GGUF
Not-For-All-Audiences
nsfw
Inference Endpoints
Undi95 commited on
Commit
e73b072
1 Parent(s): ff636f8

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +69 -4
README.md CHANGED
@@ -1,7 +1,72 @@
1
- Temporary repo, will delete it/move it later.
 
 
 
 
 
2
 
3
- This model have way less refusal.
4
 
5
- Tested on temp 0 with neutral sampler, no system prompt, no character card.
 
 
6
 
7
- Goal would be to have no refusal at temp 0, will see later.
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: cc-by-nc-4.0
3
+ tags:
4
+ - not-for-all-audiences
5
+ - nsfw
6
+ ---
7
 
8
+ ## Lumimaid 0.1
9
 
10
+ <center><div style="width: 100%;">
11
+ <img src="https://cdn-uploads.huggingface.co/production/uploads/630dfb008df86f1e5becadc3/d3QMaxy3peFTpSlWdWF-k.png" style="display: block; margin: auto;">
12
+ </div></center>
13
 
14
+ This model uses the Llama3 **prompting format**
15
+
16
+ Llama3 trained on our RP datasets, we tried to have a balance between the ERP and the RP, not too horny, but just enough.
17
+
18
+ We also added some non-RP dataset, making the model less dumb overall. It should look like a 40%/60% ratio for Non-RP/RP+ERP data.
19
+
20
+ This model includes the new Luminae dataset from Ikari.
21
+
22
+ This model have received the Orthogonal Activation Steering treatment, meaning it will rarely refuse any request.
23
+
24
+ If you consider trying this model please give us some feedback either on the Community tab on hf or on our [Discord Server](https://discord.gg/MtCVRWTZXY).
25
+
26
+ ## Credits:
27
+ - Undi
28
+ - IkariDev
29
+
30
+ ## Description
31
+
32
+ This repo contains GGUF files of Lumimaid-8B-v0.1-OAS.
33
+
34
+ Switch: [8B](https://huggingface.co/NeverSleep/Llama-3-Lumimaid-8B-v0.1-GGUF) - [70B](https://huggingface.co/NeverSleep/Llama-3-Lumimaid-70B-v0.1-GGUF) - [70B-alt](https://huggingface.co/NeverSleep/Llama-3-Lumimaid-70B-v0.1-alt-GGUF) - [8B-OAS](https://huggingface.co/NeverSleep/Llama-3-Lumimaid-8B-v0.1-OAS-GGUF) - [70B-OAS](https://huggingface.co/NeverSleep/Llama-3-Lumimaid-70B-v0.1-OAS-GGUF)
35
+
36
+ ## Training data used:
37
+ - [Aesir datasets](https://huggingface.co/MinervaAI)
38
+ - [NoRobots](https://huggingface.co/datasets/Doctor-Shotgun/no-robots-sharegpt)
39
+ - [limarp](https://huggingface.co/datasets/lemonilia/LimaRP) - 8k ctx
40
+ - [toxic-dpo-v0.1-sharegpt](https://huggingface.co/datasets/Undi95/toxic-dpo-v0.1-sharegpt)
41
+ - [ToxicQAFinal](https://huggingface.co/datasets/NobodyExistsOnTheInternet/ToxicQAFinal)
42
+ - Luminae-i1 (70B/70B-alt) (i2 was not existing when the 70b started training) | Luminae-i2 (8B) (this one gave better results on the 8b) - Ikari's Dataset
43
+ - [Squish42/bluemoon-fandom-1-1-rp-cleaned](https://huggingface.co/datasets/Squish42/bluemoon-fandom-1-1-rp-cleaned) - 50% (randomly)
44
+ - [NobodyExistsOnTheInternet/PIPPAsharegptv2test](https://huggingface.co/datasets/NobodyExistsOnTheInternet/PIPPAsharegptv2test) - 5% (randomly)
45
+ - [cgato/SlimOrcaDedupCleaned](https://huggingface.co/datasets/cgato/SlimOrcaDedupCleaned) - 5% (randomly)
46
+ - Airoboros (reduced)
47
+ - [Capybara](https://huggingface.co/datasets/Undi95/Capybara-ShareGPT/) (reduced)
48
+
49
+
50
+ ## Models used (only for 8B)
51
+
52
+ - Initial LumiMaid 8B Finetune
53
+ - Undi95/Llama-3-Unholy-8B-e4
54
+ - Undi95/Llama-3-LewdPlay-8B
55
+
56
+ ## Prompt template: Llama3
57
+
58
+ ```
59
+ <|begin_of_text|><|start_header_id|>system<|end_header_id|>
60
+
61
+ {system_prompt}<|eot_id|><|start_header_id|>user<|end_header_id|>
62
+
63
+ {input}<|eot_id|><|start_header_id|>assistant<|end_header_id|>
64
+
65
+ {output}<|eot_id|>
66
+ ```
67
+
68
+ ## Others
69
+
70
+ Undi: If you want to support us, you can [here](https://ko-fi.com/undiai).
71
+
72
+ IkariDev: Visit my [retro/neocities style website](https://ikaridevgit.github.io/) please kek