It was great catching up with (BG) Bhaskar Ghosh @ SFBA yesterday! Thank you for your constant support through our journey at PeerDB (YC S23) ! 🙏

Excited to offer a sneak peek at what we have been working on. Check out the LLM-AggreFact leaderboard [1] for factuality and hallucination detection, and the demo of our model that tops the leaderboard [2]. [1] llm-aggrefact.github.io [2] playground.bespokelabs.ai More info to

Excited to launch the first model from our startup: Bespoke Labs. Bespoke-Minicheck-7B is a grounded factuality checker: super lightweight and fast. Outperforms all big foundation models including Claude 3.5 Sonnet, Mistral-Large m2 and GPT 4o and its only 7B. Also, I want to

GPT is having a profound effect on how students write. Its verbose style, full of cliches and 'fancy', out of place vocabulary is in every paper and draft I read. A few years back, there were grammar errors and awkwardness -- but at least people had their own voice. Now,

If you suffer from nightmares induced by hallucinations, take this guardrails pill and see remedy in milliseconds. Act today now.

ollama run bespoke-minicheck .Bespoke Labs released Bespoke-Minicheck, a 7B fact-checking model is now available in Ollama! It answers with Yes / No and you can use it to fact check claims on your own documents. How to use the model with examples: ollama.com/blog/reduce-ha…

My amazing colleague Kevin Niechen is hosting an AI design event on Nov 18 at SF offices of 8VC; should be a monster one! Please ask Kevin on how to join or nominate people. 8VC Vivek Gopalan

Introducing Bespoke-Stratos-32B, our reasoning model distilled from DeepSeek-R1 using Berkeley NovaSky’s Sky-T1 recipe. The model outperforms Sky-T1 and o1-preview in reasoning (Math and Code) benchmarks and almost reaches the performance of DeepSeek-R1-Distill-Qwen-32B while

Link to the blog post: bespokelabs.ai/blog/bespoke-s… This has links to the model, code, and most importantly the open reasoning dataset!

We are one of the top trending datasets on HuggingFace today!

We're incredibly excited to be launching the 8VC Llama Stack Challenge today for university students in the US!! The first challenge is about building desktop, field, and edge AI applications using Meta's open source Llama models: 3/7 - 4/19! Join now at llamastackchallenge.com

Reliable data powers better decisions—learn how to ensure data quality at scale! Join Jonny Dixon for "Get Data Quality You Can Trust" and see how DataHub Cloud makes data health & compliance effortless. 🗓️ March 20 | 9 AM PT 👉 Register: lu.ma/5i0njefk

Every job has tedious and fun parts. What if you could offload all the tedious parts to someone and only keep the fun parts? If you are a recruiter or hiring manager, you can now do just that. Announcing general availability of Tezi's AI recruiting agent Max!

🚀🚀🚀🚀🚀 Big launch: Tezi's agentic AI recruiter, Max, is now available to all companies. Bye bye beta, hello GA. Max filled all my open headcount in < 30 days per role. Excited to share it with even more companies!

Exciting announcement & vision from our Co-Founder & CEO Raghavendra Prabhu: Tezi's agentic AI recruiter, Max, is now released for general availability. No more waitlist. The future is now. Request a demo at tezi.ai.

Top talent is the key driver behind our best companies. Raghavendra Prabhu & Tezi are revolutionizing how we find it w/ Max, an AI recruiter that saved customers 10000+ hours in just months. Max sources candidates, sends emails and schedules interviews 24/7; an awesome AI colleague!

In need of general debugging help? Our troubleshoot DataHub Slack channel is the place to go! Sign up for expert help: pages.acryl.io/slack

GenAI tools are revolutionizing coding, but hard data on their impacts is scarce. We’ve partnered w/ Software.com & Jedidiah Yueh for the 1st empirical study of how specific tools raise dev team productivity. Learn more from 8VC Partner and CTO Asanka Jayasuriya (and sign up!)

We are launching a reasoning dataset competition with Hugging Face and Together AI. With OpenThinker, we democratized access to reasoning models, but more importantly with the Open Thoughts dataset, we allowed 200+ models to be created (that are public!). Most of us have

Thank Nathan Lambert for the shout out about the work we did at Bespoke Labs.