Add Evolla model

[zhoubay]

Model description

Model Name: Evolla

Model Specifications

• Model Type: Protein-language generative model
• Parameters: 80 billion
• Training Data: AI-generated dataset with 546 million protein question-answer pairs and 150 billion word tokens

Architecture

Multimodal model integrating a protein language model (PLM) as the encoder, a large language model (LLM) as the decoder, and a sequence compressor/aligner module.

Key Features

• Decodes the molecular language of proteins through natural language dialogue
• Generates precise, contextually nuanced insights into protein function
• Trained on extensive data to capture protein complexity and functional diversity

Applications

• Protein Function Annotation: Provides detailed functional insights for proteins
• Enzyme Commission (EC) Number Prediction: Assists in classifying enzymatic activities
• Gene Ontology (GO) Annotation: Helps in understanding protein roles in biological processes
• Subcellular Localization Prediction: Predicts where proteins are located within a cell
• Disease Association Analysis: Identifies potential links between proteins and diseases
• Other Protein Function Characterization Tasks: Supports various research needs in proteomics and functional genomics

Performance

Demonstrates expert-level insights, advancing research in proteomics and functional genomics.

Availability

• Evolla-10B Weights: Hugging Face
• Code Repository: GitHub
• Webserver: Chat-Protein

License

MIT License

Contact

For inquiries, contact the corresponding author(s) via email (e.g., [email protected]).

Open source status

• The model implementation is available
• The model weights are available

Provide useful links for the implementation

No response