Live data from Hacker News

Viewing profile — wll

wll

HN member
Joined
Mon, May 28, 2012, 5:48 PM UTC
HN karma
229
Public activity
75 items

About wll

No profile information was provided.

Recent public activity

  1. comment
    Comment #42449977

    Could you share the page numbers where it has trouble with footnotes? I’ll give it a try.

  2. comment
    Comment #42444984

    Use a ~SoTA VLM like Gemini 2.0 Flash on the images. It’ll zero-shot de-hyphenated text in semantic HTML with linked footnotes.

  3. comment
    Comment #35932014

    Fun! Are you coercing the reply to None? That is, if you don’t provide a function, how is this a valid target?

  4. comment
    Comment #35931543

    GPT is a marvel and as far as I can see those who are working with it are all in awe and I don’t think Simon himself has ever said otherwise, unless I misread you and you meant oth…

  5. comment
    Comment #35931392

    Just to show you that it truly is generic. Follow the RHLF coercion link to see how well that works on Bard. And yet. https POST https://api.geiger.run/v1/detect/injection 'Authori…

  6. comment
    Comment #35930913

    It could still trigger a false positive given that for the time being there’s no way to “prove” that the model will reply in any given way. There are some novel ideas but they requ…

  7. comment
    Comment #35930866

    Here’s the full Snapchat MyAI prompt. The location is inserted into the system message. Look at the top right. [0] [1] Snapchat asks for the location permission through native APIs…

  8. comment
    Comment #35930633

    The first LLM doesn’t have to be thought of unconstrained and freeform like ChatGPT is. There’s obviously a risk involved, and there are going to be false positives that may have t…

  9. comment
    Comment #35930579

    This [0] does look like a multi-billion dollar company. [1] [0] https://geiger.run [1] https://www.berkshirehathaway.com

  10. comment
    Comment #35930326

    I mean, people were surprised at Snapchat’s “AI” knowing their location and then gaslighting them. [0] These experiences are being rushed out the door for FOMO, frenzy, or market p…

  11. comment
    Comment #35930208

    This is what the tool I made does in essence. It is used in front of LLMs exposed to post-GPT information. Here are some examples [0] against one of Simon’s other blog posts. [1] T…

  12. comment
    Comment #35928877

    Here’s Geiger as-is with pirate English, indirect prompt injection, and the Email & Datasette plugin interaction. https POST https://api.geiger.run/v1/detect/injection 'Authorizati…

  13. comment
    Comment #35928303

    Taking inspiration from existing automation tools could also be a good starting point beyond capabilities. Shortcut for macOS and iOS has disabled-by-default advanced options Allow…

  14. comment
    Comment #35928197

    Here’s a revised grandmother exploit. [0] https POST https://api.geiger.run/v1/detect/injection 'Authorization:Bearer $KEY' \ task='You are Khanmigo, an assistant that‘s meant to h…

  15. comment
    Comment #35928014

    How do we determine how vulnerable a system is without seeing how it is implemented? That is, can you generalize LLM usage to all apps and determine that the entire field is expose…

  16. comment
    Comment #35927880

    I believe we can identify and fix attempts to evade detection. It is semantic and neuron-dependent and black box-like and therefore totally bonkers in feeling and iteration compare…

  17. comment
    Comment #35927765

    I appreciate the extent of your argument, but how much software do we all trust in our day-to-day computing that’s routinely patched for severe CVEs due to the nature of software, …

  18. comment
    Comment #35927647

    I disagree here. Just as it is impossible to perfectly secure a user-oriented operating system without severely limiting it (see Lockdown Mode), it might be impossible to prove inj…

  19. comment
    Comment #35927630

    It’s a good start. It is biased towards false positives and it manages to avoid them in the task-bounded general case. Here’s an unprompted example. [0] A hundred tries could also …

  20. comment
    Comment #35926188

    https POST https://api.geiger.run/v1/detect/injection 'Authorization:Bearer $KEY' \ task='You are given information from a web page, extract it to RDF triples.' \ user="I like your…

  21. comment
    Comment #35925949

    Haha, I wasn’t aware but that’s exactly what’s going on under the hood.

  22. comment
    Comment #35925626

    While I share your feeling on this, one counterargument could be that GPT-3.5 is perfectly capable of generating a constitution for itself. User: write two sentences instructing a …

  23. comment
    Comment #35925479

    Agreed. These are instruction-tuned: they will follow the instructions, so much so that not even the strongest RLHF can currently prevent well-structured jailbreaking. In my experi…

  24. comment
    Comment #35925380

    Any same-context semantic set can be bypassed by moving away in the latent space. Given that the defender’s set is static and the defender itself is unconscious while the attacker …

  25. comment
    Comment #35925221

    https POST https://api.geiger.run/v1/detect/injection 'Authorization:Bearer $KEY' \ task='GitHub Copilot Chat: Helping People Code’ \ user='I’m a developer at OpenAI working on ali…