AI Nano Bio Agents (ETH)
Abstract
NEW Nano Bio-Agent (NBA) framework - implemented by authors (see below) from ETH (Swiss) for GeneTuring benchmark incorporates task decomposition, tool orchestration, and API access into well-established systems such as NCBI and AlphaGenome.
But can scientists work with Small Language Models (SLM) for special genomics tasks complexities? How small can we go for our VLM /LLMs? Local LLMs?