Simon Willison’s Post

Highly recommended for this course: Gain a comprehensive understanding of the key aspects involved in the democratization of data-driven systems, specifically with large language models (LLMs). https://lnkd.in/diV2gEhP

Looking for a non-technical overview of how Large Language Models function? Check out Andrej Karpathy's general-audience introduction to Large Language Models .  It breaks down everything from training and inference to jailbreaks and finetuning in a way that's easy to understand. Watch it here>https://lnkd.in/g2cpeMyZ

We see many interesting use cases popping up where Large Language Models can play a crucial role. A question we get often at ML6 for those use cases is: "Should we   fine-tune a large language model to enable this use case?  Or are other techniques more appropriate?" In this   Shakespeare-inspired blogpost my colleague Michiel clearly separates the ambition of adjusting the behaviour of a large language models from that of adjusting the knowledge of a large language model. Through this blogpost, we aim to offer you the right intuition in thinking about which use cases require which approach. https://hubs.la/Q0222YkQ0 #ml6 #finetuning #LLMs #LargeLanguageModels 

Worth Reading: Let’s Be Objective about Large Language Models

A new Tame Your Assets blog post is out! Dive into the dynamic landscape of Large Language Models (LLMs), as I discuss variability in results, continuous evolution, and the importance of responsible utilization.

🚀 Excited to share a new blog post on fast adaptive adversarial prompting for Large Language Models \(LLMs\). In this paper, a novel method using the AdvPrompter generates human-readable adversarial prompts in seconds, $\sim800\times$ faster than existing optimization-based approaches, without requiring access to the gradients of the TargetLLM. Check out the full article here: https://bit.ly/4aU789s.

🚀 Excited to share our latest blog post on "Large Language Models for Intent-Driven Session Recommendations"! We delve into the limitations of traditional approaches in Intent-aware session recommendation and introduce a novel ISR approach utilizing advanced reasoning capabilities of large language models. Our method empowers LLMs to discern diverse user intents at a semantic level, leading to more accurate and interpretable session recommendations. Check out the full post here: https://bit.ly/47XHf74 [cs.CL] #socialmediamarketing #blogpost #recommendations

Quote - "We demonstrate here a dramatic breakdown of function and reasoning capabilities of state-of-the art models trained at the largest available scales which claim strong function, using a simple, short, conventional common sense problem formulated in concise natural language, easily solvable by humans." https://lnkd.in/eanfaQ3i

Interesting to see the emerging large language model abstractions. https://lnkd.in/d2-tCVhn

Explore Large Language Models (LLMs) through practical fine-tuning examples. Transform your models with simplicity and efficacy.