Cohere event runbook
Cohere posted a string of near‑term technical events — Generative Modeling (Mar 18), AI Safety for Healthcare (Mar 19), and Domain Adaptation in CV (Mar 24) — offering timely hooks for engineering‑level engagement. The schedule suggests active research teams and opportunities to offer benchmark sessions or kernel tuning workshops. (x.com) (x.com) (x.com)
Cohere’s public event roster lists dedicated speaker pages for individual Cohere Labs talks — the site shows separate entries for the guest speakers referenced in the runbook. (cohere.com) Mingyang Deng is listed as a PhD student at MIT and the author of the arXiv paper “Generative Modeling via Drifting,” which his recent talk reports reached ImageNet one‑step generator results with an FID of 1.61 in pixel space. (youtube.com) Yasser Benigmim is identified as a PhD candidate at Télécom Paris with multiple recent papers on domain generalization and domain adaptation for semantic segmentation, including CVPR/ICCV submissions and arXiv preprints. (yasserben.github.io) Cohere Labs maintains an Open Science Community with program subgroups (Safety & Alignment, ML Healthcare, Computer Vision, etc.), a Google‑Groups/calendar workflow for event invites, and a resource guide that points to recordings and community leads. (labscommunity.cohere.com) Cohere publicly joined the U.S. AI Safety Institute Consortium in February 2024, and the company’s community pages include an ML Healthcare group used to schedule clinical‑safety and healthcare‑specific sessions. (cohere.com) (groups.google.com) Cohere publishes developer cookbooks and documentation for model customization and lists cloud partnerships that make its models deployable on third‑party clouds (Oracle OCI) while historically working with Google Cloud TPUs for infrastructure. (docs.cohere.com) (oracle.com) (intuitionlabs.ai) Cohere Labs’ community talks and event pages routinely link to arXiv papers and external code repositories, providing concrete artifacts that can be used for benchmark sessions or reproducible kernel‑tuning exercises; Mingyang Deng’s talk page links directly to the arXiv preprint and experiment metrics as an example. (labscommunity.cohere.com) (youtube.com)
