ML-based science is facing a reproducibility crisis. We think clear reporting standards for researchers can help. Today, we're introducing REFORMS, a consensus-based checklist authored by 19 researchers across many disciplines. aisnakeoil.com/p/introducing-…

Retrieval Augmented Generation vs. LLM Finetuning.

Anti-hype LLM reading list. Pretty good list. gist.github.com/veekaybee/be37…

Embracing diverse candidates in our internship program isn't to tick boxes; it's about igniting a wave of creativity and fresh viewpoints. Learn more in our blog 🌍🤝 vectara.com/embracing-dive… #AIInternships #AIRecruitment #DEI

Excited to share our NeurIPS Conference '23 paper "Exact Optimality of Communication-Privacy-Utility Tradeoffs in Distributed Mean Estimation": arxiv.org/pdf/2306.04924… Looking forward to presenting it in person and seeing many of you in New Orleans! 🙂🎷🎶 Details 👇

Today, we’re unveiling Boomerang, the LLM redefining Generative AI accuracy. Click below to learn how our ML team developed #Boomerang to outperform #Cohere and #OpenAI and solve the barriers to business adoption of #GenAI: globenewswire.com/news-release/2… #GroundedGeneration #RAG #AI

“There was a lot of research, a lot of trial and error, a lot of things that didn’t work and things that did work, that got us to this point where we now can exceed a couple of the most advanced companies in this space,” CEO Amr Awadallah 🤖 on #Boomerang

Today Vectara launched Boomerang, a SoTA neural retrieval model, competitive with OpenAI Ada, Cohere, and open source embedding models like e5 and GTR-XXL. Details in our blog post: vectara.com/introducing-bo… Production-grade RAG pipelines have never been easier to integrate.

Boomerang is live 🪃💥 Get up to speed on our new #embedding model and benefits like better search and summarizations! Read the blog by Head of Product Shane Connelly and try Boomerang today ⤵️ vectara.com/how-boomerang-… #GroundedGeneration #RAG #VectorSearch #AIDeveloper #ML

With many 🧩 dropping recently, a more complete picture is emerging of LLMs not as a chatbot, but the kernel process of a new Operating System. E.g. today it orchestrates: - Input & Output across modalities (text, audio, vision) - Code interpreter, ability to write & run

RSVP to join Vectara 10/16 & celebrate🎉 together - Let's co-build art of possible in GenAI #Ecosystem🪃 Luminary📣 Matei Zaharia Databricks Eric Savitsky SonoSim, Inc @LauraPadillaSF Sapphire Ventures Yobi Race Capital Partnership Leaders Amr Awadallah 🤖 Amin Ahmad

I wish I could be at the Muslims in ML Workshop affinity workshop at #NeurIPS2023 in-person! Cannot wait till I watch the recorded talks.

Grateful to be recognized by Forbes as one of America's Best Companies for 2024. This mention reflects the dedication of our incredible team, who works tirelessly to create a world where healthcare knows no limits. But our greatest reward? Knowing our innovations are helping to

Ben (no treats) Training on the MTEB training splits and evaluating on MTEB for the leaderboard submission was never intended. Embedding models need to be evaluated out of domain, anything else doesn't make sense (see BEIR paper). It was a big mistake to publish these training splits for MTEB.

We present a new theoretical analysis of the CLIP model, focusing on controlling the estimation error of its representations in multimodal downstream tasks. The key is the notion of approximate sufficient statistics. In generative hierarchical models, we derive an end-to-end