MLX Community · First MLX build 20 Aug 2024
Phi-3.5-mini-instruct
MLX
Text mit
A 3.8B parameter text generation model, with 4 MLX builds, from 6-bit to bf16.
- Parameters
- 3.8B From Hugging Face
- Quantisation
- 4-bit Phi-3.5-mini-instruct-4bit
- Memory needed
- 5 GB Estimate, unified memory
- Download
- 2.1 GB Weights, estimated
- Downloads
- 5.9K 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 5 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/Phi-3.5-mini-instruct-4bit --prompt "Hello"MLX builds
4 on mlx-community| Phi-3.5-mini-instruct-6bit | 6-bit | 821 MB | 3 GB | 32 |
|---|---|---|---|---|
| Phi-3.5-mini-instruct-4bit | 4-bit | 2.1 GB | 5 GB | 5.9K |
| Phi-3.5-mini-instruct-8bit | 8-bit | 4.1 GB | 7 GB | 90 |
| Phi-3.5-mini-instruct-bf16 | bf16 | 7.6 GB | 11 GB | 62 |
Sizes are estimated from Hugging Face’s parameter counts and the quantisation. Downloads are for the last 30 days.