Nobody Likes a Know-It-All: Smaller LLMs are Gaining Momentum
Was this email forwarded to you? Sign up here Nobody Likes a Know-It-All: Smaller LLMs are Gaining MomentumPhi-3 and OpenELM, two major small model releases this week.Next Week in The Sequence:
You can subscribed to The Sequence below:📝 Editorial: Nobody Likes a Know-It-All: Smaller LLMs are Gaining MomentumLast year, Microsoft coined the term 'small language model' (SLM) following the publication of the influential paper 'Textbooks Are All You Need', which introduced the initial Phi model. Since then, there has been a tremendous market uptake in this area, and SLMs are starting to make inroads as one of the next big things in generative AI. The case for SLMs is pretty clear. Massively large foundation models are likely to dominate generalist use cases, but they remain incredibly expensive to run, plagued with hallucinations, security vulnerabilities, and reliability issues when applied in domain-specific scenarios. Add to that environments such as mobile or IoT, which are computation-constrained by definition. SLMs are likely to fill that gap in the market with hyper-specialized models that are more secure and affordable to execute. This week we had two major developments in the SLM space:
Large foundation models have commanded the narrative in generative AI and will continue to do so while the scaling laws hold. But SLMs are certainly going to capture an important segment of the market. After all, nobody likes a know-it-all ;)" 🔎 ML ResearchPhi-3Microsoft Research published the technical report of Phi-3, their famous small language model that excel at match and computer science task. The new models are not that small anymore with phi-3-mini at 3.8B parameters and phi-3-small and phi-3-medium at 7B and 14B parameters respective —> Read more. The Instruction HierarchyOpenAI published a paper introducing the instruction hierarchy which defines the model behavior upon confronting conflicting instructions. The method has profound implications in LLM security scenarios such as preventing prompt injections, jailbreaks and other attacks —> Read more. MAIAResearchers from MIT published a paper introducing Multimodal Automated Interpretability Agent (MAIA), an AI agent that can design experiments to answer queries of other AI models. The method is an interesting approach to interpretability to prove generative AI models to undestand their behavior —> Read more. LayerSkipMeta AI Research published a paper introducing LayerSkip, a method for accelerated inference in LLMs. The method introduces modification in both the pretraining and inference process of LLMs as well as a novel decoding solution —> Read more. GeckoGoogle DeepMind published a paper introducing Gecko, a new benchmark for text to image models. Gecko is structured as a skill-based benchmark that can discriminate models across different human templates —> Read more. 🤖 Cool AI Tech ReleasesOpenELMApple open sourced OpenELM, a family of small LLMs optimized to run on devices —> Read more. ArticSnowflake open sourced Artic, an MoE model specialized in enterprise workloads such as SQL, coding and RAG —> Read more. MeditronResearchers from EPFL’s School of Computer and Communication Sciences and Yale School of Medicine released Meditron, an open source family of models tailored to the medical field —> Read more. Cohere ToolkitCohere released a new toolking to accelerate generative AI app development —> Read more. PenzaiGoogle DeepMind open sourced Penzai, a research tookit for editing and visualizing neural networks and inject custom logic —> Read more. 🛠 Real World MLFixing Code BuildsGoogle discusses how they trained a model to predict and fix build fixes —> Read more. Data Science Teams at LyftLyft shared some of the best practices and processes followed for building its data science teams —> Read more. 📡AI Radar
You’re on the free list for TheSequence Scope and TheSequence Chat. For the full experience, become a paying subscriber to TheSequence Edge. Trusted by thousands of subscribers from the leading AI labs and universities. |
Older messages
Edge 390: Diving Into Databricks' DBRX: One of the Most Impressive Open Source LLMs Released Recently
Thursday, April 25, 2024
The model uses an MoE architecture which exhibits remarkable perfromance on a relatively small budget. ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏
Edge 389: Understanding Large Action Models
Tuesday, April 23, 2024
One of the most important concepts in autonomous agents. ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏
Some Cool Details About Llama 3
Sunday, April 21, 2024
Solid performance, new tokenizer, fairly optimal training and other details about Meta AI's new model. ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏
Edge 388: Google DeepMind's SIMA can Follow Language Instructions in 3D Games Just Like Humans
Thursday, April 18, 2024
The AI agent represents a major improvement relative to expensive reinforcement learning methods. ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏
Edge 387: Tool Learning in Autonomous Agents
Tuesday, April 16, 2024
Agents that master tools and APIs, UC Berkeley's Gorilla and Microsoft's TaskWeaver ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏
You Might Also Like
WP Weekly 193 - Core Focus - Newsletter Blocks, Custom Woo Emails, WC Fellowship
Monday, May 13, 2024
Read on Website WP Weekly 193 / Core Focus Focusing on core WordPress, things are going strong – be it the refresh button solution or the contributors' feedback. On the positive side, the proposal
SRE Weekly Issue #424
Monday, May 13, 2024
View on sreweekly.com A message from our sponsor, FireHydrant: FireHydrant is now AI-powered for faster, smarter incidents! Power up your incidents with auto-generated real-time summaries,
📺 Getting Around Netflix Screen Limits — Don't Overlook These Smartphone Specs
Sunday, May 12, 2024
Also: Alienware's Wireless Mechanical Gaming Keyboard, and More! How-To Geek Logo May 12, 2024 📩 Get expert reviews, the hottest deals, how-to's, and more delivered directly to your inbox by
Daily Coding Problem: Problem #1439 [Easy]
Sunday, May 12, 2024
Daily Coding Problem Good morning! Here's your coding interview problem for today. This problem was asked by Microsoft. Given a 2D matrix of characters and a target word, write a function that
C#505 Refactor your code with C# collection expressions
Sunday, May 12, 2024
Explore various C# 12 refactoring scenarios for a variety of target types
RD#455 The Forensics Of React Server Components
Sunday, May 12, 2024
Deep look at the impact they have on the page load timeline
PD#573 Uncovering the Seams in Mainframes for Incremental Modernisation
Sunday, May 12, 2024
Mainframe systems continue to run much of the world's computing workload
Sunday Digest | Featuring 'The Top 20 Countries in Debt to China' 📊
Sunday, May 12, 2024
Every visualization published this week, in one place. Visual Capitalist Sunday Digest logo May 12, 2024 | View Online | Subscribe | VC+ The Best of This Week's Visuals Presented by Voronoi: The
Android Weekly #622 🤖
Sunday, May 12, 2024
View in web browser 622 May 12th, 2024 Articles & Tutorials Sponsored Workshop: Integrate Embrace's Android SDK Join us Thursday, May 23rd for a 30 minute session with solution engineers from
No one likes the new iPad ad
Sunday, May 12, 2024
Plus: Tesla's Supercharger network is in limbo View this email online in your browser By Anthony Ha Sunday, May 12, 2024 Image Credits: Apple Apple announced new iPad Pro devices this week, as well