New base models with Llama 2, pretraining LLMs with ReLoRA, training with less data via AlpaGasus, and many more … There's been lots of interesting research in the last month! I summarized the research highlights in three sentences each:
LLMS
-
Text Splitting Techniques for LLM Applications
By
–
Text splitting is a crucial step in creating #LLM apps, where different text splitting methods can greatly impact retrieval quality. Explore various text-splitting techniques via @langchain
's `Text Splitting Playground`, built with @Streamlit
! ➤ https://
langchain-text-splitter.streamlit.app -
Economic potential of generative AI: The next productivity frontier
By
–
👌The economic potential of generative AI: The next productivity frontier👈
— Dr. Debashis Dutta (@debashis_dutta) 13 août 2023
by @McKinseyDigital#OPENAI #GenerativeAI #LLM #ChatGPT #LLM #DataScientist #Tech #BigData #Analytics #innovation #Crypto #MachineLearning #TensorFlow #coding #IoT #100DaysOfCode #NodeJS #Python… pic.twitter.com/4DGQvIAHJfThe economic potential of generative AI: The next productivity frontier by @McKinseyDigital #OPENAI #GenerativeAI #LLM #ChatGPT #LLM #DataScientist #Tech #BigData #Analytics #innovation #Crypto #MachineLearning #TensorFlow #coding #IoT #100DaysOfCode #NodeJS #Python
-
Building With LLMs: Knowledge Without Training Required
By
–
Not necessary, but the knowledge is good to have given how things are going today. You don't need to build or train LLMs, just understand how you can build with them.
-
Criminal AI Twin: FraudGPT Emerges as Unrestricted Cybercrime Weapon
By
–
Sinister 'twin' of #ChatGPT used by criminals as AI weapon 'without limitations’ https://
dailystar.co.uk/news/latest-ne
ws/evil-twin-chat-gpt-used-30687316?utm_source=twitter.com&utm_medium=social&utm_campaign=sharebar
… #GPT #LLM #LLMs #GenAI #FraudGPT #AI #RCE #ZeroTrust #ZeroDay #cybercrime #hacker #privacy #APT #bot #CISO #DDoS #hacking #phishing #CyberAttack #cybersecurity #Security -
LLMs Lack Justification Compared to Machine Learning Applications
By
–
So, machine learning not LLMs — there's **no justification** to calling out to these potential applications as an upside of pursuing LLMs (and all their attendant harms).
-
PoisonGPT: How Hugging Face LLM Spreads Misinformation
By
–
PoisonGPT: Hugging Face LLM Spreads Fake News: Large Language Models (LLMs) have gained significant popularity worldwide, but their adoption… #DataAnalytics #DataScience #DataDriven #ITDirector #ITManager #Blockchain #DeepLearning #BusinessIntelligence https://
analyticsvidhya.com/blog/2023/07/p
oisongpt-hugging-face-llm-spreads-fake-news/?utm_source=dlvr.it&utm_medium=twitter
… -
LLM Hallucinations and Public Safety Regulation Challenges
By
–
I’d say LOL but along with the myth of “watermarking” comes the reality that nobody who is actually regulating will just say, “Programs that have “hallucinations” are programs with ERRORS like telling people to consume poison.” Take them away from the general public. Period.
-
AI Agents Pipeline User Code Data ML Models LLM
By
–
With our AI agents, you can pipeline: – any user code
– any data transformations
– ML models & LLM prompts and execute and monitor these pipelines on our SOTA MLOps platform. Check it out here – -
Analyze Data Like a Pro With ChatGPT for AI Insights
By
–
Analyze data like a pro with ChatGPT! Uncover key insights and improve your conversational AI interactions. #ChatGPT #DataAnalysis #AIInsights via @ingliguori