ChuckMcSneed
commited on
Commit
•
d6915d4
1
Parent(s):
3d7966e
Update README.md
Browse files
README.md
CHANGED
@@ -5,7 +5,7 @@ tags:
|
|
5 |
---
|
6 |
This is a merge of [Xwin](https://huggingface.co/Xwin-LM/Xwin-LM-70B-V0.1) and [WinterGoddess](https://huggingface.co/Sao10K/WinterGoddess-1.4x-70B-L2). Made using [mergekit](https://github.com/cg123/mergekit).
|
7 |
|
8 |
-
Smarter than Goliath, but
|
9 |
|
10 |
# Benchmarks
|
11 |
### NeoEvalPlusN_benchmark
|
@@ -21,4 +21,12 @@ Smarter than Goliath, but maybe a bit more aligned? Not sure. Needs testing.
|
|
21 |
|
22 |
### Kanye Test
|
23 |
WinterGoliath kinda gets the rhyme, Goliath doesn't.
|
24 |
-
![Kanye test](kanye_test_winter_vs_goliath.png)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
5 |
---
|
6 |
This is a merge of [Xwin](https://huggingface.co/Xwin-LM/Xwin-LM-70B-V0.1) and [WinterGoddess](https://huggingface.co/Sao10K/WinterGoddess-1.4x-70B-L2). Made using [mergekit](https://github.com/cg123/mergekit).
|
7 |
|
8 |
+
Smarter than Goliath, but a bit more aligned. Sidegrade rather than upgrade. Sacrifices neutrality and fun for smartness.
|
9 |
|
10 |
# Benchmarks
|
11 |
### NeoEvalPlusN_benchmark
|
|
|
21 |
|
22 |
### Kanye Test
|
23 |
WinterGoliath kinda gets the rhyme, Goliath doesn't.
|
24 |
+
![Kanye test](kanye_test_winter_vs_goliath.png)
|
25 |
+
|
26 |
+
### Politiscales test
|
27 |
+
[Politiscales for llama](https://huggingface.co/datasets/ChuckMcSneed/politiscales_for_llama_results)
|
28 |
+
|name|whacky|left/right|
|
29 |
+
|alpindale/goliath-120b|1.066739456|1.544969782|
|
30 |
+
|ChuckMcSneed/WinterGoliath-123b|0.518277513|2.735962|
|
31 |
+
|Xwin-LM/Xwin-LM-70B-V0.1|1.463521162|1.491684328|
|
32 |
+
|Sao10K/WinterGoddess-1.4x-70B-L2|0.384151757|4.747980293|
|