Update README.md
Browse files
README.md
CHANGED
|
@@ -11,6 +11,29 @@ model-index:
|
|
| 11 |
results: []
|
| 12 |
---
|
| 13 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 14 |
|
| 15 |
*Me during failed runs*
|
| 16 |
|
|
|
|
| 11 |
results: []
|
| 12 |
---
|
| 13 |
|
| 14 |
+
Awe snap. Another Qwen 2.5 14b by the lord and savior, Sao.
|
| 15 |
+
|
| 16 |
+
I'm still refining my own settings for Qwen but for those of you who are interested in my most recent settings:
|
| 17 |
+
|
| 18 |
+
Temp: 1.1-1.2 OR .75-.85
|
| 19 |
+
<br>
|
| 20 |
+
Min P: 0.02 Min P - 0.05 (Min P seems to help with 'oddities' in responses) 0.035 Seems a decent midpoint.
|
| 21 |
+
<br>
|
| 22 |
+
Rep Penalty: 1.08
|
| 23 |
+
<br>
|
| 24 |
+
DRY: 0.3m 1.75, 2
|
| 25 |
+
<br>
|
| 26 |
+
|
| 27 |
+
[This is the 6bpw EXL2 version of this model. For the original model, go here](https://huggingface.co/Sao10K/14B-Qwen2.5-Freya-x1)
|
| 28 |
+
<br>
|
| 29 |
+
[For the 8bpw version, go here](https://huggingface.co/Statuo/Sao10K_14B-Qwen2.5-Freya-v1-EXL2-8bpw)
|
| 30 |
+
<br>
|
| 31 |
+
[For the 4bpw version, go here](https://huggingface.co/Statuo/Sao10K_14B-Qwen2.5-Freya-v1-EXL2-4bpw)
|
| 32 |
+
<br>
|
| 33 |
+
|
| 34 |
+
|
| 35 |
+
---
|
| 36 |
+
|
| 37 |
|
| 38 |
*Me during failed runs*
|
| 39 |
|