Upload README.md with huggingface_hub
Browse files
README.md
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
tags:
|
| 4 |
+
- polarquant
|
| 5 |
+
- qwen3.5
|
| 6 |
+
- claude-opus
|
| 7 |
+
- quantized
|
| 8 |
+
base_model: Jackrong/Qwopus3.5-27B-v3
|
| 9 |
+
pipeline_tag: text-generation
|
| 10 |
+
arxiv: "2603.29078"
|
| 11 |
+
---
|
| 12 |
+
|
| 13 |
+
# 🧊 Qwopus3.5-27B-v3-PolarQuant-Q5
|
| 14 |
+
|
| 15 |
+
**27B Claude Opus distill** on consumer GPUs.
|
| 16 |
+
|
| 17 |
+
| Metric | Value |
|
| 18 |
+
|---|---|
|
| 19 |
+
| **VRAM** | 16.9 GB |
|
| 20 |
+
| **Speed** | 21.7 tok/s |
|
| 21 |
+
| **Download** | 16.2 GB |
|
| 22 |
+
| **KV Cache** | Q3 = 5.3x compression |
|
| 23 |
+
| **Dequant** | 32s |
|
| 24 |
+
|
| 25 |
+
## GPU Support
|
| 26 |
+
|
| 27 |
+
| GPU | Fits? |
|
| 28 |
+
|---|---|
|
| 29 |
+
| RTX 4090 (24 GB) | ✅ |
|
| 30 |
+
| L4 (24 GB) | ✅ |
|
| 31 |
+
| A100 (40-80 GB) | ✅ |
|
| 32 |
+
|
| 33 |
+
```bash
|
| 34 |
+
pip install polarquant[all]
|
| 35 |
+
polarquant chat Jackrong/Qwopus3.5-27B-v3
|
| 36 |
+
```
|
| 37 |
+
|
| 38 |
+
📄 [Paper](https://arxiv.org/abs/2603.29078) · 💻 [GitHub](https://github.com/caiovicentino/polarengine-vllm) · 📦 [PyPI](https://pypi.org/project/polarquant/)
|