DEV Community

#llm

Posts

đź‘‹ Sign in for the ability to sort posts by relevant, latest, or top.
Your AI Has No Long-Term Memory. You Need an External Brain

Your AI Has No Long-Term Memory. You Need an External Brain

Comments 1
7 min read
I almost added an em-dash remover to my LLM library. Then I tested whether local models even produce em-dashes.

I almost added an em-dash remover to my LLM library. Then I tested whether local models even produce em-dashes.

3
Comments
5 min read
Mutation testing, but for LLM evals — early experiment, would love feedback

Mutation testing, but for LLM evals — early experiment, would love feedback

6
Comments
2 min read
The Cheaper API Was 2.5x Cheaper. It Cost 1.6x More.

The Cheaper API Was 2.5x Cheaper. It Cost 1.6x More.

Comments
10 min read
Has Claude Code changed your role from coding to making engineering decisions?

Has Claude Code changed your role from coding to making engineering decisions?

1
Comments
5 min read
Ralph Loops Are Not About Throwing Away Context

Ralph Loops Are Not About Throwing Away Context

3
Comments 2
12 min read
How to Build a Multi-Step Agent Stress Test: Adversity Sandboxes and Oracle Checks

How to Build a Multi-Step Agent Stress Test: Adversity Sandboxes and Oracle Checks

1
Comments
1 min read
FlashMemory Cuts DeepSeek-V4's KV Cache to 13.5%: Lookahead Sparse Attention

FlashMemory Cuts DeepSeek-V4's KV Cache to 13.5%: Lookahead Sparse Attention

Comments
6 min read
The Context Window Isn't Memory. It's the CPU Cache of AI.

The Context Window Isn't Memory. It's the CPU Cache of AI.

8
Comments
4 min read
The Model Does Not Need Memory. The Situation Does.

Value of information absent from weights

The Model Does Not Need Memory. The Situation Does.

51
Comments 38
11 min read
SWIRL 5 is GA: knowledge authority for enterprise AI

SWIRL 5 is GA: knowledge authority for enterprise AI

2
Comments 1
5 min read
I built a Threat Intelligence RAG System from scratch — here's what actually broke

I built a Threat Intelligence RAG System from scratch — here's what actually broke

Comments 1
1 min read
The Citation Lied Without Lying: The Hard Limit of My Memory Gate

Publicly pre-registered failure predictions

The Citation Lied Without Lying: The Hard Limit of My Memory Gate

23
Comments 126
8 min read
OrientaIA: diseñar un orientador vocacional conversacional para adolescentes

OrientaIA: diseñar un orientador vocacional conversacional para adolescentes

Comments
2 min read
Generative AI vs Agentic AI vs AI Agents [2026 Compared]

Generative AI vs Agentic AI vs AI Agents [2026 Compared]

Comments
16 min read
đź‘‹ Sign in for the ability to sort posts by relevant, latest, or top.