SchGen: PCB Schematic Generation with Semantic-Grounded Code Representations
Paper • 2605.30345 • Published • 1
SchGen-Qwen3.6-27B-EU — MLX 4bit
Apple-Silicon (MLX) 4bit quantization of
Ailiance-fr/SchGen-Qwen3.6-27B-EU,
a sovereign EU KiCad schematic-generation model (SchGen method,
Luo et al., 2026) by
ailiance.
MLX quantization, group_size=64, 4.5 bits per weight (~14 GB).
Usage (mlx-lm)
pip install mlx-lm
mlx_lm.generate --model Ailiance-fr/SchGen-Qwen3.6-27B-EU-MLX-4bit \
--prompt "Generate a schematic for an LM358 inverting amplifier, gain -10." \
--max-tokens 2048
from mlx_lm import load, generate
model, tok = load("Ailiance-fr/SchGen-Qwen3.6-27B-EU-MLX-4bit")
print(generate(model, tok, prompt="...", max_tokens=2048))
Served with enable_thinking=false. See the
full model card
for evaluation, limitations, and citation.
License
Apache-2.0. Derivative of Qwen/Qwen3.6-27B (Apache-2.0), trained on
microsoft/SchGen_dataset (MIT). See NOTICE.
- Downloads last month
- 43
Model size
27B params
Tensor type
BF16
·
U32 ·
Hardware compatibility
Log In to add your hardware
4-bit
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support