Writing a Contract Test Suite for Your Own LLM Gateway
The assertions that belong to the proxy layer rather than to the model behind it, tested against a stub upstream so they run in milliseconds.
#LLM
9604 articles tagged with LLM
Prompt injection is still king, but “excessive agency” just jumped to #3 in OWASP’s Top 10 for LLM apps. Stop chasing un...
Prompt injection is still king, but “excessive agency” just jumped to #3 in OWASP’s Top 10 for LLM apps. Stop chasing unbreakable models—start containing fooled agents. https://jpm...
Researchers at IIT Bombay and Adobe Research built a method called Previous-Token Prediction that reconstructs an LLM’s ...
Researchers at IIT Bombay and Adobe Research built a method called Previous-Token Prediction that reconstructs an LLM’s original prompt from its output with near-perfect accuracy. ...
Qwen3.8-2.4T-A95B is finally out. Does anyone have a setup that can run this thing at home? lol #LLM #AI https://hugging...
Qwen3.8-2.4T-A95B is finally out. Does anyone have a setup that can run this thing at home? lol #LLM #AI https://huggingface.co/Qwen/Qwen3.8-2.4T-A95B
What LLM Cost Calculators Get Wrong
Is the number these calculators show you actually right? Not whether the model is in the catalog —...
New arXiv paper introduces exact Likert-scale framework to isolate LLM biases and attitudes, enabling controlled behavio...
New arXiv paper introduces exact Likert-scale framework to isolate LLM biases and attitudes, enabling controlled behavioral evaluation for autonomous agentsSource: arXiv cs.CLhttps...
VibeLifeBench introduces a benchmark of 200 long-horizon tasks to test if LLM agents can act proactively and persistentl...
VibeLifeBench introduces a benchmark of 200 long-horizon tasks to test if LLM agents can act proactively and persistently in a changing simulated world. Seven leading models all sc...
New benchmark finds automated evaluation of tool-using LLM agents often unreliable; GPT-4o-mini surpasses heuristic judg...
New benchmark finds automated evaluation of tool-using LLM agents often unreliable; GPT-4o-mini surpasses heuristic judging, and runtime interceptors cut hallucinations by 24 perce...
Living-Harness is a self-evolving agent framework that updates procedural knowledge from failure patterns to improve LLM...
Living-Harness is a self-evolving agent framework that updates procedural knowledge from failure patterns to improve LLM agent reliability, showing double-digit performance gains i...
The nice thing about OpenTelemetry for LLM apps: instrumentation is decoupled from the backend, so traces from your Pyth...
The nice thing about OpenTelemetry for LLM apps: instrumentation is decoupled from the backend, so traces from your Python, Go, and Java services land in one queryable place. Multi...
And it's not the only place from which I've been banned, btw. Using #LLM is not only dangerous for your privacy, but als...
And it's not the only place from which I've been banned, btw. Using #LLM is not only dangerous for your privacy, but also for your sociality !
Does anyone know of an #LLM friendly open #Mastodon instance ? I was banned from floss.social but I'm too lazy to self-h...
Does anyone know of an #LLM friendly open #Mastodon instance ? I was banned from floss.social but I'm too lazy to self-host something ( or I wouldn't be using LLMs right? )Fully st...
I really think that at this point, the brains of some of the anti-LLM people are exactly as fried as the worst AI bros'....
I really think that at this point, the brains of some of the anti-LLM people are exactly as fried as the worst AI bros'.Do they actually not have any senior devs around them who so...
@captaincalliope.at it's irresponsible to use #LLM tools.
@captaincalliope.at it's irresponsible to use #LLM tools.
"Help help! We accidentally fed a book about the Trojan Horse into our LLM, and it generated a bunch of angry and warlik...
"Help help! We accidentally fed a book about the Trojan Horse into our LLM, and it generated a bunch of angry and warlike Greeks!"https://freeradical.zone/@funnymonkey/117083125243...
Anyone else reading this or working on production-grade AI architecture as we speak?#SystemDesign #LLM #AI #SoftwareEngi...
Anyone else reading this or working on production-grade AI architecture as we speak?#SystemDesign #LLM #AI #SoftwareEngineering
The frontier #AI model security vulnerability finding orgy is the Internet foreclosing on tech debt.#infosec #LLM
The frontier #AI model security vulnerability finding orgy is the Internet foreclosing on tech debt.#infosec #LLM
Choosing the Right LLM-as-a-Judge: A Practical Guide with Model Recommendations
How to pick the best LLM judge for your RAG, generation, and OCR tasks. Data-driven recommendations, debiasing strategies, and cost-performance tradeoffs.
Third silent truncation in Lookspan in five review passes. This one was the worst.Before anything reaches the LLM judge ...
Third silent truncation in Lookspan in five review passes. This one was the worst.Before anything reaches the LLM judge it gets cut to 12,000 characters. The cut carried no marker....
https://www.youtube.com/watch?v=kON2ZI2BNj8#DataCenters #LLM is still not #AI
https://www.youtube.com/watch?v=kON2ZI2BNj8#DataCenters #LLM is still not #AI
How I feel about #AI "Detectors"#LLM
How I feel about #AI "Detectors"#LLM
Deploying DeepSeek V3 (LLM) Using SGLang
DeepSeek V3 is a 671B-parameter Mixture-of-Experts language model: Multi-head Latent Attention and...
Deploying Langfuse – Open-Source LLM Observability Platform
Langfuse is an open-source observability platform for LLM applications including traces...
I assume this sounded better in the original Chinese.#AI #LLM #shitpost
I assume this sounded better in the original Chinese.#AI #LLM #shitpost