Fun times with nvidia sparks and LLM.
https://f4n6.co.uk/tp-4-qwen3-5-397b-on-dgx-spark-from-silent-zeros-to-41-5-tok-s/
Remote
csec
@csec@infosec.exchange
mastodon 4.8.0-alpha.3+glitch2 Followers
53 Following
4 Posts
Joined November 08, 2022
@GossiTheDog@cyberplace.social
Almost all of the soar platforms don’t help the actual analysis. They want to “enrich” pointless things, like Whois information or just a lookup to virus total.
None of that matters, if it’s flagging on an ip that it saw as being doing something bad, I personally care about what services are on that ip if any and immediately this can steer the analysis, but sadly they don’t do that and just a whole host of useless things so they can show a dashboard and claim it has saved X dolllars.
The best automation is automating your most repetitive tasks, can you write a script to do those for you, and this is basically free and more useful than any soar platforms, yet sadly most companies do not encourage this for their staff and buy some terrible soar platform
Replying to
@DavidJBianco@infosec.exchange open source LLM should be what companies use for incident response, giving the ai companies access to sensitive data from an incident can’t be good in the first place
Replying to
@DavidJBianco@infosec.exchange it is interesting, I have been building my own AI analysis toolkit for some of the reasons that they highlight in the investigation. You can run glm 5.2 on 4 nvidia sparks and get good results, so the costs are tiny in comparison to the larger nvidia gpu’s that would require several hundred thousand spend on gpu’s