Project title
Understanding the role and complexities of agentic workflows and local models in life-science applications
Collaborators and funding
Australian BioCommons
Contact(s)
Benjamin Goudey benjamin@biocommons.org.au
Project description and aims
We will be using this project to:
1) critically compare and contrast existing benchmarks of agentic systems in AI
2) contrast the performance of local and frontier models on these datasets across distinct metrics
3) understand how the different components of the system (model, reasoning budget, SKILLs and other setup configuration) influence performance in the syste, and how we can measure and optimise this.
4) understand the strengths and limitations of the outputs and processes in these agentic workflows for users of different technical skill levels.
How is ABLeS supporting this work?
This work is supported through the Production Bioinformatics scheme provided by ABLeS.
Expected outputs enabled by participation in ABLeS
We expect to produce research and whitepapers for this explored area. This wil be released as preprints and al source code will be made public.
These details have been provided by project members at project initiation. For more information on the project, please consult the contact(s) or project links above.