llamafile v0.10.0

[aittalam]

llamafile versions starting from 0.10.0 use a new build system, aimed at keeping our code more easily
aligned with the latest versions of llama.cpp. This means they support more recent models and functionalities,
but at the same time they might be missing some of the features you were accustomed to (check out this doc for a high-level description of what has been done).

If you liked the "classic experience" more, you will always be able to access the previous versions from our releases page. Our pre-built llamafiles show which version of the server they have been bundled with (0.9.* example, 0.10.* example), so you will always know which version of the software you are downloading.

What's Changed

• Accept array in chat message content field by @henfiber in Accept array in chat message content field #760
• chore: Update README.md to include call for community feedback on llamafile by @njbrake in chore: Update README.md to include call for community feedback on llamafile #812
• chore: integrate whisper.cpp as a submodule by @njbrake in chore: integrate whisper.cpp as a submodule #813
• chore: convert stable diffusion to submodule by @njbrake in chore: convert stable diffusion to submodule #818
• chore: llama.cpp as submodule by @njbrake in chore: llama.cpp as submodule #819
• feat: move docs to mkdocs by @njbrake in feat: move docs to mkdocs #824
• chore: Add update-llama-cpp workflow. by @daavoo in chore: Add update-llama-cpp workflow. #846
• fix(update-llama-cpp): Use new_build_wip as base ref. by @daavoo in fix(update-llama-cpp): Use new_build_wip as base ref. #850
• Fixed broken llamafile URL in docs by @aittalam in Fixed broken llamafile URL in docs #873
• update supported OpenBSD versions by @sthen in update supported OpenBSD versions #897
• llamafile reloaded (v0.10.0) by @aittalam in llamafile reloaded (v0.10.0) #867

New Contributors

• @henfiber made their first contribution in Accept array in chat message content field #760
• @njbrake made their first contribution in chore: Update README.md to include call for community feedback on llamafile #812
• @daavoo made their first contribution in chore: Add update-llama-cpp workflow. #846
• @sthen made their first contribution in update supported OpenBSD versions #897

Full Changelog: 0.9.3...0.10.0

---

This discussion was created from the release llamafile v0.10.0.

[yjftsjthsd-g]

With this version, the zip file is much smaller than the raw binary. With previous releases, it was the other way around. What happened?

llamafile-0.10.0  721 MB
llamafile-0.10.0.zip  203 MB
llamafile-0.9.3  293 MB
llamafile-0.9.3.zip  477 MB
llamafile-0.9.2  293 MB
llamafile-0.9.2.zip  529 MB

[aittalam]

Hi @yjftsjthsd-g! This is due to two different factors:

• if you look into the two zip files, you'll see that 0.9.3 contains three binaries (llamafile, localscore, whisperfile) which are larger than the rest, and that's because they are all bundled with cuda libs. In 0.10.0, only llamafile is bundled with cuda because we still have to test cuda support for whisperfile. The single llamafile executable is smaller, in 0.9.3, than the zip file, because the directory contents were more than 3 times the size of the llamafile binary, and even after compression they take more space than llamafile alone.

• the new cuda dylibs are way larger than they were, which is the reason why the 0.10.0 llamafile is even larger than the previous version's zip file. And as it accounts for most of the contents of the 0.10.0 zip file, once compressed it takes much less space.