5 ms·
when is the new mistral coming out and at what size?
by purplecats 3y ago
when is the new mistral coming out and at what size?
- tarruda 3y agoI'm hoping that they make it 13B, which is the size I can run locally in 4-bit and still get reasonable performance
- kybernetikos 3y agoWhat kind of system do you need for that?
- tarruda 3y agoMine is a laptop with i7-11800h CPU + RTX 3070 Max-Q 8GB VRAM + 64GB RAM (though you can get probably get away with 16GB RAM). I bought this system for work and causal gaming, and was happy when I found out that GPU also enabled me to run LLMs locally at good performance. This laptop costed me ~= $ 1600, which was a bargain considering how much value I get out of it. If you are not on a budget, I highly recommend getting one of the high end laptops that have RTX 4090 and 16GB VRAM. With my system, Llama.cpp can run Mistral 7B 8-bit quantized by offloading 32 layers to the GPU (35 total) at about 25-30 tokens/second, or 6-bit quantized by offloading all layers to the GPU at ~ 35 tokens/second. I've tested a few 13B 4-bit models such as Codellama and got about 10 tokens/second by offloading 37 layers to the GPU. Got me about 10-15 tokens/second.
- disiplus 3y agoi have lenovo legion with 3070 8GB and was wondering should i use that instead of my macbook m1pro.
- tarruda 3y agoThe main focus of llama.cpp has been Apple silicon, so I suspect M1 would be more efficient. The author recently published some benchmarks: https://github.com/ggerganov/llama.cpp/discussions/4167 https://github.com/ggerganov/llama.cpp/discussions/4167
- michaelt 3y agoIf your GPU has ~16GB of vram, you can run a 13B model in "Q4_K_M.gguf" format and it'll be fast. Maybe even ~12GB. It's also possible to run on CPU from system RAM, to split the workload across GPU and CPU, or even from a memory-mapped file on disk. Some people have posted benchmarks online [1] and naturally, the faster your RAM and CPU the better. My personal experience is running from CPU/system ram is painfully slow. But that's partly because I only experimented with models that were too big to fit on my GPU, so part of the slowness is due to their large size. [1] https://www.reddit.com/r/LocalLLaMA/comments/14ilo0t/extensive_llamacpp_benchmark_more_speed_on_cpu_7b/ https://www.reddit.com/r/LocalLLaMA/comments/14ilo0t/extensi...
- tarruda 3y agoI get 10 tokens/second on a 4-bit 13B model with 8GB VRAM offloading as much as possible to the GPU. At this speed, I cannot read the LLM output as fast as it generates, so I consider it to be sufficient.
- johnbellone 3y agoWhich video card?
- tarruda 3y agoRTX 3070 Max-Q (laptop)
- gardnr 3y agoI can fit 13B Q4 K M models on a 12GB RTX 3060. It OOMs when the context window goes above 3k. I get 25 tok/s.
- ekianjo 3y agoa CPU would work fine for the 7B model, and if you have 32GB RAM and a CPU with a lot of core you can run a 13B model as well while it will be quite slow. If you dont care about speed, it's definitely one of the cheapest ways to run LLMs.
- vunderba 3y agoOn my Mac M1 Max 32GB of ram, Vicuna 13b (GGUF model) at 4bit consumes around 8GB of ram in Oobabooga. Tried turning on mlock and upping thread count to 6, but it's still rather slow at around 3 tokens / sec.
- ekianjo 3y agoQ5_M on Mistral 7B has good accuracy and performs decently on a CPU too