Not-For-All-Audiences
Create README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,59 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: cc-by-nc-4.0
|
| 3 |
+
tags:
|
| 4 |
+
- not-for-all-audiences
|
| 5 |
+
---
|
| 6 |
+
|
| 7 |
+
Quantized using 200 samples of 8192 tokens from an RP-oriented [PIPPA](https://huggingface.co/datasets/royallab/PIPPA-cleaned) dataset.
|
| 8 |
+
|
| 9 |
+
Branches:
|
| 10 |
+
- `main` -- `measurement.json`
|
| 11 |
+
- `3.5b6h` -- 3.5bpw, 6bit lm_head
|
| 12 |
+
- `3.7b6h` -- 3.7bpw, 6bit lm_head
|
| 13 |
+
- `5b6h` -- 5bpw, 6bit lm_head
|
| 14 |
+
- `6b6h` -- 6bpw, 6bit lm_head
|
| 15 |
+
|
| 16 |
+
|
| 17 |
+
Requires ExllamaV2 version 0.0.12 and up.
|
| 18 |
+
|
| 19 |
+
Original model link: [Envoid/Fish-8x7B](https://huggingface.co/Envoid/Fish-8x7B)
|
| 20 |
+
|
| 21 |
+
Original model README below.
|
| 22 |
+
|
| 23 |
+
***
|
| 24 |
+
|
| 25 |
+
# Caution: This model may output adult content.
|
| 26 |
+
|
| 27 |
+
|
| 28 |
+
|
| 29 |
+
|
| 30 |
+
# FISH - Fusion of Intuitive Semantic Heuristics
|
| 31 |
+
|
| 32 |
+
For this model I took the following LoRAs
|
| 33 |
+
|
| 34 |
+
[SeanWu25/Mixtral_8x7b_Medicine](https://huggingface.co/SeanWu25/Mixtral_8x7b_Medicine)
|
| 35 |
+
|
| 36 |
+
[SeanWu25/Mixtral_8x7b_WuKurtz](https://huggingface.co/SeanWu25/Mixtral_8x7b_WuKurtz)
|
| 37 |
+
|
| 38 |
+
and
|
| 39 |
+
|
| 40 |
+
[wandb/Mixtral-8x7b-Remixtral](https://huggingface.co/wandb/Mixtral-8x7b-Remixtral)
|
| 41 |
+
|
| 42 |
+
And merged them onto their base model.
|
| 43 |
+
|
| 44 |
+
I then did a simple linear merge between them, an experimental unreleased 8x7B model, and an unreleased model that was a intermediate step in creating [Envoid/BondBurger-8x7B](https://huggingface.co/Envoid/BondBurger-8x7B?not-for-all-audiences=true)
|
| 45 |
+
|
| 46 |
+
The end results are a surprisingly good model for role palying style entertainment.
|
| 47 |
+
|
| 48 |
+
At first I was disappointed with the results but ended up settling on the following sampler parameters which really bring it to life.
|
| 49 |
+
|
| 50 |
+
|
| 51 |
+
|
| 52 |
+
|
| 53 |
+
## It does eccentuate characters that have particularly aggressive personalities.
|
| 54 |
+
|
| 55 |
+
## Because this is part of the whole SensualNousInstruct family of models it still suffers from the same tokenizer/special tokens weirdness.
|
| 56 |
+
|
| 57 |
+
## This model has only been tested in Q8 GGUF form due to hardware limitations.
|
| 58 |
+
|
| 59 |
+
It responds well to [INST] do a thing [/INST] instruct style formatting (although uses the ChatML special tokens)
|