MLX Community · First MLX build 22 May 2024
dolphin-2.9.1-mixtral-1x22b
MLX
apache-2.0
A 22B parameter model, with 3 MLX builds, from 2-bit to 8-bit.
- Parameters
- 22B From Hugging Face
- Quantisation
- 2-bit dolphin-2.9.1-mixtral-1x22b-2bit
- Memory needed
- 10 GB Estimate, unified memory
- Download
- 7.0 GB Weights, estimated
- Downloads
- 32 This build, last 30 days
- Task
- —
- MLX builds
- 3 On mlx-community
- Devices
- Mac Apple silicon
Memory
Will it fit?Needs about 10 GB at 2-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 · Hugging Face CLI$ pip install -U huggingface_hub
$ hf download mlx-community/dolphin-2.9.1-mixtral-1x22b-2bitMLX builds
3 on mlx-community| dolphin-2.9.1-mixtral-1x22b-2bit | 2-bit | 7.0 GB | 10 GB | 32 |
|---|---|---|---|---|
| dolphin-2.9.1-mixtral-1x22b-4bit | 4-bit | 12.5 GB | 16 GB | 27 |
| dolphin-2.9.1-mixtral-1x22b-8bit | 8-bit | 23.6 GB | 29 GB | 26 |
Sizes are estimated from Hugging Face’s parameter counts and the quantisation. Downloads are for the last 30 days.