DEV Community

#llm

Posts

đź‘‹ Sign in for the ability to sort posts by relevant, latest, or top.
Small language models (1B–3B): what they're good for after fine-tuning
pranjulrathour profile

Small language models (1B–3B): what they're good for after fine-tuning

#smallmodels #finetuning #llm #deployment
Comments
3 min read
LoRA vs QLoRA vs full fine-tuning: cost, quality and when each makes sense
pranjulrathour profile

LoRA vs QLoRA vs full fine-tuning: cost, quality and when each makes sense

#finetuning #lora #qlora #llm
Comments
3 min read
MCP Tool Descriptions Are Instructions, Not Metadata -- and That Is the Exploit
rxkov profile
Rxkov

MCP Tool Descriptions Are Instructions, Not Metadata -- and That Is the Exploit

#ai #llm #mcp #security
1
Comments 3
5 min read
QLoRA hyperparameters that actually matter (rank, alpha, learning rate, epochs)
pranjulrathour profile

QLoRA hyperparameters that actually matter (rank, alpha, learning rate, epochs)

#qlora #hyperparameters #finetuning #llm
Comments
3 min read
OPSEC for Agents: Scope, Caps, Monitoring, and Kill Switches
rxkov profile
Rxkov

OPSEC for Agents: Scope, Caps, Monitoring, and Kill Switches

#agentic #llm #opsec #security
Comments 1
4 min read
LLM Agent Sandboxes Have Two Escape Surfaces. Only One Gets CVEs.
rxkov profile
Rxkov

LLM Agent Sandboxes Have Two Escape Surfaces. Only One Gets CVEs.

#agents #llm #mcp #security
Comments 2
5 min read
Qwen3.8-Max Is Huge. The Agent Harness Still Decides
zira125 profile
Zira

Qwen3.8-Max Is Huge. The Agent Harness Still Decides

#devops #ai #agents #llm
5
Comments 1
3 min read
The Orchestrator's Deputy Has No Scope: Ambient Authority Is the Root Bug in Multi-Agent AI
rxkov profile
Rxkov

The Orchestrator's Deputy Has No Scope: Ambient Authority Is the Root Bug in Multi-Agent AI

#agents #cybersecurity #llm #security
Comments 2
5 min read
5 Things That Make an AI Agent Actually Useful (Not Just Cool)
aninmukhe profile

5 Things That Make an AI Agent Actually Useful (Not Just Cool)

#ai #llm #programming #productivity
Comments
4 min read
Giving AI Agents the Same RBAC Rules as Your Users: Building a Laravel Permission Layer LLMs Actually Respect
hosseinhezami profile

Giving AI Agents the Same RBAC Rules as Your Users: Building a Laravel Permission Layer LLMs Actually Respect

#ai #laravel #permission #llm
5
Comments
11 min read
RAG Poisoning Is a Supply Chain Attack, Not a Prompt Injection Variant
rxkov profile
Rxkov

RAG Poisoning Is a Supply Chain Attack, Not a Prompt Injection Variant

#ai #llm #rag #security
Comments
4 min read
AI Agents Are eval(userInput) With Extra Steps
rxkov profile
Rxkov

AI Agents Are eval(userInput) With Extra Steps

#agents #ai #llm #security
Comments
5 min read
Open-Source LLMs Are Unverified Dependencies: The Model Backdoor Gap npm Already Solved
rxkov profile
Rxkov

Open-Source LLMs Are Unverified Dependencies: The Model Backdoor Gap npm Already Solved

#cybersecurity #llm #opensource #security
Comments
5 min read
OAuth Token Leakage in Agentic AI: When the Context Window Becomes a Credential Store
rxkov profile
Rxkov

OAuth Token Leakage in Agentic AI: When the Context Window Becomes a Credential Store

#ai #cybersecurity #llm #security
Comments
4 min read
LLM API Keys Fail Faster: Five Exposure Vectors Your Secret Hygiene Guide Has Not Caught Up With
rxkov profile
Rxkov

LLM API Keys Fail Faster: Five Exposure Vectors Your Secret Hygiene Guide Has Not Caught Up With

#ai #cybersecurity #llm #security
Comments
4 min read
đź‘‹ Sign in for the ability to sort posts by relevant, latest, or top.