Building knowledge base for #RAG . Any advice on the database, files types, building process? The repository of local PDF files on research designs already exists. Do we need the server?

Introducing our new guide: 'How to use Overton to plan for impact' - bit.ly/3W8diyn

This resource by Kat Hart 🌍🏞 demonstrates how Overton can help you find evidence gaps identified by policymakers and assess how your research might fit within current policy debates 🌏

Great news for those doing literature reviews. Our literature matrix export to Excel now includes Evidence strength for each study, along with commentary, so you can compare studies side-by-side more easily for aspects such as randomization, participant numbers, and effect size.

Really pleased with this. Along with our new async upload (thanks Bruno !), one of the last pieces of the puzzle is now in place. ChatGPT can't do this.

I am looking for the example of LLM fine tuning by libraries #AI #libraries venturebeat.com/ai/openai-rele…

University Policies on Generative AI, a Padlet collection
padlet.com/cetl6/universi… Thanks to Maha Bali, PhD مها بالي 🌵🇵🇸 for the heads up on this.

For me as a librarian and a researcher who specializes in research synthesis this section is important AI shouldn’t ➵ undermine information integrity and access to information;

Kamal Nahas Nature Index Last month, the article below was published in @PLoSONE comparing abstracts for medical articles written by humans and written by LLM's. Seems very relevant to your article (but not ref'd, and so not read?).
journals.plos.org/plosone/articl…

Important in my field , research synthesis: “Generative AI, like ChatGPT, will not be classified as high-risk but will have to comply with transparency requirements and EU copyright law. Some obligations:
- Disclosing that content was generated by AI;

'Systematic review search reporting is poor...Only 22 (4.9%) database searches reported all six PRISMA-S items. Forty-seven (10.4%) database searches could be reproduced within 10% ..'
bit.ly/3wP8liZ
Melissa Rethlefsen Cochrane PRISMA Statement

Just found out about Anthropic's free database with prompt templates for Claude 3.

What's interesting, is how you prompt Claude for optimized responses isn't always the same as ChatGPT.

8 must-try prompts from the database and how to access:

Pro-Palestinian activist vandalizes antique portrait of Arthur Balfour

In Britain's Cambridge, a masked pro-Palestinian activist spray-painted and slashed an antique portrait of one of the founders of the Jewish state, former British Prime Minister Arthur Balfour.

Today, we're announcing Claude 3, our next generation of AI models.

The three state-of-the-art models—Claude 3 Opus, Claude 3 Sonnet, and Claude 3 Haiku—set new industry benchmarks across reasoning, math, coding, multilingual understanding, and vision.

Play with OLMo-7B-Instruct for yourself on Together AI!
api.together.xyz/playground/cha…

#AI : ProQuest, Part of Clarivate, Launches AI-Powered Research Assistant & More News Headlines ow.ly/A2N750QJrLk #gpt #scholcomm #libraries @clarivate @proquest

✍️'A significant portion, approximately 28%, of academic journal articles with DOIs appear entirely unpreserved'

#ScholComm wp.me/p4m9em-cRd