metadata
license: apache-2.0
tags:
- stripedhyena
- long context
- deep signal processing
- hybrid
- biology
- genomics
- phage
- microviridae
Evo 1 - Microviridae
About
This is a fine-tuned version of Evo 1, a biological foundation model capable of long-context modeling and design, specifically adapted for Microviridae phage sequences.
Evo uses the StripedHyena architecture to enable modeling of sequences at a single-nucleotide, byte-level resolution with near-linear scaling of compute and memory relative to context length.
Model Details
- Base Model: Evo 1 7B 131K base
- Fine-tuning: 5,000 steps on Microviridae bacteriophage genomes
- Architecture: StripedHyena
- Context Length: 131,072 tokens
- Parameters: 7B
Usage
This model uses the StripedHyena architecture. For loading and generation, please refer to the Evo repository.
# Example usage with Evo framework
from evo import Evo
# Load base model and apply fine-tuned weights
model = Evo('evo-1-131k-base')
# Load the fine-tuned checkpoint
checkpoint = torch.load('pytorch_model.bin')
model.load_state_dict(checkpoint)
Training
Fine-tuned from evo-1-131k-base on Microviridae phage sequences for 5,000 steps.
License
Apache 2.0