Alibaba Qwen · First MLX build 18 Sept 2024
Qwen2.5-72B-Instruct
MLX
Text qwen
A 73B parameter text generation model from Alibaba Qwen, with 4 MLX builds, from 3-bit to 6-bit.
- Parameters
- 73B From Hugging Face
- Quantisation
- 4-bit Qwen2.5-72B-Instruct-4bit
- Memory needed
- 48 GB Estimate, unified memory
- Download
- 40.9 GB Weights, estimated
- Downloads
- 1.0K This build, last 30 days
- Task
- Text generation Text
- MLX builds
- 4 On mlx-community
- Devices
- Mac Apple silicon
Memory
Will it fit?Needs about 48 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/Qwen2.5-72B-Instruct-4bit --prompt "Hello"MLX builds
4 on mlx-community| Qwen2.5-72B-Instruct-3bit | 3-bit | 31.8 GB | 38 GB | 204 |
|---|---|---|---|---|
| Qwen2.5-72B-Instruct-4bit | 4-bit | 40.9 GB | 48 GB | 1.0K |
| Qwen2.5-72B-Instruct-4bit-AWQ | 4-bit | 40.9 GB | 48 GB | 56 |
| Qwen2.5-72B-Instruct-6bit | 6-bit | 59.1 GB | 69 GB | 108 |
Sizes are estimated from Hugging Face’s parameter counts and the quantisation. Downloads are for the last 30 days.