Commit
•
7c95d34
1
Parent(s):
be82791
Update README.md
Browse files
README.md
CHANGED
@@ -22,4 +22,20 @@ Stylized writing. Models are asked to explain a concept in a distinct writing st
|
|
22 |
# What does each of the tests measure I dont understand111!!!11!
|
23 |
BCD=following commands
|
24 |
|
25 |
-
PS=creative writing
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
22 |
# What does each of the tests measure I dont understand111!!!11!
|
23 |
BCD=following commands
|
24 |
|
25 |
+
PS=creative writing
|
26 |
+
|
27 |
+
# RESULTS AND DISCUSSION
|
28 |
+
![This table shows the results](llm-results.png)
|
29 |
+
|
30 |
+
Here they are. The results of each test. You can see pure data in file [LLM-test.csv](LLM-test.csv)
|
31 |
+
|
32 |
+
What they show is quite interesting:
|
33 |
+
- Goliath in the best at following commands, followed by Qwen and Nous-Hermes
|
34 |
+
- Goliath, Xwin and Mixtral are the best at creative writing
|
35 |
+
- Qwen is terrible at creative writing, but good at following commands, Mixtral is the opposite
|
36 |
+
- Size matters! 34B Nous-Capubara is the worst on average, likely due to it's size
|
37 |
+
- Xwin, Goliath and Mixtral are the best at stylized writing
|
38 |
+
- Goliath, Euryale, Xwin and Mixtral are the only ones who were capable to write coherent poems most of the time
|
39 |
+
|
40 |
+
# More tests?
|
41 |
+
Feel free to suggest more models for testing by opening new discussion. Mention model name, size and why do you want to test it.
|