Viraj Mehta
@thebigmehtaphor
cofounder and cto @tensorzero. previously did a PhD at CMU doing sample efficient reinforcement learning for tokamaks and math and cs @stanford before that
You might like
read the full analysis, methodology, and code examples: tensorzero.com/blog/is-openai… thanks to @gabrielbianconi @thebigmehtaphor Alan Mishler and Shuyang Li for their helpful feedback what's your experience with rft? cooked or let it cook?
tensorzero.com
Is OpenAI's Reinforcement Fine-Tuning (RFT) Worth It? · TensorZero
Is OpenAI's Reinforcement Fine-Tuning (RFT) Worth It?
how do inference services ever output control tokens? can't you unit test for this? shouldn't you have monitoring set up for this? I don't want to name names but sheeeeesh
the responses API is a way for @OpenAI to lock people in. It’s trivially true that any stateful interface can become stateless if you are allowed to return an encrypted blob of data (like Anthropic does today). Gating new features behind something like this is a wedge
10000 and 100 on the same day is nice & round it is great to see the project grow up big & strong!
Thank you for 10,000 stars and 100 contributors — on the same day! We're excited to continue building a next-generation open-source stack for LLM applications. Find us on GitHub ↓
Al Chen (@bigal123) from Grammarly recently built an agent with a continuous improvement loop using TensorZero + LangGraph. Watch his presentation at the NYC LangChain Meetup last week!
We need to make good decisions about an applications’s LLM stack the same way we make good decisions about parameters of other high-throughput systems. Alan’s work will bring automation and rigor to the process for TensorZero users. Couldn’t be more excited to work with him.
Alan Mishler is an ML researcher with a background in causal inference, sequential decision making, uncertainty quantification, and algorithmic fairness (1.2k+ citations). Previously, he was an AI Research Lead at JPMorgan AI Research and received a PhD in Statistics from CMU,…
🫡🤘🏾🚀
YOU INTRO → WE HIRE = $25k REWARD How to pitch your smart friends: 1. Recently hit '# 1 trending repository of the week' globally on GitHub → vast majority of the work will be open source 2. Team: 3 ML researchers (Stanford, CMU, Oxford, Columbia → 6k+ citations), Rust…
my 🐐@YoungseogC
Congratulations @youngseogc for successfully defending his ML PhD at CMU! (& thank you for the TensorZero shoutout!)
WANTED: Founding MTS with a background that combines product and engineering We can't keep up with TensorZero's open-source traction. You can help. Join our founding team in a cross-functional role that combines product, engineering, and GTM. Team: 3 ML researchers (Stanford,…
This release to me highlights the power of @rustlang and the PyO3 / NAPI-RS interfaces on top. It’s so nice to have a portable and canonical codebase to maintain consistency. The experimental features speak to the crazy stuff we’ll be building later this year. 🤔
TensorZero 2025.7.0 is out! 📌 This release revamps TensorZero's optimization workflows. Supervised fine-tuning now supports multimodal data (vision, documents, etc.), multi-turn tool use, and more. You can also launch and monitor optimization workflows programmatically with our…
It’s been a blast collaborating with Andrew over the past couple months and we couldn’t be more excited to have him on board. We’re excited for our developer community to benefit from the stuff he’s cooking!
Andrew is an ML researcher with deep expertise in Bayesian ML, causal inference, RL, and LLMs. He’s finishing his postdoc at Columbia and previously received a PhD from Oxford, during which he interned at Meta. He has 3.3k+ citations and several first-author papers at NeurIPS and…
I switched from a decade of vim to @cursor_ai. Now we can peek at how it works 🕵️. check out how we got under the hood of this bad boy
What happens under the hood of a $9.9B coding assistant? We reverse-engineered Cursor's LLM client to observe the requests being made (including prompts!), A/B test different models, and eventually optimize our own prompts + models. ↓ Link in thread with code to reproduce +…
really excited for Dynamic Evaluations - a key milestone for some of the crazy RL features we have planned
TensorZero 2025.5.8 is out! Highlights 📌 This release includes Dynamic Evaluations, a new feature that enables you to evaluate complex workflows that combine multiple inference calls with arbitrary application logic (e.g. agents, RAG, and more). Full Changelog 🔨 Remove…
United States Trends
- 1. Dodgers 544K posts
- 2. #WorldSeries 283K posts
- 3. Yamamoto 157K posts
- 4. Blue Jays 108K posts
- 5. Will Smith 40.8K posts
- 6. Miguel Rojas 36.2K posts
- 7. BACK TO BACK 1.05M posts
- 8. Kershaw 20K posts
- 9. Yankees 13.3K posts
- 10. Baseball 147K posts
- 11. Dave Roberts 10.8K posts
- 12. Ohtani 76.2K posts
- 13. Vladdy 20K posts
- 14. Kendrick 14.5K posts
- 15. #Worlds2025 26.2K posts
- 16. Jeff Hoffman 3,146 posts
- 17. Auburn 13.8K posts
- 18. Cubs 7,168 posts
- 19. Nike 30.7K posts
- 20. Ernie Clement 12.1K posts
You might like
-
David Held
@davheld -
Devendra Chaplot
@dchaplot -
Yiding Jiang
@yidingjiang -
Arthur Mensch
@arthurmensch -
Samuel Albanie 🇬🇧
@SamuelAlbanie -
Aurelien Lucchi
@AurelienLucchi -
Princeton Visual AI lab
@VisualAILab -
will grathwohl
@wgrathwohl -
Sanja Fidler
@FidlerSanja -
Zico Kolter
@zicokolter -
Davide Scaramuzza
@davsca1 -
Phillip Isola
@phillip_isola -
Deepak Pathak
@pathak2206 -
Rohin Shah
@rohinmshah -
Guanya Shi
@GuanyaShi
Something went wrong.
Something went wrong.