Replies: 1 comment
-
|
@HanClinto I've been experimenting with new versions of Llama.cpp and am currently downloading the binaries from the Llama-dot-cpp repo using the gh cli tool. I think it would add a bit more security to do some checksum comparisons. I'm also starting to notice the fast pace of Llama.cpp creates a kind of whack-a-mole for this project. I can't tell if this project has a build process yet or if the "Releases" are essentially tagged commits in main. I'm curious to hear from the core team on this as well. |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
-
Currently we're checking in llama.cpp binaries directly into source control. We keep a version tag to indicate which version of llama.cpp we're copy-pasting in here, but that's a layer of trust.
Feels like a good level of cross-checking might be to validate the md5 or other hash sums with the official llama.cpp release binary checksums, to give another layer of verification that a rogue binary didn't actually get slipped into the mix.
We could even make part of the build process to be to take the version code from our version.txt and download / inflate the zip as part of the script.
Could possibly be a feature that would go hand-in-hand with #28 as well.
Beta Was this translation helpful? Give feedback.
All reactions