DeepSeek-V3-3bit-bf16
mlx-community/DeepSeek-V3-3bit-bf16
Uncategorised · mlx
3-bit Open on Hugging Face
Parameters
671B
Weights (est.)
≈ 294 GB
Mac memory (est.)
512 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.
Details
- Base model
- deepseek-ai/DeepSeek-V3 (quantized) · DeepSeek V3 on the radar →
- Vendor
- DeepSeek
- Task
- Uncategorised
- Quantisation
- 3-bit
- Licence
- —
- Downloads
- 58 in the last 30 days · 3.4K all time · 2 likes
- Published
- 2025-01-05 · updated 2025-01-05 (1y ago)
- Tags
- deepseek_v3, custom_code
Other MLX builds of deepseek-ai/DeepSeek-V3
| Build | Quantisation | Weights | Mac memory | Downloads |
|---|---|---|---|---|
| DeepSeek-V3-3bit | 3-bit | ≈ 294 GB | 512 GB+ | 78 |
| DeepSeek-V3-3bit-bf16 (this one) | 3-bit | ≈ 294 GB | 512 GB+ | 58 |
| DeepSeek-V3-4bit | 4-bit | ≈ 378 GB | 512 GB+ | 294 |