LFM2.5-230M-MLX-4bit
MLX export of LFM2.5-230M for Apple Silicon inference.
LFM2.5-230M is a compact multilingual model built on LiquidAI's hybrid architecture, combining convolutional and attention layers for efficient long-context processing.
Model Details
| Property | Value |
|---|---|
| Parameters | 230M |
| Precision | 4-bit |
| Group Size | 64 |
| Size | 139 MB |
| Context Length | 128K |
Use with mlx
pip install mlx-lm
from mlx_lm import load, generate
from mlx_lm.sample_utils import make_sampler
model, tokenizer = load("LiquidAI/LFM2.5-230M-MLX-4bit")
response = generate(
model,
tokenizer,
prompt="The capital of France is",
max_tokens=100,
sampler=make_sampler(temp=0.7),
verbose=True,
)
Other Precisions
- LFM2.5-230M-MLX-bf16 (438 MB)
- LFM2.5-230M-MLX-8bit (233 MB)
- LFM2.5-230M-MLX-6bit (178 MB)
- LFM2.5-230M-MLX-5bit (159 MB)
- LFM2.5-230M-MLX-4bit (139 MB)
License
This model is released under the LFM 1.0 License.
- Downloads last month
- 110
Safetensors
Model size
40.1M params
Tensor type
BF16
·
U32 ·
MLX
Hardware compatibility
Log In to add your hardware
4-bit