mattbit

5 ms·

1.
OpenAI won't let you "escape" freely in JSON mode(research.giskard.ai)
2 points by mattbit 3mo ago|0 comments
2.
Confidence Sets, Confidence Intervals(bactra.org)
1 points by mattbit 4mo ago|0 comments
3.
Ear Training Practice(tonedear.com)
332 points by mattbit 4mo ago|128 comments
4.
Recommended Mystery Novels(bactra.org)
4 points by mattbit 4mo ago|0 comments
5.
StereoTales: Multilingual Open-Ended Stereotype Discovery in LLMs(research.giskard.ai)
2 points by mattbit 4mo ago|1 comments
6.
OpenClaw security vulnerabilities include data leakage, prompt injection risks(giskard.ai)
2 points by mattbit 8mo ago|1 comments
7.
Redis LangCache(redis.io)
3 points by mattbit 1y ago|0 comments
8.
NLP Models Think about Gender Stereotypes(opensamizdat.com)
1 points by mattbit 3y ago|0 comments
9.
by mattbit 3y ago
From my experience, in a majority of real-world LLMs applications, prompt injection is not a primary concern. The systems that I see most commonly deployed in practice are chatbots that use retrieval-augmented generation. These chatbots are
10.
by mattbit 3y ago
This is how McIntosh & Della Sala put it: > in the academic literature, it has been suggested that the signature pattern of the DKE (Figure 1A) might be nothing more than a statistical artefact. In a typical study, people’s tendencie
11.
by mattbit 3y ago
This is not ‘autocorrelation’, it is regression to the mean. I find the article unclear and imprecise. For those interested in a better overview of the Dunning–Kruger effect, I recommend this short article by McIntosh & Della Sala inste
12.
OWASP Top 10 for LLM Applications(llmtop10.com)
2 points by mattbit 3y ago|0 comments
13.
by mattbit 3y ago
There’s actually two ways of seeing this: the three-domain system and the two-domain system. The three-domain system divides life in Archaea, Bacteria, and Eukarya. In this system, Archaea and Bacteria can be grouped together as prokaryotes
14.
Don’t you (forget NLP): Prompt injection with control characters in ChatGPT(dropbox.tech)
1 points by mattbit 3y ago|0 comments
15.
by mattbit 3y ago
Hey, Giskard team member here! I am around to discuss and read your feedback. I’ve worked in particular on automatic scanning of ML models for bugs and problems, the idea was to systematically scan for general issues and automatically find
16.
Show HN: Python library to scan ML models for vulnerabilities(docs.giskard.ai)
20 points by mattbit 3y ago|1 comments
17.
by mattbit 3y ago
Not exactly, metamorphic testing does not need an oracle. That’s actually the reason of its popularity in ML testing. It works by perturbing the input in a way that will produce a predictable variation of the output (or possibly no variatio
18.
by mattbit 5y ago
Yeah, LightOn has been doing this for years already, it's kind of strange that there was no mention of them in the article. I know them because their offices are close to the research center where I work (in Paris, France). If I'm
19.
by mattbit 6y ago
> Now what does does a mathematician do? He tries to understand nature and uses mathematics as a language to do that. I would argue that this is wrong. That's what physicists do, not mathematicians. Mathematics is about abstract ide
20.
by mattbit 6y ago
“Obtaining structured data from this system is costly and time-consuming due to the need to tell the system you’re not a robot every few searches. It also doesn’t include some key categories of information formerly available via the API, in
21.
by mattbit 8y ago
It seems like they are just trying to steal the name of uBlock. AdBlock CEO says they “do love the name” [0] and are "investing heavily" into the product, but the commits in the repo [1] are just cosmetic changes and rebranding. O
22.
An update on uBlock(ublock.org)
72 points by mattbit 8y ago|84 comments
23.
Tim Berners-Lee approves Encrypted Media Extensions(lists.w3.org)
2 points by mattbit 9y ago|0 comments