mims-harvard/bio-posttrain-qwen3-1.7b-dna-sft
The mims-harvard/bio-posttrain-qwen3-1.7b-dna-sft model is a 1.7 billion parameter Qwen3-based language model developed by mims-harvard, specifically fine-tuned for biological reasoning tasks. It integrates a frozen Evo2 DNA encoder via a linear projection, enabling it to process and reason with DNA sequence information. This model is designed for applications requiring the interpretation and generation of text in conjunction with biological sequence data.
Loading preview...
Model Overview
The mims-harvard/bio-posttrain-qwen3-1.7b-dna-sft is a specialized 1.7 billion parameter language model built upon the Qwen3 architecture. Developed by mims-harvard, this model is a DNA supervised fine-tuning (SFT) checkpoint from the "How Post-Training Shapes Biological Reasoning Models" collection.
Key Capabilities
- Biological Reasoning: Optimized for tasks that require understanding and processing biological information, particularly in the context of DNA sequences.
- DNA Integration: Incorporates a frozen Evo2
evo2_1b_baseDNA encoder, mapping its hidden states to the text LLM's hidden states via adna_projection.ptlinear layer. - Merged Architecture: The repository provides the merged text LLM with LoRA weights fused into the base Qwen3-1.7B model, alongside the necessary
dna_projection.ptfile.
Intended Use Cases
This model is particularly well-suited for research and development in bioinformatics and computational biology. It can be used for:
- Analyzing and interpreting biological text in conjunction with DNA sequences.
- Developing applications that require a deep understanding of biological reasoning.
- Further fine-tuning on specific biological tasks that benefit from integrated DNA sequence context.