4-bit Open on Hugging Face
Parameters
8.0B
Weights (est.)
≈ 4.5 GB
Mac memory (est.)
8 GB+
Size is estimated from the parameter count and quantisation. Memory allows a fifth extra for context plus 2 GB for the system; long contexts need more.
Run it
With mlx-lm on an Apple silicon Mac:
pip install mlx-lmmlx_lm.generate --model mlx-community/Llama-3.1-8B-Instruct-4bit --prompt "Hello"
Details
- Base model
- meta-llama/Llama-3.1-8B-Instruct (quantized) · Llama 3.1 8B Instruct on the radar →
- Vendor
- Meta · Llama
- Task
- Text generation
- Quantisation
- 4-bit
- Licence
- llama3.1
- Downloads
- 4.9K in the last 30 days · 1.1M all time · 11 likes
- Published
- 2025-02-15 · updated 2025-02-15 (1y ago)
- Languages
- en, de, fr, it, pt, hi, es, th
- Tags
- llama, facebook, meta, pytorch, llama-3
Other MLX builds of meta-llama/Llama-3.1-8B-Instruct
| Build | Quantisation | Weights | Mac memory | Downloads |
|---|---|---|---|---|
| Meta-Llama-3.1-8B-Instruct-3bit | 3-bit | ≈ 3.5 GB | 8 GB+ | 99 |
| Llama-3.1-8B-Instruct-4bit (this one) | 4-bit | ≈ 4.5 GB | 8 GB+ | 4.9K |
| Meta-Llama-3.1-8B-Instruct-abliterated-4bit | 4-bit | ≈ 4.5 GB | 8 GB+ | 229 |
| Meta-Llama-3.1-8B-Instruct-abliterated-Q4-MLX | 4-bit | ≈ 4.5 GB | 8 GB+ | 98 |
| Meta-Llama-3.1-8B-Instruct-4bit-fakenews | 4-bit | ≈ 4.5 GB | 8 GB+ | 62 |
| Llama-3.1-SuperNova-Lite-4bit | 4-bit | ≈ 4.5 GB | 8 GB+ | 19 |
| Llama-3.1-SuperNova-Lite-8bit | 8-bit | ≈ 8.5 GB | 16 GB+ | 28 |
| Meta-Llama-3.1-8B-Instruct-bf16 | bf16 | ≈ 16.1 GB | 24 GB+ | 914 |
| Llama-3.1-8B-Instruct | — | ≈ 16.1 GB | 24 GB+ | 762 |
| Meta-Llama-3.1-8B-Instruct-abliterated-bfloat16 | — | ≈ 16.1 GB | 24 GB+ | 136 |
| Llama-3.1-SuperNova-Lite-bf16 | bf16 | ≈ 16.1 GB | 24 GB+ | 27 |