MLX Community · First MLX build 24 Jul 2024
Meta-Llama-3.1-405B
MLX
Text llama3.1
A 410B parameter text generation model, with 3 MLX builds, from 2-bit to 8-bit.
- Parameters
- 410B From Hugging Face
- Quantisation
- 4-bit Meta-Llama-3.1-405B-4bit
- Memory needed
- 260 GB Estimate, unified memory
- Download
- 231 GB Weights, estimated
- Downloads
- 674 This build, last 30 days
- Task
- Text generation Text
- MLX builds
- 3 On mlx-community
- Devices
- Mac Apple silicon
Memory
Will it fit?Needs about 260 GB at 4-bit
8 GB
16 GB
24 GB
32 GB
64 GB
128 GB
Fits with headroomTight, close other appsWon’t fit
Estimated from the size of the weights, plus a fifth for context and working memory and 2 GB for macOS and your apps. Long contexts need more.
Run it
Terminal · mlx-lm$ pip install -U mlx-lm
$ mlx_lm.generate --model mlx-community/Meta-Llama-3.1-405B-4bit --prompt "Hello"MLX builds
3 on mlx-community| Meta-Llama-3.1-405B-2bit | 2-bit | 128 GB | 146 GB | 57 |
|---|---|---|---|---|
| Meta-Llama-3.1-405B-4bit | 4-bit | 231 GB | 260 GB | 674 |
| Meta-Llama-3.1-405B-8bit | 8-bit | — | — | 0 |
Sizes are estimated from Hugging Face’s parameter counts and the quantisation. Downloads are for the last 30 days.