Update README.md
Browse files
README.md
CHANGED
@@ -1,19 +1,46 @@
|
|
1 |
-
|
2 |
|
3 |
-
|
4 |
|
5 |
-
---
|
6 |
|
7 |
# The LLM Creativity benchmark
|
8 |
|
9 |
Last updated: 26 Feb 2024
|
10 |
|
11 |
-
The goal of this benchmark is to evaluate the ability of Large Language Models to be used
|
|
|
|
|
12 |
|
13 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
14 |
|
15 |
![image/png](https://huggingface.co/datasets/froggeric/creativity/resolve/main/benchmark_results_2024-02-26.png)
|
16 |
|
17 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
18 |
- https://eqbench.com/
|
19 |
- https://chat.lmsys.org/
|
|
|
1 |
+
# -=- UNDER CONSTRUCTION -=-
|
2 |
|
3 |
+
_Please note that even though this repo is public, the information provided below is not yet complete._
|
4 |
|
|
|
5 |
|
6 |
# The LLM Creativity benchmark
|
7 |
|
8 |
Last updated: 26 Feb 2024
|
9 |
|
10 |
+
The goal of this benchmark is to evaluate the ability of Large Language Models to be used
|
11 |
+
as an **uncensored creative writing assistant**. Human evaluation of the results is done manually,
|
12 |
+
by me, to assess the quality of writing.
|
13 |
|
14 |
+
There are 24 questions, some standalone, other follow-ups to previous questions for a multi-turn conversation.
|
15 |
+
The questions can be split half-half in 2 possible ways:
|
16 |
+
|
17 |
+
## First split: sfw / nsfw
|
18 |
+
* sfw: 50% are safe questions that should not trigger any guardrail
|
19 |
+
* nsfw: 50% are questions covering a wide range of NSFW and illegal topics, which are testing for censorship
|
20 |
+
|
21 |
+
## Second split: story / smart
|
22 |
+
* story: 50% of questions are creative writing tasks, covering both the nsfw and sfw topics
|
23 |
+
* smart: 50% of questions are more about testing the capabilities of the model to work as an assistant, again covering both the sfw and sfw topics
|
24 |
+
|
25 |
+
## What is _not_ included
|
26 |
+
* roleplay
|
27 |
+
* mathematics
|
28 |
+
* coding
|
29 |
+
* trick questions
|
30 |
+
|
31 |
+
# Results
|
32 |
|
33 |
![image/png](https://huggingface.co/datasets/froggeric/creativity/resolve/main/benchmark_results_2024-02-26.png)
|
34 |
|
35 |
+
# Remarks about some of the models
|
36 |
+
|
37 |
+
|
38 |
+
# Questions type
|
39 |
+
|
40 |
+
I will not provide the exact questions are used for various reasons, but I can provide some general ideas about which areas they cover:
|
41 |
+
|
42 |
+
|
43 |
+
# Other interesting benchmarks
|
44 |
+
|
45 |
- https://eqbench.com/
|
46 |
- https://chat.lmsys.org/
|