Meet ReasonIR-8B✨the first retriever specifically trained for reasoning tasks! Our challenging synthetic training data unlocks SOTA scores on reasoning IR and RAG benchmarks. ReasonIR-8B ranks 1st on BRIGHT and outperforms search engine and retriever baselines on MMLU and GPQA🔥

RT @VictoriaWGraf: Worried about overfitting to IFEval? 🤔 Use ✨IFBench✨ our new, challenging instruction-following benchmark!. Loved workin….

RT @qi2peng2: Seven years ago, I co-led a paper called 𝗛𝗼𝘁𝗽𝗼𝘁𝗤𝗔 that has motivated and facilitated many #AI #Agents research works since. T….

RT @Benjamin_eecs: We've always been excited about self-play unlocking continuously improving agents. Our insight: RL selects generalizable….

RT @ChengleiSi: Are AI scientists already better than human researchers?. We recruited 43 PhD students to spend 3 months executing research….

RT @thao_nguyen26: Web data, the “fossil fuel of AI”, is being exhausted. What’s next?🤔.We propose Recycling the Web to break the data wall….

It reminds me of the cognitive behaviors that have been found to help reasoning—backtracking, subgoal setting, verifications, etc.—they all seem to fit this parallel generation pattern better than linearly chaining them. Looking forward to trying it out!.

Honored to be part of organizing the LM4Sci workshop at #COLM2025! 🔬🤖 We invite submissions that demonstrate innovative approaches to scientific reasoning and discovery. Submit by June 23! 🚀.

RT @ziruirayliu: 🔥Exited to share our new work on reproducibility challenges in reasoning models caused by numerical precision. Ever run t….

Arxiv: Clearly a lot more work is needed to understand what’s really happening with RL and prompting. We hope that our experiments with spurious rewards and spurious prompts, as well as the released code, data, checkpoints, etc. will help with this! 🔍.

🎉Our Spurious Rewards is available on ArXiv! We added experiments on.- More prompts/steps/models/analysis. - Spurious Prompts!.Surprisingly, we obtained 19.4% gains when replacing prompts with LaTex placeholder text (\lipsum) 😶‍🌫️. Check out our 2nd blog:

RT @HanGuo97: One key takeaway from recent work on test-time compute: even a small weight update can make a big difference. So, what happen….

Tip: remember to install massive-serve first: . pip install -U massive-serve. The first-time run is slow because it downloads the datastore and the model from HF. You may need to update transformers version to support Qwen. 😃.

Qwen3-0.6B x Wikipedia datastore is now supported in massive-serve! Serve a local API in one line:. massive-serve serve --domain --domain_name dpr_wiki_qwen3_0.6b_ivfpq. Use `dpr_wiki_qwen3_0.6b` for flat index. Examples of sending single/batch queries:

RT @HanGuo97: We know Attention and its linear-time variants, such as linear attention and State Space Models. But what lies in between?. I….

RT @jxmnop: new paper from our work at Meta!. **GPT-style language models memorize 3.6 bits per param**. we compute capacity by measuring t….

RT @lschmidt3: Very excited to finally release our paper for OpenThoughts!. After DataComp and DCLM, this is the third large open dataset m….

RT @HannaHajishirzi: Check out who the 2025 ACM Dissertation Award honorees are this year — our very own @sewon__min and @sharma_ashish_2!….

RT @AkariAsai: ‘Bold,’ ‘positive’ and ‘unparalleled’: Allen School Ph.D. graduates Ashish Sharma and Sewon Min recognized with ACM Doctoral….

RT @StellaLisy: Excited to share more about Spurious Rewards! Also keep an eye out for some new experiments and arxiv coming soon 👀🔜.

RT @jihan_yao: We introduce MMMG: a Comprehensive and Reliable Evaluation Suite for Multitask Multimodal Generation. ✅ Reliable: 94.3% agre….