6-bit Open on Hugging Face
Parameters
22B
Weights (est.)
≈ 17.7 GB
Mac memory (est.)
24 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/ERNIE-4.5-21B-A3B-PT-6bit --prompt "Hello"
Details
- Base model
- baidu/ERNIE-4.5-21B-A3B-PT (quantized)
- Vendor
- baidu
- Task
- Text generation
- Quantisation
- 6-bit
- Licence
- apache-2.0
- Downloads
- 40 in the last 30 days · 380 all time · 0 likes
- Published
- 2025-07-04 · updated 2025-07-04 (1y ago)
- Languages
- en, zh
- Tags
- ernie4_5_moe, ERNIE4.5, custom_code
Other MLX builds of baidu/ERNIE-4.5-21B-A3B-PT
| Build | Quantisation | Weights | Mac memory | Downloads |
|---|---|---|---|---|
| ERNIE-4.5-21B-A3B-PT-5bit | 5-bit | ≈ 15.0 GB | 24 GB+ | 49 |
| ERNIE-4.5-21B-A3B-PT-6bit (this one) | 6-bit | ≈ 17.7 GB | 24 GB+ | 40 |
| ERNIE-4.5-21B-A3B-PT-8bit | 8-bit | ≈ 23.2 GB | 32 GB+ | 50 |
| ERNIE-4.5-21B-A3B-PT-bf16 | bf16 | ≈ 43.7 GB | 64 GB+ | 37 |