Update README.md
Browse files
README.md
CHANGED
@@ -6,10 +6,14 @@ base_model:
|
|
6 |
- Undi95/Llama-3-LewdPlay-8B
|
7 |
library_name: transformers
|
8 |
tags:
|
9 |
-
-
|
|
|
10 |
- merge
|
11 |
---
|
12 |
-
|
|
|
|
|
|
|
13 |
# LewdPlay-8B
|
14 |
|
15 |
This is a merge of pre-trained language models created using [mergekit](https://github.com/cg123/mergekit).
|
|
|
6 |
- Undi95/Llama-3-LewdPlay-8B
|
7 |
library_name: transformers
|
8 |
tags:
|
9 |
+
- not-for-all-audiences
|
10 |
+
- nsfw
|
11 |
- merge
|
12 |
---
|
13 |
+
* <span style="color:orange">I'm just tinkering. All credit to the original creator: [Undi](https://huggingface.co/Undi95).</span>
|
14 |
+
* <span style="color:orange">This model is a more traditionally quantized EXL2 compared to my usual ["rpcal" versions](https://huggingface.co/zaq-hack/Llama-3-LewdPlay-8B-evo-bpw800-h8-exl2-rpcal). Llama-3-8B seems to get markedly dumber by using the "rpcal" method. In previous models, it was difficult to tell, but the margin of error increase from quantizing Llama-3-8B makes it obvious which method is better. I deleted the lower quants of rpcal because they are pretty dumb by comparison. I recommend this version, and this version only for this model. Lower quants are significantly worse.</span>
|
15 |
+
* <span style="color:orange">This model: EXL2 @ 8.0 bpw.</span>
|
16 |
+
---
|
17 |
# LewdPlay-8B
|
18 |
|
19 |
This is a merge of pre-trained language models created using [mergekit](https://github.com/cg123/mergekit).
|