Create README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,41 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
language:
|
| 4 |
+
- en
|
| 5 |
+
tags:
|
| 6 |
+
- nsfw
|
| 7 |
+
- not-for-all-audiences
|
| 8 |
+
- roleplay
|
| 9 |
+
---
|
| 10 |
+
|
| 11 |
+
## InfinityKumon-2x7B
|
| 12 |
+
|
| 13 |
+
|
| 14 |
+
|
| 15 |
+
GGUF - Imatrix quant of [InfinityKumon-2x7B](https://huggingface.co/R136a1/InfinityKumon-2x7B)
|
| 16 |
+
|
| 17 |
+
Another MoE merge from [Endevor/InfinityRP-v1-7B](https://huggingface.co/Endevor/InfinityRP-v1-7B) and [grimjim/kukulemon-7B](https://huggingface.co/grimjim/kukulemon-7B).
|
| 18 |
+
|
| 19 |
+
The reason? Because I like InfinityRP-v1-7B so much and wondering if I can improve it even more by merging 2 great models into MoE.
|
| 20 |
+
|
| 21 |
+
|
| 22 |
+
## Perplexity
|
| 23 |
+
|
| 24 |
+
Using llama.cpp/perplexity with private roleplay dataset.
|
| 25 |
+
|
| 26 |
+
| Format | PPL |
|
| 27 |
+
| --- | --- |
|
| 28 |
+
| FP16 | 3.1748 +/- 0.11928 |
|
| 29 |
+
| Q8_0 | 3.1734 +/- 0.11935 |
|
| 30 |
+
| Q6_K | 3.1752 +/- 0.11899 |
|
| 31 |
+
| Q5_K_M | 3.1731 +/- 0.11892 |
|
| 32 |
+
| IQ4_NL | 3.1752 +/- 0.11943 |
|
| 33 |
+
| IQ3_M | 3.1773 +/- 0.11528 |
|
| 34 |
+
| Q2_K | 3.2309 +/- 0.11996 |
|
| 35 |
+
|
| 36 |
+
I don't really recomend using Q2_K based on the ppl, the other quants are fine.
|
| 37 |
+
|
| 38 |
+
### Prompt format:
|
| 39 |
+
Alpaca or ChatML
|
| 40 |
+
|
| 41 |
+
Switch: [FP16](https://huggingface.co/R136a1/InfinityKumon-2x7B) - [GGUF](https://huggingface.co/R136a1/InfinityKumon-2x7B-GGUF)
|