From training to inference: The new role of web data in LLMs
Data has always been key to LLM success, but it's becoming key to inference-time performance as well.
Loading…
contributed
Community-generated articles submitted for your reading pleasure.
Related Tags
Variants of LoRA
Want to train a specialized LLM on your own data? The easiest way to do this is with low rank adaptation (LoRA), but many variants of LoRA exist.
Why all developers should adopt a safety-critical mindset
Is anyone designing software where failures don't have consequences?
Generative AI is not going to build your engineering team for you
It’s easy to generate code, but not so easy to generate good code.
You should keep a developer’s journal
A developer’s journal is a place to define the problem you’re solving and record what you tried and what worked.
This developer tool is 40 years old: can it be improved?
Would updating a tool few think about make a diff(erence)?
Four approaches to creating a specialized LLM
Wondering how to go about creating an LLM that understands your custom data? Start here.
Masked self-attention: How LLMs learn relationships between tokens
Masked self-attention is the key building block that allows LLMs to learn rich relationships and patterns between the words of a sentence. Let’s build it together from scratch.
The hidden cost of speed
It’s tempting to push projects out the door to woo and impress colleagues and supervisors, but the stark truth is that even the smallest projects should have proper review periods.
Best practices for cost-efficient Kafka clusters
In today's data-driven world, Apache Kafka has emerged as a cornerstone of modern data streaming, particularly with the rise of AI and the immense volumes of data it generates.
LLMs evolve quickly. Their underlying architecture, not so much.
The decoder-only transformer architecture is one of the most fundamental ideas in AI research.
Practical tips for retrieval-augmented generation (RAG)
Retrieval-augmented generation (RAG) is one of the best (and easiest) ways to specialize an LLM over your own data, but successfully applying RAG in practice involves more than just stitching together pretrained models.
Navigating cities of code with Norris Numbers
Settling down in a new city (or codebase) is a marathon, not a sprint.
How data are reshaping society: “Datafication” and socioeconomic transformations
More and more of our lives are becoming data-driven. Is that a good thing?
Explaining generative language models to (almost) anyone
Here’s a simple, three-part framework that explains generative language models.
How to succeed as a data engineer without the burnout
The key strategies for building a headache-free data platform.
How do mixture-of-experts layers affect transformer models?
This new LLM technique has started improving the results of models without additional training.
The hardest part of building software is not coding, it's requirements
Why replacing programmers with AI won’t be so easy.
Stop saying “technical debt”
Everyone who says "tech debt" assumes they know what we’re all talking about, but their individual definitions differ quite a bit.
Is software getting worse?
With all the advancements in software development, apps could be much better. Why aren't they?
Three types of AI-assisted programmers
What matters isn’t just whether you use it, but how.
Retrieval augmented generation: Keeping LLMs relevant and current
Retrieval augmented generation (RAG) is a strategy that helps address both LLM hallucinations and out-of-date training data.
From prototype to production: Vector databases in generative AI applications
What exactly is a vector database? And how does it relate to generative AI?
To improve as an engineer, get better at requesting (and receiving) feedback
It’s easy to ask for, and even want, feedback in a sort of theoretical sense. But soliciting and responding to feedback are, themselves skills.
