Aaron Tay (@aarontay) 's Twitter Profile
Aaron Tay

@aarontay

I'm librarian (Head, Data services) + blogger from Singapore Management University. Library science, Bibliometrics, academic discovery tech.

ID: 15196090

linkhttp://musingsaboutlibrarianship.blogspot.com/ calendar_today22-06-2008 08:48:58

41,41K Tweet

7,7K Followers

1,1K Following

Ai2 (@allen_ai) 's Twitter Profile Photo

Run olmOCR-bench yourself: github.com/allenai/olmocr… OCR your own documents: github.com/allenai/olmocr Try the olmOCR online demo: olmocr.allenai.org Read our updated technical report: olmocr.allenai.org/papers/olmocr.…

Aaron Tay (@aarontay) 's Twitter Profile Photo

Have to read this. But seems to basically reinforce results from Apples' "illusion of thinking" but without the flaws?

Aaron Tay (@aarontay) 's Twitter Profile Photo

Going to watch the video live soon but this is a great summary on what RAG (retrieval augmented generation) is and why it's (still) important - great analogies.

Aaron Tay (@aarontay) 's Twitter Profile Photo

Interesting. In theory libraries could set up MCP servers for their own special context (eg paywalled content or special restricted content) and use o3 deep research api to use them for generating results.

Aaron Tay (@aarontay) 's Twitter Profile Photo

I haven't tried this but if Elicit alert relevance level is similar to undermind alerts (and it probably is if they using LLM to judge relevance), these alerts will indeed be qualitatively better than the usual alerts you get from Google Scholar etc.

Aaron Tay (@aarontay) 's Twitter Profile Photo

They ran queries via Ai2's Scholar QA retrieval pipeline then fed the top 30 results as context to different LLM to generate the answer and crowd sourced user preference. o3 seem to be clearly the best - not surprising, I've noticed o3 seems to be the most "academic"

Aaron Tay (@aarontay) 's Twitter Profile Photo

[Blogged] A Deep Dive into EBSCOhost's Natural Language Search and Web of Science Smart Search - Two bundled "Ai-powered"search (I) open.substack.com/pub/aarontay/p…

SciSpace (@scispace_) 's Twitter Profile Photo

🚨 New on SciSpace: Introducing Deep Review Report We’ve taken Deep Review to the next level. The all-new Deep Review Report gives you a comprehensive, structured output that mirrors how top-tier researchers review literature. 📄 What’s inside the Deep Review Report? ✅

Consensus (@consensusnlp) 's Twitter Profile Photo

Introducing: Consensus Deep Search 🔬 Deep Search is a research agent that conducts literature reviews across 200M academic papers in two minutes.

Consensus (@consensusnlp) 's Twitter Profile Photo

We've loved to see everyone's response to our Deep Search launch last week đź’ś We've gotten a lot of questions about what it is and how it works, so our amazing AI engineer, Aakash (Aakash Adesara) made this video to walk you through the basics.