiring task-specific training labels.
Watch out for
- Requires execution of custom code to run the model architecture.
- Training data sources, context windows, and specific plant taxa coverage are not detailed.
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 · other
This model is gated by its authors — its licence has to be accepted on the source repository before it will run.
Questions
Can I use Botanic1-L commercially?
Check the licence before commercial use. This model is gated by its authors — its licence has to be accepted on the source repository before it will run.
What architecture does Botanic1-L use?
Botanic1-L is built from 72 BiMamba2 blocks, comprising 2.1 billion parameters in total.
What primary tasks is Botanic1-L designed for?
It is designed for extracting sequence embeddings, zero-shot scoring of variant effects, and fine-tuning on plant genomic prediction tasks.
Who created Botanic1-L?
The model was developed by living-models.
Written from Botanic1-L'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.