Downloads · 30 days
0
Qwen3.8-27B-NQ-SVD-IQ2
agiws/Qwen3.8-27B-NQ-SVD-IQ2
Qwen3.8-27B-NQ-SVD-IQ2 is a machine learning model from agiws. Use it for the machine learning task on the model card, and read the license before you ship it in a product. The card lists the license as apache-2.0.
⚠️ EXPERIMENTAL REPOSITORY — the model is NOT ready yet and is under active development.
Access
Public
Updated Sep 26, 2026
Repo size
1.1 GB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.pt135 MB · 96%
At a glance
- License
- apache-2.0
- Access
- Public
- Created
- Aug 23, 2026
- Updated
- Sep 26, 2026
- SHA
- 573d4b4c
Base models
Model card
From the Hugging Face model README
Qwen3.8-27B-NQ-SVD-IQ2
⚠️ EXPERIMENTAL REPOSITORY — the model is NOT ready yet and is under active development.
Quantization of Qwen3.8-27B using SVD low-rank factorization with quantized factors, produced with NeuralQuant (NQ).
- NQ — NeuralQuant (our quantizer engine)
- SVD — low-rank factorization of Linear weights:
W ≈ A·B + E(all factors are quantized:A→ IQ2/IQ3,B→ IQ2, residualE→ IQ1/ternary) - IQ2 — 2-bit IQ codebooks for the low-rank factors
Current status: the SVD support in the NeuralQuant library is being implemented. Model weights will be published once the model is validated.
- License
- apache-2.0
- Created
- Aug 23, 2026
- Updated
- Sep 26, 2026