openathena.ai
We help academic labs building AI models to solve technology challenges so they can focus on scientific impact
👤 Joined on 20 November 2024
Followers
Following
Total Posts
@openathena.bsky.social@openathena.bsky.social → @openathena.aiIn a new essay, David Hall @dlwh.bsky.social argues that pretraining consumes documents, while posttraining consumes tasks. Open language models could run on Wikipedia and ArXiv. Posttraining has no equivalent commons. David reflects on the choice we get to make: openathena.ai/blog/there-i...
Show PostBefore ChatGPT, there was AlphaFold winning CASP in 2020. In our latest Meet Our Team, MTS Tim O'Donnell (@timodonnell.bsky.social) talks MarinFold: distilling structure prediction into a vanilla LLM architecture, built fully in the open. Plus: what a single drop of blood remembers. 🔗 bit.ly/oa-tim
Show PostMarinDNA follows the open development approach of Marin marin.community. Beyond the outputs—data, models, and code—the research process is public on github.com/Open-Athena/....
Show PostIt is still unclear how much value single-sequence gLMs add on top of models trained with whole-genome alignments or functional genomics data. We think their best near-term opportunity is the far larger set of species without those resources.
Show PostAfter further data mixing experiments, our 1B checkpoint finished with the best overall Mendelian VEP performance among single-sequence gLMs. It slightly leads Evo 2 40B on zero-shot performance (larger gap with probing), while remaining simple and fast.
Show Post