Create README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,20 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
base_model:
|
| 3 |
+
- Qwen/Qwen3.6-35B-A3B
|
| 4 |
+
---
|
| 5 |
+
## Description
|
| 6 |
+
This repo contains specialized MoE-quants for Qwen3.6-35B-A3B. The idea being that given the huge size of the FFN tensors compared to the rest of the tensors in the model,
|
| 7 |
+
it should be possible to achieve a better quality while keeping the overall size of the entire model smaller compared to a similar naive quantization.
|
| 8 |
+
To that end, the quantization type default is kept in high quality and the FFN UP + FFN GATE tensors are quanted down along with the FFN DOWN tensors.
|
| 9 |
+
|
| 10 |
+
| Quant | Size | Mixture | PPL | 1-(Mean PPL(Q)/PPL(base)) | KLD |
|
| 11 |
+
| :--------- | :--------- | :------- | :------- | :------- | :------- |
|
| 12 |
+
| Q8_0 | 34.36 GiB (8.52 BPW) | Q8_0 | 6.719733 ± 0.043673 | +0.0000% | 0.005914 ± 0.000097 |
|
| 13 |
+
| Q5_K_M | 24.44 GiB (6.06 BPW) | Q8_0 / Q5_K / Q5_K / Q6_K | 6.728925 ± 0.043742 | +0.1368% | 0.008198 ± 0.000112 |
|
| 14 |
+
| Q4_K_M | 20.61 GiB (5.11 BPW) | Q8_0 / Q4_K / Q4_K / Q5_K | 6.741414 ± 0.043822 | +0.3227% | 0.013899 ± 0.000169 |
|
| 15 |
+
| IQ4_XS | 16.40 GiB (4.06 BPW) | Q8_0 / IQ3_S / IQ3_S / IQ4_XS | 6.888604 ± 0.044992 | +2.5131% | 0.033477 ± 0.000265 |
|
| 16 |
+
| IQ3_S | 12.65 GiB (3.13 BPW) | Q6_K / IQ2_S / IQ2_S / IQ3_S | 7.177309 ± 0.047398 | +6.8095% | 0.084848 ± 0.000588 |
|
| 17 |
+
|
| 18 |
+
|
| 19 |
+
|
| 20 |
+
|