81–90 of 302 posts
If my reading is correct, this literally just distribute an LLM model and code, and you need to do some tasks - like building - to make it actually run, right? And for this, you need to have additional tools installed?
There are some gotchas to watch out for though: https://github.com/mozilla-Ocho/llamafile#gotchas
I've been playing with various models in llama.cpp's GGUF format like this. git clone https://github.com/ggerganov/llama.cpp cd llama.cpp make # M2 Max - 16 GB RAM wget -P ./models https://huggingface.co/TheBloke/OpenHermes-2.5-Mistral-7B-16k-GGUF/resolve/main/openhermes-2.5-mistral-7b-16k.Q8_0.gguf ./server -m models/openhermes-2.5-mistral-7b-16k.Q8_0.gguf -c 16000 -ngl 32 # M1 - 8 GB RAM wget -P ./models https://hu…
Extremely cool and Justine Tunney / jart does incredible portability work [0], but I'm kind of struggling with the use-cases for this one. I make a small macOS app [1] which runs llama.cpp with a SwiftUI front-end. For the first version of the app I was obsessed with the single download -> chat flow and making 0 network connections. I bundled a model with the app and you could just download, open, and start using it.…
The binaries themselves are available standalone https://github.com/Mozilla-Ocho/llamafile/releases
I see it unfortunately requires the CUDA developer toolkit to be installed. It's totally possible to distribute CUDA apps that run without any dependencies installed other than the Nvidia driver. If they could figure that out it would be a game changer.
Earlier quoted context omitted.
I guess in every world I've worked in, deployment involved deploying a small executable which would run millions of times on thousands of servers, each instance loading a different model (or models) over its lifetime, and the weights are stored in a large, fast filesystem with much higher aggregate bandwidth than a typical local storage device. The executable itself doesn't even contain the final model- just a descri…
> a large, fast filesystem with much higher aggregate bandwidth than a typical local storage device that assumption gets wrong very fast with nvme storage, even before you add herding effects
Earlier quoted context omitted.
You're right. The limit is 4 gibibytes . Astonishingly enough, the llava-v1.5-7b-q4-server.llamafile is 0xfe1c0ed4 bytes in size, which is just 30MB shy of that limit. https://github.com/Mozilla-Ocho/llamafile/commit/81c6ad3251f...
Not at my windows machine to test this out right now, but wondering what you mean by having to store the weights in a separate file for wizardcoder, as a result of the 4gb executable limit. How does one go about this? Thank you!
curl -Lo llamafile.exe https://github.com/Mozilla-Ocho/llamafile/releases/download/0.1/llamafile-server-0.1
curl -Lo wizard.gguf https://huggingface.co/TheBloke/WizardCoder-Python-13B-V1.0-GGUF/resolve/main/wizardcoder-python-13b-v1.0.Q4_K_M.gguf
.\llamafile.exe -m wizard.ggufI think the best way to try this out is with LLaVA, the text+image model (like GPT-4 Vision). Here are steps to do that on macOS (which should work the same on other platforms too, I haven't tried that yet though): 1. Download the 4.26GB llamafile-server-0.1-llava-v1.5-7b-q4 file from https://huggingface.co/jartine/llava-v1.5-7B-GGUF/blob/main/... : wget https://huggingface.co/jartine/llava-v1.5-7B-GGUF/resolve/main/…
A couple of steps away from getting weaponized.
Extremely cool and Justine Tunney / jart does incredible portability work [0], but I'm kind of struggling with the use-cases for this one. I make a small macOS app [1] which runs llama.cpp with a SwiftUI front-end. For the first version of the app I was obsessed with the single download -> chat flow and making 0 network connections. I bundled a model with the app and you could just download, open, and start using it.…
Well, that's on the MAS/TestFlight for not doing delta updates.
Extremely cool and Justine Tunney / jart does incredible portability work [0], but I'm kind of struggling with the use-cases for this one. I make a small macOS app [1] which runs llama.cpp with a SwiftUI front-end. For the first version of the app I was obsessed with the single download -> chat flow and making 0 network connections. I bundled a model with the app and you could just download, open, and start using it.…
At least Makeayo asks where my models are now. It's obnoxious that I have to use symlinks for comfy/automatic....
All they need to do is ask me where my stuff is on first run, and an area in the config to update that setting. Not so hard!