Hi Everyone. It's been a long-due update but happy to announce that I'll be starting my MSCS at @Stanford and my RAship at @StanfordCRFM. This was a long and fun year and this would have never been possible without the support of @lateinteraction, @sarahookr, and @krishnanc_ .

Credits and big shoutout to @WesEklund for coming up with a similar inline design first!.

Moving forward is adding dataloaders and chalking up an ownership architecture which tracing in mind. Alongside I'll be patching some caching mechanism for the calls and some thinking on supporting multimodal input output. It's a nice week up ahead🙃.

DSRs(DSPy in Rust) Updates!!. Had tons of good conversation in @DSPyOSS Discord with @joshmo_dev @tech_optimist and others about the way forward. [1] There are new marco designs for long and inline signatures. [2] Finally have BYOK support for OAI, Claude and Gemini Models.

@DSPyOSS Much thanks to @tech_optimist for the discussions and shout out :).

It's not public yet but thanks for the support folks!!. Do join @DSPyOSS discord, I ramble/provide updates there every evening about this :P

RT @LakshyAAAgrawal: Paper: GEPA will be open-sourced soon as a new DSPy optimizer. Stay tuned!. Incredibly gratef….

RT @nicochristie: Shortcut – the first superhuman excel agent – is live. While not perfect, Shortcut beats first year analysts from McKins….

RT @ShirleyYXWu: Introducing 🔥Optimas🔥: The first unified framework to optimize compound AI systems composed of multiple components like tr….

RT @nicochristie: Introducing Shortcut — the first superhuman Excel agent. Shortcut one-shots most knowledge work tasks on Excel. It even….

RT @ishandutta0098: Project Launch 🚀: mukh v0.1.14 - The Face Analysis Library. Hello Everyone,. I am happy to launch my first python packa….

RT @ishandutta0098: Thousands of lines of code later finally inching towards building my first python package “mukh”. Mukh is a comprehensi….

Stoked to shared that Marin has officially launched!!! 🚀. Personally, it’s been a great privilege to have been working on this project at @StanfordCRFM alongside some brilliant folks. I got to work hands-on how different pieces fit together to build an LLM, and got my hands.

RT @AshwaryaMaratha: 🚨🚨 Sensitive Data Exposure !!. Hey team @spawn , it seems you haven't taken care of API keys getting exposed. Your pro….

RT @lateinteraction: DSPy downloads in Feb 2025 just crossed 1M, thanks to 270 incredible contributors: Chen Qian @isaacbmiller1 @kristahop….

RT @CShorten30: Hey everyone! I am super excited to share our new research report is live on ArXiv! 🎉. Querying Databases with Function Cal….

RT @AshwaryaMaratha: 🚨🚨 ANNOUNCEMENT. Agents Testing is now more important than ever before. While I was working with Elad on AutoPrompt, i….

RT @francescoswiss: second round .➡️ @joaomdmoura and @bhancock_ai from @crewAIInc.➡️ @hwchase17 and @eyfriis from @LangChainAI .➡️ @latein….

Blog: Colab: Codebase:

🔍 What We Achieved. Our model outperformed standard GNN models on the ZINC dataset, a widely used molecular drug solubility prediction benchmark. We provide a notebook to replicate these experiments, a codebase to use RingSage for your own use case, and a blog elaborating on our.

💡 Why This Matters. Traditional GNNs treat molecular graphs uniformly, missing the unique properties of rings. Our approach introduces virtual nodes to capture ring-specific information. However, this leads to an increase in the number of learned parameters, so for a fair.