MLX Community · First MLX build 23 Jul 2024
Meta-Llama-3.1-70B
MLX
Text llama3.1
A 71B parameter text generation model, with 3 MLX builds, from 4-bit to bf16.
- Parameters
- 71B From Hugging Face
- Quantisation
- bf16 Meta-Llama-3.1-70B-bf16
- Memory needed
- 160 GB Estimate, unified memory
- Download
- 141 GB Weights, estimated
- Downloads
- 91 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 160 GB at bf16
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-3.1-70B-bf16 --prompt "Hello"MLX builds
3 on mlx-community| Meta-Llama-3.1-70B-4bit | 4-bit | 39.7 GB | 47 GB | 71 |
|---|---|---|---|---|
| Meta-Llama-3.1-70B-8bit | 8-bit | 75.0 GB | 86 GB | 57 |
| Meta-Llama-3.1-70B-bf16 | bf16 | 141 GB | 160 GB | 91 |
Sizes are estimated from Hugging Face’s parameter counts and the quantisation. Downloads are for the last 30 days.