File size: 2,654 Bytes
b8cf9ba 42e549d b8cf9ba ab21db3 b8cf9ba 09066f3 b8cf9ba |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 |
---
license: cc-by-nc-4.0
---
MLewd is a model created to be... Lewd. That's all. Based on ReMM.
There was so much attempt on this model that I can't count them all. Bear with me lmao.
The OG plan: https://pastebin.com/hfJ80rKL
Command useds and explaination :
```shell
Due to hardware limitation, some merge was done in 2 part.
Last mix :
- ReMM (Base) (0.57)
- Doctor-Shotgun/llama-2-13b-chat-limarp-v2-merged (Llama Chat Uncensored) (0.35)
- KoboldAI/LLAMA2-13B-Holodeck-1 (0.08)
Part 1: python ties_merge.py TheBloke/Llama-2-13B-fp16 ./MLewdBase-L2-13B-part1 --merge Undi95/ReMM-L2-13B --density 0.88 --merge KoboldAI/LLAMA2-13B-Holodeck-1 --density 0.12 --cuda
Part 2: python ties_merge.py TheBloke/Llama-2-13B-fp16 ./MLewdBase-L2-13B --merge Undi95/MLewdBase-L2-13B-part1 --density 0.65 --merge Doctor-Shotgun/llama-2-13b-chat-limarp-v2-merged --density 0.35 --cuda
(MLewd-L2-13B-v1-2 got disqualified)
- Applying LoRA: nRuaif/Kimiko-v2-13B at (0.24) weight on MLewd-L2-13B-v1-1
=> Result: MLewd-L2-13B-v1-3
================== ERP RANKING TEST ===========================
19.42 | MLewd-L2-13B-v1-3.q5_K_M.gguf (-> Best)
19.25 | MLewd-L2-13B-v1-1.q5_K_M.gguf
18.25 | MLewd-L2-13B-v1-2.q5 K M.gguf
================== RETRY ===========================
Mix:
- Undi95/MLewd-L2-13B-v1-3 (0.82)
- Sao10K/Stheno-Inverted-L2-13B (0.18)
!python ties_merge.py TheBloke/Llama-2-13B-fp16 ./MLewd-L2-13B-v1-7 --merge Undi95/MLewd-L2-13B-v1-3 --density 0.82 --merge Sao10K/Stheno-Inverted-L2-13B --density 0.18 --cuda
=> Result: MLewd-L2-13B-v1-7
Final touch (trying my best here) :
MLewd-L2-13B-v1-7 (0.77) + zarakiquemparte/PIPPA-ShareGPT-Subset-QLora-13b (LoRA 0.23)
=> MLewd-L2-13B-v1-7-TRY2
FINAL : MLewd-L2-13B-v1-7-TRY2 (0.82) + BluemoonRP (0.18)
=> MLewd-L2-13B-v1-8-3
RIP to all the version that got trashed.
```
<!-- description start -->
## Description
This repo contains quantized files (Q4_K_S and Q5_K_M) of MLewd-L2-13B, a trying-to-be lewd LLM model.
<!-- description end -->
<!-- description start -->
## Models used
- Undi95/ReMM (Base)
- Doctor-Shotgun/llama-2-13b-chat-limarp-v2-merged (Llama Chat Uncensored)
- KoboldAI/LLAMA2-13B-Holodeck-1
- Sao10K/Stheno-Inverted-L2-13B
## Loras used
- nRuaif/BluemoonRP-L2-13B-This-time-will-be-better/tree/main/lora-out-13b-final-BM/checkpoint-15/adapter_model
- zarakiquemparte/PIPPA-ShareGPT-Subset-QLora-13b
<!-- description end -->
<!-- prompt-template start -->
## Prompt template: Alpaca
```
Below is an instruction that describes a task. Write a response that appropriately completes the request.
### Instruction:
{prompt}
### Response:
```
Special thanks to Sushi kek |